A stealth adhesion factor contributes to Vibrio vulnificus pathogenicity: Flp pili play roles in host invasion, survival in the blood stream and resistance to complement activation

The tad operons encode the machinery required for adhesive Flp (fimbrial low-molecular-weight protein) pili biogenesis. Vibrio vulnificus, an opportunistic pathogen, harbors three distinct tad loci. Among them, only tad1 locus was highly upregulated in in vivo growing bacteria compared to in vitro culture condition. To understand the pathogenic roles of the three tad loci during infection, we constructed single, double and triple tad loci deletion mutants. Interestingly, only the Δtad123 triple mutant cells exhibited significantly decreased lethality in mice. Ultrastructural observations revealed short, thin filamentous projections disappeared on the Δtad123 mutant cells. Since the pilin was paradoxically non-immunogenic, a V5 tag was fused to Flp to visualize the pilin protein by using immunogold EM and immunofluorescence microscopy. The Δtad123 mutant cells showed attenuated host cell adhesion, decreased biofilm formation, delayed RtxA1 exotoxin secretion and subsequently impaired translocation across the intestinal epithelium compared to wild type, which could be partially complemented with each wild type operon. The Δtad123 mutant was susceptible to complement-mediated bacteriolysis, predominantly via the alternative pathway, suggesting stealth hiding role of the Tad pili. Complement depletion by treating with anti-C5 antibody rescued the viable count of Δtad123 in infected mouse bloodstream to the level comparable to wild type strain. Taken together, all three tad loci cooperate to confer successful invasion of V. vulnificus into deeper tissue and evasion from host defense mechanisms, ultimately resulting in septicemia.

Introduction Vibrio vulnificus is an opportunistic Gram-negative marine pathogen that causes fatal septicemia and necrotizing wound infections in susceptible individuals with underlying hepatic diseases and other immunocompromised conditions. V. vulnificus is halophilic and found worldwide in warm coastal and brackish waters in association with shellfish such as oysters and other sea animals. In humans, this pathogen frequently causes rapidly progressing fatal sepsis with a mortality rate of greater than 50% within a few days post-infection after eating raw seafood and contamination of preexisting wounds [1][2][3][4]. During the infectious process, V. vulnificus must cope with dramatic environmental changes by sensing changes in the host milieu [5]. To establish successful infections in vivo, V. vulnificus must manage spatiotemporally coordinated changes in the expression levels of various virulence genes.
To understand the genome-wide gene expression changes in V. vulnificus after infection, we recently performed a transcriptomic analysis of cells grown in vivo using a rat peritoneal infection model. Notably, among the newly identified in vivo-expressed genes, a Flp/Tad pilusencoding gene cluster (the tad1 locus) was found to be highly upregulated. Flp pili are polymers of the mature Flp pilin protein, and they are assembled and secreted by a complex of proteins encoded by the tad operon. Flp pili were reported to be abundantly expressed, extremely adhesive, and bundled in Aggregatibacter (previously Actinobacillus) [6][7][8][9]. The Tad proteins have been reported to be essential for adherence, biofilm formation, colonization, and pathogenesis in a number of genera and are considered to be instrumental in the colonization of diverse environmental niches [6,7,[10][11][12].
The genus Vibrio includes three main human pathogens (V. cholerae, V. parahaemolyticus and V. vulnificus), all of which were reported to carry genes of pili biogenesis. The type IV pili were most well characterized in pathogenic vibrios and studied for pathogenic roles. The mannose-sensitive hemagglutinin (MSHA) and chitin-regulated pilus (ChiRP), members of the type IV pilus, were reported to contribute to biofilm formation. V. cholerae toxin co-regulated pilus (TCP) is critical for host colonization and serves the cognate receptor for cholera toxin (CTX) phage [13][14][15]. The Flp subtype pilus related to the TCP is associated with the tight adherence (Tad), from which the gene locus name originated [16]. Homologs of the tad locus are widely distributed in the Vibrionaceae and many Vibrio genomes encode multiple tad loci [17].
V. vulnificus CMCP6 harbors three distinct tad loci [18] (S1 Fig), among which the tad1 locus has been identified as a possible virulence factor because of its ubiquity in sequenced virulent V. vulnificus strains [19][20][21]. The tad1 expression was preferentially induced under iron-rich conditions [18], whilst the tad3 locus was expressed in artificial seawater [22]. Recently, through transposon (Tn) insertion mutation analysis, the tad2 locus (VV2_0084 to VV2_0095) was reported to be important in initial surface attachment, auto-aggregation and resistance to mechanical clearance of bacterial biofilms [12]. However, the pathogenic roles of Tad pili have not yet been addressed for Vibrionaceae. This study attempted to investigate the contribution of the high in vivo expression of the tad1 operon to the V. vulnificus pathogenicity and to understand why three similar tad operons were maintained throughout the long history of evolution. We evaluated how each tad operon contributes to V. vulnificus virulence. Since the three tad operons share genes with similar function, single-gene-mutation analyses could not rule out overlapping functions of the remaining genes in the same operon or in the other tad operons. Thus, we constructed mutant strains with single and multiple complete tad loci deletions and then complemented them with individual cosmid clones harboring each tad operon. Using a variety of mouse infection models coupled with molecular genetic analyses, we demonstrate here that all three tad operons are required for V. vulnificus pathogenicity as the cryptic pili contribute to host cell and tissue invasion, survival in the blood, and resistance to complement activation.

Transcriptional analyses of the three structural flp genes in V. vulnificus infecting the rat peritoneal cavity
To understand how host signals modulate tad operon expression in V. vulnificus, we analyzed the in vivo transcriptional levels of three structural flp genes using a rat peritoneal infection model. Real time RT-PCR results indicated a significantly higher flp-1 mRNA level when the bacteria were grown in vivo, corresponding to an approximately 878-fold increase ( Fig 1A) (P < 0.001). Conversely, both the flp-2 and flp-3 transcript levels were slightly decreased when V. vulnificus was grown in vivo ( Fig 1A) (P < 0.05 for flp-2 and P < 0.001 for flp-3). The expression levels of the flp genes were also measured using conventional RT-PCR. Using different numbers of amplification cycles, we confirmed that the flp-2 and flp-3 genes were transcribed at low levels under both tested conditions; in particular, flp-2 expression was detected only after 35 cycles (Fig 1B).
In a wide variety of bacteria, type IV pili expression is solid-surface dependent [23,24], and the tad1 locus was recently found to be expressed under iron-limited conditions [18]. Thus, we measured the expression of the flp-1 gene under these growth conditions. As shown in Fig  1C and 1D, both the iron-limited and surface-associated growth conditions clearly stimulated flp-1 transcription, increasing its expression levels by approximately 131-and 210-fold, respectively (P < 0.001 compared to that of the expression level in HI broth). Combining these two conditions significantly increased the flp-1 transcript level by 367-fold (P < 0.01), but the transcription level was still much lower than that observed in vivo. This finding indicates that changing one or two growth parameters in culture does not mirror the in vivo environment, where multiple host factors and growth conditions would simultaneously influence tad1 operon expression. RNA was isolated from bacteria grown in the rat peritoneal cavity (IV, in vivo) or in 2.5% NaCl HI broth (IT, in vitro) and the transcript levels of the structural flp genes were analyzed via real-time (A) and conventional RT-PCR (B). RNA was isolated from bacteria grown under 2.5% NaCl HI broth, iron-limited, solid surface and in vivo conditions and the transcript levels of the structural flp genes were analyzed via realtime (C) and conventional RT-PCR (D). DP (80 μM) was added to 2.5% NaCl HI broth for iron limitation. The real-time RT-PCR data were normalized to gyrA and expression relative to the in vitro level. Data shown represent the mean ± SEM of three independent experiments performed in triplicate. Statistical analysis was carried out using Student's t test ( � , P < 0.05; ��� , P < 0.001).

All three tad operons are required for full V. vulnificus virulence
To explore the contribution of each tad operon to V. vulnificus pathogenicity, we performed mouse lethality assays employing intraperitoneal (i.p.) and intragastric (i.g.) infection routes. Interestingly, in the i.p. infection model, the Δtad123 mutant showed a 41-fold increase in the LD 50 , while the single and double mutants showed no differences (Table 1). Significantly prolonged survival was observed in the Δtad123 mutant-administered mice, which received infectious doses of 1.0 × 10 7 and 1.0 × 10 6 CFU/mouse. At a dose of 10 7 CFU/mouse, all of the mice infected with wild-type cells died within 5 hours post-infection, whereas approximately 60% of the mice infected with the Δtad123 mutant survived up to 48 h after the challenge (S2 Fig) (P < 0.01). However, after i.g. infection, which leads to slower translocation of the bacteria into blood circulation, we observed only a 10-fold LD 50 increase ( Table 1). The lethality varied depending on the route of infection, which influences the rate of bacterial invasion, growth and/or clearance at both the primary infection site and in the blood stream. Taken together, all three tad operons must be deleted to significantly ameliorate V. vulnificus virulence.

The Δtad123 mutant exhibits significantly decreased adhesion to HeLa cells
Since common pili are generally involved in the attachment of bacteria to surfaces in nature, we hypothesized that deletion of the three tad operons might influence the adhesive ability of V. vulnificus. To test this hypothesis, we performed an adhesion assay in which HeLa cells were infected with V. vulnificus at an MOI of 250 followed by quantification of the number of bacteria adhered to the host cells. After incubation for 45 min, the number of Δtad123 mutant cells adhered to the HeLa cells was 15-fold less than that of the parental wild-type strain ( Fig  2A) (P < 0.001). The wild-type strain formed small clusters of aggregated bacteria on the surfaces of the HeLa cells, eventually leading to cell lysis. In contrast, only a few Δtad123 mutant cells attached to the surfaces of the HeLa cells, and the infected host cells maintained cell contours similar to those of the uninfected cells. However, the adhesion of Δtad123 mutant cells to the host cells gradually increased in a time-dependent manner (

Structure of the Tad pili on the surface of V. vulnificus cells
To observe the morphology of the Tad pili, we prepared in vivo grown wild-type, Δtad123, and Δtad123 cells carrying pLAFR3::tad1 locus then performed scanning electron microscopy (SEM) observation (Fig 3). In the wild-type strain, the cell surface appeared to be covered with slime-like material. Corrugated elevations and grooves ran along the longitudinal axis. In the Δtad123 mutant strain, which was devoid of the slime-like materials, the grooves and elevations were more conspicuous compared with those of the isogenic wild-type strain. Moreover, the directionality of the convexity of the surface structure was absent on the mutant surface. Interestingly, the cell surfaces of the in trans tad1 complemented strains showed similar structural characteristics to the wild type strain, suggesting that the Tad pili contribute to the formation of the slime-like surface structure. However, the typical surface groove and convexity was less obvious in the complemented strains. These results suggest that the Tad pili of V. vulnificus might also contribute to the cell envelops biogenesis by including the slime-like outer structure. We tried to further characterize the thin fimbrial projections of the putative Flp pili via immunogold electron microscopy. Firstly, we produced recombinant Flp pilin proteins and attempted to raise specific antibodies against them in animals. However, we could not obtain any appreciably immunogenic antisera even after many repeated trials. Peptide-based immunizations were also unsuccessful in raising specific antibody responses. We came to conclude that the Flp pilin have very low immunogenicity. To solve this problem, we constructed V. vulnificus strains carrying a pBAD24::FlpV5 plasmid expressing a hybrid protein of Flp pilin fused to the highly immunogenic V5 tag (S5 Fig) with the expectation that the plasmidencoded FlpV5 subunits would assemble into growing pilus fibers under the control of an arabinose-inducible promoter. By using dot blot analysis with an anti-V5 antibody, we confirmed V5-positive signals in the wild type V. vulnificus transconjugants (WT-FlpV5) under the inducing conditions ( Fig 4A). The immunogold-labeling analysis revealed that wild-type cells displayed gold particles on their surface (white arrow), indicating expressed FlpV5 pilins assembled into the authentic pili ( Fig 4B). However, the transconjugated mutant cells (Δtad123-FlpV5) rarely displayed gold particles and this phenotype was complemented in trans by the cosmid harboring the wild-type tad1 locus ( Fig 4B). Interestingly, aggregated gold particles were detected on the grid near bacterial cells (orange arrow), suggesting FlpV5 proteins overexpressed under the arabinose induction did not assemble into pilus structure and exported from the bacterial cells. For quantitative analysis, we enumerated the gold particles associated with 5 bacterial cells per group (total 15 bacteria/group) from representative photos under x10,000 magnifications. The number of gold particles associated with the Δtad123 mutant cells was significantly scantier than that of the WT (P < 0.01) (Fig 4C). The decreased number of gold particles was significantly rescued by the tad1 operon complementation (P < 0.001) (Fig 4C). Supporting the results of the immunogold staining, immunofluorescence detection via confocal microscopy also revealed positive fluorescent signals for V5p in the FlpV5-expressing wild-type strain ( Fig 4D). In contrast, the Δtad123 mutant cells did not show any fluorescent signals, and this deficiency was complemented in trans by cosmids encoding the wild-type tad1 locus (Fig 4D). Many broken filaments were found in the backgrounds of the EM photos, suggesting detachment of the brittle pilus structures during the sample preparation procedures. Taken together, the EM analyses demonstrated the obvious existence of extracellular Flp structures in wild-type V. vulnificus CMCP6.

The Δtad123 mutant cells display defective RtxA1 production, leading to delayed cytotoxicity toward HeLa cells
RtxA1 is a crucial cytotoxin involved in cellular damage and necrosis of infected tissues [25][26][27][28][29]. We previously reported that host cell contact is required for RtxA1 production and cytotoxicity [25]. Thus, we speculated that attenuated adherence to host cells should hamper RtxA1 production and consequently attenuate host cell killing and tissue invasion. We performed a Western blot analysis to assess RtxA1 production after HeLa cell infection. The toxin was detected using an anti-GD domain antibody targeting the C-terminal fragment (RtxA1-C; approximately 130 kDa), which is internalized in the host cell cytoplasm [30]. As a result of its impaired ability to maintain contact with its host cells, the Δtad123 mutant exhibited significantly lower toxin production compared with that of its parental strain ( Fig 5A). RtxA1 was secreted in a time-delayed manner in the mutant cells, and its secretion gradually increased over time. This delay was significantly rescued by the tad operon complementation (Fig 5A). We next assessed the cytotoxicity of V. vulnificus toward HeLa cells over a time course. As shown in Fig

The Δtad123 mutation results in decreased biofilm formation
The contribution of each tad operon was investigated for the biofilm formation. The Δtad123 mutant showed significantly decreased biofilm formation compared with wild type strain (P < 0.001) (Fig 6A & 6B). The Δtad1 mutation appeared to play the most dominant role in the inhibition of V. vulnificus adhesion to host cells (S3C Fig). The biofilm formation defect of the Δtad123 mutant was complemented in trans by the cosmid harboring each of wild type tad loci (Fig 6A & 6B).

Tad pili are essential for intestinal invasion by V. vulnificus
Bacterial pili are used to attach to host cells and tissues, and confer invasive competence [16,[31][32][33][34]. Furthermore, secretion of the RtxA1 cytotoxin, which is induced by adhesion of the  Three tad operons are required for invasion and serum resistance bacterial cells to the host cells, is highly correlated with host tissue invasion [25]. To investigate the effects of mutation of the tad123 loci on V. vulnificus invasion, we carried out an in vivo invasion assay using a mouse ligated ileal loop infection model. The viable bacterial cells in the blood of the infected mice were quantified to evaluate tissue invasiveness of bacteria. The number of bacterial cells in the blood samples from the Δtad123 mutant-infected mice was significantly lower than that in the mice infected with the wild-type strain, even 6 hours postinfection ( Fig 7A) (P < 0.001).
In addition to the in vivo invasion assay, bacterial invasiveness was further confirmed using an in vitro intestinal epithelial barrier system. Polarized HCA-7 cells grown on Transwells were apically infected with bacteria, leading to physical apical-to-basolateral trans-epithelial migration of the bacteria. After 3 to 5 hours of incubation, we detected significantly fewer Δtad123 mutant cells than wild-type cells in the basolateral chamber ( Fig 7B) (P < 0.05). The single and double mutants showed no changes (S3D Fig). The cell count of the Δtad123 mutant reached that of the wild-type strain after 6 hours of incubation. The defect of transepithelial translocation of the Δtad123 mutant was complemented in trans by the cosmid harboring each of wild type tad loci ( Fig 7C). Taken together, the in vitro data and the in vivo invasion results demonstrate a function for Tad pili in conferring invasive competence to V. vulnificus.

Tad pili are important for V. vulnificus survival in blood
It is likely that the impaired invasion alone could not fully account for the higher LD 50 observed for the Δtad123 mutant in the intraperitoneal infection model (Table 1); therefore we hypothesized that mutation of the tad123 loci could compromise V. vulnificus survival in the bloodstream. To investigate this possibility, we monitored the number of viable bacteria in the blood over a time course following intraperitoneal (i.p.) or intravenous (i.v.) infection. Interestingly, the triple mutant cells were defective at surviving in mouse blood. Significantly fewer mutant cells were recovered from the blood of mice infected via both routes (Fig 8A and 8B). In particular, very few mutant cells were detected after direct introduction of the bacteria into the blood stream via i.v. injection, resulting in an approximately 3-log reduction in the number of CFUs compared with the number of CFUs detected for the wild-type strain (Fig 8A). The viable bacteria in the blood following i.p. infection should represent V. vulnificus cells that succeeded at both invasive translocation and resisting the serum bactericidal activities. On the other hand, the i.v. infection model shows how well the wild-type and mutant bacteria survived the serum bactericidal activities. These findings clearly indicate that Tad pili also play important roles in the survival of V. vulnificus in the blood stream.

Tad pili are required for V. vulnificus serum resistance
Serum bactericidal activity is an important innate immune defense against intravascular invasion by bacterial pathogens [35,36]. Thus, we hypothesized that Tad pili might play a protective role against serum components. To address this hypothesis, we tested the susceptibility of triple mutant cells to normal human serum (NHS). Bacterial viability was assessed after 2 hours of incubation with different NHS concentrations. Notably, the bacteria lacking all three tad loci were extremely sensitive to human serum (Fig 9A). For the Δtad123 mutant, 20% NHS led to dramatically decreased viability, and exposure to 40% NHS resulted in 4-log scale decrease of the viable cells. In contrast, the wild-type cells showed resistance against human serum (Fig 9A). A time course assay with 60% NHS was carried out to further compare the serum resistance levels of the isogenic mutant and the wild-type strains. During the first hour of incubation, more than 3-log scale of the Δtad123 mutant lost viability (Fig 9B). The single and double mutants showed no changes (S3E Fig). The serum susceptibility of the Δtad123 mutant was complemented in trans by the cosmid harboring each of wild type tad loci (Fig 9A & 9B). This result explains the significant difference in the survival rates observed between the wild type and Δtad123 mutant strains, and further confirms that Tad pili are required for V. vulnificus serum resistance. In vivo invasion. Log-phase bacteria were inoculated into ligated ileal loops of mice. Blood samples were acquired from infected mice by cardiac puncture at the indicated times. Viable cells were counted by plating on 2.5% NaCl HI agar plates. Data shown represent the mean ± SEM of three independent experiments performed in five mice. (B) In vitro invasion. HCA-7 cells grown on Transwell filters were apically exposed to log-phase bacteria. Invasiveness was determined by measuring the number of bacterial cells that translocated from the apical to the basolateral compartment of the Transwells. WT (pLAFR3), wild type harboring pLAFR3; Δtad123 (pLAFR3), Δtad123 mutant harboring pLAFR3; Δtad123 (pLAFR3:: tad1), Δtad123 mutant in trans complemented with pLAFR3::tad1 locus; Δtad123 (pLAFR3::tad2), Δtad123 mutant in trans complemented with pLAFR3::tad2 locus); Δtad123 (pLAFR3::tad3), Δtad123 mutant in trans complemented with pLAFR3::tad3 locus. Data shown represent the mean ± SEM of three independent experiments performed with triplicates. (C) Restoration of in vitro invasion in tad-complemented strains at 6 h post-incubation in the presence of 2 μg/ml tetracycline. Viable bacterial cells were counted by plating on 2.5% NaCl HI agar plates. Data shown represent the mean ± SEM of three independent experiments performed with triplicates. Statistical analysis was carried out using Student's t test. � , P < 0.05; �� , P < 0.01; ��� , P < 0.001; ND, not detected (detection limit; 1x10 3 CFU/mL). Given that the complement system, which is activated by pathogenic bacteria is primarily responsible for the direct killing of bacteria in NHS [37], we further dissected the bactericidal activity of three complement pathways activated by V. vulnificus. Indeed, the use of heat-inactivated serum (HIS) lacking the lytic complement activity successfully rescued the viability of the mutant strain (Fig 9C). This finding indicates that heat-labile complement proteins are responsible for the killing of the Δtad123 mutant cells in NHS. Complement activation occurs via one or more of three pathways: the classical pathway, the MBL/lectin pathway and the alternative pathway [35]. To identify which complement pathway was responsible for the death of the Δtad123 mutant cells in serum, we selectively blocked the specific complement activation pathways. Remarkably, inhibition of the alternative pathway completely ablated the complement-mediated killing activity. The survival of the Δtad123 mutant cells fully recovered to the wild-type level in bentonite-absorbed NHS (Fig 9C). Furthermore, inhibition of either injected with bacteria that had been previously incubated in the rat peritoneal cavity for 6 hours for in vivo adaptation. Blood samples were acquired from infected mice via cardiac puncture at the indicated times. Viable bacterial cells were counted by plating on 2.5% NaCl HI agar plates. Very low numbers of Δtad123 mutant cells were recovered from the blood of infected mice. Data shown represent the mean ± SEM of three independent experiments performed in seven mice. Statistical analysis was carried out using Student's t test ( � , P < 0.05; �� , P < 0.01; ��� , P < 0.001).
https://doi.org/10.1371/journal.ppat.1007767.g008 the classical pathway or the lectin pathway partially recovered the survival of the mutant cells (Fig 9C). The serum susceptibility of the Δtad123 mutant was complemented in trans by the cosmid harboring each of wild type tad loci (Fig 9C). Taken together, these results indicate that Tad pili likely play an important role in protecting V. vulnificus from direct complementmediated bacteriolysis resulting predominantly from activation of the alternative pathway.

Depletion of mouse complement system support in vivo survival of Δtad123 mutant
To test whether serum susceptibility of the Δtad123 mutant is reverted in vivo by inhibition of complement system, we pretreated mice with a specific complement-inhibiting drug (anti-C5 antibody) approved by FDA [38] and then infected the mice with wild type or Δtad123 mutant cells. We monitored the number of viable bacteria in the blood in a time course following intraperitoneal infection. As shown in Fig 9D, the viable bacterial count of Δtad123 mutant could be rescued by the anti-C5 antibody treatment almost to the level of wild type strain. These data clearly indicate that reduced viability of the Δtad123 mutant is due to increased susceptibility to complement system in the mouse infection model. Given the innately lower complement level in mice compared with human [39], the Tad pili system should play more dominant roles in the pathogenesis of human infections.

Discussion
Host-pathogen interactions during microbial infections can be described as a dynamic battlefield where the microbe's clever strategies for survival and multiplication confront the formidable host immune defenses. To investigate the virulence regulation of V. vulnificus during infection, we recently performed comparative genome-wide transcriptional analyses of cells grown in vitro and in vivo. A rat peritoneal infection model was used to simulate the physiological host milieu. Interestingly, among the newly identified in vivo-expressed candidate genes, tad1 was found to be highly upregulated in vivo (unpublished data). The pathogenic potential of the tad1 cluster is also supported by previous reports of the ubiquity of the tad1 locus in sequenced virulent V. vulnificus strains [19][20][21]. Furthermore, it is notable that the genome of V. vulnificus CMCP6 contains three distinct tad loci, in which similar functional genes are organized in the same order and transcriptional orientation. In this study, our goals were to investigate why V. vulnificus CMCP6 has maintained three tad loci throughout evolution and how each tad operon contributes to V. vulnificus virulence and to determine whether all three tad operons are required for its virulence. By deleting each tad locus and complementing the deletion in trans, we attempted to address these questions (at least in part) and found that all three tad operons are required for the full virulence of V. vulnificus. Only complete abrogation of all three tad loci led to significantly decreased lethality in mice (Table 1). Based on animal and cell culture infection models coupled with molecular genetic analyses, we came (pLAFR3::tad1), Δtad123 mutant in trans complemented with pLAFR3::tad1 locus; Δtad123 (pLAFR3::tad2), Δtad123 mutant in trans complemented with pLAFR3::tad2 locus); Δtad123 (pLAFR3::tad3), Δtad123 mutant in trans complemented with pLAFR3::tad3 locus. Data shown represent the mean ± SEM of four independent experiments performed in triplicates. Statistical analysis was carried out using Student's t test (C). ��� , P < 0.001. ns, non-significant (detection limit; 1x10 3 CFU/mL). (D) Viable V. vulnificus cells in bloodstream of anti-C5 monoclonal antibody pretreated mice. Groups of mice were administered with 40mg/kg mouse anti-C5 two times as described Materials and Methods and then the mice were intraperitoneally infected with 5x10 5 wild type or Δtad123 mutant cells previously incubated in the rat peritoneal cavity for 6 hours for in vivo adaptation. Blood samples were acquired by eye bleeding. Viable bacterial cells were counted by plating on 2.5% NaCl HI agar plates. Data are shown as the mean ± SEM (n = 5). Statistical analysis was done using Student's t test ( � , P < 0.05; �� , P < 0.01; ��� , P < 0.001). IC, isotype control. https://doi.org/10.1371/journal.ppat.1007767.g009 Three tad operons are required for invasion and serum resistance to understand the coordinated contributions of the three V. vulnificus tad operons to host cell invasion as well as to survival of complement-mediated bacteriolysis. Deletion of all three tad loci impaired the adherence of the bacterial cells to the host cells (Fig 2), thus hampering RtxA1 cytotoxin production and delivery (Fig 5) and, consequently, tissue invasion (Fig 7). These results corroborate our previous findings that host cell contact is required for V. vulnificus toxin secretion and pathogenicity [25].
The bactericidal action of serum is an important component of the host defense against bloodstream infections [35,36]. As most fatal cases of V. vulnificus infection result from septicemia, serum resistance is considered an essential feature for survival in the host environment. It is well documented that clinical V. vulnificus isolates have a significantly greater survival ability in human serum compared with that of environmental isolates [40,41]. Several mechanisms have been proposed to explain this phenomenon, the most significant of which may be differences in siderophore expression and/or capsule formation [40]. In the present study, we discovered a novel function of V. vulnificus Tad pili in conferring resistance to the complement-mediated bactericidal activity of its host. The ubiquity of the tad1 cluster in virulent V. vulnificus strains suggests that the surface expression of Tad pili may be another key determinant for the survival of V. vulnificus in host milieus, which could conceivably differentiate the clinical and environmental strains. Bacteria lacking Tad pili rapidly lost viability in serum via direct complement-mediated bacteriolysis, predominantly activated via the alternative pathway. The mechanism through which Tad pili protect bacteria from complement attack should be further studied. The poor immunogenicity of Flp pili could have something to do with the serum resistance. Given that the tad triple mutant lost its slime-like surface morphology (which could be complemented in trans by cosmids harboring an individual tad operon), it is plausible that Flp pili could anchor secreted polysaccharides during formation of durable capsular lattice.
Inhibition of the alternative complement pathway by the Flp pili might be related to the low immunogenicity of the structural pilin protein. To understand the poor recognition of Tad pili by the immune system, we analyzed the antigenicity and structural characteristics of Tad pilin using bioinformatics tools. After in silico prediction of the 3D structure of Tad pilin, we compared it with the orthologs from Aggregatibacter actinomycetemcomitans, which seem to have high functional similarity with that of V. vulnificus. Bordetella pertussis Fim2 and Escherichia coli CfaB (S7 Fig). V. vulnificus Tad pilin was predicted to form an alpha helix that partially overlaps structurally with the A. actinomycetemcomitans Flp1 and Flp2, which share the Flp common motif [8], and B. pertussis Fim2 pilins, which are thought to contribute to the assembly of pilin monomers into the fimbrial ultrastructure. When compared with the immunogenic Fim2 and CfaB pilins, which function as vaccine candidates, Tad pilin appeared to be relatively hydrophobic, and only a small fraction contained the hydrophilicity required for antigenicity ( S8 Fig). The presence of pilus-like structures has been reported to be more closely associated with clinical isolates of V. vulnificus than with environmental strains [42]. Under SEM, we observed filamentous surface structures that extended from the wild-type cell bodies that were absent from the Δtad123 triple operon mutant cells (Fig 3). Interestingly, the presumable Tad pili structure became more elongated when the cells were grown in vivo, suggesting a pathogenic function during establishment of successful infections. This morphological change under in vivo culture conditions corroborates the previously suggested hypothesis that the Tad pili significantly contribute to V. vulnificus pathogenicity [12]. The thickness and size of the putative pili structures were quite elusive, unlike other Tad/Flp pili such as those of Aggregatibacter (previously Actinobacillus) [6][7][8][9]. To confirm the cell surface expression of Tad pili, we performed immunogold-labeling and fluorescence staining. To the best of our knowledge, no specific ultrastructural analysis of V. vulnificus Tad pili and their molecular pathogenic roles have been previously reported. While performing these experiments, we failed to raise functional antibodies against the V. vulnificus Flp pilin, presumably because of its very low hydrophilicity and immunogenicity as addressed above. To overcome this obstacle, we first fused a V5 tag sequence to the N-or C-terminus of the flp-1 gene on the chromosome. However, we could not detect any signal from chromosomally V5-tagged Flp in a dot blot analysis using an anti-V5 antibody. The V5-tagged Flp might have been structurally defective, leading to ineffective assembly into pili structures. Alternatively, the engineered strains may have incurred polar effects on downstream gene expression during double crossover homologous recombination. We subsequently constructed a V5-tagged Flp overexpression system encoded on the multicopy pBAD24 plasmid (S5 Fig). Our hypothesis was that, when expressed under the control of an arabinose-inducible promoter, some proportion of the overexpressed V5-tagged Flp proteins might be randomly incorporated during assembly of the chromosomally expressed native pilin subunits. As expected, we could detect positive signals from V. vulnificus cells overexpressing the Flp-V5 fusion protein in both experiments (Fig 4). However, only a fraction of the transconjugants carrying the overexpression plasmid could be stained with immunogold or fluorescence approaches, suggesting minimal incorporation of V5-tagged Flp expressed from the single-copy chromosomal locus possibly due to structural deformations by addition of the V5 tag.
Taken together, our results provide new insights into the pathogenic significance of Tad pili in V. vulnificus CMCP6. The Tad provide pathogenic V. vulnificus with the ability to adhere to and invade host cells and shield the cells against complement-mediated bacteriolysis inside the host. During these two distinct stages of infection, the Tad pili-mediated host cell adhesion and evasion of the anti-complement activity inside host provide, respectively, the signal required to induce expression of the potent RtxA1 cytotoxin and the ability of V. vulnificus to robustly grow in vivo.

Bacterial strains, plasmids and media
Bacterial strains and plasmids used in this study are listed in Table 2. V. vulnificus CMCP6 is a highly virulent clinical isolate from the Chonnam National University Hospital, South Korea [5,25]. V. vulnificus and E. coli were grown in 2.5% NaCl heart infusion (HI) and in Luria-Bertani (LB) medium, respectively. Antibiotics were used as previously described [43].

Locus deletion mutant construction and complementation
We constructed in-frame single, double, and triple deletion mutants of entire genes in the tad1, tad2 and tad3 loci by the allelic-exchange method. We designed two sets of primers to amplify~1-kb DNA fragments in the upstream or downstream region of each tad operon (S1 Table). The primers were synthesized with overhangs recognized by specific restriction enzymes (REs). The upstream and downstream amplicons of each tad operon were ligated by cross-over PCR to produce a 2-kb fragment [44]. The fusion fragments were digested with appropriate REs and subcloned into pDM4 suicide vector. The resulting recombinant vector was transformed into E. coli SM10 λ pir and subsequently transferred into V. vulnificus CMCP6 by conjugation. Stable Cm R transconjugants were selected on Vibrio-selective thiosulfate citrate bile salt sucrose (TCBS) agar plate containing Cm. Plating of the transconjugants on 2.5% NaCl HI agar plate containing 10% sucrose was performed to select clones that experienced second homologous recombination events forcing excision of the vector sequence and leaving only mutated or wild-type allele of the genes. Each in-frame deletion mutation was confirmed by PCR with the chromosomal DNA from the respective mutant as template.
For the use in genetic complementation experiments, we screened cosmid clones that contain intact tad1, tad2 or tad3 operon from a pLAFR3 cosmid library of V. vulnificus CMCP6 [25,45]. The selected cosmid library clone containing an individual tad operon was transferred to the triple tad operon deletion mutant by triparental mating with a conjugative helper plasmid pRK2013. The transconjugants were screened on TCBS agar plates containing tetracycline

Conventional and real-time RT-PCR
The transcriptional levels of the three structural flp genes, which encoded the major structural components of Flp pili, were measured by conventional and real-time RT-PCR. gyrA was chosen as the reference gene forn the qRT-PCR as previously reported [45]. Forward and reverse primer pairs were designed and are provided in S2 Table. Total RNA was isolated from logphase bacterial cells grown in the rat peritoneal cavity or in 2.5% NaCl HI broth using the RNeasy minikit (Qiagen). One microgram of purified RNA was converted into cDNA using QuantiTect Reverse Transcription Kit (Qiagen) in accordance with the manufacturer's protocol. qRT-PCR was performed to quantify each target transcript using QuantiTect SYBR green PCR kit (Qiagen). The relative gene expression was normalized to the expression of gyrA using the threshold cycle (ΔΔC T ) method [46]. For conventional RT-PCR, 16S rRNA was used as the internal standard. After 25 to 35 cycles, the amplicons were separated on 2% (wt/vol) agarose gels and stained with ethidium bromide. The transcription levels of flp-1 under iron-limited and solid surface growth conditions were also analyzed by qRT-PCR. For the iron limitation experiment, dipyridyl (Sigma-Aldrich) was added to the 2.5% NaCl HI broth at a final concentration of 80 μM for iron limitation.

Ethics statement
All animal experimental procedures were performed with approval from the Chonnam National University Institutional Animal Care and Use Committee under protocol CNU IACUC-H-2015-44. Animal research facility maintenance and experimental procedures were carried out strictly keeping the guideline in the Animal Welfare Act legislated by Korean Ministry of Agriculture, Food and Rural Affairs.

LD 50 determination
The intraperitoneal 50% lethal dose (i.p. LD 50 ) of V. vulnificus was determined using 7-weekold, randomly bred specific-pathogen-free (SPF) female ICR mice (Daehan Animal Co., Daejeon, South Korea). Five mice per group were intraperitoneally inoculated with 10-fold serial dilutions of fresh bacterial suspensions (10 9 to 10 5 CFU/mouse). The intragastric (i.g.) LD 50 was determined using six-day-old randomly bred SPF CD-1 suckling mice (Daehan Animal Co., Daejeon, South Korea). Seven mice per group were intragastrically administered with 10-fold serial dilutions of fresh bacterial suspensions containing 0.1% Evans blue (Sigma-Aldrich) to ensure correct i.g. administration. The control animals received 100 μl of PBS containing 0.1% Evans Blue. The challenged mice were monitored for 48 h. LD 50 values were calculated based on probit analysis, using IBM SPSS 21.0 software (IBM).

Construction of C-terminally V5-tagged Flp (Flp-V5)
DNA fragments of the structural flp-1 pilus gene without its stop codon were amplified and subcloned into the pBAD24 vector. Subsequently, double-stranded oligonucleotides encoding the V5 peptide, "GGTAAGCCTATCCCTAACCCTCTCCTCGGTCTCGATTCTACGTAA", were fused to the C-terminus of the flp-1 gene in the pBAD24-Flp plasmid. At the end of the V5 sequence, a TAA codon was added to terminate translation. The pBAD24 plasmids containing the Flp-V5 fusion protein were transformed into E. coli DH5α competent cell. The sequence of the cloned fragment was confirmed by DNA sequencing. The resulting vectors were transferred into V. vulnificus via triparental mating with a conjugative helper plasmid pRK2013. The transconjugants were screened on 2.5% NaCl HI agar plates containing ampicillin.

Dot blot analysis
V. vulnificus strains carrying pCMM2103 (pBAD24::Flp-V5) were grown for 4 h on 2.5% NaCl HI agar plates containing ampicillin. Flp-V5 expression was induced for 4 h via addition of 0.1% L-arabinose. The cell suspensions were applied to nitrocellulose membrane and fixed with 4% paraformaldehyde for 20 min. The membrane was blocked for 1 h using 5% skim milk in PBS and then incubated with anti-V5 polyclonal antibodies (diluted 1:5000, Abcam) for 2 h. After washing, the membrane was developed with HRP-conjugated goat anti-rabbit IgG secondary antibody (Dako). Stained dots on a white background indicated positive results.

Scanning electron microscopy
To compare surface structure of the wild-type, Δtad123 and Δtad123 (pLAFR3::tad1), we performed SEM observation. Bacteria were grown in vivo using a rat peritoneal infection model as previously described [45]. To minimize shearing force during bacterial preparations, all procedures were carefully performed. And we also applied osmotic adaptation with fixative by 3 staged applying a step-down approach from 2.5% NaCl containing fixative to 0.9% and 0% NaCl containing solution with a gentle agitation. Bacterial cells were fixed at room temperature for 4 h in a fixation solution containing 0.5% glutaraldehyde and 4% paraformaldehyde in 0.05 M sodium cacodylate buffer (pH 7.2). After three washes with 0.05 M cacodylate buffer, all of the samples were mounted on nickel grids coated with carbon film (150 mesh) (EMS, USA). After blocking nonspecific binding sites with 1% BSA in EM-immunogold (EMG) buffer (0.05% Tween, 0.5 M NaCl, 0.01 M phosphate buffer, pH 7.2), the samples were incubated at 4˚C for 24 h with anti-V5 tag monoclonal antibody (ab27671, Abcam, UK) at a 1:20 dilution at 4˚C, followed with incubation for 1 h in goat anti-rat antibody (1:50) conjugated to 6 nm gold particles. The grids were washed in EMG buffer, PBS, and distilled water and stained for 12 min with 4% uranyl acetate in deionized distilled water. The surfaces of all of the samples were observed using a field emission scanning electron microscope (Helios G3 CX, FEI Co., Hillsboro, Oregon, USA) at 1 kV acceleration with TED mode.

Immunogold labellng
A drop of V. vulnificus cell suspension was applied to a nickel grid coated with carbon film for 1 min. Because of its structural fragility of V. vulnificus, we prepared the bacterial sample with very gentle manner such as limited frequencies of pipetting and washing processes. Moreover, to minimize the insults from the critical points drying, we firstly fixed the in vivo grown cells with osmolarity-modified fixative (which contains 2.5% NaCl) and changed the solution to conventional fixative for SEM study (0.5% glutaraldehyde and 4% paraformaldehyde in 0.05 M sodium cacodylate buffer (pH 7.2)) under room temperature. Moreover, to reduce damages from electron and enhance the resolution beam during SEM analysis, we used the focused ion scanning electron microscope (FIB). Subsequently, the samples were incubated with the anti-V5 polyclonal antibody (diluted 1:20, Abcam) and labeled with 5-nm colloidal gold-conjugated goat anti-rabbit IgG secondary antibody (diluted 1:20, BritishBio-Cell, UK).

Confocal microscopy
To induce V5-tagged pilin expression, mid-log phase V. vulnificus cells were grown for 4 h on 2.5% NaCl HI-ampicillin agar plates supplemented with 0.1% L-arabinose. Bacterial pellet was then gently suspended in PBS buffer. The induced bacterial cells were directly immobilized on poly-L-lysine-coated coverslips. Samples were fixed with 4% formaldehyde for 30 min and then incubated with primary anti-V5 antibodies (1:300) for 2 h. After three washes with PBS buffer, the cells were incubated for 1 h with a Texas Red-conjugated anti-rabbit secondary antibody (Molecular Probes) and DAPI (Invitrogen). The samples were observed under a laser scanning confocal microscope (LSM 510, Zeiss, Oberkochen, Germany), and the obtained images were analyzed by using the ZEN Lite software (Zeiss, Oberkochen, Germany).

Adhesion assay
We quantitatively analyzed bacterial adhesion to host cell by using viable cell counting and microscopic observation. HeLa cell monolayers (5x10 5 /well) were grown on 24 well plates (SPL, cat#30024) and then infected for 45 minutes with log-phase V. vulnificus cells at MOI 250. The monolayer was washed twice with PBS to remove non-adherent bacteria. The wells were then suspended with 200 μl of PBS containing 0.1% Triton and incubated for 10 minutes at room temperature. The number of colony forming units (CFU) that adhered to Hela cells was enumerated by ten fold serially diluting in PBS and spotted on HI agar plates. For microscopic observation, the cells were fixed in methanol after the PBS washing and stained with 0.1% Giemsa solution (Sigma-Aldrich). The number of V. vulnificus cells that adhered to single HeLa cells was counted under a light microscope at 400× and 1,000× magnification (Nikon Eclipse 50i, Japan).

Determination of biofilm and confocal imaging
To induce the biofilm formation, freshly cultured V. vulnificus cells (5x10 5 CFU/ml) were applied into each well of 24 well plates (SPL, cat#30024). The plates were further incubated at 37˚C for 24 hours and then gently washed once with PBS. The wells were stained with 200 μl of 0.3% crystal violet for 15 minutes and gently washed with PBS. The stained biofilm was extracted with 200 μl of 100% ethanol and two fold diluted with PBS to measure the absorbance at 595 nm by a microplate reader (Molecular Devices Corp., Menlo Park, CA). To acquire the confocal microscopic images of biofilm, the biofilm was induced in 4 Well Cell Culture Slide (SPL, cat#30124) for 24 hours and then gently washed once with PBS. The biofilm was then stained with acridine orange and observed by a confocal microscope as previously reported [47]. The samples were observed under a LSM510 confocal microscope (Zeiss, Oberkochen, Germany), and the obtained images were analyzed by using the ZEN 2012 x32 blue software.

Western blot analysis
To detect RtxA1, HeLa cells grown in 6-well plates were infected for 35 and 45 min with logphase V. vulnificus strains at an MOI of 100. The bacteria attached to the HeLa cells were lysed using a lysis buffer (Cell Signaling), followed by concentration using an Amicon Ultra-0.5 centrifugal filter apparatus (Merck KgaA). The samples were then subjected to 10% SDS-PAGE. RtxA1 proteins were detected using an anti-GD domain antibody (RtxA1-C, a band of approximately 130 kDa) [30].

Cytotoxicity assay
To determine the effect of tad operon mutations on cytotoxicity against HeLa cells, we performed the lactate dehydrogenase (LDH) release assay as previously described [43].

In vivo invasion assay
Bacterial cells that translocated from the intestine to the bloodstream were measured as previously described [48]. Seven-week-old randomly bred SPF female ICR mice were starved for 16 h. The ileum was tied off in a 5-cm segment and log-phase V. vulnificus cells (4.0 × 10 6 CFU/ 400 μl) were inoculated into the ligated segment. Blood samples were acquired from the infected mice via cardiac puncture. The number of viable bacterial cells was counted by plating on 2.5% NaCl HI agar plates. In parallel, viable V. vulnificus cells in the ligated ileal loops were also enumerated by plating on TCBS agar plates.

In vitro invasion assay
Polarized HCA-7 cells grown in Transwell filter chambers (8 μm pore size; CoStar, Cambridge, MA, USA) were apically exposed to log-phase V. vulnificus cells at an MOI of 5. Invasiveness was determined by measuring the number of bacterial cells that translocated from the apical to basolateral compartment of the Transwells. Viable bacterial cells were counted by plating on 2.5% NaCl HI agar plates.

Determination of in vivo V. vulnificus growth
In vivo growth of V. vulnificus was measured using the dialysis tube implantation model as previously described [48]. CelluSep H1 dialysis tubing (MWCO 12,000~14,000; Membrane Filtration Products, Inc. Texas) was incubated with PBS overnight. The dialysis tube was disinfected with 70% alcohol for 1 h and washed three times with sterile PBS before use. Sevenweek-old female Sprague Dawley (SD) rats (DBL. Co. Ltd, Daejeon, Korea) were anesthetized with a mixture of 10% Zoletil and 5% Rumpun dissolved in PBS. Three 10-cm dialysis tubes containing 2 ml of 5 × 10 5 CFU/ml V. vulnificus cells were surgically implanted into the rat peritoneal cavity. The bacterial growth at each time point was analyzed using three rats. Culture samples were harvested for viable cell counting on 2.5% NaCl HI agar plates 2, 4 and 6 h after implantation.

V. vulnificus growth in mouse blood
To assess bacterial growth in blood, seven mice per group were intravenously (i.v.) or i.p. injected with 100 μl of 5 × 10 5 CFU cells that had been incubated in the rat peritoneal cavity for 6 h for induction of tad1 locus expression and in vivo adaptation. Blood samples were acquired from the infected mice via cardiac puncture at the indicated times. Viable bacterial cells were counted by plating on 2.5% NaCl HI agar plates.

Determination of serum bactericidal activity against V. vulnificus
Log-phase V. vulnificus cells (1.0 × 10 7 CFU/10 μl) were added to 200 μl of PBS containing various NHS concentrations. The samples were incubated at 37˚C for appropriate times. Viable bacterial cells were counted by plating on 2.5% NaCl HI agar plates. To block activation of the classical pathway, NHS was pretreated with 10 mM ethylene glycol-bis (2-aminoethylether)-N, N, N 0 , N 0 -tetraacetic acid and 5 mM MgCl 2 for 30 min at 37˚C (EGTA/Mg 2+ ) [49]. To prepare the MBL-depleted serum, mannose-agarose beads (Sigma-Aldrich) were washed three times with sterile PBS and then incubated with NHS at 4˚C for 1 h with gentle rotation [50]. The alternative pathway is inhibited via properdin absorption with bentonite [51]. 10 mg of bentonite was washed three times with PBS and incubated with NHS at 37˚C for 10 min to absorb the properdin.

Anti-C5 pretreatment and determination of V. vulnificus in bloodstream
Seven-week-old randomly bred SPF female ICR mice were intraperitoneally administered with 40 mg/kg/day of anti-C5 monoclonal antibody (Hycult BiotechInc, Pa, USA) or isotype control two times in two-days interval following manufacturer's instruction. Twenty-four hours after the second administration of the anti-C5, mice were intraperitoneally infected with 5 x 10 5 CFU cells that had been incubated in the rat peritoneal cavity for 6 h for in vivo adaptation. Blood samples were acquired from the infected mice by eye puncture in a time course. Viable bacterial cells were counted by plating on 2.5% NaCl HI agar plates.

Statistical analysis
The results are expressed as the mean ± standard error of the mean (SEM) unless otherwise stated. Each experiment was repeated a minimum of three times, and the results from representative experiments are shown. Statistical analyses were performed using the Prism 6.00 software for Windows (GraphPad software, San Diego, CA). Multiple comparisons were performed using Student's t test and analysis of variance (ANOVA) followed by Bonferroni post hoc tests. P-values < 0.05 was considered statistically significant.