Characterization of Opposing Responses to Phenol by Bacillus subtilis Chemoreceptors

ABSTRACT

Bacillus subtilis employs 10 chemoreceptors to move in response to chemicals in its environment. While the sensing mechanisms have been determined for many attractants, little is known about the sensing mechanisms for repellents. In this work, we investigated phenol chemotaxis in B. subtilis. Phenol is an attractant at low, micromolar concentrations and a repellent at high, millimolar concentrations. McpA was found to be the principal chemoreceptor governing the repellent response to phenol and other related aromatic compounds. In addition, the chemoreceptors McpC and HemAT were found to govern the attractant response to phenol and related compounds. Using chemoreceptor chimeras, McpA was found to sense phenol using its signaling domain rather than its sensing domain. These observations were substantiated in vitro, where direct binding of phenol to the signaling domain of McpA was observed using saturation transfer difference nuclear magnetic resonance. These results further advance our understanding of B. subtilis chemotaxis and further demonstrate that the signaling domain of B. subtilis chemoreceptors can directly sense chemoeffectors.

IMPORTANCE Bacterial chemotaxis is commonly thought to employ a sensing mechanism involving the extracellular sensing domain of chemoreceptors. Some ligands, however, appear to be sensed by the signaling domain. Phenolic compounds, commonly found in soil and root exudates, provide environmental cues for soil microbes like Bacillus subtilis. We show that phenol is sensed as both an attractant and a repellent. While the mechanism for sensing phenol as an attractant is still unknown, we found that phenol is sensed as a repellent by the signaling domain of the chemoreceptor McpA. This study furthers our understanding of the unconventional sensing mechanisms employed by the B. subtilis chemotaxis pathway.

INTRODUCTION

Flagellated bacteria can swim up gradients of chemicals favorable to their growth and survival, known as attractants, and down ones inhibitory to their growth and survival, known as repellents (1). They sense these chemicals using chemoreceptors. Bacillus subtilis, a Gram-positive soil bacterium, possesses 10 chemoreceptors. Eight are transmembrane and two are soluble chemoreceptors (2). The transmembrane chemoreceptors possess an extracellular sensing domain and an intracellular signaling domain. The extracellular sensing domain is coupled to the intracellular signaling domain by two transmembrane helices, known as TM1 and TM2, and an intracellular HAMP domain (3). The soluble chemoreceptors also have signaling and sensing domains but lack the transmembrane helices and HAMP domain.

The B. subtilis chemoreceptors form stable complexes with the histidine kinase CheA and the coupling proteins CheV and CheW (4, 5). These complexes form large, hexagonally structured clusters that preferentially localize at the poles of the cell (6, 7). These clusters are able to amplify the signaling response through allosteric interactions with neighboring chemoreceptors (8). The binding of attractants to the B. subtilis chemoreceptors increases the rate of CheA autophosphorylation (5). The phosphoryl group is then transferred to CheY, a soluble response regulator. Phosphorylated CheY can then bind the flagellar motors and induce them to spin counterclockwise, which causes B. subtilis to swim in a straight direction (5, 9). Presumably, the binding of repellents to B. subtilis chemoreceptors decreases the rate of CheA autophosphorylation. This, in turn, decreases the concentration of phosphorylated CheY, which reduces the likelihood that the motors spin counterclockwise. When not bound with phosphorylated CheY, the motors spin clockwise, which causes B. subtilis to tumble about. Chemical gradients are sensed through a temporal mechanism involving sensory adaptation. B. subtilis uses three complementary mechanisms for sensory adaptation involving receptor methylation, allosteric regulation by a soluble protein known as CheD, and phosphorylation of the CheV adaptor protein (4, 10, 11).

Multiple studies have investigated the mechanisms for sensing attractants by chemoreceptors in B. subtilis and other diverse species of bacteria (see reference 12 for a comprehensive review). However, far less is known about the sensing mechanisms involving repellents. In this study, we investigated phenol chemotaxis in B. subtilis. Phenol is an attractant at low, micromolar concentrations and a repellent at high, millimolar concentrations. By analyzing different mutants, McpA was found to be the principal chemoreceptor governing the repellent response to phenol. In addition, the chemoreceptors McpC and HemAT were found to govern the attractant response, although the sensing mechanisms remain unknown. Using receptor chimeras, McpA was found to sense phenol using its signaling domain rather than its sensing domain. These observations were substantiated in vitro, where direct binding of phenol to the signaling domain of McpA was observed using saturation transfer difference nuclear magnetic resonance (STD-NMR). Along with phenol, we found that B. subtilis exhibits opposing responses to additional aromatic compounds, with McpA governing the repellent response and HemAT and McpC governing the attractant response. While many studies on aromatic compounds focus on metabolism-dependent chemotaxis (13–17), our work shows that B. subtilis exhibits a metabolism-independent response to these compounds.

RESULTS

B. subtilis senses phenol as an attractant at low concentrations and a repellent at high concentrations.

The chemotactic response to phenol was measured using the capillary assay (1). In this assay, glass capillaries containing test compounds were introduced into wells filled with bacteria in chemotaxis buffer. The diffusion of these test compounds out of the capillary generates a chemical gradient. Bacteria that entered the capillary after 1 h of incubation were counted to quantify the chemotaxis response. The base response in the absence of a chemical gradient was determined using a control capillary containing buffer only. A colony count higher than the base count indicates an attractant response, while a lower colony count indicates a repellent response. Phenol was chosen as the model compound to elucidate the chemotactic response to aromatic compounds. Phenolic compounds are abundant in nature, and chemotaxis to phenol has been studied in other bacteria (18).

The capillary assay response to increasing concentrations of phenol is shown in Fig. 1. The number of bacteria accumulating in the capillaries was higher than the base count for capillaries containing phenol at concentrations of <100 μM. A drop in the response was observed at higher concentrations. The decrease in the response at higher concentrations could be due to a repellent response to phenol, loss of motility, or decreased viability of cells. Previous studies have shown that the MIC of phenol for B. subtilis is 16 mM (19), suggesting a repellent response instead of decreased viability. To determine whether the reduced colony count was due to a repellent response, we used the “chemical-in-pond” modification of the capillary assay (20). In the modified assay, referred to as the repellent assay here, capillaries are filled with chemotaxis buffer and inserted into bacterial wells, called ponds, containing the repellent. The chemotaxis repellent response was quantified as the number of bacteria entering the capillary. A potent repellent response to phenol was observed at a concentration of 316 μM in the pond (Fig. 2A). These results suggest that phenol is sensed as an attractant at low concentrations and as a repellent at high concentrations.

FIG 1.

Chemotaxis response of wild-type B. subtilis to phenol. A dose-dependent response of the wild-type strain to increasing concentrations of phenol was measured using the attractant capillary assay. The dashed line indicates the base response to the buffer control.

FIG 2.

Repellent and attractant chemoreceptors for sensing phenol. (A) Repellent chemotaxis responses of strains expressing single chemoreceptors to phenol (316 μM) measured in the repellent capillary assay. The inset shows the repellent chemotaxis response of a mutant lacking the McpA chemoreceptor (ΔmcpA) to phenol. (B) Attractant chemotaxis responses of strains expressing single chemoreceptors to phenol (100 mM) measured in the attractant capillary assay. The inset shows the attractant chemotaxis response of the mutant strains lacking phenol chemoreceptors. Error bars indicate standard deviations obtained from three biological replicates performed on separate days.

McpA is the major chemoreceptor for sensing phenol as a repellent.

Many repellents are membrane-active compounds, and the possibility that the repellent response is due to direct action on the membrane was dismissed in a previous study (21). The dose-dependent response of phenol in the capillary assays also hinted at a receptor-specific response. B. subtilis has 10 chemoreceptors (2). To identify the chemoreceptors involved in repellent sensing, we tested B. subtilis mutants expressing a single chemoreceptor with the other nine deleted (22) (Fig. 2A). In the repellent assay, the phenol concentration in the pond was set to 316 μM, where the chemotaxis responses were more reproducible than at higher phenol concentrations, where the cells were less motile. Cell motility was even more impaired in the case of mutant strains, leading to lower chemotaxis responses overall.

In these experiments, only the strain expressing McpA as its sole chemoreceptor exhibited a repellent response to phenol. To confirm this result, an mcpA knockout (ΔmcpA) mutant was tested. The ΔmcpA mutant’s repellent response to phenol was almost completely eliminated, confirming that McpA is the main repellent chemoreceptor (Fig. 2A, inset). Other chemoreceptors may sense phenol as a repellent, but at native expression levels, their contribution is minor, and McpA dominates the response.

Functional redundancy of attractant chemoreceptors.

B. subtilis is attracted to phenol at low concentrations. To identify the chemoreceptors that sense phenol as an attractant, we tested single-receptor mutants using the attractant capillary assay (Fig. 2B). In the attractant assay, the actual concentrations that cells are exposed to are 50- to 100-fold lower than the concentrations inside capillaries (23). Therefore, we tested the single-receptor mutants using higher phenol concentrations (100 mM compared to micromolar concentrations for the wild type) to compensate for the lack of signal amplification associated with allosteric interactions between chemoreceptors (24, 25). Mutant strains with McpC or HemAT as their sole chemoreceptors exhibited attractant responses to phenol. These responses were comparable to that of the control strain (ΔmcpA) deleted for the repellent response (Fig. 2B, inset). It is notable that the attractant responses were not additive. To confirm this, we also measured the attractant responses of strains lacking mcpA and mcpC and strains lacking mcpA and hemAT to phenol. As expected, the attractant responses of these strains were similar to that of the ΔmcpA control strain (Fig. 2B, inset). The reason for the functional redundancy of attractant chemoreceptors for phenol is not clear. However, functional redundancy of chemoreceptors for sensing aromatic compounds is also observed in other bacteria (13).

Chemotaxis responses to other aromatic compounds.

To understand the ligand range and preference better, we tested the chemotactic responses to a number of aromatic compounds that differ in structures using the attractant and repellent capillary assays. Benzene and toluene were selected due to their simple structures. Physiologically relevant aromatics such as salicylate, benzoate, resorcinol, and p-hydroxybenzoic acid, which are found in plant root exudates, were also tested (26).

Concentrations were optimized for each assay for the wild-type and mutant strains separately. This was done to avoid the compounded effects of multiple chemoreceptors in the wild-type strain and to generate a higher response from the single-receptor strains, as discussed above. Attractant capillary assays were performed for the wild type at a concentration of 1 μM in the capillary (Fig. 3A). Low concentrations were tested to avoid repellent responses associated with McpA. At a concentration of 1 μM, toluene had a higher attractant response than the other tested compounds, while p-hydroxybenzoic acid showed no response (P > 0.05). The overall responses to aromatic compounds are lower than the responses for more conventional ligands like amino acids (27, 28).

FIG 3.

Wild-type chemotaxis response to other aromatic compounds. (A) Attractant chemotaxis response of the wild-type strain to 1 μM aromatic compounds in the attractant capillary assay. (B) Attractant chemotaxis responses of strains expressing McpC or HemAT as their sole chemoreceptors to different aromatic compounds. (C) Repellent chemotaxis response of the wild-type strain to 1 mM aromatic compounds in the repellent capillary assay. (D) Repellent chemotaxis responses of strains expressing McpA as their sole chemoreceptor to different aromatic compounds at a concentration of 316 μM measured in the repellent capillary assay. Error bars indicate standard deviations obtained from three biological replicates performed on separate days. Statistical analysis was performed (n = 3), and P values are calculated against the buffer control. * represents a P value of <0.05. n.s, not significant. Table S2 in the supplemental material lists the attractant and repellent chemoreceptors for the compounds tested in this work.

To test whether McpC and HemAT are also the attractant chemoreceptors for these other compounds, we tested single-receptor strains using the attractant capillary assay. As the repellent chemoreceptor McpA was absent in these strains, we tested higher phenol concentrations in capillaries (100 mM) to obtain a greater response than with the buffer control. For compounds with limited solubility, the highest soluble concentrations were used (see Data Set S1 in the supplemental material). The strain with McpC as its sole chemoreceptor was able to sense other aromatic compounds as well, except for p-hydroxybenzoic acid (P > 0.05) (Fig. 3B). The strain with HemAT as its sole chemoreceptor sensed benzene, toluene, phenol, and resorcinol as attractants (P < 0.05) (Fig. 2B).

Repellent responses of the wild-type strain were tested for the same compounds with a 1 mM concentration in the pond (Fig. 3C). At this concentration, compounds that showed a high repellent response, other than phenol, were resorcinol, sodium benzoate, and sodium salicylate. No statistically significant responses were observed for benzene, toluene, p-coumaric acid, and p-hydroxybenzoic acid (P > 0.05). The strain expressing McpA alone was also able to mediate repellent responses to all compounds tested in this study, except for p-hydroxybenzoic acid (P > 0.05) (Fig. 3D). Again, a lower concentration of 316 μM was used for testing single-receptor mutant strains because the cells were directly in contact with the repellent in the pond.

Cytoplasmic sensing of phenol.

Transmembrane chemoreceptors have a modular structure consisting of an extracellular sensing domain and a conserved, intracellular signaling domain. They also have a cytoplasmic HAMP domain for relaying signals between the two domains (3). It was previously shown that phenol sensing by Escherichia coli chemoreceptors is mediated by the transmembrane helices and HAMP domain (29). In order to determine the regions of McpA involved in sensing phenol as a repellent in B. subtilis, we used chimeric receptors generated by swapping domains between McpA and McpB (23, 30). McpB senses asparagine as an attractant using its extracellular sensing domain and is not involved in phenol sensing (28). This makes it a good control in order to confirm the functionality of the chimeras as well as to identify the regions of McpA involved in phenol sensing (Fig. S1A). All chimeras were expressed under the control of the mcpA promoter in a strain deleted for the native chemoreceptors.

Strains expressing an McpA-McpB chimera with the sensing domain from McpA replaced with the one from McpB (mcpA₄₄B₂₆₇A) showed a repellent response to phenol and an attractant response to asparagine (Fig. 4 and Fig. S1A). This suggests that the extracellular sensing domain of the McpA chemoreceptor does not sense phenol. Likewise, strains expressing an McpB-McpA chimera with the McpB signaling domain replaced with the one from McpA (mcpB₃₅₉A) showed a repellent response to phenol, indicating that the transmembrane helices and HAMP domain of McpA are not required for sensing phenol. The repellent response to phenol was eliminated in a chimera where the McpA signaling domain was swapped with the McpB signaling domain (mcpA₃₅₈B). This mutant strain, however, responded to acidic pH mediated by the McpA extracellular sensing domain (22), proving the functionality of the chimera (Fig. S1B). These observations suggest that the repellent response to phenol is mediated by the intracellular signaling domain of McpA.

FIG 4.

The signaling domain of McpA is involved in sensing phenol as a repellent. (A) Domain structure of McpA. The chemoreceptor consists of an extracellular sensing domain (dCache_1), two transmembrane helices (TM1 and TM2) followed by an intracellular HAMP domain, and an intracellular signaling domain (MCPsignal). The signaling domain is comprised of three subdomains known as the methylation helices (MH), flexible bundle (FB), and conserved signaling tip (CS). (B) Cartoon structure of the McpA chemoreceptor monomer. (C) Repellent chemotaxis responses of strains expressing different chimeras between McpA and McpB as their sole chemoreceptor to 316 μM phenol measured in the repellent capillary assays. The nomenclature for chimeras indicates the chemoreceptor name followed by the residue number at which the junction is created. Error bars indicate standard deviations obtained from three biological replicates performed on separate days.

The signaling domains of chemoreceptors consist of coiled-coil helices with high sequence conservation across bacteria and archaea. They have three structurally distinct subdomains: the methylation helices, the flexible bundle, and the signaling tip (3). We tested receptor chimeras with fusions near the junction of the methylation helices and flexible bundle (mcpB₃₉₇mcpA and mcpB₄₃₃A). Cells expressing mcpB₃₉₇A showed a repellent response to phenol, while cells expressing mcpB₄₃₃A failed to respond to phenol but responded to asparagine normally (Fig. 4 and Fig. S1A). These results suggest that phenol is sensed by McpA using the region spanning residues 397 to 433, which corresponds to the bottom of the methylation helix and the top of the flexible bundle on the N terminus of the McpA signaling domain (Fig. 4). Interestingly, the corresponding region on McpB is also involved in sensing short-chain aliphatic alcohols (C₁ to C₅), including ethanol (23).

Docking experiments confined to the phenol-sensing region on McpA were carried out to gain some insights into putative residues involved in phenol binding. Computational predictions showed that phenol conformations with the lowest docking energy scores cluster within the dimer (cluster 1) (ranging from Glu³⁹⁷to Ile⁴¹⁰ on the N-terminal helix [N-helix] and Glu⁶⁰⁰ to Ser⁶¹³ on the C-helix). Interestingly, other repellent ligands that are sensed by McpA also clustered in the same region, except for salicylate (Fig. S2B and C). p-Hydroxybenzoic acid was used as a negative control for the docking experiments as it failed to induce chemotaxis responses in the capillary assays. p-Hydroxybenzoic acid docked in a different region (cluster 2) (residues Leu⁴¹⁷ to Ser⁴³⁰ on the N-helix and Asp⁵⁷⁹ to Gln⁵⁹³ on the C-helix) (Fig. S2C). The docking scores are provided in Table S2. To assess whether these residues are important for phenol binding, we first aligned the amino acid sequences spanning residues 397 to 410 on the N-helix and the neighboring residues 600 to 613 on the C-helix of all 10 chemoreceptors of B. subtilis (Fig. S2D). Note that chemoreceptors are highly conserved at their signaling domain, and hence, the alignment was used to guide amino acid substitution experiments. Unfortunately, single substitutions of nonconserved residues in McpA for the corresponding amino acids in McpB (N402K, A406S, S425A, A424T, M421V, S431A, and S606A) did not show any change in the response (data not shown).

The attractant chemoreceptors for phenol, McpC and HemAT, are phylogenetically distant from the other chemoreceptors in B. subtilis and, hence, are not suitable for chimeric analysis (Fig. S3). Indeed, chimeric receptors with the cytoplasmic domains swapped between McpA and McpC failed to respond to proline, which is sensed by the sensing domain of McpC (27). HemAT also proved to be unamenable to chimeric analysis using either McpA or YfmS (the only other soluble B. subtilis chemoreceptor). However, we were able to identify an aromatic attractant chemoeffector for McpB.

Cells expressing McpB as their sole chemoreceptor were able to sense benzene as an attractant. We were further able to identify the region involved in sensing benzene as an attractant using chimeric receptors between McpA (repellent chemoreceptor for benzene) and McpB (attractant chemoreceptor for benzene). Strains expressing the chimera mcpB₃₉₇A, involving the sensing domain of McpB and the signaling domain of McpA, showed a repellent response to benzene. However, strains expressing the chimera mcpB₄₃₃A, which contains the above-mentioned 36-amino-acid region (residues 397 to 433) of the McpB signaling domain, showed an attractant response to benzene (Fig. S4), implying that the same region of the signaling domain used by McpA as a repellent is also used by McpB to sense benzene as an attractant.

In vitro characterization of phenol-chemoreceptor interactions using STD-NMR.

To determine whether the chemoreceptors directly bind phenol, we used ¹H STD-NMR (proton saturation transfer difference nuclear magnetic resonance) (31). STD-NMR is a small-molecule-based NMR method for identifying ligands binding with medium to weak affinities (100 μM < K_d [dissociation constant] < 10 mM). This technique exploits the transfer of magnetization from a selectively irradiated protein to a ligand in its proximity (32). Ligand protons closest to the protein will receive magnetization more efficiently (33). As the protein-ligand complex is in equilibrium, the magnetization is transferred to the bulk solution upon ligand dissociation. This is observed as a decrease in the ligand peak in the NMR spectrum (on-resonance spectrum) over that of the reference spectrum with no selective saturation (off-resonance spectrum). The difference between the on- and off-resonance spectra will have peak intensities corresponding only to the ligand protons that bound to the protein.

Figure 5A shows the ¹H and STD-NMR spectra for the signaling (McpA_C) and the sensing (McpA_S) domains of McpA incubated with 2 mM phenol. Phenol protons have resonance peaks at 7.2 ppm (meta), 6.9 ppm (para), and 6.8 ppm (ortho). Strong STD-NMR signals for all three proton groups were observed with the McpA signaling domain in the presence of phenol, while STD-NMR signals for the McpA sensing domain incubated with the same concentration of phenol showed negligible peaks, corroborating the data from the chimeric receptor experiments. Additionally, no phenol peaks were observed for the control experiment in the absence of proteins.

FIG 5.

Estimation of the dissociation constant (K_d) of binding between the McpA signaling domain and phenol using STD-NMR. (A) ¹H and STD-NMR spectra obtained from incubation of 20 μM McpA signaling (top) and sensing (bottom) domains with 2 mM phenol. (B) Time evolution of STD amplification factor (STD-AF) values obtained from incubation of 20 μM McpA signaling domain with different concentrations of phenol (0.5 mM to 4 mM). The initial slope (STD-AF₀) values were calculated from the fitted curves to estimate the binding dissociation constant. (C) STD-AF₀ values from the previous step were fitted against different phenol concentrations to estimate the binding dissociation constant using the Langmuir isotherm model [STD-AF₀ = αL/(L + K_d), where K_d = 4.8 mM, α = 18.9, and R² = 0.98].

To gain insight into the direction of phenol binding to the McpA signaling domain, we analyzed the STD amplification factor (STD-AF) for phenol protons. Briefly, the STD-AF for a proton is calculated as the ratio of the integrated peak signal in the STD spectrum over that of the reference spectrum multiplied by the ligand excess. A difference of at least 10% between ligand epitopes is recommended for classification as a preferred binding orientation (Table 1) (34). These analyses show that the para proton has a slightly higher STD effect than other protons (15% over ortho protons and 18.9% over meta protons), suggesting that the para proton is closer to the chemoreceptor.

TABLE 1.

STD amplification factors for phenol-protein interactions

Protein	STD-AF (%)
	H(2) meta proton	H(3) para proton	H(1) ortho proton
McpAC	11.09	13.69	11.61
McpAS	3.52	5.54	3.72
McpCC	3.11	4.15	2.89
McpCS	0.82	1.33	1.28
HemATC	23.85	26.90	24.29
HemATS	17.41	19.50	15.87
McpBC	3.64	4.02	3.22

Competitive binding experiments have been used to distinguish between specific and nonspecific binding in conjunction with STD-NMR (35, 36). However, these experiments require a reference ligand for which the binding site on the protein and the binding affinity between the two are known. Since such a ligand was unknown for these experiments, we instead conducted the STD-NMR experiments for the McpB signaling domain in the presence of excess phenol to provide an additional negative control. The rationale for this experimental design was that the McpB signaling domain has a coiled-coil structure with 74% sequence identity to the McpA signaling domain. As expected, negligible phenol peaks (STD-AF, <4%) were observed in the STD spectrum (Fig. S5A). We also conducted a control STD-NMR experiment with p-hydroxybenzoic acid, which showed no response in the capillary assays. No signals corresponding to the ligand were observed in the STD spectrum of the McpA signaling domain incubated with 2 mM p-hydroxybenzoic acid (Fig. S5B).

We next used STD-NMR to determine the affinity of phenol for the signaling domain of McpA. STD-NMR titration experiments at different ligand concentrations can be used to determine the binding affinity (K_d) of a protein-ligand complex. However, STD signals also rely on the kinetics of protein-ligand rebinding. The method for initial growth rates, where STD factors are calculated at the limit of zero saturation time (no protein-ligand rebinding), has been successfully used in a previous report for determining K_d values (37). Briefly, STD-AF values were first calculated at different saturation times to obtain the initial slope for each concentration (STD-AF₀) (Fig. 5B). Initial slope values were then plotted against ligand concentrations to determine the K_d using the Langmuir isotherm curve. The binding affinity for phenol was calculated as 4.8 mM with McpA’s signaling domain (Fig. 5C). It is important to note that the binding curve for the phenol and McpA signaling domain pair followed the hyperbolic curve model with a decreasing slope at higher phenol concentrations, indicative of a specific binding event. However, at higher phenol concentrations (>4 mM), nonspecific binding dominates, and the curve increased linearly instead of reaching a plateau. As a consequence, the K_d is overestimated in this study. The overestimation of binding K_d values has also been reported in another STD-NMR study, and orthogonal techniques are recommended for obtaining accurate values (34, 37). As such, we conducted isothermal titration calorimetry experiments to assess the binding between the McpA signaling domain and phenol. Unfortunately, these experiments were not conclusive because the affinity is weak (data not shown). In the case of HemAT, the phenol binding mechanism appears to be rather complex as phenol directly interacts with both the sensing and signaling domains of HemAT.

Chimeric receptor experiments suggested that the McpB signaling domain is involved in sensing benzene. The STD spectrum of the McpB signaling domain in the presence of 2 mM benzene showed a peak associated with the benzene moiety with an STD-AF of 16.4%, implying a direct interaction of benzene with the McpB signaling domain (Fig. S5C).

In the case of HemAT, phenol peaks were observed in the STD spectra for both the sensing and signaling domains (Fig. 6A). The para proton had the highest STD-AF values; however, the difference between STD-AF values within different moieties was too close to the 10% threshold described above. Therefore, it is difficult to classify this as a preferred binding direction. HemAT is a soluble chemoreceptor that contains a heme group for sensing molecular oxygen (38, 39). The sensing domain dimer interface consists of a four-helical bundle at its core with other helices packed around it. HemAT sensing domain helices are hypothesized to be involved in sensing ethanol (23), and phenol likely follows a similar mechanism. The signaling domain of HemAT has a coiled-coil conserved structure similar to that of McpA. Surprisingly, the STD spectra for phenol and the McpC sensing and signaling domains were negligible (Fig. 6B). This suggests that phenol is probably sensed by the transmembrane helices or the HAMP domain of McpC.

FIG 6.

In vitro characterization of binding between phenol and attractant chemoreceptors, McpC and HemAT. (A) ¹H and STD-NMR spectra obtained from incubation of 20 μM HemAT signaling (top) and sensing (bottom) domains with 2 mM phenol. (B) ¹H and STD-NMR spectra obtained from incubation of 20 μM McpC signaling (top) and sensing (bottom) domains with 2 mM phenol. ¹H peaks for ortho, meta, and para moieties of phenol are shown within dashed boxes.

DISCUSSION

Previous studies reported that several membrane-active agents are repellents for B. subtilis (40). Initially, it was unknown whether repellents are sensed by receptors or the membrane through changes in fluidity. Later, it was shown that repellents are sensed by chemoreceptors; however, their identities were unknown (41, 42). In this work, we report that the repellent receptor for phenol is McpA, a transmembrane chemoreceptor that also senses external acidic pH as an attractant (22). McpA was found to be a broad-range chemoreceptor capable of sensing multiple other aromatic compounds as repellents. Our data indicate that phenol is sensed by the cytoplasmic signaling domain of McpA. This is in contrast to the more common mechanism where chemoeffectors bind the extracellular sensing domain to induce signaling. That said, multiple examples of sensing by the signaling domain have been documented in the literature. For example, the region below the HAMP domain was also found to be important for sensing toluene and o-xylene by the Tar chemoreceptor in E. coli (43). Phenol sensing in E. coli also follows an unconventional route where phenol appears to bind the transmembrane helices and HAMP domain (29). In addition, the PctA chemoreceptor from Pseudomonas aeruginosa senses chlorinated compounds as repellents, but no binding was observed with the extracellular sensing domain of the chemoreceptor (44). This suggests the possibility of binding to alternative sites or indirect activation by other ligand binding proteins. Finally, ethanol is sensed by the signaling domain of McpB in B. subtilis (23).

Chimeric receptor analysis revealed the importance of amino acid residues near the junction of N-terminal methylation helices and the flexible bundle in McpA. In the case of attractant sensing, benzene also appears to be sensed by the same region in McpB. Swapping the 36-amino-acid regions between the McpA and McpB chemoreceptors converted the repellent response of McpA to an attractant response, and vice versa for McpB. This region was also important for sensing ethanol by the signaling domain of McpB (23). Molecular dynamics simulations and mutation studies correlated changes in the coiled-coil packing of the signaling domain with the ability of McpB to sense ethanol. While the exact mechanism for signal propagation in chemoreceptors is still being resolved, kinase activity is thought to be modulated by changes in chemoreceptor dynamics that propagate the signal through HAMP and signaling domains (45–47). Recent studies have characterized chemoreceptor flexibility and dynamics in different signaling states. Electron paramagnetic resonance (EPR) studies have shown that the N-terminal methylation helices have different dynamics than the C-terminal tail of the chemoreceptor in both methylated and demethylated states (48–50). The mobility of the N-terminal methylation helices was proposed to be a key signaling element in hydrogen exchange and solid-state NMR studies (51, 52). Cryo-electron tomography and molecular dynamics simulation studies of the E. coli Tsr chemoreceptor showed the importance of the glycine hinge (present in the flexible bundle region) in controlling chemoreceptor compactness and flexibility to allow for different signaling states (53, 54). It is possible that ligands like phenol and ethanol interact with signaling domain helices and cause changes in chemoreceptor packing or mobility, thus mimicking traditional signals that would have been propagated from the sensing domain via the HAMP region. However, the exact mechanism of phenol sensing by the McpA chemoreceptor remains unknown.

The attractant response to phenol is mediated by McpC and HemAT. McpC is a transmembrane chemoreceptor involved in sensing amino acids and sugars (27, 55). It is unclear how phenol induces signaling in McpC as no interaction between phenol and the McpC sensing or signaling domains was observed by STD-NMR. These experiments fail to detect tight binding events (affinities in the nanomolar-to-micromolar range) (37). Because the exchange of the ligand is slow, the transfer to bulk solution is low, and weak signals are observed. Another possibility is that phenol interacts with the transmembrane helices or HAMP domain of McpC in a manner similar to those of Tar and Tsr in E. coli (29); however, the possibility of an indirect sensing mechanism cannot be ruled out. The McpC sensing domain senses many amino acids indirectly through membrane-associated proteins (27), while the signaling domain is involved in sugar taxis through interactions with the phosphotransferase system (PTS) (55). Thus, phenol could induce signaling through an indirect interaction. HemAT, on the other hand, is a cytoplasmic heme-containing chemoreceptor known to sense molecular oxygen (56). Both the sensing and signaling domains of the HemAT chemoreceptor bound phenol in the STD-NMR experiments. The sensing domain of HemAT is also involved in recognizing ethanol, although the heme group is not involved (23). How phenol induces different output responses for different chemoreceptors is still an open-ended question.

Phenol is a complex chemoeffector and induces both attractant and repellent responses in E. coli and Salmonella enterica (18, 57). The overall response is dependent on the relative abundances of different chemoreceptors (58). HemAT and McpA are the two most abundant chemoreceptors in B. subtilis (19,000 ± 3,900 molecules/cell and 15,900 ± 3,000 molecules/cell, respectively), while McpC chemoreceptor concentrations are low (2,800 ± 640 molecules/cell) (2). Unlike E. coli and S. enterica, the overall response to phenol in B. subtilis does not appear to be dictated by the relative abundance of its chemoreceptors. It is possible that a different mechanism, such as the biphasic adaptation response described previously for indole, is at play (59). In particular, E. coli cells previously adapted to roughly 700 μM indole showed an attractant response to high indole concentrations (2 mM) in chemotaxis assays, while a repellent response was seen for unprimed cells. However, B. subtilis was not previously adapted to phenol in our capillary assays. One possibility is that the overall response depends on the kinetics and conformational changes induced upon the binding of phenol to the three chemoreceptors. Unfortunately, capillary assays are limited in quantifying the exact effect of phenol addition or removal. For example, we previously found that the chemoeffector concentrations experienced by the cells near the capillary are 10 to 50 times lower than the initial concentration inside the capillary (23). The attractant and repellent capillary assays are also optimized with different cells and chemoeffector concentrations, which limits direct comparison of the opposing responses induced by a chemoeffector. The mechanism of inversion of an attractant to a repellent response is a topic of ongoing study.

Phenolic compounds are ubiquitous in nature (60). Phenol cannot be metabolized by B. subtilis (19) and likely serves as a cue to direct the cells toward physiologically favorable environments. This hypothesis is in line with previous evidence that showed that B. subtilis chemotaxis to chemicals is not necessarily associated with their metabolism (23, 61). Plant root exudates contain phenolic compounds that act as signaling molecules for plant-microbe interactions (62, 63). Phenol is found at low, micromolar concentrations in soil and water (64–66). In nature, it is likely that B. subtilis senses low concentrations of phenol as attractants to locate plant roots. Chemotaxis to plant root exudates has been shown to be important for root colonization (67). B. subtilis is a member of the plant-growth-promoting rhizobacteria (68). Root exudates from soybean and rice plants are known to attract Bacillus amyloliquefaciens and Bacillus spp. (26, 69). Chemotaxis was also found to be essential for early root colonization of Arabidopsis thaliana by B. subtilis (70). This study also reported an increase in the chemotaxis response to root exudates by a ΔmcpA strain. As McpA was found to be the major repellent chemoreceptor for phenol and other aromatic compounds investigated in this study, the increased chemotaxis response in the ΔmcpA strain can be explained as being due to the loss of the repellent chemoreceptor. Other compounds, including p-hydroxybenzoic acid, vanillyl alcohol, and isoflavones, are known chemoattractants for Agrobacterium and Rhizobium species of the Rhizobiaceae family (63). Metabolism-independent chemotaxis is also seen for Pseudomonas putida strain KT2440 toward phenolic compounds such as salicylate (15). Certain phenolic compounds such as salicylic acid and hydroxycoumarins are also produced by plants in response to pathogen attacks (62). Salicylic acid was reported to be a repellent for B. amyloliquefaciens (26). The presence and concentration of phenolic compounds in root exudates thus change according to many factors. Changing the attractant and repellent chemotaxis responses to aromatics at different concentrations likely helps B. subtilis navigate altering rhizosphere conditions. Some yeasts, such as Saccharomyces cerevisiae and Candida albicans, also release phenolic compounds as quorum signaling molecules (71–73). Aromatic compounds have also been detected in culture supernatants of Pseudomonas fluorescens (74). B. subtilis exhibits complex predator-prey relationships with many microorganisms and is known to release many antifungal and antibiotic compounds (75–77). Therefore, the chemotaxis response to aromatics may also help B. subtilis find prey and evade predators.

Phenolic compounds are also classified as pollutants and cause various harmful effects on humans and animals. These compounds are primarily found in industrial wastes and forest fires and are released into the environment through anthropogenic activities. Bioremediation strategies based on the microbial decomposition of phenolic compounds are being used widely to remove phenols from wastewater (78, 79). Chemotaxis toward environmental pollutants can be exploited to guide bacteria and locate pollutants in the environment. As more reports of novel bacterial species responding to pollutants are being reported (13, 17), an understanding of the underlying mechanisms of chemotaxis is essential to enable the design of new strategies facilitating the bioavailability of pollutants for degradation.

MATERIALS AND METHODS

Chemicals, media, and growth conditions.

B. subtilis strains were routinely grown on tryptose blood agar base (TBAB) (1% tryptone, 0.3% beef extract, 0.5% NaCl, and 1.5% agar) plates at 30°C for 16 h. Chemotaxis experiments were performed with capillary assay minimal medium (CAMM) [50 mM potassium phosphate buffer (pH 7.0), 1.2 mM MgCl₂, 0.14 mM CaCl₂, 1 mM (NH₄)₂SO₄, 0.01 mM MnCl₂, and 42 μM ferric citrate]. Chemotaxis buffer consists of 10 mM potassium phosphate buffer (pH 7.0), 0.14 mM CaCl₂, 0.3 mM (NH₄)₂SO₄, 0.1 mM EDTA, 5 mM sodium lactate, and 0.05% (vol/vol) glycerol. All aromatic compounds were of reagent grade and above. Chemicals were purchased from Sigma and Fisher.

Strains and plasmids.

All B. subtilis strains were derived from the chemotactic strain OI1085 (80). The strains and plasmids used in this work are listed in Tables 2 and 3, respectively. All cloning was performed using NEB (New England BioLabs) 5-alpha competent E. coli.

TABLE 2.

Strains used in this study

Strain	Relevant genotype or description	Source or reference
5-alpha	E. coli cloning host	New England BioLabs
BL21(DE3)	E. coli protease-deficient expression host	Novagen
OI1085	trpF7 hisH2 metC133 che+	80
PTS324	ΔmcpA	22
PTS325	ΔtlpA	22
PTS334	ΔmcpA ΔtlpA	22
GB032	ΔmcpA ΔmcpC	This work
GB054	ΔmcpA ΔhemAT	This work
OI3545	Δ10mcp; Ermr Cmr Kanr che+	56
OI3921	OI3545 amyE5720::mcpA; Spcr	96
OI3605	OI3545 amyE5720::mcpB; Spcr	55
OI3974	OI3545 amyE5720::mcpC; Spcr	55
OI4474	OI3545 amyE5720::tlpA; Spcr	22
OI4475	OI3545 amyE5720::tlpB; Spcr	22
OI4483	OI3545 amyE5720::tlpC; Spcr	22
OI4476	OI3545 amyE5720::yfmS; Spcr	22
OI4477	OI3545 amyE5720::yvaQ; Spcr	22
OI4482	OI3545 amyE5720::hemAT; Spcr	22
OI4479	OI3545 amyE5720::yoaH; Spcr	22
GBS141	OI3545 amyE5720::PmcpA-mcpB; Spcr	This work
GBS041	OI3545 amyE5720::PmcpA-tlpA; Spcr	This work
PTS155	OI3545 amyE5720::PmcpA-mcpA[M1–L44] mcpB[D45–T267] mcpA[M267–E661]; Spcr	This work
PTS157	OI3545 amyE5720::PmcpA-mcpB[M1–Q359] mcpA[D359–E661]; Spcr	This work
GBS119	OI3545 amyE5720::PmcpA-mcpB[M1–N397] mcpA[E397–E661]; Spcr	This work
GBS118	OI3545 amyE5720::PmcpA-mcpB[M1–I433] mcpA[Q433–E661]; Spcr	This work
GBS208	OI3545 amyE5720::PmcpA-mcpA[M1–Q358] mcpB[D359–E662]; Spcr	This work

TABLE 3.

Plasmids used in this study

Plasmid	Description	Source or reference
pET28a(+)	His-tagged cloning vector for protein purification; Kanr	Novagen
pAIN750	B. subtilis empty vector for integration at amyE; Ampr Spcr	96
pAIN750mcpA	pAIN750::mcpA	55
pAIN750mcpB	pAIN750::mcpB	55
pPT059	pAIN750::mcpA[M1–L44] mcpB[D45–T267] mcpA[M267–E661]	This work
pPT086	pAIN750::mcpA[M1–Q358] mcpB[D359–E662]	23
pPT061	pAIN750::mcpB[M1–Q359] mcpA[D359–E661]	This work
pGB43	pAIN750::mcpB[M1–N397] mcpA[E397–E661]	23
pGB34	pAIN750::mcpB[M1–I433] mcpA[Q433–E661]	23
pGB131	6×His-N-terminal McpB expression plasmid, pET28(a)::mcpBC	This work
pGB130	6×His-N-terminal McpA expression plasmid, pET28(a)::mcpAC	This work
pPT021	6×His-C-terminal McpA expression plasmid, pET28(a)::mcpAS	This work
pGB46	6×His-N-terminal HemAT expression plasmid, pET28(a)::hemATC	23
pSP03	6×His-N-terminal HemAT expression plasmid, pET28(a)::hemATS	23

Receptor chimeras were constructed according to a procedure described previously (23). Briefly, the region outside the fusion points was amplified by whole-plasmid PCR with the pAIN750mcpA plasmid as the template. The desired fragments from mcpB were amplified from pAIN750mcpB with a short overlap on both ends. The PCR products were purified after gel extraction, assembled using Gibson assembly (81), and transformed into E. coli. After sequence verification, the correct plasmids were linearized at the XhoI restriction site, religated, and transformed into B. subtilis OI3545 (receptorless mutant) using the two-step Spizizen method (82). Colonies were screened for spectinomycin resistance, and integration into the amyE locus was verified using gram iodine solution (0.33% iodine, 0.66% potassium iodide, and 1% sodium bicarbonate) on starch plates. Correct clones were unable to hydrolyze starch and form clear zones.

For the construction of chimeric receptors under the control of the mcpA promoter, whole-plasmid PCR was used to amplify the pAIN750mcpA plasmid excluding the mcpA region to be swapped by the desired region from another receptor. The desired region from another receptor was also PCR amplified with overlapping primers, and both DNA fragments were purified and assembled using Gibson assembly. Point mutations were performed as described previously (23).

Cloning for recombinant protein production was performed in E. coli BL21(DE3). The DNA fragments corresponding to the signaling domains of McpA (residues 359 to 661) and McpB (residues 359 to 662) were PCR amplified and cloned in frame with an N-terminal His₆ tag in the pET28a(+) plasmid at the NheI restriction site using Gibson assembly. The McpA sensing domain (residues 21 to 278) was cloned in frame with a C-terminal His₆ tag in the pET28a(+) plasmid between the XhoI and NcoI restriction sites. The McpC sensing domain (residues 33 to 278) was cloned in frame with an N-terminal His₆ tag in pET28a(+) at the NdeI restriction site using Gibson assembly. Assembled plasmids were transformed into E. coli. After isolation and sequence verification, all plasmids were transformed into the E. coli BL21(DE3) strain for protein expression and purification.

Capillary assay for chemotaxis.

The chemical-in-pond modification of the capillary assay was used for quantifying repellent chemotaxis responses (20). Briefly, B. subtilis strains were grown for 16 h at 30°C on TBAB plates. The cells were scraped from the plates and resuspended to an optical density at 600 nm (OD₆₀₀) of 0.03 in 5 mL CAMM supplemented with 50 μg/mL histidine, 50 μg/mL methionine, 50 μg/mL tryptophan, 20 mM sorbitol, and 2% tryptone broth (TB 1% tryptone, 0.5% NaCl). The cultures were then grown to an OD₆₀₀ of 0.4 to 0.45 at 37°C with shaking at 250 rpm, after which 50 μL of GL solution (5% [vol/vol] glycerol and 0.5 M sodium lactate) was added, and the cells were incubated for another 15 min. The cells were then washed twice with chemotaxis buffer and incubated for an additional 25 min at 37°C with shaking at 250 rpm to ensure that the cells were motile. Cells were then diluted to an OD₆₀₀ of 0.01 in chemotaxis buffer containing the appropriate concentrations of repellents. The culture was shaken for 10 min at room temperature at 150 rpm and then aliquoted into 0.3-mL ponds on a slide warmer at 37°C. Closed-end capillary tubes filled with chemotaxis buffer were inserted into the ponds. After 1 h, cells in the capillaries were harvested, transferred to 3 mL of top agar (1% tryptone, 0.8% NaCl, 0.8% agar, and 0.5 mM EDTA), and plated onto TB agar (TB and 1.5% agar) plates. These plates were incubated for 16 h at 37°C, and colonies were counted. Experiments were performed in triplicate each day and repeated on three different days. Attractant assays for aromatic compounds were carried out at an OD₆₀₀ of 0.001 for 1 h. For testing the functionality of the mutant strains expressing chimeric receptors, attractant assays were performed for 30 min with cells diluted to an OD₆₀₀ of 0.001 in the pond and an asparagine solution (3.16 μM) in the capillary. For measuring the chemotaxis response to external acidic pH, capillaries filled with chemotaxis buffer at pH 7.0 were inserted into the pond containing cells at an OD₆₀₀ of 0.001 preadapted to pH 8.0 in chemotaxis buffer, and cells in the capillaries were harvested after 1 h and counted as described above (22).

Protein purification.

E. coli BL21(DE3) cells harboring the His₆-tagged expression plasmids were grown in a 2-L flask containing LB medium supplemented with 30 μg/mL kanamycin at 37°C with shaking at 250 rpm until an A₆₀₀ of 0.7 was reached, at which point expression was induced by the addition of 1 mM isopropyl β-d-1-thiogalactopyranoside (IPTG). The cultures were grown for 12 h at 25°C and then harvested by centrifugation at 7,000 × g at 4°C for 10 min. Cells were resuspended in lysis buffer (50 mM NaH₂PO₄, 300 mM NaCl, 10 mM imidazole [pH 8]) and sonicated (5 10-s pulses). After centrifugation at 40,000 × g for 1 h, the supernatant was loaded onto a 5-mL GE Hi-Trap chelating column charged with 0.1 M NiSO₄ and binding buffer (50 mM NaH₂PO₄, 300 mM NaCl, 20 mM imidazole [pH 8]). The protein-bound column was then washed with 10 column volumes of binding buffer, and proteins were eluted with an imidazole gradient of 20 to 500 mM. The collected protein fractions were pooled, concentrated using an Amicon ultrafiltration cell (Millipore), and dialyzed into phosphate-buffered saline (PBS) (10 mM Na₂HPO₄, 1.8 mM KH₂PO₄, 137 mM NaCl, 2.7 mM KCl [pH 7.4]) at 4°C. HemAT domains were purified as described previously (23). Aliquots were stored at −80°C. The concentration of purified proteins was measured by the Pierce bicinchoninic acid (BCA) protein assay kit.

Saturation transfer difference nuclear magnetic resonance spectroscopy.

All nuclear magn3etic resonance (NMR) spectroscopy measurements were performed on a Varian VNMRS instrument at 750 MHz with a 5-mm Varian HCN probe at 298 K without sample spinning. Prior to measurements, protein samples were buffer exchanged into PBS (50 mM KH₂PO₄, 20 mM NaCl [pH 7.4]) in D₂O using Micro Bio-Spin columns with Bio-Gel P-6 (Bio-Rad Laboratories). Protein (final concentration, 20 μM) and ligand (final concentration, 2 mM) samples were added to the NMR tube in a 500-μL solution volume. For dissociation constant (K_d) measurements, ligand concentrations were varied from 0.5 mM to 4 mM. ¹H spectra were obtained from 32 scans with a 90° pulse and a 2-s relaxation delay. In saturation transfer difference NMR (STD-NMR) experiments, the protein samples were selectively saturated at 0.5 ppm with a train of Gaussian pulses of a 50-ms duration with a 0.1-ms delay and a 5-s relaxation delay for a total saturation time of 1 to 4 s and 256 scans. Off-resonance irradiation was applied at 30 ppm. A trim pulse of 50 ms was used to reduce protein background. All STD spectra were obtained by phase cycling after a block size of 8 to reduce artifacts resulting from temperature variation and magnet instability. Manual subtraction was performed to obtain the difference spectra. Control experiments were performed on samples containing only ligands without protein. All areas were calculated using MNova V14.1 (Mestrelab Chemistry Solutions) in stacked mode.

The STD amplification factor (STD-AF) is defined as the fractional saturation of a given proton multiplied by the excess of the ligand over the protein (37):

$$\text{STD-AF}=\left(\frac{I_0-I_{\mathit{sat}}}{I_0}\right)\frac{L}{P}$$

The method of initial slopes was used to determine the K_d (37, 83). The STD amplification factors at initial slopes (STD-AF₀) for each ligand concentration were calculated by plotting the STD-AF evolution with saturation times and fitting the equation

$${\text{STD-AF}}_t={\text{STD}}_{\text{max}}\left(1\hspace{0.17em}-\hspace{0.17em}{e}^{-k_{\mathit{sat}}t}\right)$$

The initial slope was then obtained by taking the derivative at time zero, yielding

$${\text{STD-AF}}_0={\text{STD}}_{\text{max}}{k}_{\mathit{sat}}$$

Plotting STD-AF₀ values against the increasing ligand concentrations (L) would yield a Langmuir hyperbolic dose‐response curve. Finally, the K_d was calculated by fitting the data to the following equation:

$${\text{STD-AF}}_0=\frac{\text{α}L}{L\hspace{0.17em}+\hspace{0.17em}{K}_d}$$

Phylogenetic tree and structural analyses of chemoreceptor domains.

B. subtilis (RefSeq accession number GCF_000009045.1) chemoreceptor sequences were obtained from the MiST 3.0 database (84). Tree analysis was carried out using the TREND platform (85). Protein sequences were first aligned with the online version of MAFFT (86), and the alignment result was then used as an input in the FastTree program (87) to construct the tree. Domain predictions were performed using HMMER3 (88).

In silico docking experiments.

The dimer structure of the McpA signaling domain (residues 352 to 661) was constructed using Modeller (v-9.23) (89). The homology model was based on the Thermotoga maritima Tm113 chemoreceptor (PDB accession number 2CH7) (90). Side chain conformations were refined using SCWRL4 (91), and the resulting protein structure was minimized using the YASARA server (92). Three-dimensional (3D) structures of the ligands were obtained from PubChem (93). Docking experiments were carried out using Autodock Vina (v-1.1.2) (94), in which the grid size was set to 58 by 50 by 62 points with 1-Å spacing surrounding residues 391 to 436, and exhaustiveness was set at 10. Clusters were visualized using Autodock Tools GUI (95).

Data availability.

Raw data are provided as Data Set S1 in the supplemental material.