ABSTRACT
Vibrio cholerae, the causative agent of cholera, is an abundant environmental bacterium that can efficiently colonize the intestinal tract and trigger severe diarrheal illness. Motility, and the production of colonization factors and cholera toxin, are fundamental for the establishment of disease. In the aquatic environment, V. cholerae persists by forming avirulent biofilms on zooplankton, phytoplankton and chitin debris. Here, we describe the formation of artificial, biofilm-like communities, driven by exposure of planktonic bacteria to synthetic polymers. This recruitment is extremely rapid and charge-driven, and leads to the formation of initial “seed clusters” which then recruit additional bacteria to extend in size. Bacteria that become entrapped in these “forced communities” undergo transcriptional changes in motility and virulence genes, and phenotypically mimic features of environmental biofilm communities by forming a matrix that contains polysaccharide and extracellular DNA. As a result of this lifestyle transition, pathogenicity and in vivo host colonization decrease. These findings highlight the potential of synthetic polymers to disarm pathogens by modulating their lifestlye, without creating selective pressure favoring the emergence of antimicrobial resistant strains.
SIGNIFICANCE Vibrio cholerae is an important human pathogen and causes watery diarrhea after consumption of contaminated water. Its reservoir are aquatic environments, where it persists in an avirulent biofilm state. Upon ingestion, it escapes biofilms and expresses virulence factors, leading to colonization and pathogenicity within the human host. Here, we show that capture by charged polymers rapidly immobilizes V. cholerae and artificially forces it into a sessile state. This mimics environmental cues for biofilm formation and leads to repression of virulence factors through loss of motility. This work highlights a novel artificial lifestyle and an efficient way to neutralize virulent V. cholerae and block disease and transmission.
INTRODUCTION
Vibrio cholerae is a Gram-negative bacterium responsible for several million incidences of enteric disease and up to 142,000 deaths every year (1). Many of these cases are attributable to the V. cholerae El Tor biotype, which is the cause of an ongoing global epidemic, the 7th to sweep our planet in recorded history. A natural inhabitant of aquatic environments, El Tor’s success has been attributed to genetic systems for quorum sensing, chitin breakdown and virulence. Within the human host, the bacterium initiates a virulence programme including the induction of colonization factors and toxins. The two major virulence factors expressed by El Tor strains are cholera toxin and the toxin-coregulated pilus (TCP). Cholera toxin is an ADP-ribosyltransferase of the AB5 family, which leads to the profuse watery diarrhea and electrolyte loss characteristic of the disease. Its subunits are encoded by ctxA and ctxB which are organized in an operon(2). TCP, a type IV pilus, is required for formation of bacterial microcolonies in the small intestine and leads to a local enhancement in toxin concentration at the site of infection. The major pilus subunit is encoded by tcpA(3, 4). In aquatic environments, V. cholerae persists by forming biofilms on the surfaces of phytoplankton, zooplankton and chitin debris (5, 6). Biofilms, which are composed of a matrix of exopolysaccharide and extracellular DNA surrounding the bacteria, offer a protective environment against aquatic predators as well as the host environment. Thus, biofilm formation is an important contributing factor to human disease (7, 8). Efficient colonization of the host intestine, however, requires disassembly of biofilms, and a switch to active motility allowing the bacterium to search out and attach to the host epithelium (8, 9). Indeed, escape from the biofilm is a prerequisite for induction of the virulence programme (8) and failure to escape from the biofilm state results in decreased host colonization fitness in vivo (10). Clearly, the ability to switch between motile and sessile lifestyles, along with the carefully controlled induction of virulence factors, is central to the establishment of disease and the emergence of Cholera epidemics (11).
Although efforts to develop a widely effective vaccine against V. cholerae are ongoing, the efficacy of existing vaccines is low. Attenuated V. cholerae, for example, elicits protective immunity in as few as 16% of patients in developing countries (12). In addition, current vaccines offer protection for only two years, a time period shorter than many epidemics. Currently, and in the absence of a vaccination program, the best way to prevent such outbreaks is by water treatment and provision of sanitation infrastructure. As such, low-tech measures to facilitate decontamination of drinking water, such as simple means of filtration, are desirable and can have a dramatic impact on disease incidence (13). Moreover, there is a lack of understanding of how V. cholerae responds to these strategies, and how this pathogen regulates virulence and motility upon immobilization onto a filtration device. To this end, we demonstrate here that linear, cationic polymers can rapidly capture V. cholerae from aqueous environments and, upon sequestration into these artificially precipitated communities, induce an avirulent phenotype in V. cholerae that mimics environmental biofilm formation. Overall, the polymers force V. cholerae into an artificial sessile lifestyle, which inhibits virulence factor production, colonization and dissemination.
RESULTS
Cationic polymers rapidly form three-dimensional clusters upon contact with V. cholerae
Many polycationic polymers have been designed to maximize their antimicrobial effects (14–16) and previously published work demonstrated a trade-off between the charge and hydrophobicity in cationic polymers and their ability to cluster bacteria, and/or affect bacterial viability within clusters (17-19). Based on our previous work with closely related species Vibrio harveyi (17, 18), we decided to investigate the potential of (poly(N-(3-aminopropyl)methacrylamide), pAPMAm – P1 and (poly(N-[3-(dimethylamino)propyl]methacrylamide), pDMAPMAm - P2, to remove V. cholerae from aqueous environments (Figure 1a). These polymers are both cationic under neutral aqueous conditions and were synthesized via free radical polymerization with high purity (Figure S1 and Figure S2). Upon contact with V. cholerae, both polymers rapidly formed clusters in situ in a concentration dependent manner (Figure 1b), and cluster formation reached equilibrium within minutes (Figure 1c). Cluster formation proceeded via initial nucleation of small layers or sheets of bacteria, which increased in size both by lateral interaction with additional bacteria, as well as stacking of bacteria on existing sheets to form clusters over the first 15 minutes, and then remained stable over the duration of the experiment (Figure 1b,d). No significant differences were observed between both polymers, suggesting that clustering was mainly dominated by electrostatic interactions between the positively charged polymers and the negatively charged bacteria. Bacterial clusters were stable for at least 24 hours and had a well-defined three-dimensional structure (Figure 1d and Figure S3). Both polymers induced bacterial clustering with high efficiency, and the endpoints were practically indistinguishable in terms of numbers of particles and cluster size (Figure 1b).
Effect of polymer-induced clustering on bacterial growth and membrane integrity
The effect of cationic polymers on bacterial viability varies significantly with charge, hydrophobicity, polymer concentration and the bacterial species tested (17, 18). Thus, we explored if and how clustering of V. cholerae affects bacterial proliferation and viability under physiological conditions. Bacterial proliferation during co-incubation of GFP expressing V. cholerae with different concentrations of polymers was measured, monitoring GFP fluorescence over 25 hours (Figure 2a). Bacterial proliferation was generally unaffected, except at very high polymer concentrations (0.5 mg/ml of P1, Figure 2a). However, it was unclear from these experiments whether the slower increase in fluorescence was due to inhibition of bacterial growth, as is observed within subsets of cells within bacterial biofilms (11), or due to cellular damage commonly observed with highly charged cationic polymers (19). Flow cytometry of bacterial samples exposed to polymers and LIVE/DEAD® cell viability stains allowed us to determine bacterial viability via measuring membrane integrity of V. cholerae sequestered in clusters at the experimental endpoint (Figure 2b). Viability and membrane integrity were largely unaffected by clustering even following overnight incubation, except at very high polymer concentrations (0.5 mg/ml P1, Figure 2b). We also investigated the effect of both polymers on host cells, in this case cultured Caco-2 intestinal epithelial cells, using lactate dehydrogenase release (LDH) assays to probe cellular membrane integrity (Figure S4). P1 and P2 both compromised membrane integrity of epithelial cells at 5·10−3 and 5·10−4 mg/mL or above, respectively, compared to untreated control cells. Thus, for functional experiments, we focused on investigating the effects of the highest effective, non-toxic concentration of both polymers.
Sequestration of V. cholerae into polymer clusters induces a biofilm-like state and suppresses bacterial virulence at the transcriptional level
The switch to a sessile lifestyle and biofilm formation in V. cholerae is initiated by surface sensing and downregulation of bacterial motility (11, 20). Since we observed during imaging experiments that bacterial clustering abrogated bacterial motility, we investigated whether polymer-induced clustering would also affect in vitro biofilm formation. Polymer-induced clustering lead to a significant induction of biofilm formation in V. cholerae, as measured using crystal violet assays and imaging of GFP-V. cholerae (Figure 3a). Polymer-induced biofilms also released higher levels of extracellular DNA into the biofilm, compared to untreated V. cholerae (Figure 3b-d).
Since sequestration into polymer clusters promotes a sessile state, we investigated what impact this environmental signal would have on virulence regulation. For this purpose, we created a series of transcriptional reporter strains, by introducing, via conjugation, a variant of the pRW50 plasmid containing oriT into V. cholerae (Figure 4a). Thus, we could follow the induction of V. cholerae promoters during growth in medium mimicking inducing conditions within the host environment (DMEM, 37 °C), using β-galactosidase assays. AphA is a master regulator of virulence that is required for the activation of tcpP. TcpP, in turn, activates toxT, which activates downstream virulence genes, including those responsible for the production of cholera toxin and the toxin-coregulated pilus (TCP). In the absence of polymer, the aphA promoter was strongly induced (~9-fold compared to vector control) in mid-log phase cells (Figure 4b). Sequestration of V. cholerae into clusters significantly repressed the aphA promoter activity (approx. 70% and 55% suppression with P1 and P2, respectively). The toxT promoter was also strongly induced (~12-fold over vector control) in the absence of polymer but significantly repressed in bacterial clusters (~30% and 20% inhibition by P1 and P2, respectively). Similarly, the promoters of the two key virulence factors, ctxAB and tcpA, were strongly induced under conditions mimicking the host environment (12-fold and 4-fold induction over vector control, respectively) with both P1 and P2 suppressing ctxAB transcription by approximately 60% and 70%, respectively. The effect on tcpA was less pronounced, with P1 showing only mild suppression, and P2 suppressing transcription by ~50% (Figure 4b). We also investigated the effect of physical immobilization within artificial clusters on transcriptional regulation of genes encoding for components of the flagellar systems (flaE and flaA, respectively). Both of these were comparatively weak promoters (~2 to 2.5-fold induction compared to vector control). Interestingly, both flaE and flaA were downregulated by P1 mediated clustering, while P2, which is more hydrophobic than P1, had no effect on flaA or flaE transcription levels (Figure S5).
Polymer sequestration of bacteria abolishes V. cholerae infection in vitro and intestinal colonization in vivo
Since polymer-induced clustering repressed the induction of key virulence factors, we asked what impact clustering would have on V. cholerae infection. Initially, we tested the effect of clustering on colonization and toxicity towards cultured intestinal epithelial cells. Sequestration of V. cholerae by polymers led to a significant reduction in bacterial attachment to host cells, as determined both by dilution plating/CFU counts and imaging of infected cells (Figure 5a-c). Imaging of Caco-2 cells infected with either planktonic or clustered V. cholerae also revealed a decrease in V. cholerae mediated toxicity as a result of bacterial clustering. Host cells infected with clustered bacteria showed less cell death and more intact cell-cell junctions than cells infected with planktonic bacteria (Figure 5c). This protective effect of the polymers was also observed when cytotoxicity was measured by LDH release assays, following infection with planktonic or clustered V. cholerae (Figure 5b).
Finally, we tested whether non-bactericidal concentrations of polymers would protect against intestinal colonization by V. cholerae in vivo. Previously, zebrafish (Danio rerio) have been established as an aquatic host which can be colonized and infected by V. cholerae in a concentration dependent manner, and infection eventually leads to mortality (21, 22). We tested if bacterial clustering would affect subsequent colonization of zebrafish larvae by V. cholerae following a 6 hour exposure. Zebrafish larvae exposed to either 107 or 108 CFU/mL of planktonic or polymer-clustered GFP-expressing V. cholerae were first imaged and then sacrificed, and intestinal V. cholerae were extracted from the tissue and enumerated by dilution plating on selective TCBS agar (Figure 6a-b). Images of infected fish showed that GFP-expressing V. cholerae had specifically colonized the gastrointestinal tract, with the majority of bacteria attached to the mid-intestine (Figure 6c). The bacterial burden was visibly lower in fish infected with clustered bacteria, compared to fish infected with planktonic bacteria. Bacterial sequestration was more efficient in blocking colonization at an infectious dose of 107 (Figure 6a) where bacterial burden was reduced more than 100-fold. At a dose of 108 V. cholerae, colonization was still significantly inhibited by polymer-induced clustering, albeit the effect was much smaller (Figure 6b).
DISCUSSION
The interaction between bacteria and polycationic molecules plays an important role in nature. Many antimicrobial peptides are polycationic, and their interactions with and impact on bacterial physiology is well characterized (23, 24). Polycationic synthetic polymers have been researched as a cheaper and more stable alternative to mimic the effect of naturally occurring AMPs, and their design and use in previous studies is often targeted at maximizing their antimicrobial effects (25-27). Recent work has however highlighted the potential of such materials to bind bacteria with the aim to manipulate bacterial behaviors, such as quorum sensing, without impacting bacterial viability (17, 18). In the face of increasing problems with the emergence of antibiotic resistant bacterial strains in clinical settings, such strategies, which would potentially provide less selective pressure on the emergence of drug-resistance, become more and more relevant (28). For V. cholerae, alternative approaches targeting virulence, rather than bacterial viability, have been the topic of previous research for some time, underpinning the need for novel ways to prevent and treat infections with this globally important human pathogen (29, 30). Anti-adhesion therapies are often proposed as attractive anti-virulence strategies, that compromise the ability of the pathogen to colonise the host and thus establish an infection (31, 32). Similarly, removal of the pathogen through sequestration is often proposed as a cost-effective approach to decontaminate water sources. However, little is known about V. cholerae response to these artificial environments and how binding to these materials may affect regulation of virulence in these pathogens.
With this in mind, we set out to synthesize a set of two polymers with different cationic groups, but identical backbones, to evaluate their impact on bacterial clustering and behavior in V. cholerae. Based on our previous research with V. harveyi (17, 18), pAPMAm – P1 and pDMAPMAm - P2, were investigated. We anticipated that P1 would present higher toxicity towards both bacteria and host cells (Figure 2 and Figure S4) (14-16). We found that both polymers were able to induce clustering of V. cholerae irrespective of their cationic nature, and the clusters quickly became big enough to precipitate out of solution (Figure 1). Both polymers had little impact on bacterial growth, viability and membrane permeability, in particular at concentrations below 0.05 mg/mL (Figure 2). Overall, P1 has a bigger effect on bacterial growth and viability than P2, (Figure 2), but even P1 showed bactericidal activity only at the highest concentration tested (0.5 mg/ml). Similarly, charge and buffering impacted on eukaryotic membrane integrity, with both P1 and P2 able to disrupt host cell membranes at high concentrations (Figure S4). For both materials, this cytotoxicity would result in a narrow therapeutic window, and thus future applications of polymers inducing bacterial clustering by electrostatic interactions alone would likely lean more towards an ex vivo preventative application, for example as part of a low-tech water decontamination/filtration strategy. Although this is in itself a promising approach, our future efforts will also focus on the synthesis of materials with decreased toxicities that exhibit high affinity towards the bacteria by other means (e.g. incorporation of natural ligands for V. cholerae such as N-acetyl-glucosamine into the polymer). This approach may open up new avenues to extend future applications of such materials towards a prophylactic or therapeutic use in patients.
Infectious V. cholerae are often taken up as small biofilms, from which bacteria escape to colonize the epithelium. Once bound, bacteria initiate microcolony formation, before eventually exiting the host’s GI tract, often following re-organization into biofilms (33, 34), to cause environmental dispersal and onward-transmission. The ability to transition between motile and sessile states is thus key to V. cholerae’s virulence regulation upon entering the human host and initialization of its colonization programme. Active bacterial motility and induction of virulence factors are both crucially required for V. cholerae pathogenesis. In natural environments, the transition of V. cholerae to a sessile lifestyle and inhibition of motility is accomplished both by transcriptional repression of flagellar genes, as well as induction of extracellular polysaccharide production, both of which are mediated by c-di-GMP (35). The cues triggering these motile to sessile transition in V. cholerae are still subject to investigations, but recent work showed that both lowered temperatures as well as type IV pili-mediated surface sensing can feed into c-di-GMP signaling and thus biofilm formation (36, 37). With our polymers bacterial motility is also largely abolished, albeit by physical deposition into polymer-based clusters. Interestingly, while both polymers cause immobilization to a similar extent, P1 but not P2 caused transcriptional repression of flagellar genes (Figure S5). This suggests that physicochemical properties of the adhesive surface, rather than the mechanical process of surface sensing alone, also impact the transition to a sessile lifestyle. Interestingly, despite their different effects on gene regulation in response to immobilization, both polymers lead to an increase in bacterial deposition on an abiotic surface upon bacterial clustering, which was accompanied by an increased release in extracellular DNA (Figure 3b-d). This increase would suggest that these polycationic polymers may act as an alternative cue to promote a transition toward a sessile, community-based lifestyle for V. cholerae. At the same time, clustering of V. cholerae lead to a decrease in virulence factors at the transcriptional level (Figure 4b). Regulation of virulence genes in V. cholerae is a complex process and several pathways converge at this point. Crucially, both high cell density, via quorum sensing, and c-di-GMP dependent signaling can act to repress virulence genes (38, 39). Based on the fact that a “biofilm-like state” is induced by clustering in our system, while quorum sensing leads to HapR-dependent suppression of biofilm formation, we conclude that the transcriptional repression of virulence genes we observe here is triggered by a cue that mimics more closely the transition towards a sessile lifestyle in aquatic environments, rather than high cell density dependent signaling. The net effect of this polymer-induced phenotypical switch towards avirulence is a decrease in colonization and a decrease in cytotoxicity towards cultured cells (Figure 5).
Finally, we evaluated the potential of these dual-action polymers to inhibit infection in an in vivo model. Zebrafish are a suitable natural host model for V. cholerae colonization and transmission (21, 22) as their gastrointestinal development and physiology closely mimics that of mammalian organisms (40). Additionally, ease of propagation and live imaging made them a good choice of host for our in vivo studies. Due to license restrictions on the experimental duration in the zebrafish infection model, we were unable to characterize the effect of V. cholerae on zebrafish survival. However, the observed decrease in initial colonization (Figure 6) supports the notion that clustering would be an effective way to “neutralize” V. cholerae in vivo. Overall, our results show that the tested materials mainly act to modulate bacterial behavior in a way that positively impacts on the outcome of infection (Figure 7).
Conclusions
Here, we have shown that linear polymers that can sequester the human pathogen V. cholerae into clusters, downregulate virulence and mitigate colonization and toxicity in relevant in vitro and in vivo models. Using cationic polymers and a combination of phenotypic and transcriptional assays, we demonstrate that this reduction in virulence is a result of V. cholerae switching to a non-pathogenic environmental-like phenotype upon clustering. Our observations suggest that polymeric materials can underpin the development of novel cost-effective strategies to minimize V. cholerae pathogenicity without promoting antimicrobial resistance. As such we anticipate that these materials can act as a blueprint for the development of novel cost-effective prophylactic or therapeutic polymers, but to this end, a clear understanding of how these materials trigger phenotypic responses in these pathogens is essential. Our efforts to optimize affinity toward V. cholerae while minimizing toxicity towards the host will be reported in due course.
MATERIALS AND METHODS
Polymers used in this study were poly-N-(3-aminopropyl)methacrylamide p(APMAm), P1 and poly-N-[3-(dimethylamino)propyl]methacrylamide p(DMAPMAm), P2. Their synthesis and characterization, as well as their use in biological assays, are described in detail in the Supporting Materials and Methods.
CONTRIBUTIONS
All authors contributed to the experimental set-up and discussed the results. A.M.K, F.F.-T. and K.V. secure funding. N.P.-S., L.M., I.I. and D.N.C. synthesised and characterised the polymers. N.P.-S., L.M., K.V., A.M.K. performed the biological assays. N.P.-S., K.V., A.M.K. and F.F.-T. analysed the data, and N.P.-S., K.V., A.M.K. and F.F.-T. wrote the manuscript, with all other authors contributing to its final version.
Declaration of competing interests
None to declare.
ACKNOWLEDGEMENTS
We thank D. Grainger and his group for their advice on the construction of transcriptional reporter plasmids. We thank members of the Krachler and Fernandez-Trillo labs for critical reading and comments on the manuscript. This work was supported by University of Birmingham Fellowships (to A.M.K. and F.F.-T.), Wellcome Trust grant 177ISSFPP (to A.M.K. and F.F.-T), BBSRC grants BB/M021513/1 (to K.V. and A.M.K.) and BB/L007916/1 (to A.M.K.), BBSRC MIBTP studentships (to L.M.) and a CONICYT fellowship (to N.P.-S.).