In situ proteolysis of the Vibrio cholerae matrix protein RbmA promotes biofilm recruitment

Significance

Adhesion to surfaces or biofilm formation is used by bacteria to build symbiotic relationships, cause disease, and survive outside of eukaryotic hosts. This process is often tightly controlled at the level of transcription. One type of biofilm formed by the diarrheal pathogen Vibrio cholerae depends on an ECM composed of exopolysaccharide and matrix-associated proteins, such as RbmA. RbmA has dual functions in reinforcement of the biofilm structure and recruitment of new cells. Here, we show that the recruitment function of RbmA is activated not at the transcriptional level but by a proteolytic event that depends on association of RbmA with the biofilm. This regulatory mechanism allows for the sequential implementation of the dual functions of RbmA.

Abstract

The estuarine gram-negative rod and human diarrheal pathogen Vibrio cholerae synthesizes a VPS exopolysaccharide-dependent biofilm matrix that allows it to form a 3D structure on surfaces. Proteins associated with the matrix include, RbmA, RbmC, and Bap1. RbmA, a protein whose crystallographic structure suggests two binding surfaces, associates with cells by means of a VPS-dependent mechanism and promotes biofilm cohesiveness and recruitment of cells to the biofilm. Here, we show that RbmA undergoes limited proteolysis within the biofilm. This proteolysis, which is carried out by the hemagglutinin/protease and accessory proteases, yields the 22-kDa C-terminal polypeptide RbmA*. RbmA* remains biofilm-associated. Unlike full-length RbmA, the association of RbmA* with cells is no longer VPS-dependent, likely due to an electropositive surface revealed by proteolysis. We provide evidence that this proteolysis event plays a role in recruitment of VPS⁻ cells to the biofilm surface. Based on our findings, we propose that association of RbmA with the matrix reinforces the biofilm structure and leads to limited proteolysis of RbmA to RbmA*. RbmA*, in turn, promotes recruitment of cells that have not yet initiated VPS synthesis to the biofilm surface. The assignment of two functions to RbmA, separated by a proteolytic event that depends on matrix association, dictates an iterative cycle in which reinforcement of recently added biofilm layers precedes the recruitment of new VPS⁻ cells to the biofilm.

The gram-negative bacterium Vibrio cholerae is both an inhabitant of estuarine and marine environments and a human diarrheal pathogen (1). In the aquatic environment and in the intestines of eukaryotic hosts, V. cholerae is most often found attached to surfaces (2). Surface attachment has been proven essential for intestinal colonization and is generally accepted to be important for survival in the environment (3–6). The genetic underpinning of surface attachment in natural aquatic environments has not been carefully investigated. However, there is evidence from the laboratory that pili and adhesins play a role in attachment to chitinous surfaces (5, 7–9), whereas synthesis of an exopolysaccharide-dependent biofilm matrix is required for colonization of the arthropod intestine (3).

Due to its unique small-molecule inducers, complex regulatory cascade, multicomponent matrix composition, and hypothesized environmental importance, the V. cholerae exopolysaccharide-dependent biofilm has been the subject of intense scrutiny. This biofilm requires matrix synthesis, a process that is regulated by environmental nutrients and secreted metabolites (10–14). Many of these environmental signals alter the cytoplasmic level of cyclic-dimeric-GMP (c-di-GMP), most likely through the action of one of over 60 putative signal transduction proteins (15). The c-di-GMP, in turn, activates synthesis of the biofilm matrix.

The V. cholerae biofilm matrix includes the proteins RbmA, RbmC, and Bap1, as well as the VPS polysaccharide whose synthesis is encoded in two large operons, VpsI and VpsII, containing the genes vpsA–vpsK and vpsL–vpsQ, respectively (16, 17). Transcription of rbmA, rbmC, and bap1 is coregulated with the vps genes, and the encoded proteins are required for the 3D structure of the biofilm (16, 18). In particular, RbmA surrounds cells to augment biofilm accumulation and reinforce the biofilm structure, whereas Bap1 and RbmC adhere the biofilm to a surface (16, 19).

Two recent studies have elucidated the structure of RbmA (20, 21). This protein, which forms a dimer in solution, consists of two type III fibronectin (Fn III) domains joined by a flexible linker. RbmA dimers present two potential binding faces coined the “wide” and “narrow” grooves. The wide groove consists of two lobes of an RbmA monomer (Fig. 1A), whereas a surface formed by the union of two RbmA monomers comprises the narrow groove (Fig. 1B). In crystallographic studies, D-loop and O-loop conformations have been reported for the narrow groove of RbmA (20, 21). However, only the D-loop conformation is thought to be important for RbmA function (20).

Fig. 1.

RbmA is cleaved during biofilm formation. The structure of RbmA, showing the wide groove (A) and the D-loop and O-loop conformations of the narrow groove (B), is illustrated. RbmA monomers are blue and aqua, and the loop is yellow. (C) Images of WT V. cholerae MO10 (WT) or ΔrbmA mutant 24-h biofilms after vortexing with glass beads. (Scale bars: 0.5 cm.) (D) Quantification of biofilm-associated and planktonic cells in 24-h static cultures of WT V. cholerae (WT) and a ΔrbmA mutant (ΔA). *P < 0.05. (E) Western blot analyses of RbmA-FLAG in WT V. cholerae biofilms at various times. A 35-kDa band representing RbmA (R) and a 22-kDa band representing RbmA* (R*) are present. (F) Quantification of biofilm formation by WT V. cholerae over time.

Studies of V. cholerae biofilm development in real time using both conventional and superresolution microscopy have shown that RbmA accumulates only on the surface of cells that are competent to synthesize a VPS-dependent matrix (22). We noted that matrix-associated RbmA undergoes limited proteolysis to the C-terminal peptide RbmA* (16). We hypothesized that this proteolysis event might play a role in biofilm formation. Here, we show that association with VPS-producing cells promotes limited proteolysis of RbmA by the hemagglutinin protease (HAP) and accessory proteases, thus revealing an electropositive surface within the wide groove. We provide evidence that RbmA* gains the ability to bind VPS⁻ cells, thereby recruiting these cells to the biofilm surface. We propose a model in which proteolysis of RbmA increases recruitment of planktonic cells to the biofilm surface. These newly attached cells then continue the process of multilayer biofilm formation by initiating matrix synthesis.

Although a role for proteolysis of adhesins has previously been described in biofilm formation by gram-negative organisms (23, 24), the association of this phenomenon with biofilm accumulation rather than dispersal is unusual. We propose that the dependence of biofilm recruitment on RbmA proteolysis dictates an iterative cycle in which the existing biofilm matrix is reinforced by full-length RbmA before the addition of new layers of matrix-less planktonic cells. This sequence of events ensures the construction of a stable, multilayer biofilm.

Results

RbmA Promotes Both Biofilm Integrity and Recruitment of Planktonic Cells.

The biofilm matrix-associated protein RbmA maintains the integrity of the biofilm in the face of the mechanical stress that is likely to be encountered in aquatic environments (19). To demonstrate this function of RbmA, we imaged the aggregates produced by mechanically disrupting a biofilm formed over 24 h by WT V. cholerae strain MO10 or an isogenic ΔrbmA mutant. Although this treatment yielded large fragments of the WT V. cholerae biofilm, treatment of mutant biofilms lacking RbmA led to biofilm dispersal (Fig. 1C). A second function of RbmA is to augment biofilm accumulation. To demonstrate this function of RbmA, we cultured both WT V. cholerae and a ΔrbmA mutant for 24 h under static conditions and then quantified biofilm-associated and planktonic cells. Deletion of rbmA led to a reduction in biofilm-associated cells accompanied by an increase in cells remaining free in the medium (Fig. 1D). Therefore, RbmA has dual functions in preservation of biofilm integrity and recruitment of planktonic cells to the biofilm.

RbmA Proteolysis Does Not Induce Biofilm Dissolution.

In a previous study, we detected limited proteolysis of RbmA to RbmA*, a C-terminal, biofilm-associated peptide (16). To investigate the progression of proteolysis, we isolated biofilms at various times and detected affinity-tagged RbmA and RbmA* by Western blot analysis. Limited proteolysis of RbmA progressed over time without significant dissolution of the biofilm (Fig. 1 E and F).

HAP, PrtV, and VC0157 Participate in RbmA Proteolysis.

To study the impact of RbmA proteolysis on biofilm formation, our first goal was to inhibit this event by inactivating the responsible protease or proteases. We first added a variety of protease inhibitors to WT V. cholerae biofilms formed over 24 h and assessed cleavage of RbmA in biofilms after 48 h (Fig. S1A). The serine protease inhibitor PMSF, the general divalent cation chelator EDTA, and the Ca²⁺-specific chelator EGTA all diminished RbmA proteolysis. Supplementation with Ca²⁺, Mn²⁺, or Zn²⁺ in the presence of EDTA partially restored proteolysis. Based on these observations, we conclude that both metalloproteases and serine proteases are involved in cleavage of RbmA and that this cleavage is activated by specific divalent cations.

Fig. S1.

RbmA is cleaved by metalloproteases and serine proteases dependent on the ions Mn²⁺, Zn²⁺, and Ca²⁺, and it associates with cells that synthesize VPS. Western analysis of RbmA-FLAG in the 48-h biofilms formed by WT MO10 V. cholerae treated with the noted protease inhibitors or divalent cation supplements for 24 h (A); in 24-h and 48-h biofilms formed by C6706 carrying transposon insertions (Tn) or in frame deletions in known or putative secreted endoproteases of V. cholerae (B); in an MO10 Δhap mutant treated with the noted protease inhibitors for 24 h (C); in 24-h and 48-h biofilms formed by MO10 (M) or C6706 (C) carrying deletions of hap, prtV, VC0157, combinations of two, or all three (ΔT) (D); and associated with cells or cell-free supernatants (Sup) harvested from LB broth cultures incubated for 6 h at 27 °C with shaking (E). Duplicate cell pellets were washed with PBS. The ΔrbmA and ΔrbmAΔvpsL mutants expressing RbmA from a plasmid were tested. AEBSF, 4-(2-aminoethyl) benzenesulfonyl fluoride hydrochloride. R indicates full-length RbmA, while R* indicates RbmA*. PI mix is protease inhibitor cocktail for bacterial cell lysate (P8465; Sigma).

Because RbmA is proteolyzed extracellularly, we targeted predicted or known extracellular endoproteases for mutagenesis and overexpression (Table S1). We first took advantage of an existing library of transposon insertion mutants in the C6706 strain background (25). We verified the transposon insertion sites for all protease mutants except the transposon insertion site reported to be within locus VC0157. Therefore, we constructed a strain carrying an in-frame deletion in VC0157 by allelic exchange. We then engineered a chromosomal C-terminal RbmA affinity tag into the parental strain and each protease-deficient strain, and detected RbmA in 24-h biofilms by Western blot analysis (16). Decreased cleavage of RbmA was noted in the biofilm formed by the hapA mutant (Fig. S1B).

Table S1.

Predicted or known extracellular endoproteases of V. cholerae

Locus	Name	Type
VC0157		Ser protease
VC0820	TagA	Metalloprotease
VC1200	VesB	Ser protease
VC1649	VesC	Ser protease
VC1650	VchC	Metalloprotease
VCA0148	TagA-related	Metalloprotease
VCA0223	PrtV	Ca2+-dependent metalloprotease
VCA0803	VesA	Ser protease
VCA0865	Hap	Ca2+-dependent zinc metalloprotease

We exploited the delayed cleavage of RbmA to determine the effect of ectopic expression of each secreted protease on RbmA cleavage at 24 h. Only overexpression of hapA resulted in early cleavage of RbmA (Fig. 2A). We then constructed strains carrying in-frame deletions in hapA, prtV, and VC0157 and found that much less RbmA proteolysis was present in the biofilms formed by the ΔhapA mutant after 48 h, but not in those biofilms formed by the other protease mutants (Fig. 2B). We conclude that the Ca²⁺-dependent Zn²⁺ metalloprotease HAP plays a dominant role in RbmA proteolysis. This finding is in agreement with the observation that the metal chelator EDTA prevents RbmA cleavage and the cations Zn²⁺ and Ca²⁺ restore it (26).

Fig. 2.

Proteases HAP, PrtV, and VC0157 cleave RbmA in a VPS-dependent manner near amino acid residue 75. Western blot analysis of RbmA-FLAG in 24-h and 48-h biofilms formed by WT V. cholerae overexpressing each of the known or putative secreted endoproteases (A); 48-h WT and protease mutant biofilms (B); ∆hapA mutant biofilms over time (C); 24-h and 48-h biofilms formed by a ∆hapA mutant expressing each of the known or putative secreted proteases of V. cholerae (D); 24-h, 48-h, or 72-h biofilms formed by V. cholerae with deletions of hap, prtV, VC0157, combinations of two, or all three (ΔT) (E); and biofilms formed by a VPS mutant overexpressing rbmA-FLAG cocultured with VPS⁻ or ΔrbmA mutant cells carrying a control vector or a vector allowing overexpression of HAP (F). EDTA (1 mM) was added where indicated. R, full-length RbmA; R*, RbmA*. (G) Sequence of RbmA indicating the secretion signal in green and the portion of RbmA that is cleaved in orange. The Cys residues predicted to participate in a disulfide bond are shown in blue.

The presence of a small amount of RbmA* in ΔhapA mutant biofilms at 48 h indicated that RbmA might be a substrate for additional proteases (Fig. 2B). To assess this possibility, we used Western blot analysis to follow RbmA cleavage over time in a ΔhapA mutant. As shown in Fig. 2C, RbmA* appeared at time points beyond 48 h. Therefore, we conclude that other proteases, in addition to HAP, are able to cleave RbmA.

To identify the accessory RbmA proteases, we first assessed RbmA* abundance in ΔhapA mutant biofilms in the presence and absence of the serine protease inhibitor 4-(2-aminoethyl) benzenesulfonyl fluoride hydrochloride (Fig. S1C). This inhibitor completely abrogated cleavage of RbmA, suggesting the involvement of a serine protease. We then engineered ΔhapA mutant strains carrying the previously constructed plasmids for overexpression of each of the putative secreted proteases and assessed the presence of RbmA* in biofilms after 24 h and 48 h. After 24 h, overexpression of HAP alone yielded RbmA*, whereas at 48 h, overexpression of VC0157, PrtV, and VesA also increased levels of RbmA* (Fig. 2D). To determine which of these proteases was responsible for RbmA cleavage in the setting of native levels of expression, we assessed RbmA cleavage in protease mutant strains. As shown in Fig. 2E, although deletion of prtV partly decreased RbmA cleavage at 48 h and 72 h in a ΔhapA mutant background, only deletion of VC0157 was sufficient to abrogate RbmA* production. A similar result was observed in a C6706ΔhapA mutant (Fig. S1D). We conclude that PrtV and VC0157 are secondary RbmA proteases. Because supplementation with serine protease inhibitors and mutation of VC0157, but not prtV, completely arrested RbmA proteolysis in the ΔhapA mutant background, we hypothesize that the action of PrtV on RbmA proteolysis may be indirect, possibly through activation of VC0157. Furthermore, because these proteases yielded C-terminal cleavage products of similar size and no additional degradation products were observed, we speculate that this region of biofilm-associated RbmA is particularly susceptible to proteolysis.

Cell Association Protects RbmA* from Degradation by HAP.

In LB broth cultures, RbmA associates with surface-attached and planktonic V. cholerae in a VPS-dependent fashion, suggesting that VPS is synthesized not only by biofilm cells but also by planktonic cells (22, 27) (Fig. S1E). Because RbmA* was consistently more abundant in cell pellets than in cell-free supernatants, we hypothesized that VPS might modulate RbmA proteolysis. To test this possibility, we combined a VPS⁻ strain expressing RbmA with a ΔrbmA strain carrying a control vector or a vector encoding HAP and examined RbmA proteolysis after 48 h of static incubation. In this experiment, VPS is supplied by the ΔrbmA mutant, and RbmA is therefore predicted to be cell-associated. We noted limited proteolysis of RbmA to RbmA*, which was accelerated by overexpression of HAP (Fig. 2F). We then cocultured a VPS⁻ strain expressing RbmA from a plasmid with a VPS⁻ strain carrying an empty vector. In this case, full-length RbmA, but not RbmA*, was visible by Western blot analysis. We questioned whether this finding was the result of decreased HAP synthesis. To eliminate this possibility, we again cocultured two VPS⁻ strains; however, in this experiment, one strain carried a plasmid enabling overexpression of HAP. In this case, neither RbmA nor RbmA* was detected by Western blot analysis. To show that the disappearance of RbmA was the result of complete degradation by the metalloprotease HAP, we repeated this experiment with the addition of EDTA. Under these conditions, only full-length RbmA was observed. Our results demonstrate that HAP can access RbmA both in the presence and absence of VPS synthesis. However, when VPS is present, HAP cannot further proteolyze RbmA*. These findings suggest that association with cells that synthesize VPS protects RbmA* from degradation.

RbmA Is Cleaved Near Residue 75 to Yield RbmA*.

To identify the site of RbmA cleavage, we separated biofilm proteins on an SDS/PAGE gel, excised the band corresponding to RbmA*, and submitted this sample for liquid chromatography (LC)/MS. The most proximal RbmA-specific tryptic fragment identified in this analysis aligned with the sequence beginning at Lys residue 75 (Fig. 2G and Fig. S2A). These results indicate that, in addition to the 30-aa signal sequence, 44 amino acids are released by proteolysis to yield RbmA*, a 197-aa peptide with a predicted mass of 22 kDa. Three observations suggest that the cleavage site we identified by LC/MS is at or close to the actual cleavage site. First, the mass of RbmA* isolated from native biofilms, as estimated by Western blot analysis, was ∼22 kDa. Second, SDS/PAGE of a purified polypeptide corresponding to the sequence predicted for RbmA* yielded a 22-kDa band (Fig. S2B). Finally, RbmA is predicted to have a disulfide bond between residues C34 and C69. If cleavage removes this disulfide bond, SDS/PAGE performed under reducing conditions should alter the migration of RbmA but not RbmA*. This was, in fact, what we observed (Fig. S2C).

Fig. S2.

Identification and verification of RbmA cleavage site. (A) Peptides identified by LC/MS after excision of RbmA* from an SDS/PAGE gel. (B) Visualization of purified RbmA and RbmA* by Coumassie Blue staining of an SDS/PAGE gel. (C) Western blot analysis of RbmA-FLAG and the corresponding RbmA* cleavage product under nonreducing and reducing conditions.

Proteolysis of RbmA Augments Biofilm Accumulation.

We then used ectopic expression of HAP and RbmA* to explore the effect of RbmA proteolysis on surface association. Overexpression of HAP had no effect on biofilm integrity even after 2 d, when RbmA was completely proteolyzed to RbmA* (Figs. 2D and 3A). However, this proteolysis did result in a twofold increase in biofilm accumulation after 24 h, which was not observed in the absence of RbmA (Fig. 3B). This result suggested to us that RbmA* might play a role in biofilm accumulation. To determine if RbmA* could promote biofilm accumulation in the absence of full-length RbmA, we rescued a ΔrbmA mutant with RbmA* fused to the C-terminal type l secretion sequence (T1SS) of the V. cholerae RTX toxin to allow for secretion of RbmA* (28) (Fig. S3). As controls, we included plasmids expressing full-length RbmA, the T1SS alone, a Sec secretion-defective form of full-length RbmA fused to the T1SS, and RbmA* with no T1SS. Western blot analysis of biofilm cells expressing these constructs showed adequate expression with proteolytic removal of the T1SS motif following secretion (Fig. S4). RbmA* fused to the T1SS, but not RbmA* alone, rescued the biofilm accumulation defect of a ΔrbmA mutant (Fig. 3C). As a final test, we added similar amounts of purified RbmA or RbmA* to a ΔrbmA mutant. Again, we found that both full-length RbmA and RbmA*, but not heat-denatured RbmA*, rescued biofilm accumulation (Fig. 3D).

Fig. 3.

RbmA* augments biofilm association but does not disrupt biofilm integrity. (A) Images of 1- and 2-d-old biofilms formed by WT V. cholerae carrying either a control plasmid (pCTL) or a plasmid encoding HAP after vortexing with glass beads. (Scale bars: 0.5 cm.) Quantification of 24-h biofilms formed by WT V. cholerae or a ΔrbmA mutant carrying a control vector pBADlacZ (pCTL) or the same vector encoding HAP (B); carrying a vector encoding native RbmA (pA), the T1SS of V. cholerae RTX (pX), RbmA with a C-terminal T1SS instead of an N-terminal secretion signal sequence (psec-AX), RbmA* with a C-terminal T1SS (pA*X), or RbmA* with no means of secretion (pA*) (C); or supplemented with purified RbmA (A), RbmA* (A*), or boiled RbmA* (A*ⁱⁿ) (D). Quantification of biofilm-associated (E) and planktonic (F) cells in 24-h cultures of WT V. cholerae, a ΔrbmA mutant, and ΔT mutant either alone or supplemented with purified RbmA*. (G) Quantification of 2-d-old biofilms formed by WT V. cholerae, a ΔrbmA mutant, and ΔT mutant either alone or supplemented with purified RbmA*. * indicates statistical significance.

Fig. S3.

N terminus (N-term) of RbmA is required for secretion. (A) Western blot analysis of RbmA peptides in cells (C) and supernatants (S) harvested from shaking cultures of ΔrbmA mutant cells carrying the indicated plasmids. Cells were collected by centrifugation at 10,000 × g for 10 min. (B) Diagram of RbmA truncations and fusions constructed for rescue experiments.

Fig. S4.

RbmA and RbmA* fused to the type 1 secretion motif of RTX results in removal of the secretion motif and minimal degradation at high expression levels. Western blot analysis of biofilms formed by ΔrbmA mutant cells carrying plasmids encoding RbmA-FLAG (pRbmA), RbmA-FLAG lacking the Sec secretion signal but C-terminally fused to the type 1 secretion motif of RtxA (psec-RbmA-RTX), and RbmA*-FLAG fused to the C-terminal type l secretion motif of RtxA (pRbmA*-RTX). Bands corresponding to the full-length fusion proteins, as well as peptides with the type 1 secretion motif cleaved, are shown.

To explore the role of native levels of RbmA proteolysis, we measured biofilm accumulation after 24 h and 48 h by WT V. cholerae, a ΔrbmA mutant, and a ΔhapΔVC1057ΔprtV triple-protease (ΔT) mutant either alone or supplemented with RbmA*. Because RbmA proteolysis had not yet begun, the biofilms formed by WT V. cholerae and the ΔT mutant after 24 h were not significantly different. Addition of purified RbmA* to the ΔT mutant culture resulted in a twofold increase in recruitment of planktonic cells to the biofilm (Fig. 3 E and F), similar to the effect produced by HAP overexpression (Fig. 3B). After 48 h, both ΔrbmA and ΔT biofilms showed significantly decreased accumulation compared with WT V. cholerae (Fig. 3G). However, unlike the ΔrbmA mutant biofilm, upon vortexing, the ΔT biofilm maintained its integrity (Fig. S5). Furthermore, the addition of purified RbmA* rescued the accumulation defect of the ΔT mutant. Taken together, these results support an independent role for RbmA* in augmenting biofilm accumulation by WT V. cholerae and suggest that cleavage of RbmA does not affect biofilm integrity.

Fig. S5.

Triple-protease mutant biofilm resists dispersion. Images of biofilms were formed over 48 h and then vortexed in the presence of glass beads. WT V. cholerae (WT); a ΔrbmA mutant; a mutant with the genes encoding the HAP, PrtV, and VC0157 proteases deleted (ΔT); or a mutant with the genes encoding all three proteases and rbmA deleted (ΔrbmAΔT) are shown.

Cleavage of RbmA Enables Binding to VPS⁻ Cells.

N-terminal proteolysis of RbmA removes the binding surface of one Fn III lobe, shifting the electrostatic potential of the wide groove from negative to positive (Fig. 4A). We therefore hypothesized that RbmA cleavage might alter its binding affinity. To test this possibility, we incubated purified full-length RbmA and RbmA* with a VPS⁺ or VPS⁻ ΔrbmA mutant for 1 h and assessed cell association. As shown in Fig. 4B, full-length RbmA bound only to VPS⁺ cells, but not to VPS⁻ cells. In contrast, RbmA* bound to the surface of both VPS⁺ and VPS⁻ cells. We propose that proteolysis of RbmA alters the surface potential of the wide groove, such that it binds to WT cells that have not initiated VPS synthesis. However, we cannot rule out the possibility that the cell envelope of a VPS mutant is altered such that it is not equivalent to the cell envelope of a WT cell.

Fig. 4.

Proteolysis of RbmA to RbmA* promotes biofilm recruitment but not invasion. (A) Wide grooves of RbmA and RbmA*. Electronegative surfaces are red, whereas electropositive surfaces are blue. The circle indicates the electropositive patch exposed in RbmA*. (B) Western blot analysis of affinity-tagged and purified RbmA or RbmA* attached to ΔrbmA (ΔA) or ΔrbmAΔvpsL (ΔAL) mutant cells. (C–E) Projections onto the X–Z or Y–Z plane of preformed WT tomato-labeled V. cholerae biofilms ectopically expressing HAP as indicated after incubation with GFP-labeled VPS⁻ mutant cells alone or with purified RbmA*. (Scale bars: 10 μm.) (F) Quantification of VPS⁻ cells per field for conditions shown in C–E.

RbmA Proteolysis Increases Recruitment of VPS⁻ Cells to the Biofilm Surface.

We reasoned that biofilm-associated RbmA* might play an important role in recruitment of VPS⁻ cells to the biofilm surface. To test this possibility, we labeled WT and VPS⁻ cells with tomato and GFP, respectively, in similar chromosomal locations. We then incubated preformed biofilms containing either WT V. cholerae alone or expressing HAP with VPS⁻ cells and imaged these biofilms by confocal microscopy. On average, ∼10-fold more VPS⁻ cells were distributed along the surface of WT biofilms expressing HAP, but no VPS⁻ cells were observed within the biofilm (Fig. 4 C, D, and F). Because VPS⁻ cells invaded a preformed ΔrbmA mutant biofilm consistent with previous reports (27), such biofilms were not useful in studying the dependence of the recruitment phenotype on RbmA (Fig. S6). Instead, we incubated preformed WT biofilms with VPS⁻ cells supplemented with purified RbmA*. Again, on average, in the presence of RbmA*, ∼16-fold more VPS⁻ cells were recruited to the biofilm surface (Fig. 4 E and F).

Fig. S6.

VPS⁻ cells invade a ΔrbmA mutant biofilm. A z-axis projection of a preformed ΔrbmA mutant biofilm incubated with GFP-labeled ΔvpsL cells (VPS⁻). Cells in the ΔrbmA mutant biofilm were stained with DAPI. (Scale bar: 10 μm.)

To determine whether increased recruitment of VPS⁻ cells to the biofilm surface correlated with increased incorporation into the biofilm, we formed biofilms by coincubating WT V. cholerae and VPS⁻ cells alone and with ectopically expressed HAP or purified RbmA*. After 24 h, entire pellicles were harvested and imaged, and the number of VPS⁻ cells incorporated per volume of biofilm-associated WT cells was quantified. We found that both ectopic expression of HAP and addition of purified RbmA* increased biofilm incorporation of VPS⁻ cells (Fig. S7). Taken together, these data support a model in which generation of RbmA* by the action of HAP increases recruitment of VPS⁻ cells to the surface of the forming biofilm but does not facilitate biofilm invasion.

Fig. S7.

RbmA proteolysis increases incorporation of VPS⁻ cells into forming biofilms. Projections onto the X–Y plane (above) and X–Z or Y–Z plane (below) of 3D reconstructions of biofilms formed by coculture of a GFP-labeled ΔvpsL mutant and tomato-labeled WT V. cholerae (WT) alone (A), supplemented with purified RbmA* (B), and carrying a vector encoding HAP (C). Mutant and WT were combined at a ratio of 1:100. (Scale bars: 10 μm.) (D) Quantification of the number of ΔvpsL cells per volume of WT cells. Two random fields of view were quantified in each of three independently formed pellicles.

Discussion

RbmA is a biofilm matrix-associated structural protein that functions both in biofilm accumulation and in reinforcement of the biofilm matrix. We were intrigued by the observation that RbmA undergoes limited proteolysis to the C-terminal peptide RbmA* without biofilm dissolution or loss of biofilm integrity. Here, we provide evidence that association of RbmA with VPS-producing cells promotes limited proteolysis of RbmA to RbmA*, which is carried out by HAP, PrtV, and VC0157. Unlike full-length RbmA, which binds only to cells that synthesize a VPS-dependent matrix, RbmA* can also bind to VPS⁻ cells. As a result, RbmA* increases recruitment of VPS⁻ cells to the surface of the forming biofilm.

In the laboratory, V. cholerae biofilms are formed in rich media that activate biofilm matrix synthesis in both surface-associated and, to a lesser extent, planktonic cells. However, in many natural environments, matrix synthesis is activated only after association with digestible surfaces, such as chitin, which provide the requisite nutritive signals. In such environments, recruitment of cells that have not initiated matrix synthesis to the biofilm surface may be particularly important for biofilm growth. We propose a model in which association of RbmA with the biofilm matrix after secretion from cells reinforces the matrix. RbmA then undergoes limited proteolysis of RbmA to RbmA*, thus promoting the recruitment of a new layer of planktonic cells. These cells, in turn, initiate matrix synthesis and continue the cycle (Fig. 5).

Fig. 5.

RbmA proteolysis in V. cholerae biofilm formation. (I) Cells secrete VPS and matrix proteins to form a new layer of the biofilm. (II) This new layer is reinforced by binding of RbmA to cells in a VPS-dependent manner. RbmA undergoes limited proteolysis to RbmA*. (III) RbmA* recruits new VPS⁻ cells to the surface of the biofilm. (IV) New cells secrete VPS and matrix proteins, and the cycle begins anew.

In bacterial biofilm formation, many roles for extracytoplasmic proteases have been described. In most cases, these proteases stimulate biofilm dispersion. For example, the V. cholerae chitin-binding protein GbpA is cleaved by HAP and PrtV, resulting in decreased attachment to chitin-coated beads (5, 24). Pseudomonas fluorescens biofilm formation is mediated by the outer membrane protein LapA (29, 30). When activated by the c-di-GMP–binding protein LapD, the periplasmic protease LapG cleaves LapA. Release of LapA from the cell surface results in biofilm dispersion (23, 31–33). In contrast, proteolysis of RbmA augments biofilm accumulation.

In Staphylococcus epidermidis and Staphylococcus aureus, a proteolysis event that augments biofilm formation has been described in detail. The targets of proteolysis are the large homologous proteins SasG of S. aureus and AaP of S. epidermidis (34). These proteins consist of an N-terminal A domain, a series of B domains, and a C-terminal LPXTG motif, which anchors these proteins to the cell wall. The A domain mediates surface adhesion (34, 35). After cleavage from the A domain by a variety of bacterial and host proteins, the B domain promotes intercellular interactions by means of the formation of homodimers (36–38). Although there are similarities between the roles of proteolysis in V. cholerae and staphylococcal biofilm formation, there are also important differences. Although RbmA function requires the VPS polysaccharide, Aap and SasG mediate biofilm formation in the absence of the staphylococcal biofilm polysaccharide (38). Furthermore, unique domains of Aap and SasG perform their respective functions, whereas RbmA consists of two similar domains, one of which is modified to carry out its second function. Finally, the A domains of Aap and SasG mediate surface attachment, a function that is carried out by the homologous proteins RbmC and Bap1 in V. cholerae.

When two functions are confined to one protein as they are in Aap, SasG, and RbmA, regulation of one by a conditional cleavage event allows these functions to be carried out sequentially. For instance, in the case of Aap and SasG, proteolysis enforces an order of operations in which surface attachment occurs before the formation of intercellular interactions. Partial proteolysis of RbmA occurs after association with VPS-producing cells. Thus, matrix association and reinforcement must precede recruitment of new cells. Because larger biofilms are more susceptible to disruption by mechanical stress, proteolytic regulation of recruitment allows recent recruits to reinforce their association with the biofilm matrix before attachment of new cells. This developmental program may yield a biofilm that is better able to withstand the high-shear estuarine environments in which V. cholerae is found.

Materials and Methods

Bacterial Strains, Plasmids, and Media.

Details of the bacterial strains and plasmids used in this study are listed in Tables S2 and S3. Media and strain construction are described in SI Materials and Methods and Table S4.

Table S2.

Strains

Strains	Genotype	Source
Escherichia coli
SM10λpir	thi thr leu tonA lacY supE recA::RP4-2-Tc::Muλpir R6K; Kmr	J. Mekalanos, Harvard Medical School, Boston
V. cholerae
PW357	O139 MO10 lacZ::vpsLp→lacZ; Smr	(14)
PW397	PW357 ΔvpsIΔvpsIl	(3)
PW707	PW357 Δbap1ΔrbmC	(16)
PW1397	C6706 lacZ	J. Mekalanos
PW1616	PW357 rbmA-FLAG	This study
PW1617	PW357 rbmA-FLAG ΔhapA	This study
PW1618	PW357 rbmA-FLAG ΔhapΔprtVΔVC0157	This study
PW1704	PW357 rbmA-FLAG ΔVC0157	This study
PW1705	PW357 rbmA-FLAG ΔprtV	This study
PW1706	PW357 rbmA-FLAG ΔhapΔprtV	This study
PW1707	PW357 rbmA-FLAG ΔhapΔVC0157	This study
PW1708	C6706 lacZ rbmA-FLAG	This study
PW1709	C6706 lacZ rbmA-FLAG ΔhapA	This study
PW1710	C6706 lacZ rbmA-FLAG ΔVC0157	This study
PW1711	C6706 lacZ rbmA-FLAG ΔhapΔprtV	This study
PW1712	C6706 lacZ rbmA-FLAG ΔhapΔVC0157	This study
PW1713	C6706 lacZ rbmA-FLAG ΔhapΔprtVΔVC0157	This study
PW1714	C6706 lacZ ΔrbmA	This study
PW1715	O139 MO10 ΔvpsL-QΔrbmA	This study
PW1716	O139 MO10 ΔvpsL-QΔrbmA-ΔwbfR	This study
PW856	O139 MO10 with GFP in lacZ site	(3)
PW1701	O139 MO10 with tomato in lacZ site	This study

Km^r, kanamycin resistance; Sm^r, streptomycin resistance.

Table S3.

Plasmids

Plasmids	Description	Source
pRbmA	FLAG-tagged full-length RbmA in pFLAG-CTC	(16)
pRbmA*	FLAG-tagged RbmA* in pFLAG-CTC	This study
p[Sec−]RbmA	FLAG-tagged RbmA without the Sec signal in pFLAG-CTC	This study
p[Sec+]RbmA*	FLAG-tagged RbmA* with the Sec signal in pFLAG-CTC	This study
pRbmA*-RTX	RbmA fused to RtxA T1SS in pFLAG-CTC	This study
pRbmA-RTX	RbmA* fused to RtxA T1SS in pFLAG-CTC	This study
pFLAG-CTC	Control vector	Sigma
pBADLacZ	Control vector	Invitrogen
pTagA	VC0820 in pBAD-TOPO	This study
pVCA0148	VCA0148 in pBAD-TOPO	This study
pHAP	VCA0865 in pBAD-TOPO	This study
pPrtV	VCA0223 in pBAD-TOPO	This study
pVC1650	VC1650 in pBAD-TOPO	This study
pVC0157	VC0157 in pBAD-TOPO	This study
pVesA	VCA0803 in pBAD-TOPO	This study
pVesB	VCA1200 in pBAD-TOPO	This study
pVesC	VC1649 in pBAD-TOPO	This study
pWM91	oriR6KmobRP4 lacI pTac tnp miniTn10Km	B. Wanner, Purdue University,  West Lafayette, IN
pWM91::ΔrbmA	pWM91 carrying a fragment of rbmA harboring an unmarked internal deletion	(16)
pWM91::rbmA-FLAG	pWM91 encoding the C terminus of RbmA fused to a FLAG-tag with flanking DNA for recombination onto the chromosome	This study
pWM91::ΔhapA	pWM91 carrying a fragment of hapA harboring an unmarked internal deletion	This study
pWM91::ΔrbmA	pWM91 carrying a fragment of rbmA harboring an unmarked internal deletion	This study
pWM91::ΔprtV	pWM91 carrying a fragment of prtV harboring an unmarked internal deletion	This study
pWM91::ΔVC0157	pWM91 carrying a fragment of VC0157 harboring an unmarked internal deletion	This study
pGP704::rbmA-FLAG	pGP704 carrying the 3′ end of rbmA encoding a FLAG-tag; Apr	(16)

Table S4.

Primers

Primers	Description
Construction of V. cholerae hapA deletion
VCA0865P1	GCAGGTGCAGATGCCAAATCC
VCA0865P2	TAACGAGCGGCCGCACATTTCTCAATCCTAGAGATGTTG
VCA0865P3	TGCGGCCGCTCGTTATAACTTCCTTGCCACCTACCTG
VCA0865P4	GCAGAAAAAATGGTGGGCAATATCG
Construction of V. cholerae prtV deletion
VCA0223P1	GTCATTTATTCGATTGATGATTTG
VCA0223P2	TAACGAGCGGCCGCACATTTTATTTCCTTAATATTTC
VCA0223P3	TGCGGCCGCTCGTTATAATTCTTCCTTCTCCTTCCATG
VCA0223P4	GCGAATTGGCGATCGAAAAGC
Construction of V. cholerae VC0157 deletion
VC0157P1	GATATAGTTTAATTGCGGCAAG
VC0157P2	TAACGAGCGGCCGCACATGATAAGTTTCCTTAAAAAG
VC0157P3	TGCGGCCGCTCGTTATAACATGACGCATTAAACAATAAAAGC
VC0157P4	CCAAAACGTAATAGGTGAGAGC
Construction of V. cholerae rbmA chromosomal fusion to FLAG-tag
RbmA-FlagInsP1	CTTACTGATGGTCGTATGTTG
RbmA-FlagInsP2	CTTGTCGTCATCGTCCTTGTAGTCTTTTTTTACCACTGTCATTGACTG
RbmA-FlagInsP3	GACTACAAGGACGATGACGACAAGTAAATTTACCTAGTCACTTAGTCG
RbmA-FlagInsP4	GCATCAATGACCCAAACAACC
Construction of RbmA* protein expression plasmid
Forward	AGAGGT CCATGGGT AAG TAT TGG CTG AGC ATT AAA GGC
Reverse	GAGAGT CTCGAG CTT CTT CAC CAC GGT CAT TGA TTG TTC C

Mixing Experiments to Determine the Role of VPS in Proteolysis.

Suspensions of the indicated V. cholerae strains were added in equal amounts to LB to yield an OD₆₅₅ of 0.1. After 24 h, EDTA was added to the indicated samples. At 48 h, 150 μL of Laemmli buffer was added to the entire culture, boiled for 10 min, and analyzed by Western blotting.

Biofilm Assays.

Biofilm formation was quantified as previously described (10).

Cell-Binding Assays.

The ΔrbmA or ΔrbmAΔvpsL mutants were incubated with shaking until an OD₆₅₅ of 3.5–4 was reached. Cells were pelleted and resuspended in fresh LB broth. Purified RbmA or RbmA* was added to cell suspensions, and the mixture was incubated for 1 h. Cells were transferred to a new tube, washed three times with PBS, and collected by centrifugation. Cell-associated RbmA or RbmA* was assessed by Western blot analysis.

Biofilm Recruitment Assays.

The indicated strains were incubated statically in wells containing coverslips for 24 h. Coverslips were washed twice with PBS and then incubated for 1 h with a GFP-labeled ΔvpsL cell suspension, as well as purified RbmA* where indicated. Coverslips were washed six times, mounted, and viewed with a Zeiss LSM 700 microscope. For analysis of ΔrbmA biofilms, cells were stained with DAPI (0.5 μg/mL). Volocity software (PerkinElmer) was used to quantify biofilm-attached ΔvpsL cells per field. At least two random fields of view in two biofilms were evaluated for each condition. This experiment was repeated twice with similar results.

Coincubation Assays.

Tomato-labeled V. cholerae carrying either an empty vector or a plasmid encoding HAP was cocultured in wells with a GFP-labeled ΔvpsL mutant. Where indicated, RbmA* was also added to a final concentration of 25 μg/mL. After 24 h, a portion of the pellicle was gently transferred to a slide, covered with a coverslip, and imaged using an LSM 700 microscope. Volocity software was used to quantify the number of ΔvpsL mutant cells per volume of WT cells. The ΔvpsL mutant cells were identified using a minimum size of 1 μm³. Cell clusters were subsequently separated using an object size guide of 1 μm³. Two randomly selected fields of view were evaluated in each of three pellicles. This experiment was repeated twice with similar results.

SI Materials and Methods

Growth Media.

Bacteria were grown in LB broth. When required, streptomycin (Sm; 100 μg/mL) and ampicillin (50 or 100 μg/mL, as noted) were added to the growth medium. Protein expression was induced with isopropyl-d-1-thiogalactopyranoside or 0.02% or 0.05% l-arabinose as indicated. EGTA, EDTA, a protease inhibitor mix mixture (Sigma), 1 mM 4-(2-aminoethyl) benzenesulfonyl fluoride hydrochloride, or 2 mM PMSF was added to biofilms for protease inhibition experiments.

Generation of Deletion Mutants and Chromosomal rbmA with a 3′ FLAG-Tag Sequence.

For protease surveys in the C6706 background, RbmA was tagged by insertion of a suicide plasmid (15). Gene deletions and unmarked chromosomal protein tags were engineered using the splice overlap extension (SOE) method and double homologous recombination (15). The primers used for mutant generation are listed in Table S4.

Construction of Expression Plasmids.

The protease expression plasmids were constructed by cloning genes from V. cholerae MO10 into the pBadTOPO vector using a 5′ primer containing a stop codon followed by a ribosome binding site and a 3′ primer with a STOP codon. SOE was used to construct the rbmA/rtxA fusions, which were then cloned into pFLAG-CTC expression vector (Sigma).

Biofilm Assays.

Three hundred microliters of a cell suspension diluted to an OD₆₅₅ of 0.05 in LB broth was dispensed into three borosilicate tubes, and these tubes were incubated at 27 °C. At the indicated time, the density of planktonic cells was assessed by OD₆₅₅. The remaining biofilm or pellicle was dispersed by vortexing for 1 min with 1-mm-diameter glass beads in PBS [30% (vol/vol); BioSpec Products, Inc.], and the OD₆₅₅ of the resulting cell suspension was measured.

Western Blot Analysis.

For detection of RbmA-FLAG in biofilms, the planktonic cells were removed, 300 μL of Laemmli buffer was added, samples were boiled for 10 min, and 10 μL of this lysate was loaded onto a 4–20% acrylamide gel (Pierce). After separation, proteins were transferred to a PVDF membrane using a semidry transfer method and detected using an anti-FLAG antibody. The membrane was then incubated overnight at 4 °C in a blocking solution consisting of TBS with 0.1% Tween 20 (TBS-T) and 5% (wt/vol) skim milk. Blocking solution was removed and replaced with fresh blocking solution supplemented with an HRP-conjugated anti-FLAG–binding antibody (Sigma) at a dilution of 1:5,000 or 1:10,000. After incubation for 1 h, the membrane was washed three times with TBS-T and detected with ECL Plus Western blotting detection reagent (Pierce).

Purification of RbmA and RbmA*.

These proteins were purified as previously described (20).

Calculation of Surface Electrostatic Potentials and Preparation of Structures.

Structures were prepared using RbmA model 4BE6 in PyMOL (20). The RbmA* model was manually generated by editing the 4BE6 structure and removing all amino acids from the N terminus up to K75. Then, the Adaptive Poisson Boltzman Software (APBS) package was used as for the unedited RbmA structure, whereas surface electrostatic potentials were calculated via the APBS package and PDB2PQR conversion software. Using the PyMOL APBS tools plug-in with standard settings, solvent-accessible surface potentials were then displayed at ±2 kT/e, where k is the Boltzman constant, T is temperature, and e is the electron charge (red for negative, blue for positive).

Cell-Binding Assays.

The ΔrbmA or ΔrbmAΔvpsL mutants were diluted to an OD₆₅₅ of 0.001 and incubated at 27 °C with shaking until an OD₆₅₅ of 3.5–4 was reached (∼18 h). Cells were pelleted and resuspended in fresh LB broth. Purified RbmA or RbmA* was added to cell suspensions in a 10-μg/mL ratio, and the mixture was incubated for 1 h at room temperature. Cells were transferred to a new tube, washed three times with PBS, and collected by centrifugation. Cell-associated RbmA or RbmA* was assessed by Western blot analysis. A smaller band was consistently present after incubation of RbmA* with cells, which likely represents a degradation product.

Biofilm Recruitment Assays.

The indicated strains were incubated statically starting from an OD₆₅₅ of 0.05 in wells containing 18 × 18-mm coverslips for 24 h at 27 °C. Coverslips were removed, washed twice with PBS, and then incubated for 1 h with a GFP-labeled ΔvpsL cell suspension adjusted to an OD₆₅₅ of 0.5 and supplemented with purified RbmA* (40 μg/mL) where indicated. Following incubation, the coverslip was washed six times, mounted, and viewed with a Zeiss LSM 700 microscope. For analysis of ΔrbmA mutant biofilms, cells were stained with DAPI (0.5 μg/mL). Volocity software (PerkinElmer) was used to quantify biofilm-attached ΔvpsL cells in a z-axis projection. At least two random fields of view in two biofilms were evaluated for each condition. This experiment was repeated twice with similar results.

Coincubation Assays.

Tomato-labeled V. cholerae carrying either an empty vector or a plasmid encoding HAP was cocultured in wells with a GFP-labeled ΔvpsL mutant at a ratio of 100:1 in LB broth supplemented with Sm and arabinose. Where indicated, RbmA* was also added to a final concentration of 25 μg/mL. After 24 h, a portion of the pellicle was gently transferred to a slide, covered with a coverslip, and imaged using an LSM 700 microscope. Volocity software was used to quantify the number of ΔvpsL mutant cells per volume of WT cells. The ΔvpsL mutant cells were identified using a minimum size of 1 μm³. Cell clusters were subsequently separated using an object size guide of 1 μm³. Two randomly selected fields of view were evaluated in each of three pellicles. This experiment was repeated twice with similar results.

Statistical Methods.

At least three experimental replicates were performed for each quantitative analysis. Error bars represent the SD, and statistical significance (P < 0.05) was calculated using a Student’s t test.