Flagellar motility, extracellular proteases and Vibrio cholerae detachment from abiotic and biotic surfaces

Abstract

Vibrio cholerae of serogroups O1 and O139, the causative agent of Asiatic cholera, continues to be a major global health threat. This pathogen utilizes substratum-specific pili to attach to distinct surfaces in the aquatic environment and the human small intestine and detaches when conditions become unfavorable. Both attachment and detachment are critical to bacterial environmental survival, pathogenesis and disease transmission. However, the factors that promote detachment are less understood. In this study, we examine the role of flagellar motility and hemagglutinin/protease (HapA) in vibrio detachment from a non-degradable abiotic surface and from the suckling mouse intestine. Flagellar motility facilitated V. cholerae detachment from abiotic surfaces. HapA had no effect on the stability of biofilms formed on abiotic surfaces despite representing > 50 % of the proteolytic activity present in the extracellular matrix. We developed a balanced lethal plasmid system to increase the bacterial cyclic diguanylate (c-di-GMP) pool late in infection, a condition that represses motility and HapA expression. Increasing the c-di-GMP pool enhanced V. cholerae colonization of the suckling mouse intestine. The c-di-GMP effect was fully abolished in hapA isogenic mutants. These results suggest that motility facilitates detachment in a substratum-independent manner. Instead, HapA appears to function as a substratum-specific detachment factor.

Introduction

Bacterial adhesion to abiotic and biotic surfaces is mediated by hair-like appendages known as pili. Vibrio cholerae of serogroup O1 and O139, the causative agent of the diarrheal disease cholera, expresses multiple pili, which mediate its adherence to different substrata in the aquatic environment and the human small intestine (1). For instance, the mannose-sensitive hemagglutinin (MSHA) participates in V. cholerae attachment to borosilicate (2, 3) and to the exoskeleton of planktonic crustacean (4); the N-acetylglucosamine binding protein GbpA mediates attachment of vibrios to chitin and intestinal mucin (5, 6); the chitin-regulated pilus ChiRP promotes attachment to chitinous surfaces (7) and the toxin-coregulated pilus (TCP) mediates attachment to cultured intestinal cells, the intestinal microvilli and microcolony formation (8–10). This latter type IV pilus is essential for V. cholerae colonization of the suckling mouse and human intestine (11–13).

V. cholerae attachment to chitinous surfaces in aquatic ecosystems can progress to the formation of biofilm communities that can persist for longer periods in the environment (1). In the small intestine, initial attachment of vibrios to the protective mucus layer can be followed by multiple cycles of mucus gel penetration, adherence to the microvilli, microcolony formation and detachment [reviewed in (14, 15)]. Vibrios have also been suggested to form biofilm-like structures during infection, which can be excreted in a hyperinfective stage that promotes cholera fecal-oral transmission (10, 16, 17). Flagellar motility plays an important role in surface attachment wherein bacterial near-surface motility allows vibrios to scan surfaces for regions favoring strong pili-surface interactions (18). Following attachment, the intracellular level of cyclic diguanylic acid (c-di-GMP) regulates the transition between the motile and sessile lifestyles by (i) binding to and inhibiting the activity of FlrA to suppress the transcription of the flagellar biosynthesis regulon (19) and (ii) binding to receptor proteins VpsR and VpsT to activate the biosynthesis of the biofilm exopolysaccharide and protein extracellular matrix (20).

The ability of V. cholerae to detach from a substratum when the environmental conditions are unfavorable is equally critical to its persistence in the environment, pathogenesis and disease transmission. Detachment of vibrios late in infection returns ≈ 10⁹ virulent bacteria per ml of cholera stool to the aquatic environment and facilitates their transmission to secondary hosts. Detachment could result from the cleavage of an adhesin anchoring cells to a substratum and/or degradation of the substratum itself. For example, the V. cholerae Zn-metalloprotease HapA has been shown to degrade mucin (21) and cleave the mucin binding adhesin GbpA (6). As a consequence, V. cholerae hapA mutants remain attached for longer periods to mucin-coated polystyrene plates and cultured mucin-secreting HT29-18N2 goblet cells (22, 23). Detachment could also result from the expression of hydrolytic enzymes that degrade the extracellular matrix that hold vibrios together in a biofilm. For example, V. cholerae mutants lacking the extracellular DNases Dns and Xds make enhanced biofilms in vitro presumably because they fail to detach (24). Finally, derepression of genes encoding the polar flagellum system could provide vibrios with momentum to detach and swim away toward an unspent substratum. Studies conducted in ligated rabbit ileal loops suggested that activation of motility in the stationary phase by the general stress response regulator RpoS contributes to V. cholerae detachment from the intestinal mucosa into the lumen (25). Flagellar motility, however, has also been reported to enhance the capacity of V. cholerae to establish infection (22, 26). This effect appears to be dependent on strain and assay conditions. A recent study showed that motility is required for colonization of the proximal but not the distal small intestine (10).

In this study, we focus on the role of flagellar motility and HapA in V. cholerae detachment from abiotic and biotic surfaces. HapA, the major soluble protease produced by the cholera bacterium, is a member of the M4 neutral peptidase family that displays significant amino acid sequence homology to P. aeruginosa elastase and Bacillus thermoproteolyticus thermolysin (27). This soluble protease is present in the matrix of biofilms formed under laboratory conditions (28). Further, HapA is expressed during infection and hapA mutants exhibit a modest increase in colonization of the suckling mouse intestine (22, 29). Similar to motility, the expression of HapA is negatively regulated by c-di-GMP (30) and activated by RpoS (25, 31, 32). Studies on the role of motility and HapA on V. cholerae detachment have been hindered by the fact that both activities also influence bacterial adherence (3). To circumvent this obstacle, (i) we examined the role of motility and HapA in the stability of biofilms formed under conditions in which these activities are repressed by artificially increasing the c-di-GMP pool and (ii) tested the effect of diminishing V. cholerae motility and HapA late in infection on bacterial detachment from the small intestine. We used the suckling mouse model to examine the effect of increasing the c-di-GMP pool late in infection on bacterial exit from the host. We provide evidence that motility, but not HapA, is required for V. cholerae detachment from biofilms formed on an abiotic surface. We also show that increasing the c-di-GMP pool late in infection enhances intestinal colonization and that this effect is fully abrogated in a hapA mutant.

Materials and methods

Strains and media

Escherichia coli and V. cholerae strains, plasmids and oligonucleotide primers used in this study are described in Table 1. V. cholerae mutants and transformants were all derived from the O1 El Tor biotype strain C7258. E. coli TOP10 (Life Technologies, Carlsbad, CA) and S17-1λpir (33) were used for cloning purposes. V. cholerae strains were grown in Luria-Bertani (LB) medium at 30°C. Swarm agar consisted in LB medium containing 0.3 % agar. For measuring the expression of HapA, vibrios were cultivated until stationary phase in tryptic soy broth (TSB) at 37°C. When necessary, culture media were supplemented with thymidine (200 μg /mL), ampicillin (100 μg/mL), kanamycin (25 μg/mL), polymyxin B (100 units/mL), isopropyl-β-D-thiogalactopyranoside (IPTG; 0.5 mM) or 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside (X-gal; 20 μg/mL).

Table 1.

Strains, plasmids and oligonucleotide primers

Strain	Description	Reference
E. coli
TOP10	F-mcrA (mrr-hsdRMS-mcrB C) Δ80dlacZΔM15ΔlacX74 recA1 araD139 Δ(ara leu)7697galUgalK rpsL(Str) endA1 nupG	Invitrogen
S17-1λpir	F− recA hsdR RP4-2 (Tc∷Mu) (Km∷Tn7) lysogenized with λpir	(33)
V. cholerae
C7258	O1 El Tor, Ogawa	C linical isolate, Perú 1991
C7258ΔAB	C7258 ΔpomA ΔpomB	(35)
C7258ΔABL	C7258 ΔpomA ΔpomB ΔlacZ	This study
C7258H	C7258 hapA∷celA	This study
C7258ΔT	C7258 ΔthyA	This study
C7258ΔTΔC	C7258 ΔthyA ΔcqsA	This study
C7258ΔTL	C7258 ΔthyA ΔlacZ	This study
C7258ΔTLCΔ	C7258 ΔthyA ΔlacZ ΔcqsA	This study
C7258ΔTAB	C7258 ΔthyA ΔpomA ΔpomB	This study
C7258ΔTLAB	C7258 ΔthyA ΔlacZ ΔpomA ΔpomB	This study
C7258HTΔ	C7258 ΔthyA hapA∷celA	This study
C7258HTLΔ	C7258 ΔthyA ΔlacZ hapA∷celA	This study
Plasmids
pTT3	rrnB T1T2 transcription terminator in pUC19	(40)
pCMW75	V. harveyi IPTG-inducible QrgB expression vector	(42)
pCVDΔthyA	Suicide vector with genomic DNA flanking thyA deletion	(34)
pCVDΔlacZ	Suicide vector with genomic DNA flanking lacZ deletion	(31)
pGPH6	Suicide vector with hapA∷celA insertion	(29)
pCVDΔpomAB	Suicide vector with genomic DNA flanking pomAB deletion	(35)
pCVDΔcqsA	Suicide vector with genomic DNA flanking cqsA deletion	(36)
pAJA1	rrnB T1T2-hapR promoter VCA0956 cassette in pBR322	This study
pAJA1thyA-8	Complementing thyA gene cloned in pAJA1	This study
pAJA1thyA-8*	pAJA1thyA-8 containing GGDEF → AADEF mutation in VCA0956	This study
Primer	Sequence (5′–3′)
HapR39	GCCTCTAGATGAATTTGACGAGCAA
HapR493	TAAGGATCCGCGTCCATAGGGGTA
VCA0956-15	GCCGGATCCATGACAACTGAAGATT
VCA0956-1043	GGAGAATTCGGGTACGTATAGCAGA
ThyA703	GCCAGCTTTATGAATTCGCCTTAGA
ThyA2045	GGCGAATTCTTAAAGGTGCTGAGCT
VCA0956mut	CGTGACGGCGTGACAGCTTATCGTTATGCCGCTGAAGAGTTGCATGATTGCTCCG

Construction of plasmids and mutants

The construction of isogenic ΔthyA, ΔlacZ, hapA::celA, ΔpomAB and ΔcqsA mutants of strain C7258 has been described previously (29, 31, 34–36). Double and triple mutants combining the above alleles were constructed by conjugal transfer of a pCVD442-based suicide vectors (37) baring the particular mutation from the permissive host E. coli S17-1λpir to the corresponding V. cholerae receptor strain followed by sucrose selection as described previously (35). To this end, suicide vector pCVDΔthyA (34) was used for introducing thyA deletions; pCVDΔlacZ (31) for lacZ; pGPH6 (29) for hapA::celA; pCVDΔcqsA (36) for cqsA, and pCVDΔpomAB (35) for pomAB. All deletion mutants were confirmed by PCR and DNA sequencing. To study the effect of enhancing the c-di-GMP pool late in infection, we developed a balanced lethal plasmid system consisting of vectors pAJAIthyA-8 and pAJAIthyA-8*. The active vector pAJAIthyA-8 contains a complementing thyA gene and VCA0956 encoding a diguanylate cyclase (DGC) expressed from the hapR promoter, which is activated when vibrios reach high cell density (> 2 × 10⁸ cells/ml) (38). We selected VCA0956 based on its proven high DGC activity that correlates with suppression of motility (39). To construct this vector, we started with plasmid pTT3 containing the rrnBT₁T₂ transcription terminator in pUC19 (40). Then, we amplified a DNA fragment containing hapR transcription and translation initiation signals using primers HapR39 and HapR493 from strain C7258 genomic DNA and cloned the resulting PCR product as an XbaI-BamHI fragment downstream from the transcription terminator in plasmid pTT3. A promoterless VCA0956 gene was amplified using primers VCA0956-15 and VCA0956-1043 and inserted as a BamHI-EcoRI fragment downstream from the hapR promoter. Next, the DNA fragment containing the rrnBT₁T₂-hapR promoter-VCA0956 cassette was cloned in pBR322 to yield the plasmid pAJAI (Table 1). Finally, an additional DNA fragment encoding a complementing thyA gene was amplified from strain C7258 genomic DNA using primers ThyA703 and ThyA2045 and inserted as an EcoR1 fragment in pAJA1 to yield pAJAIthyA-8. PCR was conducted using the Advantage 2 PCR kit (Takara Bio USA, Inc.) following the provider’s instructions and all PCR products were confirmed by DNA sequencing. As a control vector, we constructed plasmid pAJAIthyA-8* encoding an active site mutant of VCA0956. Site-directed mutagenesis was conducted using the QuickChange Lightning Multi Site-Directed Mutagenesis kit (Agilent Technology) and primer VCA0956mut following the provider’s protocol. The GGDEF active site motif of VCA0956 in pAJAIthyA-8 was changed to AAEEF in pAJAIthyA-8*, a modification known to abolish enzyme activity (41).

Manipulation of the c-di-GMP pool

To enhance the c-di-GMP pool in vitro, strains C7258 and its mutants were transformed with plasmids pCMW75 which encodes V. harveyi DGC QrgB expressed from the Tac promoter (42). Induction of QrgB in V. cholerae with IPTG results in a 10-fold increase in intracellular c-di-GMP that suppresses motility and activates matrix exopolysaccharide expression (30, 42).

In vitro adherence and detachment assay

Strains containing plasmid pCMW75 were grown in LB medium overnight and diluted 1:20 in fresh medium containing 0.5 mM IPTG. Bacterial adherence was measured in 96-well polystyrene microtiter plates incubated 8 to 64 h at 30°C using the crystal violet staining method as described previously (43). The crystal violet readings were normalized for growth and attachment was expressed as the optical density (OD) ratio OD₅₇₀/OD₆₀₀. For the differential enumeration of vibrios in mixed biofilms containing strain C7258 (motile) and C7258ΔABL (non-motile), the biofilms were developed in borosilicate glass tubes and disrupted by vortexing in the presence of 1.0 mm glass beads. Dispersal was confirmed by light microscopy and the ratio of motile to non-motile cells was determined by dilution plating in LB agar containing X-gal. Strain C7258ΔT containing plasmid pAJAIthyA-8 or pAJAIthyA-8* were also allowed to for biofilms in 96-well polystyrene microtiter plates previously coated with porcine gastric mucin (1 mg/mL) (type III; Sigma Chemical Co.).

Determination of protease activity

Proteolytic activity in the biofilm matrix was determined using the EnzChek Protease Assay Kit (Molecular Probes) following the provider’s protocol. The biofilms were developed for different time periods in black wall 96-well microtiter plates, the planktonic cells were discarded, and the wells were washed three times with phosphate-buffered saline (pH 7.4) (PBS). Then, the intact biofilm was incubated with the BODIPY FL labeled peptide for 60 min and fluorescence was read using excitation/emission wavelengths 505/589 nm. The fluorescence readout was normalized by the quantity of adherent cells estimated by crystal violet staining and the results expressed as fluorescence units (FI) per OD₅₇₀. To determine protease activity in disrupted biofilms, single colonies were inoculated into 2 mL of LB medium and the biofilms were formed in borosilicate tubes for different time periods at 30°C. Triplicate tubes were used for biofilm disruption and crystal violet staining, respectively. In both cases, planktonic cells were discarded and the biofilms were washed three times with PBS. To estimate the biofilm mass, adherent bacteria were stained with crystal violet (0.1 %), washed with PBS and the stained biomass was resuspended in 1 mL dimethylsulfoxide for OD₅₇₀ measurement. For protease activity, adherent bacteria were vortexed for 1 min in 1 mL PBS containing 1 g of glass beads (1 mm ⌀). Then, 0.1 mL aliquots were transferred to microtiter plate wells containing 0.1 mL of BODIPY FL labeled peptide (10 μg/mL) and the reaction was incubated 60 min. As above, the fluorescence readout was normalized by the quantity of adherent cells estimated by crystal violet staining (OD₅₇₀). The presence of HapA in V. cholerae cell-free supernatants was measured using an azocasein-degradation assay as described previously (44). One azocasein unit was defined as the amount of enzyme producing an increase of 0.01 optical density units at 442 nM per h.

Competitive colonization assays

Animal studies were conducted in accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health and a protocol approved by the Morehouse School of Medicine Institutional Animal Care and Use Committee. Competitive colonization assays using the suckling mouse model were performed as described previously (22). Briefly, a minimum of six 4–5 days old CD-1 mice were starved for 6 h and inoculated orally through a 21-gauge disposable feeding needle with 10⁶ colony forming units consisting of a 1:1 mixture of each challenge strain in 50 μL of PBS. Mice were maintained for 16 h in a cage with bedding at 30°C. The animals were sacrificed by decapitation and the entire small intestine was removed and homogenized in 5 mL of PBS in an Ultraturrax T8 homogenizer. For differential counting, one challenge strain contained a deletion of the lacZ gene that does not affect intestinal colonization. Colonizing vibrios were enumerated by dilution plating in LB agar containing X-gal. Finally, the competitive index (CI) was calculated as the output ratio of the competing strains after intraintestinal growth normalized by the input ratio.

Results

Motility enhances V. cholerae detachment from an abiotic surface

Flagellar motility has been shown to facilitate bacterial attachment (3, 18). However, the role of motility in detachment has remained unexplored. Since non-motile vibrios can still form robust biofilms, we took a novel approach to address the role of motility in detachment. Specifically, we allowed biofilms to develop under conditions in which the c-di-GMP pool is artificially enhanced to inhibit motility. Then, we examined the effect of preventing vibrios from returning to the planktonic lifestyle with the inhibitor Q24DA, which blocks flagellar motility by interacting with PomB, a component of the inner membrane sodium-conducting channel that drives flagellar rotation (35, 45). In Fig. 1, we show that the amount of crystal violet-stained material declines with time in the wild type strain treated with vehicle only. However, in the presence of Q24DA the mass of adherent bacteria continued to increase rather than declining. Addition of Q24DA phenotypically mimicked the effect of deleting pomAB, though the mutant as expected adhered less at the early time point (Fig.1). Q24DA did not significantly affect the stability of the biofilm in the pomAB mutant (Fig. 1). These results suggested that, although motility favors initial adhesion, it also plays a fundamental role in the detachment phase. To further support this conclusion, we developed mixed biofilms consisting of a 1:1 mixture of strain C7258 (motile) and C7258ΔABL (non-motile) that form blue and white colonies in X-gal agar, respectively. Then, we disrupted the biofilm by vortexing in the presence of glass beads and enumerated the number of motile and non-motile cells at different time points by dilution plating. As shown in Fig. 1, motile vibrios predominated in the biofilm at the early time points. On the contrary, non-motile vibrios predominated at the late time points. These findings indicate that while motile vibrios have an attachment advantage, non-motile cells have a detachment disadvantage.

Fig. 1. Role of motility in V. cholerae detachments and biofilm dispersal.

Left and middle panels. Strain C7258 (Wt) and C7258ΔAB (ΔpomAB) containing plasmid pCMW75 were allowed to form biofilms in LB medium containing IPTG for 8 h at 30°C. At this time point, the preformed biofilms were treated with Q24DA (10 μg/mL) in dimethyl sulfoxide (DMSO) or DMSO (vehicle control) and further incubated up to 64 h. Then, biofilm formation was determined as described in materials and methods. Rightmost panel. Mixtures of strain C7258 (Wt) and C7258ΔABL (ΔpomAB) at a 1:1 ratio (based on colony forming units) were allowed to form biofilms for different time periods, the biofilms were disrupted with glass beads and the fraction of motile to non-motile cells determined by dilution plating in LB agar containing X-gal. The trendline indicates the decline in motile cells in the aging biofilm. Each bar represents the average of at least three independent experiments. Error bars indicate the standard deviation (STDEV). Symbols: * p < 0.05, unpaired T test.

HapA does not promote detachment of vibrios from an abiotic surface

We next examined the role of HapA in bacterial detachment from an abiotic surface. HapA is the major soluble protease secreted by V. cholerae through the type II secretion system together with other proteins including cholera toxin (46). Furthermore, HapA has been detected in the biofilm matrix (28) though its abundance relative to other proteases present in this compartment is unknown. Thus, we first examined if the activity of HapA accounts for a significant fraction of the proteolytic activity exhibited by V. cholerae biofilms. As shown in Fig. 2a, comparison of the proteolytic activity in biofilms formed by strain C7258 and C7258H (hapA::celA) showed that HapA activity accounted for a significant fraction (≈ 50 %) of the total proteolytic activity of the biofilm matrix. To rule out the possibility of the lower protease activity detected in the hapA biofilm being due to structural differences affecting diffusion of the fluorescent peptide probe, we compared the protease activity present in wild type and hapA disrupted biofilms. The results shown in Fig 2b, confirm that HapA accounts for a significant fraction of the proteolytic activity present in the biofilm matrix.

Fig. 2. Proteolytic activity of V. cholerae biofilms.

a. Strains C7258 (Wt) and C7258H (hapA) were allowed to form biofilms for different time periods. Planktonic cells were discarded and the total proteolytic activity of the intact biofilm determined as described in materials and methods using a fluorescent probe. The biofilm mass was determined by crystal violet staining and the results expressed as fluorescence units (FI) normalized by the biofilm mass (OD₅₇₀). b. Strains C7258 (Wt) and C7258H (hapA) were allowed to form biofilms in borosilicate tubes. Biofilm mass was estimated by crystal violet staining and the proteolytic activity present in the disrupted biofilms was measured using a fluorescent probe as described in materials and methods. Each bar represents the average of at least three independent experiments. Error bars indicate the STDEV. Symbols: ** p < 0.01, unpaired T test.

HapA has been suggested to play an early role in biofilm growth through limited proteolysis of the matrix protein RbmA (47). To dissect the role of HapA in detachment from its participation in biofilm growth, we allowed biofilms to develop under conditions of high c-di-GMP to repress hapA. Then, we examined if permanent loss of HapA by mutation had any effect on biofilm stability. To accentuate the stability of the biofilms formed by wild type and hapA strains incubated for different times, the OD₅₇₀ measurements were normalized by the quantity of bacteria attached at 8 h (relative attachment). As shown in Fig. 3, we could not demonstrate differences in the decline of crystal violet-stainable material over time between the wild type and hapA strain. We conclude that HapA does not act on the V. cholerae biofilm matrix to promote detachment under the conditions tested. However, addition of the protease inhibitor cocktail cOmplete significantly increased biofilm stability in the wild type and hapA mutant (Fig. 3) suggesting that other proteases or hydrolases inhibited by cOmplete could function as detachment factors.

Fig. 3. Role of proteases in the stability of V. cholerae biofilms.

Strains C7258 (Wt) and C7258H (hapA) transformed with plasmid pCMW75 were allowed to form biofilms in the presence of IPTG to enhance the c-di-GMP pool for different time periods in microtiter plates. The amount of crystal violet stained material was measured at OD₅₇₀. The quantity of bacteria attached at each time point was normalized by the amount present at 8 h (relative attachment). The cOmplete protease inhibitor cocktail was added as recommended by the provider to the Wt and hapA strain at 8 h. Each bar represents the average of at least three independent experiments. Error bars indicate the STDEV. Statistical significance was determined using an unpaired T test (* p < 0.05; ** p < 0.01).

HapA facilitates vibrio detachment from the suckling mouse intestine

HapA exhibits mucinase activity (21) and has been shown to promote V. cholerae detachment from mucin-secreting goblet cells in cell culture (23). Thus, we considered the possibility of HapA acting as a substratum-specific detachment factor in infection. We developed the balanced lethal plasmid system shown in Fig. 4 to enhance the c-di-GMP pool late in infection, a condition that represses the expression of HapA and motility. It has been reported that increasing the c-di-GMP pool early in infection diminishes V. cholerae motility and ToxT-dependent transcription of tcpA encoding the TCP pilus, thereby inhibiting intestinal colonization (48). Thus, we engineered our balanced lethal plasmid system to enhance the c-di-GMP pool after vibrios reaching a high cell density in the small intestine. We hypothesized that, if HapA and/or motility is required for detachment, increasing the c-di-GMP pool late in infection to suppress these activities should enhance intestinal colonization. In plasmid pAJA1thyA-8, the DGC VCA0956 is expressed from hapR transcription and translation initiation signals previously proven to be activated at high cell density due to the accumulation of cholera autoinducer 1 (CAI-1) (40). We confirmed that strain C7258ΔT containing the active plasmid pAJA1thyA-8 made enhanced biofilms on polystyrene and mucin-coated surfaces compared to the transformants harboring the VCA0956 active site mutant (Fig. 5a). In Fig. 5c we show that enhancing the c-di-GMP by expressing VCA0956 suppressed motility in the swarm agar test and inhibited the expression of HapA (Fig. 5b), a phenotype with a tight requirement for the quorum sensing regulator HapR (49). As expected, the above experiment showed that deletion of pomAB does not impact quorum sensing-dependent hapA expression neither deletion of hapA impinges on motility.

Fig. 4. Balanced lethal plasmid system designed to increase the c-di-GMP pool late in infection.

The vector pAJA1thyA-8 consists of (i) a wild type thyA (thymidylate synthase) gene for positive selection in vivo when transformed into a V. cholerae strain harboring a thyA deletion and (ii) a gene encoding the DGC VCA0956 expressed from the hapR promoter, which is activated at high cell density. The rrnBT₁T₂ transcription terminator was inserted upstream the hapR promoter to prevent spurious activation of VCA0956 transcription from vector sequences. The control plasmid pAJA1thyA-8* harbors a mutation in the active site of VCA0956 abolishing enzyme activity. Symbols: bla, β-lactamase gene; hapRp, hapR promoter.

Fig. 5. Validation of balanced lethal plasmid system to increase c-di-GMP pool late in infection.

a. Strain C7258ΔT containing plasmid pAJA1thyA-8* (open bar) or pAJA1thyA-8 (filled bar) were allowed to form biofilms in polystyrene and mucin-coated polystyrene microtiter plates as described in materials and methods. b. Strains C7258ΔT, C7258ΔTAB and C7258HΔT containing plasmid pAJA1thyA-8* (open bars) or pAJA1thyA-8 (filled bars) were grown to stationary phase in TSB medium and production of HapA measured as described in methods. Each bar represents the average of three independent cultures and error bars denote the STDEV (** p < 0.01, unpaired T test). c. The above stationary cultures were stabbed into 3 mL of swarm agar in 6-well tissue culture plates and incubated 16 h at 30°C. The plate shown is representative of three experiments.

Wild type V. cholerae containing plasmid pAJA1thyA-8 effectively colonized the suckling mouse intestine confirming that activation of VCA0956 expression to increase the c-di-GMP pool is delayed until vibrios reach a high cell density in the gut. Increasing the c-di-GMP pool late in infection in wild type V. cholerae augmented colonization 3.6-fold, indicating that under these conditions vibrios tend to remain associated to the suckling mouse intestinal tissue (Fig. 6). No effect was observed in competing strains containing a deletion of cqsA that does not produce the CAI-1 signaling molecule required to activate the hapR-VCA0956 promoter fusion (Fig. 6). This result corroborates that under these experimental conditions, vibrios increase their c-di-GMP pool upon entering quorum sensing mode to activate the plasmid-encoded hapR-VCA0956 fusion. To determine if the effect of enhancing the c-di-GMP pool late in infection was due to inhibition of motility and/or HapA production, we repeated the same competition in pomAB (non-motile) and hapA genetic backgrounds. In Fig. 6 we show that increasing the c-di-GMP pool had a smaller effect on intestinal colonization in the pomAB genetic background. Similarly, the effect of enhancing the c-di-GMP pool was fully abrogated in the hapA genetic background (Fig. 6). To rule out the possibility of the pomAB and hapA mutants not reaching the cell density required to activate the hapR-VCA0956 fusion, we determined the titer of vibrios recovered from the intestinal tissue. V. cholerae strains containing pAJAIthyA-8 were recovered from the small intestine of mice with mean titers of 6.8 ± 0.3 (Wt), 6.6 ± 0.6 (pomAB) and 6.6 ± 0.5 (hapA) log units per g of tissue.

Fig. 6. Competition between strains exhibiting enhanced versus basal c-di-GMP pool.

The following strains were transformed with plasmids pAJA1thyA-8 or pAJA1thyA-8*, respectively: C7258ΔT and C7258ΔTL (Wt); C7258ΔTΔC and C7258ΔTLΔC (ΔcqsA); C7258ΔTAB and C7258ΔTLAB (ΔpomAB), C7258HΔT and C7258HΔTL (hapA). Paired strains containing pAJA1thyA-8 or pAJA1thyA-8* were mixed at 1:1 ratio and inoculated to newborn mice by oral gavage. The competitive index (CI) was calculated by normalizing the output ratio of strains containing pAJA1thyA-8 over pAJA1thyA-8* after intestinal colonization by their input ratio. The lines through each data set denote the median CI. Symbols: * p < 0.01 (unpaired T test).

Discussion

V. cholerae detachment from biofilms and the intestinal mucosa is a critical step in cholera dissemination, which remains poorly understood. In this study, we examined the role of motility and V. cholerae’s major secreted protease HapA in this process. It has been shown that non-motile vibrios use a distinct developmental pathway in biofilm formation that entails formation of aggregates in solution that could settle onto surfaces as microcolonies to further develop into mature biofilms (50, 51). We took advantage of the above behavior and the motility inhibitor Q24DA to dissect the role of motility in surface detachment. To this end, we allowed vibrios to (i) initiate biofilm formation under conditions in which vibrio exopolysaccharide expression is exacerbated and motility inhibited and (ii) examined the effect of blocking their capacity to regain motility on biofilm stability. Following this approach, we provide the first direct evidence that flagellar motility promotes the detachment of vibrios from biofilms formed on abiotic surfaces. In an aging biofilm, activation of motility could result from entry of cells into quorum sensing (high cell density) mode, a condition that lowers the c-di-GMP pool (42). Furthermore, lowering of the c-di-GMP diminishes the assembly of the MSHA pili required for surface attachment (19, 52). We hypothesize that late in the biofilm cycle, low c-di-GMP enhances motility and simultaneously disrupts the flagellum-MSHA cooperation that favors permanent attachment providing cells with momentum to detach and swim toward an unspent substratum.

Biofilm dispersal can also result from disruption of the biofilm matrix by hydrolytic enzymes. HapA is the major protease secreted by V. cholerae into the culture medium. For instance, negligible proteolytic activity can be detected in milk agar plates or using an azocasein assay in the cell-free supernatants of hapA mutants (44). Furthermore, HapA is activated at high cell density (49, 53). However, V. cholerae secretes additional proteolytic enzymes. Thus, we used a more sensitive protease fluorescence-based assay, to show that HapA accounts for a significant fraction of the total proteolytic activity present in the V. cholerae biofilm matrix. Nevertheless, the biofilms formed by a hapA mutant were not significantly more stable than wild type biofilms. A broad-spectrum protease inhibitor cocktail enhanced biofilm stability suggesting the participation of other hydrolases in biofilm dispersal. Other proteases present in the V. cholerae secretome include a collagenase (VchC) (54); amino peptidases (Lap, LapX); serine proteases VesA, VesB, VesC (55) and IvaP (56) and the metalloprotease PrtV (57). In addition, inhibition of Dns and Xds DNases activities by the cOmplete cocktail could contribute to biofilm stability.

We used the suckling mouse cholera model to determine the role of c-di-GMP, HapA and/or motility in detachment in vivo. In this model, vibrios that fail to attach to the intestinal mucosa are cleared resulting in diminished bacterial recovery post infection (58). Compared to ligated ileal loops, this model lends itself to examining the conditions that favor bacterial exit from the host. We designed a balanced lethal plasmid system to enhance the c-di-GMP pool late in infection to inhibit the expression of motility and HapA. Here we demonstrate that enhancing the c-di-GMP pool augmented colonization and that this increase was dependent on the expression of HapA and motility late in infection. We show that at the time mice were sacrificed (16 h), the wild type, pomAB and hapA strains were recovered from the small intestine with similar titers. Therefore, the lack of response of the pomAB and hapA mutants to an increase in c-di-GMP in the competitive colonization assay cannot be explained by a failure of these mutants to reach high cell density in the suckling mouse intestine. However, since non-motile vibrios may require more time to establish infection, we cannot rule out the possibility of the pomAB mutant entering quorum sensing mode latter than the wild type strain resulting in a lower expression of the hapR-VCA0956 fusion.

The above results are in agreement with a previous study using the rabbit ileal loop model proposing that detachment of vibrios from the intestinal epithelium into the lumen was attributable to activation of motility in the stationary phase (25). Our findings also concur with studies showing that HapA promotes detachment of vibrios from cultured intestinal cells (22, 23, 59) and with the observation that hapA mutants exhibit a modest increase in intestinal colonization in the suckling mouse model (22, 29). A hapA detachment phenotype was not observed, however, in the rabbit ileal loop model (25). This discrepancy could be due to the nature of the animal models used. Contrary to the suckling mouse model, a rabbit ileal loop is a closed system in which the normal flow of intestinal content is disrupted, a condition that could alter the physiology of the test animal and the host-pathogen interaction. Possible mechanisms by which HapA could contribute to detachment include the degradation of the protective mucus barrier to which vibrios are known to attach (60) or by cleaving the GbpA adhesin that mediates attachment of vibrios to mucin (6). It is likely that both motility and HapA synergistically promote detachment from the intestinal mucosa since the polar flagellum of vibrios is not an efficient locomotion organelle in a highly viscous media such as the protective mucus gel (61). Furthermore, it has been reported that the vibrio polar flagellum tends to break in the viscous mucus gel (62).

In sum, we compared the role of flagellar motility and HapA in V. cholerae detachment from two extreme surfaces: the non-degradable abiotic polystyrene or borosilicate surfaces and the degradable suckling mouse intestinal mucosa. Flagellar motility but not HapA appeared to facilitate detachment from an abiotic surface. The finding that HapA could facilitate detachment from the suckling mouse intestine but not from a polystyrene surface suggests that HapA could function as a substratum-specific detachment factor. We note that V. cholerae may also hijack host hydrolytic enzymes to detach late in infection. This would explain the finding that inactivation of hapA has a smaller effect on intestinal colonization compared to other colonization factors such as the toxin coregulated pilus (22).

Highlights.

• ■Motility facilitates detachment of Vibrio cholerae O1 from abiotic and biotic surfaces.
• ■Hemagglutinin/protease (HapA) is the major protease present in the biofilm extracellular matrix.
• ■An increase in the intracellular pool of cyclic diguanylic acid late in infection diminishes V. cholerae detachment from the intestinal mucosa.
• ■HapA contributes to V. cholerae detachment from the suckling mouse small intestine.