Bile salts and alkaline pH reciprocally modulate the interaction between the periplasmic domains of Vibrio cholerae ToxR and ToxS

Summary

ToxR is a transmembrane transcription factor that is essential for virulence gene expression and human colonization by Vibrio cholerae. ToxR requires its operon partner ToxS, a periplasmic integral membrane protein, for full activity. These two proteins are thought to interact through their respective periplasmic domains, ToxRp and ToxSp. In addition, ToxR is thought to be responsive to various environmental cues, such as bile salts and alkaline pH, but how these factors influence ToxR is not yet understood. Using NMR and reciprocal pull down assays, we present the first direct evidence that ToxR and ToxS physically interact. Furthermore, using NMR and DSF, we show that the bile salts cholate and chenodeoxycholate interact with purified ToxRp and destabilize it. Surprisingly, bile salt destabilization of ToxRp enhanced the interaction between ToxRp and ToxSp. In contrast, alkaline pH, which is one of the factors that leads to ToxR proteolysis, decreased the interaction between ToxRp and ToxSp. Taken together, these data suggest a model whereby bile salts or other detergents destabilize ToxR, increasing its interaction with ToxS to promote full ToxR activity. Subsequently, as V. cholerae alkalinizes its environment in late stationary phase, the interaction between the two proteins decreases allowing ToxR proteolysis to proceed.

Introduction

Vibrio cholerae, the causative agent of the disease cholera, is an aquatic Gram-negative bacterium that when ingested must adapt to its site of colonization, the human small intestine, in order to cause disease (Banwell et al., 1970; Angelichio et al., 1999; Millet et al., 2014). The ToxRS operon is conserved across the Vibrio family and regulates the expression of genes in response to environmental changes (Skorupski and Taylor, 1997; Welch and Bartlett, 1998; Shee Eun Lee et al., 2000; Wang et al., 2002; Bina et al., 2003). ToxR is a transmembrane transcription factor with an N-terminal DNA binding domain and a C-terminal periplasmic domain (Miller et al., 1987) and ToxS is a transmembrane periplasmic protein that potentiates ToxR activity (Miller et al., 1989). In V. cholerae, ToxR is required for virulence gene expression, bile resistance, and human colonization (Herrington et al., 1988; Hase and Mekalanos, 1998; Provenzano et al., 2000; Krukonis et al., 2000). To induce virulence, ToxRS cooperates with a second homologous pair of transmembrane proteins encoded by the tcpPH operon on the Vibrio Pathogenicity Island (VPI) to activate the expression of the master regulator of virulence ToxT (Hase and Mekalanos, 1998; Krukonis et al., 2000). In contrast to the increased activity of ToxR when the bacteria are actively colonizing the intestine, recent work has shown that under conditions that mimic the late stages of infection, i.e. alkaline pH and stationary growth phase, ToxR becomes inactive due to proteolysis (Almagro-Moreno, Kim, et al., 2015). This regulated intramembrane proteolysis of ToxR appears to be necessary for V. cholerae to enter a dormant state as it returns to the aquatic environment (Almagro-Moreno, Kim, et al., 2015).

While ToxR controls the expression of many genes (Bina et al., 2003; Kazi et al., 2016), one of the most prominent effects of ToxR activity is the reciprocal regulation of the outer membrane porins OmpU and OmpT (Champion et al., 1997; Bina et al., 2003), which occurs in response to certain environmental stimuli (Miller and Mekalanos, 1988; Mey et al., 2012). One condition that influences ToxR activity is nutrient availability. In nutrient abundant environments such as in rich medium and in the host, ToxR directly binds to the promoters of the porin genes activating the expression of ompU (Miller and Mekalanos, 1988; Crawford et al., 1998; LaRocque et al., 2008) and repressing that of ompT (Li et al., 2000). In contrast, in nutrient limiting conditions, such as in minimal medium (Mey et al., 2012) or during the late stationary phase of growth (Li et al., 2002), the expression of ompU is decreased and OmpT becomes the predominantly expressed porin. When grown in minimal media containing the amino acids asparagine, arginine, glutamate, and serine (NRES), the expression and levels of ToxR increase significantly such that OmpU becomes the dominantly expressed porin (Miller and Mekalanos, 1988). The increase in ToxR production under this condition has been shown to occur through the Var/Csr regulatory circuit (Mey et al., 2015).

The other condition that ToxR responds to is the presence of bile (Provenzano et al., 2000). Bile salts are a major component of bile and function as detergents that aid in digestion and are bactericidal (Begley et al., 2005; Dawson, 2012). When V. cholerae is grown with bile or bile salts the expression of OmpU increases dramatically in a ToxR dependent manner (Provenzano et al., 2000). This upregulation of OmpU is essential for resistance to bile acids, organic acids and antimicrobial peptides present in the small intestine of the host (Provenzano et al., 2000; Merrell et al., 2001; Mathur and Waldor, 2004). This is thought to occur as OmpU has a negative charge associated with the pore and is known to be more efficient at excluding bile salts than OmpT (Simonet et al., 2003; Duret and Delcour, 2006). It has been shown that if OmpT is upregulated in place of OmpU, the bacteria have growth and survival defects in the presence of the bile salt deoxycholate (Provenzano and Klose, 2000; Provenzano et al., 2001). Unlike the response of ToxR to nutrient availability or NRES, the presence of bile salts, cholate, and deoxycholate induce the expression of ompU in a manner that does not involve either an increase in the expression of ToxR mRNA or protein (Mey et al., 2012). This finding suggests that the mechanism by which bile upregulates the expression of ompU is different from that caused by amino acids (Mey et al., 2012).

While most work to date has focused on ToxR, it is clear that its activity is substantially impacted by ToxS. ToxS was originally identified as potentiating ToxR activity (Miller et al., 1989). Since then several studies have shown defects in ToxR regulated genes, including virulence factors, in toxS mutants (Pfau and Taylor, 1998; Mey et al., 2012; Fengler et al., 2012; Almagro-Moreno, Root, et al., 2015). ToxS also appears to play a role in ToxR stability as entry of V. cholerae into the dormant state was accelerated by deletion of toxS due to the rapid proteolysis of ToxR compared to wild type under conditions of high pH and stationary growth (Almagro-Moreno, Root, et al., 2015). In other Vibrio family members ΔtoxS mutants have increased bile sensitivity (Vibrio anguillarum, (Wang et al., 2002)) and colonization defects in an infant mouse model (Vibrio parahaemolyticus, (Hubbard et al., 2016)). Additionally, it has been observed that when ToxR is overexpressed it forms what is presumed to be disulfide linked homodimers (Ottemann and Mekalanos, 1996; Fengler et al., 2012). The disulfide linked homodimers are abolished if ToxS is also expressed (Ottemann and Mekalanos, 1996; Fengler et al., 2012). The most parsimonious explanation for the above observations is that the two proteins interact through their respective periplasmic domains as has been previously proposed (DiRita and Mekalanos, 1991). However, the evidence that they physically interact is inconclusive (Ottemann and Mekalanos, 1996; Pfau and Taylor, 1998).

The observation that bile salts, unlike amino acids, increase the expression of ompU without influencing the levels of ToxR itself raises the possibility that their mechanism of action may involve direct interaction with the periplasmic domain of ToxR. Furthermore, if the periplasmic domains of ToxR and ToxS do in fact interact, it is possible that bile salts would have an impact on their interaction. To address these possibilities, we took advantage of the modular nature of the proteins as a first step in understanding the biochemical basis of their interactions with each other and potential modulators. The periplasmic domains of both ToxR (ToxRp) and ToxS (ToxSp) were individually expressed and purified. Using NMR and reciprocal pull down assays, we show here that ToxRp and ToxSp physically interact. Furthermore, using NMR and DSF, we show that purified ToxRp was able to interact with the bile salts cholate and chenodeoxycholate, but was destabilized by the bile salts. While such destabilization might be thought to subsequently lead to a decrease in the interaction between ToxRp and ToxSp, unexpectedly, our results show that the presence of bile salts and other detergents increased the interaction between them. In contrast, alkaline pH, which is one of the factors that leads to ToxR degradation, decreased the interaction between ToxRp and ToxSp. Taken together, the data presented here suggest a model whereby the V. cholerae ToxRS system has evolved to take advantage of the destabilization of ToxR caused by bile salts to enhance its interaction with ToxS and promote full transcriptional activity.

Results

V. cholerae grown in minimal medium with bile salts require ToxR and ToxS for efficient porin switching from OmpT to OmpU

V. cholerae has recently been shown to induce the expression of OmpU without changes in ToxR expression or protein levels when grown in minimal media supplemented with the bile salts cholate and deoxycholate (Mey et al., 2012). We sought to determine if chenodeoxycholate (CDC), which along with cholate are the two bile salt backbones produced in the liver (Dawson, 2012), and crude bile could also regulate porin expression in minimal media. Vibrio cholerae N16961 wild-type, ΔtoxR, and ΔompU mutants were cultured in minimal T-media with or without ox-bile, cholate, or CDC. As shown in Figure 1A, the wild type strain expressed ompU and repressed ompT in the presence of all the compounds tested whereas the ΔtoxR strain expressed ompT in all of the conditions. The ΔompU strain expressed ompT in unsupplemented T-media, but was unable to induce expression of either porin in the presence of the compounds. These results show that crude bile and CDC regulate porin expression through ToxR, most likely in a manner similar to what has previously been shown for cholate and deoxycholate (Mey et al., 2012).

Figure 1.

Vibrio cholerae regulation of porin expression in response to cholate, ox-bile, and CDC is dependent on ToxR and ToxS. Strains were cultured in T-media alone and media supplemented with 0.1% cholate, 0.01% CDC, or 0.3% ox-bile and grown until mid log phase. A. N16961 wild type, ΔtoxR, and ΔompU. B. N16961 wild type, ΔtoxR, and ΔtoxS. C. Western blot of N16961 wild-type, ΔtoxR and ΔtoxS using anti-ToxR antibody.

To test whether ToxS is involved in porin switching in T-media, a ΔtoxS mutant was constructed in the N16961 background. The wild type, ΔtoxR and ΔtoxS strains were cultured as above in the presence of bile, cholate, or CDC. As previously shown, the compounds caused the wild type strain to undergo porin switching, while the ΔtoxR mutant failed to switch porins when exposed to the compounds. Interestingly, despite the presence of bile or the bile salts, the ΔtoxS strain robustly produced OmpT indicating deficient porin switching (Fig 1B). The lack of porin regulation in the ΔtoxS mutant was not due to a lack of ToxR protein as seen by western blot (Fig. 1C). Therefore, ToxS protein is required for efficient ToxR mediated porin switching.

ToxR and ToxS physically interact

The above finding that both ToxR and ToxS are required for the expression of ompU in bile salts strengthens the possibility that the two proteins interact through their periplasmic domains. The periplasmic domains of ToxR and ToxS have been proposed to interact (DiRita and Mekalanos, 1991) and ToxS has been shown to be important for the stability of ToxR (Almagro-Moreno, Root, et al., 2015) but the mechanism by which this occurs is not yet known. To determine whether these two proteins interact, the periplasmic domains of ToxRp (residues T199 to E294) and ToxSp (residues S25 to S173) (Fig. 2A) were cloned into a vector with a N-terminal chitin binding domain-intein tag (CBDI). This way when the CBDI-fusions were bound to chitin beads, the fused periplasmic domains would be oriented in a way that resembles the topology found at the membrane. The domains were successfully purified and shown to run at their expected molecular weights on an SDS gel (Fig. 2B).

Figure 2.

Cloning and purification of the ToxR and ToxS periplasmic domains. A. Schematic of ToxR (top) and ToxS (bottom) showing domain boundaries. ToxR contains a cytoplasmic DNA binding domain, while ToxS has a 5 amino acid cytoplasmic tail. Both proteins contain a single pass transmembrane domain along with a C-terminal periplasmic domain. DNA encoding the periplasmic domains of ToxR (T199-E294) and ToxS (S25-S173) were cloned in frame into pTYB21, which contains a N-terminal chitin binding domain intein tag. The amino acid sequences of the purified constructs are shown with the periplasmic domain sequences in bold, and the N-terminal five amino acids leftover from the intein tag after cleavage in gray. B. A Colloidal Blue stained 16% SDS-PAGE gel of 1 μg of purified ToxRp and ToxSp.

The ability of the periplasmic domains of ToxR and ToxS to interact in solution was examined using heteronuclear single quantum correlation (HSQC) NMR experiments using ¹⁵N labeled ToxRp. As shown in Fig. 3A, such a comparison revealed several broadened peaks, as well as a few shifted peaks. This was the first indication these two proteins were interacting in solution.

Figure 3.

ToxRp and ToxSp interact as determined by NMR and pull down assays. A. Overlay of HSQC experiments of 100 μM ¹⁵N-ToxRp in magenta and 100 μM ¹⁵N-ToxRp with 100 μM ToxSp in black. The circled peaks are examples of peaks that shifted or broadened and unaffected peaks. The inset shows detail of a pair of peaks that were effected. B. SDS-PAGE gel of the ToxRp pull down assay: lane 1; CBDI tag, lane 2; CBDI with 100 μM ToxRp, lane 3; beads with 100 μM ToxRp, lane 4; CBDI-ToxSp fusion, lane 5; CBDI-ToxSp with 100 μM ToxRp, and lanes 6–9; 1 μg, 0.5 μg, 0.25 μg, 0.125 μg of purified ToxRp. About 2–3% of the input ToxRp was pulled down. C. SDS-PAGE gel of the ToxSp pull down assay: lane 1; CBDI tag, lane 2; CBDI with 100 μM ToxSp, lane 3; beads with 100 μM ToxSp, lane 4; CBDI-ToxRp fusion, lane 5; CBDI-ToxRp with 100 μM ToxSp, and lanes 6–9; 1 μg, 0.5 μg, 0.25 μg, 0.125 μg of purified ToxSp. About 6–9% of the input ToxSp was pulled down. Each pull down assay was performed twice.

Confirmation of the interaction observed by NMR was obtained using reciprocal pull down assays using chitin binding domain-intein (CBDI) peiplasmic domain fusions (see M&M). CBDI-ToxSp pulled down ToxRp (Fig. 3B lane 5) and ToxSp was pulled down by CBDI-ToxRp (Fig. 3C lane 5). Densitometric analysis of the gels determined 2–3% of the input ToxRp was pulled down by CBDI-ToxSp, and 6–9% of the input ToxSp bound to CBDI-ToxRp. Neither purified domain bound significantly to the bead or CBDI tag controls. In addition, the CBDI-periplasmic domain fusions and the CBDI tag did not produce bands corresponding to the purified domain used in the individual pull downs (Fig. 3B and C). As further controls, the sialic acid periplasmic binding domain protein (SiaP) (Gangi Setty et al., 2014) and the periplasmic domain of TcpH from V. cholerae were also cloned into the N-terminal chitin biding domain intein fusion vector and used in pull downs with ToxRp. TcpH is known to stabilize TcpP (Beck et al., 2004), analogous to ToxS, and therefore is not expected to bind to ToxR. While SiaP binds to the sialic acid transporter, which helps promote infection (Almagro-Moreno and Boyd, 2009; Thomas, 2016), and therefore should not bind to ToxRp. As shown in Fig. S1, neither the CBDI-TcpH or CBDI-SiaP constructs were able to pull down ToxRp indicating that its interaction with ToxSp is specific. Thus, the results of the pull down assays are consistent with the NMR data showing that the ToxRp and ToxSp domains interact, thus, establishing this assay as a reliable means to test for modulators of this interaction.

Cholate and CDC, but not NRES, interact with and destabilize ToxRp

Since the mechanism by which bile salts influence ToxR activity is unknown, we sought to determine if they could interact with ToxRp. If bile salts directly interact and act as ligands for ToxR to influence porin gene expression, then it would be expected that cholate and CDC would stabilize a lower energy conformation of the periplasmic domain, as well as have a finite number of binding sites. We therefore tested whether cholate, CDC, or NRES could interact with ¹⁵N labeled ToxRp by NMR with HSQC experiments. NMR titrations were performed with increasing concentrations of cholate and CDC. The spectra revealed cholate and CDC interacted with ToxRp while NRES, at the concentration known to increase the amount of ToxR protein, did not (Fig. 4 A, B, C & D). Comparison of the HSQC of NRES treated ToxRp with the HSQC of ToxRp alone almost perfectly overlapped (Fig. 4D), indicating NRES does not interact with ToxRp. Conversely, the spectra of ToxRp with cholate (Fig. 4B) or CDC (Fig. 4C) showed both shifted and unshifted peaks, indicating an interaction event. However, the peaks shifted throughout the titration, indicating the compounds did not saturate the binding sites on the domain (Fig. S2). For both cholate and CDC, the affected peaks are the same, indicating the compounds interacted with ToxRp in a similar manner (compare Fig. 4B and 4C). Interestingly, ToxRp treated with 10 mM cholate remained folded (see Fig. 4B), whereas 5 mM CDC unfolded the protein (Fig. S3). These results show that both CDC and cholate, but not NRES, interact with ToxRp, potentially destabilizing the protein.

Figure 4.

¹⁵N-ToxRp interacts with cholate, and CDC, but not NRES. For each condition 100 μM ¹⁵N-ToxRp was incubated with the indicated compound. Black circles denote examples of peaks that shift, upon addition of cholate or CDC but not NRES, as well as peaks that did not shift regardless of treatment. Detail of selected peaks are shown in the insets. A. The HSQC of ¹⁵N-ToxRp alone in magenta. B. The spectra of ¹⁵N-ToxRp with 10 mM cholate in blue overlaid on the ¹⁵N-ToxRp spectra. C. Spectra of 2.5 mM CDC treated ¹⁵N-ToxRp in green overlaid on the untreated spectra. D. The HSQC’s of ¹⁵N-ToxRp treated with 12.5 mM NRES in cyan overlaid on the spectra of ¹⁵N-ToxRp alone.

Because the NMR data demonstrated an interaction between ToxRp and the bile salts that either is of low affinity, in the case of cholate, or destabilizing, in the case of CDC, we used differential scanning fluorometry (DSF) as a second technique to confirm cholate and CDC binding to ToxRp (Fig. 5). DSF assesses protein stability by monitoring changes in protein fluorescence, due to unfolding, as a function of temperature, which allows the melting temperature (T_m) to be calculated (Niesen et al., 2007). In general, the T_m of proteins increases as the concentration of a specifically binding ligand increases (Cooper et al., 2000). Interestingly, the T_m of ToxRp did not change in the presence of 1 or 2 mM cholate even though the initial fluorescence increased with increasing cholate concentration (Fig. 5). This increase in initial fluorescence as the concentration was raised indicates ToxRp was destabilized by cholate. In the case of CDC, it decreased the T_m of ToxRp at both concentrations and increased the initial fluorescence, clearly indicating that it destabilized ToxRp. The amino acids NRES showed no effect on the melting profile of ToxRp (Fig. 5). These results demonstrate that both cholate and CDC are able to directly interact with ToxR and influence its structure to destabilize the protein.

Figure 5.

DSF shows ToxRp is destabilized by cholate and CDC. A. Selected melting curves from a DSF experiment with cholate, CDC, and NRES. 100 μM of ToxRp or buffer with the compounds at the indicated concentrations were subjected to DSF using 5x Spyro Orange. There were 4 technical replicates for each independent experiment (n = 3). B. Table of melting temperatures of ToxRp treated with 1 mM and 2 mM cholate or CDC, as well as 12.5 mM NRES determined by DSF. The melting temperature of ToxRp for each condition was determined by finding the temperature that corresponded to the maximum of the first derivative as the fluorescence signal was increasing.

The interaction between ToxR and ToxS is increased in the presence of detergents, but not NRES

Because ToxRp was destabilized by cholate and CDC, we wanted to assess how this destabilization would influence the interaction between ToxRp and ToxSp. Therefore, the effect of increasing CDC concentration on the interaction between ToxRp and CBDI-ToxSp was examined. Unexpectedly, as the concentration of CDC was raised, the amount of ToxRp pulled down by CBDI-ToxSp increased (Fig. 6), suggesting that destabilization of ToxRp by other bile salts could also increase the interaction with ToxSp.

Figure 6.

The interaction between ToxRp and CBDI-ToxSp increases with increasing concentrations of CDC. A fresh supernatant of CBDI-ToxSp was incubated with magnetic chitin beads. At the same time 100 μM of purified ToxRp was incubated with 0.003–10 mM CDC (final concentrations). After washing the bound CBDI-ToxSp, ToxRp with the various concentrations of CDC were incubated with the beads. The resulting samples were run on gels that were stained with Colloidal Blue for densitometric analysis. The amount of ToxRp pulled down was normalized from 0–1 for each experiment (n = 3). The average ± SE was graphed on the vertical axis versus mM of CDC on the horizontal axis.

To test this hypothesis, and to assess the effects of non-bile detergents, pull down assays were performed in the presence of CDC, cholate, ox-bile, Triton X-100 (X-100), and NRES at the indicated concentrations. Dilution series of the untreated and treated samples were run on the same gel to determine the relative change in the amount of ToxRp pulled down due to the compound tested (Fig. 7A). The fold change in the amount of ToxRp pulled down in the presence of the treatment versus buffer was determined and is reported as the average ± SE (n=4). CDC, cholate, bile, and X-100 increased the amount of ToxRp pulled down by CBDI-ToxSp relative to buffer alone (10.7 ± 1.1; 7.9 ± 1.3; 5.0 ± 1.1; and 4.9 ± 0.6 respectively). All of these changes were statistically significant (p-values < 0.05) from a fold change of 1, which would be the expected fold change if the treatment had no effect. NRES produced a 1.3 ± 0.5 fold change, which was not significantly different from a fold change of 1 (Fig. 7B). In addition, CDC did not increase the ability of ToxRp to bind to either CBDI-SiaP, or the CBDI-TcpH periplasmic domain (Fig. S1A and B). These data suggest that destabilization of ToxRp by detergents increases the interaction between ToxRp and ToxSp.

Figure 7.

Ox-bile, CDC, cholate, and Triton X-100 increase the interaction between ToxRp and CBDI-ToxSp as assessed using the pull down assay. For each pull down, an untreated ToxRp sample was performed in parallel with ToxRp treated with the indicated compound along with appropriate controls (Fig. S3). The resulting treated and untreated samples were serially diluted in half (designated as relative dilution 1x-1/8x), run together on a gel, and stained with Colloidal Blue. A. Excerpts of the ToxRp bands from Colloidal Blue stained gels from representative experiments. The fold change in the amount of ToxRp pulled down in the treated versus the untreated sample was determined by densitometric analysis. B. Bar graph showing the average fold change ± SE in the amount of ToxRp pulled down in response to the various treatments (n = 4). The horizontal line marks a fold change of 1, which would be expected if the treatment did not affect the interaction between ToxRp and CBDI-ToxSp. The * marks treatments that were statistically different from a fold change of 1 with p-values < 0.05. The fold changes are as follows: CDC, 10.7 ± 1.1; cholate, 7.9 ± 1.3; ox-bile, 5.0 ± 1.1; X-100, 4.9 ± 0.6; NRES, 1.3 ± 0.5. Only the NRES treated sample fold change was not significantly different from a fold change of 1.

Influence of pH on the interaction between ToxR and ToxS

ToxR proteolysis is required for V. cholerae to enter a dormant state after cells have entered stationary phase and alkalinized their environment (Almagro-Moreno, Kim, et al., 2015). The proteolysis of ToxR was accelerated in a ΔtoxS mutant, leading to the conclusion that ToxS protects ToxR from degradation (Almagro-Moreno, Root, et al., 2015). Because ToxRp and ToxSp interact, it is possible that, at high pH, the interaction between ToxRp and ToxSp is reduced, allowing ToxR to be proteolyzed. To test the pH dependence of the ToxRp-ToxSp interaction, pull downs were performed at pH 5.5, 7.4, and 9.0. These pH’s were chosen because acidic pH’s have been shown to induce virulence expression (Miller and Mekalanos, 1988), most of the small intestine, the site of V. cholerae colonization, is around pH 7.4 (Evans et al., 1988), and V. cholerae has been shown to raise the pH of culture media to around 9.0 (Almagro-Moreno, Kim, et al., 2015). To test the pH dependence of the interaction, purified ToxRp was diluted ~4 fold (100 μM final concentration) into buffers at pH 5.5, 7.4 and 9.0. Lysates of CBDI-ToxSp and CBDI were prepared as above and the proteins were bound to the beads. The beads were washed with the appropriate buffer, and incubated with ToxRp at the appropriate pH. The bound ToxRp was run on a gel to determine the amount of ToxRp pulled down relative to pH 7.4. Both pH 5.5 and pH 9.0 resulted in less ToxRp pulled down 28% and 20% respectively (n = 4) of that pulled down at pH 7.4 (Fig. 8A). In addition, DSF was performed in the same buffers to determine if there was an effect on the T_m of the domain. There was no change in the melting temperature between pH 7.4 and 9.0. However, at pH 5.5 the T_m becomes much lower, indicating the domain is unstable (Fig. 8B). This might explain the slight increase of ToxRp pulled down at pH 5.5 versus pH 9.0. While these results do not shed light on how ToxR is active at slightly acidic pH’s, they do show the optimal pH for the periplasmic domain interaction is around 7.4, which is the pH of most of the human small intestine (Evans et al., 1988). Furthermore, this suggests that as V. cholerae alkalizes its environment the interaction between ToxR and ToxS decreases, and in conjunction with stationary phase growth leads to ToxR proteolysis.

Figure 8.

Effect of pH on the interaction between ToxRp and CBDI-ToxSp, as well as ToxRp stability. A. Purified ToxRp was diluted into buffers at pH 5.5, 7.4, and pH 9.0. Supernatants of CBDI-ToxSp and the CBDI tag were prepared as described and incubated with the chitin beads. The beads were washed with buffer at the appropriate pH, then ToxRp at the indicated pH was added to the beads. The resulting samples were run on gels and analyzed as described above. Gel slice of the ToxRp lanes of the dilution series at pH 5.5, 7.4, and pH 9.0 from a representative experiment. The percent of ToxRp pulled down relative to pH 7.4 is below the appropriate lanes and represents the average ± SE (n = 4). B. Melting temperatures of ToxRp as determined by DSF when diluted into the same buffers as used for the above pull downs. The melting temperatures are reported as average ± SE of n = 3 experiments with four technical replicates each.

Discussion

V. cholerae is a non-obligate pathogen that cycles between the aquatic environment, where it is either free living or attached to the chitinous surfaces of marine organisms, and the human host where it colonizes the intestinal epithelium in a ToxR dependent manner (Banwell et al., 1970; Herrington et al., 1988; Angelichio et al., 1999; Millet et al., 2014). ToxR controls the expression levels of many proteins, most prominently the outer membrane porins OmpU and OmpT. OmpT is the predominantly expressed porin when ToxR activity is low, which occurs in low nutrient conditions such as in the aquatic environment. In the host, ToxR activity is stimulated by several factors, such as nutrients and bile salts, causing OmpU to become the predominantly expressed porin. Recent experiments in minimal media have shown that ompU expression, while dependent on ToxR, is regulated differently in response to nutrients (as simulated by the amino acids NRES) than bile salts (Mey et al., 2012). Whereas NRES stimulated an increase in ToxR and ToxS expression through the Var/Csr circuit (Mey et al., 2015), when the media was supplemented with bile salts the ToxR expression level was unchanged (Mey et al., 2012). These results suggest that bile salts influence porin gene regulation by a different mechanism than amino acids. We hypothesized that one such mechanism might involve a direct effect on ToxR, influencing its ability to interact with ToxS.

Since the published evidence of an interaction between the periplasmic domains of ToxR and ToxS is circumstantial (DiRita and Mekalanos, 1991; Ottemann and Mekalanos, 1996; Pfau and Taylor, 1998), we initially sought to determine whether the isolated domains of these proteins interact. Both NMR and reciprocal pull down experiments indicated that the periplasmic domains bound to each other in solution, but not to the CBDI tag, CBDI-TcpHp, or CBDI-SiaP, representing the first direct physical evidence that ToxR and ToxS interact. These results suggest that interaction between ToxRp and ToxSp is required for full activation of ToxR (Miller et al., 1989), as well to protect ToxR from premature regulated intramembrane proteolysis (Almagro-Moreno, Root, et al., 2015).

The simplest explanation for how the bile salts affect ToxR activity is that they bind to the ToxR periplasmic domain. Both NMR and DSF experiments showed that cholate and CDC, but not NRES, directly interact with ToxRp. However, the bile salts still bound to the domain even at the highest concentrations tested suggesting non-specific binding. The DSF showed the domain was destabilized in the presence of bile salts. Therefore, the salts prefer to bind to the unfolded state of the protein indicating the detergent properties of the bile salts predominate when incubated with ToxRp. The unfolding of ToxRp by the bile salts, at first, would be thought to lead to a loss of activity and an increase in ompT expression. However, the finding that OmpU was produced in the presence of these compounds suggests the activity of ToxR unexpectedly increases when destabilized by bile salts.

The finding that ToxRp was destabilized by the bile salts led us to consider what effect the bile salts, as well as NRES or pH would have on the ToxRp-ToxSp interaction. Addition of NRES to the pull down assay did not change the interaction between ToxRp and ToxSp. This is consistent with the interaction studies showing ToxRp does not interact with NRES, as well as the fact that NRES effects porin regulation through the Var/Csr regulatory circuit where it induces production of ToxR and ToxS (Mey et al., 2012). Performing the pull down assay at alkaline pH modestly decreased the interaction between ToxRp and ToxSp, suggesting intramembrane proteolysis of ToxR is modulated by decreasing the interaction between the two proteins, thereby allowing proteases access to ToxRp.

Unexpectedly, bile salts, which are shown here to destabilize ToxRp, increased the interaction between ToxRp and ToxSp. Although the interaction of the bile salts with ToxRp is fairly weak, and the interaction of ToxSp with ToxRp is also weak, the addition of bile appears to significantly increase their interaction strength. This increased interaction appears to occur through the detergent like properties of the bile salts, as the ToxRp-ToxSp interaction also increased in the presence of the detergent X-100. The idea that detergents destabilize ToxRp and increase the ToxRp-ToxSp interaction is supported by observations that CDC destabilized ToxRp and increased the ToxRp-ToxSp interaction more than cholate. Interestingly, the increase in the interaction positively correlates with the increase in ToxR activity as assayed by porin expression. These findings suggest that the presence of bile salts in the intestinal environment destabilizes ToxRp, leading to an increase in ToxS binding and activation of ToxR transcriptional activity at the ompU promoter. This hypothesis is supported by the lack of ompU expression in the ΔtoxS mutant.

It has previously been shown that bile salts cause widespread protein unfolding in E. coli (Cremers et al., 2014). We propose that in V. cholerae ToxR is directly sensing the stress induced by bile salts by becoming partially unfolded so as to be able to more efficiently interact with ToxS leading to full transcriptional activation. Given the fact that bile salts induce a variety of different cellular stresses (Begley et al., 2005), including disulfide and protein unfolding (Cremers et al., 2014), it makes sense that bacteria have evolved proteins to take advantage of the different stresses. For example the chaperone Hsp33 becomes activated when exposed to disulfide stress (Graf et al., 2004), which also includes the disulfide stress induced by the bile salts (Cremers et al., 2014). Interestingly, the bile salts deoxycholate, chenodeoxycholate, as well as decanoate (Rosenberg et al., 2003) 2,2-, and 4,4-dipyridle (Rosner et al., 2002) activate the E. coli transcription factor Rob by binding to the regulatory domain causing an unspecified conformational change (Rosenberg et al., 2003). In V. cholerae TcpP forms homodimeric disulfide bonds, increasing the transcription of ToxT, when exposed to disulfide stress (Morgan et al., 2016). The activation of ToxR by bile salts appears to be not simply a transcription factor undergoing a conformational change to become active, but rather an evolutionary adaptation in which the detrimental effects of its signal (i.e. destabilization by detergent like molecules) has been compensated for by the chaperone like function of ToxS, which facilitates ToxR activation.

It is worth considering how bile salts fit into the picture of environmental signals that ToxR has evolved to respond to. Several members of the Vibrio genus colonize chitinous surfaces belonging to a variety of aquatic organisms (Johnson, 2013). Therefore, the Vibrios are under constant threat of being ingested by aquatic or terrestrial animals that produce bile. Thus, bile resistance is a trait that would provide a fitness advantage. This is also true with regards to virulence. Upon the acquisition of the Vibrio pathogenicity islands and CTXΦ (Waldor and Mekalanos, 1996; Karaolis et al., 1998), ToxR evolved the ability to activate the expression of the master virulence regulator ToxT together with TcpP (Krukonis et al., 2000), as well as to activate the expression of the ctx genes on its own (Miller and Mekalanos, 1984). Interestingly, it has been shown that the activation of ctx, but not that of tcp, by ToxR in V. cholerae classical biotype strains requires the presence of bile salts (Hung and Mekalanos, 2005). In light of our findings here, this suggests that the ability of bile salts to destabilize ToxR and enhance its interaction with ToxS may play a role in the activation of ctx expression as well as that of ompU. It is also interesting that the fatty acid component of bile inhibits ToxT activity, preventing the expression of tcp as well as ctx (Schuhmacher and Klose, 1999; Chatterjee et al., 2007; Lowden et al., 2010). Therefore, whereas bile salts stimulate ToxR activity inducing bile resistance and setting the stage for virulence, other environmental conditions such as the presence of fatty acids in the lumen of the intestine ensure that virulence gene expression is activated only at the appropriate time and place.

The results presented here suggest a model (Fig. 9) for how full-length ToxR and ToxS interact to activate ToxR. According to this model, the periplasmic domains of ToxR and ToxS weakly interact in the aquatic environment in the absence of any modulators. When the bacteria are ingested by bile producing organisms, bile salts partially unfold the periplasmic domain of ToxR leading to an increased interaction with ToxS, and resulting in changes in porin regulation, bile resistance, and, in classical biotype strains, an increase in ctx expression (Fig. 9 upper). Subsequently, in the late stages of infection, when the bacteria have reached stationary phase and alkalinized the environment, the interaction between ToxR and ToxS periplasmic domains decreases, allowing proteases greater access to the still destabilized ToxR, leading to its proteolysis (Fig. 9 lower). It is important to note that the work in this study was performed using purified domains from membrane proteins, and in the absence of membrane components. While this represents a first step in understanding how ToxS and ToxR interact, further work will be necessary to validate this model.

Figure 9.

Model of ToxR and ToxS interactions in response to bile and alkaline pH. ToxR and ToxS interact through their respective periplasmic domains at a basal level in the aquatic environment. When exposed to bile in the intestine (upper), ToxR is destabilized, increasing the interaction between ToxRp and ToxSp. This increases ToxR activity, leads to changes in porin regulation, bile resistance, and, in classical biotype strains, an increase in ctx expression. In the presence of alkaline pH (lower), the interaction between the periplasmic domains decreases, hastening ToxR proteolysis leading to V. cholerae entering a dormant state as it prepares to re-enter an aquatic environment.

In summary, these are the first studies demonstrating that the periplasmic domains of ToxR and ToxS interact, and that this interaction is modulated by bile salts, other detergents, and pH in a way that is correlated with ToxR activity. We conclude that: first, the interaction between ToxR and ToxS plays an important role in modulating V. cholerae adaptation to changing environments; and second, although detergent like molecules cause ToxR to unfold, this increases binding to ToxS and enhances transcriptional activity. Finally, given the conservation of the two proteins across the Vibrio family, this mechanism is likely to play an important role in related species.

Methods

Construction of a N16961 ΔtoxS mutant

To construct the in-frame ΔtoxS mutant, 500 bp of the 5′ and 3′ regions flanking toxS in V. cholerae N16961 were amplified using primers SM334 (GATCGTCTAGACGCGAGCTATGGCGTGCTGG), SM335 (GATCGGCGGCCGCAATTCTTAACCTGACTGAGC) and SM336 (GATCGGCGGCCGCTATTTTGCATAGCAAGATCCT), SM337 (GATCGAGATCTAGCTGCTCATGACATCTCTC). The amplified fragments were cloned into pKAS154 (Kovacikova and Skorupski, 2002) and the sequence of the resulting plasmid, pCRM100, was verified. The plasmid was used to transform S17-1λpir cells. The transformed cells were then mated with V. cholerae N16961 for allelic exchange. The resulting strains were sequenced using primers that flanked the insert SM338 (GCATGGCCACACGCACATTC) and SM339 (CCAATTGATCGCCCGAGTGG).

Determine porin regulation in response to bile and CDC in minimal media

To determine if sodium-cholate, sodium-chenodeoxycholate, and ox-bile (all from Sigma-Aldrich) could induce porin switching in minimal media similar to that described previously (Mey et al., 2012), V. cholerae O1 El Tor N16961 wild type, ΔtoxR, ΔtoxS and ΔompU strains, as indicated, were grown overnight at 37 °C in LB media. The next day, the cultures were diluted 1:100 in minimal T-media (100 mM TRIS, 100 mM NaCl, 50 mM KCl, 20 mM NH₄Cl₂, 2 mM KH₂PO₄, 1 mM Na₂SO₄, 1 mM CaCl₂•2H₂O, 0.5 mM MgCl₂•6H₂O, pH 7.4) (Simon and Tessman, 1963) supplemented with 0.2% (wt/vol) sucrose, 20 μM FeSO₄, and 1x vitamin mix (10 μM of each: thiamine HCl, calcium pantothenate, p-aminobenzoic acid, p-hydroxybenzoic acid, 2, 3-dihydroxybenzoic acid), as described in http://www.genome.wisc.edu/resources/protocols/ezmedium.htm, containing either 0.3% ox-bile, 0.1% cholate, 0.01% CDC, or no compound as a control. The cultures were incubated for another 8 hours at 37 °C. The cultures were then pelleted and whole cell lysates were prepared. The protein concentrations of the lysates were determined using the BCA assay (Thermo Fisher Sci.). Equal amounts of protein were run on a precast 16% SDS-PAGE gel (Novex). The gels were stained with Colloidal Blue (Invitrogen) to visualize the porins.

ToxR western blotting

To verify the expression of ToxR in the ΔtoxS mutant, western blotting was performed. A 16% gel was run as above then the proteins were transferred onto a nitrocellulose membrane using the iblot system (Invitrogen). The membrane was blocked overnight in TBST 3% BSA. Serum containing the primary antibody was incubated with the blot at 1:10,000 dilution in TBST, 3% BSA for 2 h. The blot was washed with TBST and the secondary anti-rabbit HRP antibody was diluted 1:10,000 in TBST, then incubated with the blot for 30 minutes. After washing with TBS, the antibody was detected using ECL (Thermo-Fisher) and visualized using film.

Cloning of the ToxR and ToxS periplasmic domains

DNA corresponding to the amino acid sequences of V. cholerae El Tor N16961 ToxR (VC0984) T199-E294 and ToxS (VC0983) S25-S173 were ordered from DNA2.0 and inserted into pTYB21 (NEB). The constructs were digested out of the DNA 2.0 plasmids using NdeI and BamHI (NEB) and inserted into pTYB21 in frame with the N-terminal chitin binding domain-intein tag. The plasmid sequences were verified by sequencing. This resulted in constructs that contained an additional five amino acids on the N-terminus when cleaved from the intein tag. The domains were expressed using Codon plus (DE3) pRIL cells (Agilent).

Unlabelled protein expression and purification

Protein expression was accomplished using a modified high cell density protocol. Starter cultures using ZYP-0.8G media as described by Studier (Studier, 2005) with 200 μg ml⁻¹ carbenicillin and 25 μg ml⁻¹ chloramphenicol were grown overnight at 30 °C. The next day the cultures were diluted 1:100 into a starter culture of TB with 2 mM MgSO₄ and antibiotics as above, then the culture was incubated at 37 °C. Several hours later the cultures were diluted 1:10 into a production culture of TB with 2 mM MgSO₄ with 50 μg ml⁻¹ carbenicillin. The culture was grown at 37 °C until an OD600 of 2–3 then induced with 200 μM IPTG, and 5% glycerol (v/v) was added to the culture. The culture was incubated at 18 °C overnight.

Purification buffers were kept at 4 °C at all times. Cells were pelleted at 3000 × g for 25 minutes at 4 °C, then resuspended in low salt TRIS buffer (20 mM TRIS pH 8.5, 200 mM NaCl, 3% (v/v) glycerol) with 1 mM EDTA, 500 μM PMSF, and a Complete protease inhibitor tablet EDTA free (Roche). The cells were lysed with 3 passes through a French Press. The lysate was clarified by ultracentrifugation ~100,000 × g for 45 minutes. The resulting supernatant was filtered using a 0.45 μm filter, then applied to a chitin resin (NEB) column. The column was washed with 6 CV low salt TRIS buffer with 1 mM ATP and 2 mM MgCl₂. The column was then washed with 10 CV of high salt TRIS buffer (20 mM TRIS pH 8.5, 1M NaCl, 3% (v/v) glycerol), followed by 40 CV of low salt TRIS buffer, and finally with 3 CV of cleavage buffer (low salt TRIS buffer with 108 mM DTT, 1 mM EDTA, 500 μM PMSF, and 1 Roche complete tablet). The column was then incubated at 16 °C for 3 days.

The cleaved protein was eluted from the chitin column in the low salt TRIS buffer and the elution was concentrated to a volume of about 5 ml using a 3 kDa cut-off Amicon concentrator (Millipore). The concentrated protein was further purified over a Hi Load Superdex S75 16 / 600 column (GE Healthcare). Fractions containing pure protein were either used as is or pooled and concentrated using a 3 kDa cut-off Amicon concentrator. For ToxRp, the concentrated protein was aliquoted and frozen until needed. Purified ToxSp was used within two days.

Labeled Expression and purification

¹⁵N Isotopically labeled ToxRp was expressed using a high density IPTG induction protocol. The culture was grown as described for the unlabeled expression until the production culture reached an OD of 3–4. The cells were then centrifuged at 450 × g for 10 minutes at room temperature. The cells were resuspended in M9 minimal media with 3 g L⁻¹ of ¹⁵N-NH₄Cl₂ (Cambridge Isotope Laboratories) as the sole nitrogen source with 50 μg ml⁻¹ of carbenicillin. The culture was incubated at 37 °C for at least one hour. Then IPTG was added to the culture at a final concentration of 500 μM, and the culture was incubated at 18 °C overnight.

The next day the culture was centrifuged at 3000 × g for 25 minutes and resuspended in TRIS pH 8.5 buffer (20 mM TRIS pH 8.5, 200 mM NaCl) with 1 mM EDTA, 500 μM PMSF, and a Complete protease inhibitor tablet EDTA free (Roche). The cells were lysed with 3 passes through a French Press. The lysate was clarified by centrifugation at ~100,000 × g for 45 minutes. The resulting supernatant was filtered with a 0.45 μm filter and labeled ¹⁵N-ToxRp fusion was captured using a chitin resin column that was equilibrated with 10 CV of the TRIS pH 8.5 buffer. The column was then washed with 6 CV of the TRIS pH 8.5 buffer with 1 mM ATP, 2 mM MgCl₂, followed by 10 CV of high salt phosphate buffer (20 mM KPO₄ pH 8, 1M NaCl). Then a 40 CV wash with phosphate pH 8 buffer (20 mM KPO₄ pH 8, 200 mM NaCl), followed by a 3 CV wash of cleavage buffer (20 mM KPO₄ pH 8, 200mM NaCl, 500 μM PMSF, 1 mM EDTA, Complete protease tablet). The column was then incubated at 16 °C for three days.

The cleaved protein was eluted from the column with phosphate pH 8 buffer. The elution was concentrated to a volume of about 5 ml with a 3 kDa cut-off Amicon concentrator. The concentrate was further purified with a Hi Load Superdex S75 16 / 600 gel filtration column with the phosphate pH 8 buffer. Selected fractions were concentrated using a 3 kDa cut-off Amicon concentrator and frozen at −80 °C until needed.

NMR titration of ¹⁵N-ToxRp with cholate and CDC

In order to determine if cholate and CDC could specifically interact with ToxRp, NMR titrations were performed. Stocks of 100 mM cholate, and 100mM CDC were prepared fresh in 20 mM KPO4 pH 8, 200 mM NaCl. Cholate and CDC were then serially diluted by half until the final stock concentration was 0.78 mM. The 62.5 mM (total) NRES (Sigma-Aldrich) stock was made by dissolving the individual amino acids in 20 mM KPO4 pH 8, 200 mM NaCl at 62.5 mM then adding equal volumes of each stock to a microcentrifuge tube. NMR samples were prepared by adding the stocks to tubes containing buffer and D₂O. Purified ¹⁵N-ToxRp was then added to the tubes to a final concentration of 100 μM. The final concentrations of the bile salts were 0 mM, 0.08 mM, 0.16 mM, 0.31 mM, 0.63 mM 1.25 mM, 2.50 mM, 5.00 mM, and 10.00 mM. The final NRES concentration was 12.5 mM, total amino acid concentration. All samples had a final concentration of 10% D₂O. HSQC spectrums were collected using a Bruker 600 MHz instrument at 298 K. Spectra were analyzed using NMRFAM-SPARKY 3.115 (Woonghee Lee et al., 2015).

Differential scanning fluorometry (DSF)

In order to assess protein stability in the presence of the compounds, DSF was performed. For each condition a master mix was prepared with the Sypro-orange dye (ThermoFisher Sci.) and cholate, CDC, NRES, or buffer. For cholate and CDC 10x stocks were prepared fresh in the low salt TRIS buffer. The NRES mixture was also prepared in the low salt TRIS buffer as described above. The master mixes were aliquoted into a 96 well PCR-plate to make 4 technical replicates for the protein and compound only controls. ToxRp was added to the wells to a final concentration of 100 μM. The final concentrations of cholate and CDC were 1 and 2 mM, NRES 12.5 mM, and dye 5x. DSF was performed using an Applied Biosystems StepOnePlus qRT-PCR system (ThermoFisher Sci.). The temperature was raised from 25 °C to a final temperature of 90 °C with a ramp rate of 1 °C min⁻¹. The normalized fluorescence intensities were transferred into a STATA file. The melting curves were smoothed and the first derivative was determined with STATA 13 (StataCorp LP). All graphing and statistical analysis were performed using STATA 13.

Pull down experiments

To assess the interaction between ToxRp and ToxSp, pull down assays were performed in the presence and absence of proposed modulators. All stocks were prepared in the low salt TRIS buffer described above. Stocks of cholate, Triton X-100 (Sigma), bile, and chenodeoxycholate were prepared fresh for each experiment at 10x concentrations. The amino acids NRES were prepared as described above in the low salt TRIS buffer. All pull downs were performed using Eppendorf Protein LoBind Tubes.

Where indicated working solutions of CDC, cholate, ox-bile, NRES, and X-100 were prepared as described above in the buffer used in the pull down assay. For the CDC dose response the final concentrations were from 0.003–10 mM. For the compound pull down assay the concentrations were: cholate 0.086%; CDC 0.083%; X-100 0.1%; ox-bile 0.3%; NRES 12.5 mM.

For the pull down experiments, fresh supernatants of the required periplasmic domain fusion and CBDI tag alone, as indicated, were prepared as described for protein purification. The appropriate supernatant or low salt TRIS buffer were incubated with washed magnetic chitin beads (NEB) for 1 hour at 4 °C while rotating. The purified protein was added to buffer or the compound at a final concentration of 100 μM (1.1 mg ml⁻¹ for ToxRp, and 1.8 mg ml⁻¹ for ToxSp). After the supernatants were finished incubating, the beads were washed 3 times with the low salt TRIS buffer. Then the appropriate prepared samples were added to the tubes. The tubes were then incubated for another hour at 4 °C while rotating. After the final incubation, the beads were washed 3 times with the low salt TRIS buffer. For the last wash, the beads were transferred to a clean tube and briefly spun to pellet the beads. After washing, 2x SDS sample buffer with 10 mM DTT and 1 mM EDTA was added to the tubes and the samples were boiled for 5 minutes. The resulting samples were then diluted in 1x SDS sample buffer with DTT, serial dilutions were prepared as indicated, and run on 4–15% gels (Bio Rad). The gels were stained using Colloidal Blue.

To analyze the pull down assay results, the stained gels were imaged and then processed using Image J (Schindelin et al., 2015). The background was subtracted from the images and lanes were outlined. The lanes were plotted and the area corresponding to the purified ToxRp or ToxSp was determined. The resulting measurements were exported to Stata 13. For the CDC, dose response the measurements were normalized from 0–1. To calculate the fold change, the treated band intensities and treated relative dilutions were used as a standard curve to calculate the position of the 1x untreated relative dilution on the treated relative dilution curve. The inverse of the calculated relative dilution was taken as the fold change. A similar procedure was used to determine the amount of ToxRp and ToxSp pulled down in the initial interaction assays. All graphing and statistics were performed using Stata 13.

Pull downs at different pH

To pull down ToxRp at the various pH’s, purified ToxRp at ~4 mg ml⁻¹ was diluted to 1.1 mg ml⁻¹ into buffers containing either 50 mM MES pH5.5, 50 mM HEPES pH 7.4, or 50 mM TRIS pH 9.0, and 200 mM NaCl, 3% glycerol. Lysates of CBDI and CBDI-ToxSp were prepared and incubated with the chitin beads as above. After incubating with the lysates, the beads were washed with the appropriate buffer and the diluted ToxRp was added to the tubes. After the final wash, SDS samples were prepared by boiling the beads. A dilution series of ToxRp pulled down at pH 7.4 was run on a gel with 1x and 1/2x samples of the other pH trials. Using the pH 7.4 dilution series as a standard curve the percentage of ToxRp pulled down at pH 5.5 and pH 9.0 relative to pH 7.4 was determined using STATA 13.

Table 1.

Bacterial strains used in this study

Strain	Relevant genotype	Reference
KSK 949	V. cholerae O1 El Tor N16961 wild type	Lab collection
KSK 1694	V. cholerae O1 El Tor N16961 ΔtoxR	Lab collection
SM 532	V. cholerae O1 El Tor N16961 ΔompU	Lab collection
CM 120	V. cholerae O1 El Tor N16961 ΔtoxS	This work