Vibrio vulnificus is autochthonous to estuaries and warm coastal waters. Infection occurs via open wounds or ingestion, where its asymptomatic colonization of seafood, most infamously oysters, provides a gateway into the human food chain. Colonization begins with initial surface contact, which is often mediated by bacterial surface appendages called pili. Type IV Tad pili are widely distributed in the Vibrionaceae, but evidence for a physiological role for these structures is scant. The V. vulnificus genome codes for three distinct tad loci. Recently, a positive correlation was demonstrated between the expression of tad-3 and the phenotypes of a V. vulnificus descendent (NT) that exhibited increased biofilm formation, auto-aggregation, and oyster colonization relative to its parent. However, the mechanism by which tad pilus expression promoted these phenotypes was not determined. Here, we show that deletion of the tad pilin gene (flp) altered the near-surface motility profile of NT cells from high curvature, orbital retracing patterns characteristic of cells actively probing the surface to low curvature traces indicative of wandering and diminished bacteria–surface interactions. The NT flp pilin mutant also exhibited decreased initial surface attachment, attenuated auto-aggregation and formed fragile biofilms that disintegrated under hydrodynamic flow. Thus, the tad-3 locus, designated iam, promoted initial surface attachment, auto-aggregation and resistance to mechanical clearance of V. vulnificus biofilms. The prevalence of tad loci in the Vibrionaceae suggests that they may play equally important roles in other family members.
Biofilms provide a protective environment for growth and, in nature, the majority of microorganisms live within complex biofilm communities.1 For many bacteria, Type IV pili (T4P) have been shown to play an important role in the switch from a free-swimming, planktonic lifestyle to a surface-attached, sessile existence by enhancing initial surface association and subsequent micro-colony formation.2,3 The pilin subunits that comprise T4P are synthesized as prepilins that are then processed prior to incorporation into the pilin structure. Prepilins are classified into two types (A and B) based on their characteristics and the systems used for their export and assembly.4,5 B-type prepilins can be further subdivided and the Flp subtype is associated with the tight adherence (tad) pilus locus.6,7,8
The tad locus was first reported as being required for biofilm formation by the oral pathogen, Actinobacillus actinomycetemcomitans and homologous loci were subsequently identified in several other pathogens.8 Lesions in the tad locus decreased biofilm formation or caused attenuation in several of these species,9,10 leading to the locus being dubbed a widespread colonization island due to its requirement for colonization or virulence by these pathogens and its conservation in diverse archaeal and bacterial species.8,11
The Vibrionacea includes three main human pathogens: Vibrio cholerae, V. parahaemolyticus, and V. vulnificus.12 Vibrio cholerae, which causes millions of cholera cases and 170,000 deaths each year,13 uses the A-type mannose-sensitive hemagglutinin (MSHA) and chitin-regulated competence (ChiRP) pili to attach to abiotic and chitinous surfaces, respectively.14,15 The B-type toxin coregulated pilus is essential for intestinal colonization and pathogenesis.16 The MSHA and ChiRP pili have been shown to play similar roles in V. parahaemolyticus,17,18 a major cause of gastroenteritis, and V. vulnificus,19,20,21 which has the highest death rate (>35%) and economic burden per case (>$3,000,000) of any foodborne pathogen in the United States.22 Homologs of the tad locus are widely distributed in the Vibrionaceae and many Vibrio genomes encoded multiple tad loci;8,23 however, a physiological function had not be described for a Tad pilus for any family member. The V. vulnificus genome codes for three distinct tad loci (tad-1, tad-2, and tad-3). We recently showed that tad-3 expression contributed to the increased biofilm formation and oyster colonization phenotypes of a V. vulnificus descendent (NT) relative to its parent. However, the mechanism by which tad-3 pilus expression promoted these phenotypes was not known. Here, we show that its expression promoted initial bacterial surface attachment and auto-aggregation, and increased the resistance of V. vulnificus biofilms to mechanical clearance—phenotypes associated with efficient niche (oyster) colonization.20,23,24
tad-3 expression supports the formation of large NT aggregates
The V. vulnificus genome codes for three tad loci, designated tad-1, tad-2, and tad-3, that are homologs of the A. actinomycetemcomitans tad locus11,23 (Fig. 1). The tadABC, rcpAB, flp, tadV, and tadEF genes shared the highest homology (31–56%) among the tad loci.
Unlike WT, NT grew as visible clumps (Fig. S1). To gain insight into the level at which tad-3 contributed to this aggregation phenotype, bacterial counts in 50-μm filtrates (CFUf) relative to the starting culture (CFUs) were determined for the WT, NT, NTΔflp, and complemented (NTΔflp-C) strains (cultures with an optical density at 600 nm of 0.1–0.15 were used). The CFUf/s of WT cultures was 0.992, indicating that the majority of cells in the starting culture were recovered in the 50-μm filtrate (Fig. 2a). Over half of NT cells were removed following the filtration step (CFUf/s = 0.456), suggesting that many of the cells formed aggregates too large to traverse the filter. The CFUf/s of NT cultures that were vortexed to shear surface appendages such as pili and flagella (NT-S) was 0.942, similar to WT. Deletion of flp (NTΔflp) increased the number of cells in the filtrate relative to NT (CFUf/s = 0.766) and complementation of NTΔflp with a plasmid-borne copy of flp decreased the CFUf/s to 0.506. Imaging of the starting cultures suggested that WT existed largely as single cells, while NT could self-associate into massive aggregates >100 μm in diameter (Fig. 2b). Large aggregates (>20 μm) were clearly visible for NTΔflp, but those >50 μm in size were in lower abundance relative to NT. Together, these results suggested that pilin production promoted V. vulnificus auto-aggregation.
FACS analysis of 50-μm filtrates was used to more accurately measure size distribution in the cell population. Greater heterogeneity was observed for NT cells than WT cells (Fig. 2c). WT bacteria primarily existed as single cells (0.8% aggregates), while 15% of NT cells formed larger aggregates (Fig. 2d). Deletion of flp (NT∆flp) decreased the percentage of aggregates to 4.8% and this could be complemented following flp expression from a plasmid (12% aggregates for NT∆flp-C). These results suggested that flp, and by extension the Tad pilus, contributed to the increased auto-aggregation phenotype of NT.
tad-3 expression contributes to rugose colony development
We previously demonstrated that elevated intracellular c-di-GMP concentrations induce rugose colony development in V. vulnificus.25,26 To determine if tad-3 contributed to the rugose phenotype, the diguanylate cyclase DcpA was expressed in NT and NTΔflp cells to increase intracellular c-di-GMP levels. Rugosity developed in both backgrounds but with different temporal and apical dynamics (Fig. 3). Colonies remained flat and featureless in both backgrounds over the first 12 h. By 24 h, the dramatic surface undulations that are a hallmark of rugose colony development were clearly visible for NT, while vertical expansion of NTΔflp colonies was less evident. Notably, NT exhibited a well-developed rugose phenotype by 36 h. Comparatively, rugose colony formation by NTΔflp was both slow and diminished. Although some surface corrugation was observed, rugosity failed to fully develop after 36 h (total surface area was 2.5-fold lower than NT). This suggested that a functional Tad pilus contributed to the full and timely development of the rugose phenotype.
tad-3 expression mediates initial bacterial surface interactions
Vibrio cholerae has been shown to exhibit two distinct motility behaviors, termed roaming and orbiting, when approaching a surface for initial attachment.14 Roaming is defined by low curvature traces that result from weak contacts between MshA pili and the surface, while orbiting is exemplified by high curvature, circular tracks that retrace over the same area and result from strong pili-surface contacts.27 The majority of V. cholerae cells that eventually attach were shown to originate from the orbiting cell population. To determine if the Tad pilus impacted the near-surface motility behavior of V. vulnificus, the movement of NT and NTΔflp cells was tracked and the traces were analyzed (Fig. 4a, b). The time cells spent orbiting/retracing an area was nearly fourfold higher for NT than NTΔflp (mean of 0.37 compared to 0.10, p < 6 × 10−9). These results suggested that the deletion of flp skewed the near-surface trajectory of V. vulnificus from orbiting towards roaming. We surmised that this negative effect on initial bacteria–surface interactions might also inhibit initial surface attachment. Indeed, NT readily colonized a coverslip surface in a 2 h initial attachment assay, while virtually no attachment was detected for NTΔflp (Fig. 4c, d). Initial surface adherence was recovered following complementation with flp (NTΔflp-C). However, when bacteria were allowed to adhere for 16 h, comparable levels of surface colonization were observed for NT and NTΔflp (Fig. Fig. 4d, inset), indicating that NTΔflp was eventually able to colonize given a prolonged contact time. When adherent NTΔflp cells were re-inoculated into fresh media, they again attached poorly over the initial 2 h contact window. This suggested that the ability of NTΔflp to eventually attach was likely not due to the emergence of a mutant that suppressed the initial attachment defect. Together, these results suggested that the Tad pilus plays an important role during the initial attachment phase when a surface is being probed for colonization but its loss did not prevent eventual surface association in a closed static system.
tad-3 expression promotes resistance of V. vulnificus biofilms to mechanical clearance
To better mimic environmental conditions, biofilm formation by NT and NTΔflp strains expressing gfp was monitored under hydrodynamic flow. The strains were seeded into separate micro-fluidic flow-cell chambers and biofilm development was tracked hourly. NT readily established micro-colonies at moderate flow (3 ml min−1) and progressed to macro-colony formation (Fig. 5a). Macro-colonies expanded in place or following contact with bacterial aggregates carried by the flow (Video S1). The eventual outcome was a well-developed and structured biofilm. Conversely, biofilm formation by NTΔflp was severely hampered. NTΔflp micro-colonies that were initially established during chamber inoculation were displaced under flow (3 ml min−1), preventing progression to a mature biofilm. Despite this, an appreciable NTΔflp biofilm could eventually be established under low flow (0.75 ml min−1, Fig. 5b). However, NTΔflp still required considerably more time than NT to securely attach to the surface and achieve a substantial biomass. The resulting NTΔflp biofilm was also fragile, since increasing the flow to 3 ml min−1 led to its disintegration, whereas the NT biofilm remained intact. However, the flp mutant could be incorporated into maturing biofilms that were resistant to high flow rates (5 ml min−1) when inoculated in mixed culture with NT cells (Figure S2). These results suggested that the Tad pilin stabilized the maturing biofilm. The tad locus was therefore designated iam for its critical role in initial attachment, auto-aggregation and resistance to mechanical clearance of V. vulnificus biofilms.
T4P serve diverse functions in Gram-negative bacteria.3 They are known to mediate biofilm formation, surface and twitching motility,28 natural transformation,29 transduction,30 and host–cell interactions.5 We previously isolated a V. vulnificus descendent (NT) that exhibited an increase in biofilm formation and oyster colonization relative to the parental strain.23 We found that a tad pilus locus, one of three encoded in the genome, was upregulated in NT and demonstrated that its expression promoted the key biofilm-related phenotypes of initial attachment, auto-aggregation, and resistance to mechanical clearance. We have designated this locus iam (previously tad-3) to reflect its impact of V. vulnificus physiology when expressed and to distinguish it from the two other tad loci in the genome, which may affect different aspects of V. vulnificus biology.
Pilin-dependent initial surface attachment is an important stage of biofilm formation. It was recently demonstrated in V. cholerae that flagellar-driven counter rotation of the bacterial cell body periodically brings MSHA pili into contact with a nearby surface.14 Strong pili-surface interactions extended the linger time of a cell in an area, resulting in an orbital motion pattern. Notably, only orbiting cells could irreversibly attach to a surface and form micro-colonies. We observed that an flp pilin mutant (NTΔflp) spent less time orbiting and probing the surface than NT, suggesting that deletion of flp impaired the near-surface motility behavior of V. vulnificus. As a consequence, initial surface attachment by NTΔflp was attenuated relative to NT. This was similar to the motility and biofilm phenotypes observed for a V. cholerae mshA pilin mutant14,27 and supports a role for the Iam pilus in establishing initial bacteria–surface contact in V. vulnificus.
Vibrio vulnificus biofilm formation is multifactorial (Fig. 6). Flagellar motility propels bacteria toward a surface. Capsular polysaccharide (CPS) production is essential for virulence but inhibits biofilm formation.31 The Iam pilus plays a key role during the initial attachment stage by prolonging the bacteria–surface contact time. The MSHA and ChiRP pili may also participate, as mutations in either can negatively impact biofilm formation in certain conditions.14,19,20,32 Irreversible surface attachment is dependent on expression of the c-di-GMP-regulated brp-encoded EPS.25,33 GbpA and CabA34,35,36 are key matrix-associated proteins that contribute to V. vulnificus biofilm development. GbpA mediates attachment to chitinous substrates and CabA is a calcium-dependent extracellular protein required for biofilm maturation, rather than initial attachment.34 Micro-colony and macro-colony formation ensues and the Iam pilus, in conjunction with CabA, functions to maintain the integrity of the developing biofilm. Motile single cells or aggregates may be released from the mature biofilm. Bivalves such as oysters have the ability to size-select particles for ingestion—the retention of particles of <5 µm is poor (<20%) for most bivalves but particles >5 µm are captured from the inhalant with nearly 100% efficiency.37 Planktonic Vibrio species are approximately 1 × 2 μm2. Thus, the incorporation of bacteria into marine aggregates via self-association, attachment to sediment, zooplankton, algae, or detritus can increase their retention within feeding bivalves, leading to the potential accumulation of pathogenic microorganisms.24 Vibrio vulnificus, perhaps best known for colonizing oysters, can reach densities of 105–106 bacteria g−1 in summer months.38,39,40 We have shown that NT, which exhibited an enhanced auto-aggregation phenotype, was more readily captured by suspension-feeding oysters than the planktonic parental strain.23 Oysters are also able to pump and process large volumes of the surrounding water column, up to 3 L h−1 for a single oyster.41 The ability of V. vulnificus biofilms to withstand mechanical clearance by the fluid flow of feeding oysters that would otherwise purge them from a surface is fundamental for successful host colonization in the environment.42 We observed that NT biofilms resisted mechanical clearance, whereas NTΔflp biofilms quickly disintegrated and had difficulty re-establishing under flow. The ramifications of flp deletion, diminished and fragile biofilms that were readily cleared under hydrodynamic flow, and the marked reduction in oyster colonization by a NT∆flp mutant23 underscores the importance of the Iam pilus in niche colonization. The prevalence of tad loci in the Vibrionaceae8,23 suggests that they may play a similar role in other family members.
Materials and methods
Strains and growth conditions
Media were purchased from BD Difco. Antibiotics and additives were purchased from Sigma and used as follows: ampicillin (Ap), 100 µg ml−1; gentamycin (Gm), 10 (E. coli) or 35 µg ml−1 (V. vulnificus); rifampicin (Rf), 100 µg ml−1; isopropyl-β-d-thiogalactoside, 100 µM; l-arabinose (l-ara), 0.1%. An Rf-resistant isolate of V. vulnificus ATCC27562 was used as the parental strain and E. coli S17.1λπ was used for conjugation to V. vulnificus. Vibrio vulnificus NT and NT∆flp were previously described.23
Rugose colony imaging
NT and NT∆flp strains carrying pBAD24T-dcpA25,43 were grown in LB Gm overnight at 30°C with shaking. Cultures were then diluted 1:200 in fresh LB Gm l-ara and grown to mid-log phase. A 2 µl aliquot was spotted onto LB Gm l-ara plates and incubated at 30°C for 36 h and photographed with an Axio-Cam MRc5 (Zeiss) digital camera connected to a M60 stereo microscope (Leica). A fixed radius was used to determine the surface area for colonies with the SP package44 in R.45 Corresponding 3D elevation maps were created with ImageJ.46 A representative image of samples done in triplicate is shown.
NT and NTΔflp strains expressing gfp23 were grown in LBS (LB containing 2% NaCl) medium overnight at 30 °C. The cells were passed over a 50-μm filter and flow cytometry of the filtrates was performed on a FACSCalibur flow cytometer (BD Biosciences) equipped with a blue laser (λex = 488 nm) and a band pass filter measuring green fluorescence (FL1; 530/30 nm). The sample flow rate was set at “low” (12 ± 3 μl min−1) and 50,000 events were recorded within a preset gate defining the viable cell population. Analyses were performed on biological triplicates and the data was analyzed using FlowJo X (Tree Star).
The motility of early-exponential (OD600 of 0.1) V. vulnificus cells was recorded by dark-field microscopy on an Olympus IX83 microscope using a ×20 ELWD objective. A stack of at least 150 frames (100 ms exposure time) was recorded for each sample and the movement of 15 randomly sampled motile cells per stack was traced using the MTrackJ47 plugin in Comstat.48 The number of frames a cell spent orbiting (O) within a 10 × 10 μm2 area was divided by the total number of frames (T) to give O/T. Data from three stacks of three biological replicates were analyzed for each strain. Traces shown are from a single experiment. Violin plots were created with ggplot2.49
Biofilm development in micro-fluidic chambers
Polydimethylsiloxane qglass flow-cell devices containing eight 40 × 5 × 1 mm3 chambers were fabricated as previously described50 (see Supplementary Information for details). The chambers were sterilized by sequential treatment with 50 ml each of 3% H2O2, sterile H2O, and LBS prior to inoculation. Mid-log gfp-expressing NT and NTΔflp cultures (OD600 of 0.1) were seeded into separate flow-cell chambers. For mixed culture experiments, NTΔflp cells expressing td-Tomato were used to distinguish them from gfp-expressing NT population. Initial attachment (no flow) proceeded for 20 min followed by a flow rate of 0.75 ml min−1 (low), 3 ml min−1 (normal), or 5 ml min−1 (high) where indicated. Biofilm images and z-stacks (20 × 1 μm2 slices) were captured with an Olympus IX83 microscope using a UPLSAPO ×40 silicon oil immersion objective (NA 1.25, WD 0.3 mm). Quantitative analysis to determine biomass was performed using cellSense (Olympus) and Comstat.48 Data from three biological replicates were analyzed for each strain. Images presented are from a single representative experiment.
Genomic and statistical analyses
The tad pilus sequences were searched against the Vibrionales (NCBI and Uniprot databases) using BLAST51 (E-value cut-off of <9.4e−152). Conserved genomic regions were detected using the Mauve genome aligner.52 The distance matrix surface plot for a translation alignment of the tad locus from A. actinomycetemcomitans and the three V. vulnificus tad loci was created in Excel (Microsoft). Statistical analyses for all plots were determined by one-way analysis of variance or Student’s t test where indicated.
All data are available via the npj website.
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
We thank Christiane Hassel (FACS analysis), John Frye (fabricating the micro-fluidic master mold), Dr. Cecile Berne (assembling micro-fluidic devices), and Dr. David Kysela (motility tracking) for their help. This research was supported by funds from Indian University Bloomington to D.A.R-M.