Phenol sensing by Escherichia coli chemoreceptors: a nonclassical mechanism

The four transmembrane chemoreceptors of Escherichia coli sense phenol as either an attractant (Tar) or a repellent (Tap, Trg, and Tsr). In this study, we investigated the Tar determinants that mediate its attractant response to phenol and the Tsr determinants that mediate its repellent response to phenol. Tar molecules with lesions in the aspartate-binding pocket of the periplasmic domain, with a foreign periplasmic domain (from Tsr or from several Pseudomonas chemoreceptors), or lacking nearly the entire periplasmic domain still mediated attractant responses to phenol. Similarly, Tar molecules with the cytoplasmic methylation and kinase control domains of Tsr still sensed phenol as an attractant. Additional hybrid receptors with signaling elements from both Tar and Tsr indicated that the transmembrane (TM) helices and HAMP domain determined the sign of the phenol-sensing response. Several amino acid replacements in the HAMP domain of Tsr, particularly attractant-mimic signaling lesions at residue E248, converted Tsr to an attractant sensor of phenol. These findings suggest that phenol may elicit chemotactic responses by diffusing into the cytoplasmic membrane and perturbing the structural stability or position of the TM bundle helices, in conjunction with structural input from the HAMP domain. We conclude that behavioral responses to phenol, and perhaps to temperature, cytoplasmic pH, and glycerol, as well, occur through a general sensing mechanism in chemoreceptors that detects changes in the structural stability or dynamic behavior of a receptor signaling element. The structurally sensitive target for phenol is probably the TM bundle, but other behaviors could target other receptor elements.     

Repellent response functions of the Trg and Tap chemoreceptors of Escherichia coli.

The chemoreceptors responsible for the repellent response of Escherichia coli to phenol were investigated. In the absence of all four known methyl-accepting chemoreceptors (Tar, Tsr, Trg, and Tap), cells showed no response to phenol. However, when Trg, which mediates the attractant response to ribose and galactose, was introduced via a plasmid, the cells acquired a repellent response to phenol. About 1 mM phenol induced a clear repellent response; this response was suppressed by 1 mM ribose. Thus, Trg mediates the repellent response to phenol. Mutant Trg proteins with altered sensing for ribose and galactose showed a normal response to phenol, indicating that the interaction site for phenol differs from that for the ribose- and galactose-binding proteins. Tap, which mediates the attractant response to dipeptides, mediated a weaker repellent response to phenol. Tsr, which mediates the attractant response to serine, mediated an even weaker response to phenol. Trg and Tap were also found to function as intracellular pH sensors. Upon a pH decrease, Trg mediated an attractant response, whereas Tap mediated a repellent response. These results indicate that all the receptors in E. coli have dual functions, mediating both attractant and repellent responses.     

Oxygen as attractant and repellent in bacterial chemotaxis.

Studies of bacterial chemotaxis to oxygen (aerotaxis) over a broad range of oxygen concentrations showed that at high concentrations, oxygen was a repellent of Salmonella typhimurium, Escherichia coli, and some bacilli, whereas it is known that at lower concentrations (less than or equal to 0.25 mM dissolved oxygen), oxygen is an attractant. In a temporal assay of aerotaxis, S. typhimurium in medium equilibrated with air (0.25 mM dissolved oxygen) and then exposed to pure oxygen (1.2 mM) tumbled continuously for approximately 20 s. The oxygen concentration that elicited a half-maximal negative (repellent) response was 1.0 mM for both S. typhimurium and E. coli. The receptor for the negative chemoresponse to high concentrations of oxygen is apparently different from the receptor for the positive chemoresponse to low concentrations of oxygen, since the oxygen concentration that elicits a half-maximal positive (attractant) response in S. typhimurium and E. coli is reported to be 0.7 microM. Adaptation to high concentrations of oxygen, like adaptation to low concentrations of oxygen, was independent of methylation of a transducer protein. Only the response to low oxygen concentrations, however, was altered by interaction with the amidated Tsr transducer in cheB mutants.     

Glycerol elicits energy taxis of Escherichia coli and Salmonella typhimurium.

Escherichia coli and Salmonella typhimurium show positive chemotaxis to glycerol, a chemical previously reported to be a repellent for E. coli. The threshold of the attractant response in both species was 10(-6) M glycerol. Glycerol chemotaxis was energy dependent and coincident with an increase in membrane potential. Metabolism of glycerol was required for chemotaxis, and when lactate was present to maintain energy production in the absence of glycerol, the increases in membrane potential and chemotactic response upon addition of glycerol were abolished. Methylation of a chemotaxis receptor was not required for positive glycerol chemotaxis in E. coli or S. typhimurium but is involved in the negative chemotaxis of E. coli to high concentrations of glycerol. We propose that positive chemotaxis to glycerol in E. coli and S. typhimurium is an example of energy taxis mediated via a signal transduction pathway that responds to changes in the cellular energy level.     

Ligand-specific activation of Escherichia coli chemoreceptor transmethylation

Adaptation in the chemosensory pathways of bacteria like Escherichia coli is mediated by the enzyme-catalyzed methylation (and demethylation) of glutamate residues in the signaling domains of methyl-accepting chemotaxis proteins (MCPs). MCPs can be methylated in trans, where the methyltransferase (CheR) molecule catalyzing methyl group transfer is tethered to the C terminus of a neighboring receptor. Here, it was shown that E. coli cells exhibited adaptation to attractant stimuli mediated through either engineered or naturally occurring MCPs that were unable to tether CheR as long as another MCP capable of tethering CheR was also present, e.g., either the full-length aspartate or serine receptor (Tar or Tsr). Methylation of isolated membrane samples in which engineered tethering and substrate receptors were coexpressed demonstrated that the truncated substrate receptors (trTsr) were efficiently methylated in the presence of tethering receptors (Tar with methylation sites blocked) relative to samples in which none of the MCPs had tethering sites. The effects of ligand binding on methylation were investigated, and an increase in rate was produced only with serine (the ligand specific for the substrate receptor trTsr); no significant change in rate was produced by aspartate (the ligand specific for the tethering receptor Tar). Although the overall efficiency of methylation was lower, receptor-specific effects were also observed in trTar- and trTsr-containing samples, where neither Tar nor Tsr possessed the CheR binding site at the C terminus. Altogether, the results are consistent with a ligand-induced conformational change that is limited to the methylated receptor dimer and does not spread to adjacent receptor dimers.     

Role of threonine residue 154 in ligand recognition of the tar chemoreceptor in Escherichia coli.

The Tar chemoreceptor of Escherichia coli mediates attractant responses to aspartate, maltose, and phenol, repellent responses to Ni2+ and Co2+, and thermoresponses. To understand the role of threonine residue 154, which is located in the ligand-binding domain of Tar, we replaced the residue with serine, isoleucine, and proline by site-directed mutagenesis. The replacements caused reductions in aspartate sensing but had only a small effect on maltose sensing and almost no effect on phenol sensing, repellent sensing, and thermosensing. These results indicate that Thr-154 of Tar is rather specifically involved in aspartate sensing. The reductions in the response threshold for aspartate by the replacements with serine, isoleucine, and proline were less than 1, about 2, and more than 5 orders of magnitude, respectively. When the corresponding threonine residue in the Tsr chemoreceptor was replaced with the same amino acids, roughly similar reductions in the response threshold for serine resulted. Thus, these threonine residues seem to have a common role in detecting the aspartate and serine attractant families. A mechanism by which these chemoreceptors detect the amino acid attractants is discussed.     

Inversion of thermosensing property of the bacterial receptor Tar by mutations in the second transmembrane region

The aspartate chemoreceptor Tar of Escherichia coli serves as a warm sensor that produces attractant and repellent signals upon increases and decreases in temperature, respectively. However, increased levels of methylation of the cytoplasmic domain of Tar resulting from aspartate binding convert Tar to a cold sensor with the opposite signaling behavior. Detailed analyses of the methylation sites, which are located in two separate alpha-helices (MH1 and MH2), have suggested that intra- and/or intersubunit interactions of MH1 and MH2 play a critical role in thermosensing. These interactions may be influenced by binding of aspartate, which could trigger some displacement of MH1 through the second transmembrane region (TM2). As an initial step toward understanding the role of TM2 in thermosensing, we have examined the thermosensing properties of 43 mutant Tar receptors with randomized TM2 sequences (residues 190-210). Among them, we identified one mutant receptor (Tar-I2) that functioned as a cold sensor in the absence of aspartate. This is the first example of attractant-independent inversion of thermosensing in Tar. Further analyses identified the minimal essential divergence from the wild-type Tar sequence (Q191V-W192R-Q193C) required for the inverted response. Thus, displacements of TM2 seem to influence the thermosensing function of Tar.     

Identification of methylation sites and effects of phototaxis stimuli on transducer methylation in Halobacterium salinarum.

The two transducers in the phototaxis system of the archaeon Halobacterium salinarum, HtrI and HtrII, are methyl-accepting proteins homologous to the chemotaxis transducers in eubacteria. Consensus sequences predict three glutamate pairs containing potential methylation sites in HtrI and one in HtrII. Mutagenic substitution of an alanine pair for one of these, Glu265-Glu266, in HtrI and for the homologous Glu513-Glu514 in HtrII eliminated methylation of these two transducers, as demonstrated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis autofluorography. Photostimulation of the repellent receptor sensory rhodopsin II (SRII) induced reversible demethylation of HtrII, while no detectable change in the extent of methylation of HtrI was observed in response to stimulation of its cognate sensory rhodopsin, the attractant receptor SRI. Cells containing HtrI or HtrII with all consensus sites replaced by alanine still exhibited phototaxis responses and behavioral adaptation, and methanol release assays showed that methyl group turnover was still induced in response to photostimulation of SRI or SRII. By pulse-chase experiments with in vivo L-[methyl-(3)H]methionine-labeled cells, we found that repetitive photostimulation of SRI complexed with wild-type (or nonmethylatable) HtrI induced methyl group turnover in transducers other than HtrI to the same extent as in wild-type HtrI. Both attractant and repellent stimuli cause a transient increase in the turnover rate of methyl groups in wild-type H. salinarum cells. This result is unlike that obtained with Escherichia coli, in which attractant stimuli decrease and repellent stimuli increase turnover rate, and is similar to that obtained with Bacillus subtilis, which also shows turnover rate increases regardless of the nature of the stimulus. We found that a CheY deletion mutant of H. salinarum exhibited the E. coli-like asymmetric pattern, as has recently also been observed in B. subtilis. Further, we demonstrate that the CheY-dependent feedback effect does not require the stimulated transducer to be methylatable and operates globally on other transducers present in the cell.     

The lifetime of photosensory signals in Halobacterium halobium and its dependence on protein methylation

Halobacteria spontaneously reverse their swimming direction about every 10 s. This behavioral pattern is transiently disturbed upon stimulation through sensory photosystems of different spectral sensitivity. As a result of stimulation, a single swimming interval is either prolonged (attractant response) or shortened (repellent response). Thereafter the cell returns to its autonomous reversal rhythm, i.e., it quickly adapts. Method are presented to determine the lifetime of repellent as well as of attractant cellular signals at the site of signal integration, using particular stimulation programs. Independent of the photosystem through which the signals were generated, the total lifetime of a repellent signal was 1.3 s. The decay of the signal was rapid during the first 100 ms and slow thereafter. The lifetime of an attractant signal was about 4 s and likewise did not depend on the photosystems. The degree of methylation of membrane proteins was increased by attractant stimuli and decreased by repellent stimuli. Inhibition of protein methylation by homocysteine was accompanied by a slowdown of the decay of both the repellent and attractant signal. A mutant strain with an increased demethylation also gave increased signal lifetimes. A lowered Ca2+ concentration, which activates methylation in vivo, led to shortened signal lifetimes. Methylation is proposed to be the mechanism which limits the signal lifetime and thereby allows the cells to adapt.     

Adaptational assistance in clusters of bacterial chemoreceptors

Sensory adaptation of low-abundance chemoreceptors in Escherichia coli requires assistance from high-abundance receptors, because only high-abundance receptors carry the carboxyl-terminal pentapeptide sequence NWETF that enhances adaptational covalent modification. Using membrane vesicles containing both high-abundance receptor Tar and low-abundance receptor Trg, we observed effective assistance in vitro for all three adaptational modifications: methylation, demethylation and deamidation. These results demonstrated that adaptational assistance involves not only the previously documented assistance for methylation but also assistance for the two CheB-catalysed reactions. We determined rates of assisted methylation and demethylation at many ratios of assisting to assisted receptor. Analysis by a model of assistance indicated one Tar dimer could assist seven Trg dimers in methylation or five in demethylation, defining assistance neighbourhoods. These neighbourhoods were larger than a trimer of homodimers, required only receptors and were minimally affected by formation of signalling complexes. Time courses of assisted Trg methylation in membranes with low amounts of Tar showed that assisting receptors did not diffuse beyond initial neighbourhoods for at least two hours. Taken together, these observations indicate that chemoreceptors can form stable neighbourhoods larger than trimers in the absence of other chemotaxis proteins. Such interactions are likely to occur in natural receptor clusters in vivo.     

A model of excitation and adaptation in bacterial chemotaxis.

We present a model of the chemotactic mechanism of Escherichia coli that exhibits both initial excitation and eventual complete adaptation to any and all levels of stimulus ("exact" adaptation). In setting up the reaction network, we use only known interactions and experimentally determined cytosolic concentrations. Whenever possible, rate coefficients are first assigned experimentally measured values; second, we permit some variation in these rate coefficients by using a multiple-well optimization technique and incremental adjustment to obtain values that are sufficient to engender initial response to stimuli (excitation) and an eventual return of behavior to baseline (adaptation). The predictions of the model are similar to the observed behavior of wild-type bacteria in regard to the time scale of excitation in the presence of both attractant and repellent. The model predicts a weaker response to attractant than that observed experimentally, and the time scale of adaptation does not depend as strongly upon stimulant concentration as does that for wild-type bacteria. The mechanism responsible for long-term adaptation is local rather than global: on addition of a repellent or attractant, the receptor types not sensitive to that attractant or repellent do not change their average methylation level in the long term, although transient changes do occur. By carrying out a phenomenological simulation of bacterial chemotaxis, we find that the model is insufficiently sensitive to effect taxis in a gradient of attractant. However, by arbitrarily increasing the sensitivity of the motor to the tumble effector (phosphorylated CheY), we can obtain chemotactic behavior.     

Failure of sensory adaptation in bacterial mutants that are defective in a protein methylation reaction.

Chemotactic bacteria, such as E. coli, detect changes in the chemical composition of the environment. Addition of an attractant or repellent leads to an immediate response, characterized by a change in the swimming behavior of the cells--a process known as sensory excitation. However, the response gradually disappears with time, despite the continued presence of the chemical--a process known as sensory adaptation. We report here the behavior of a class of nonchemotactic mutants (cheX) that can carry out sensory excitation but are defective in the process of sensory adaptation. These mutants are also defective in the ability to carry out a protein methylation reaction which has previously been implicated in the adaptation process (Goy, Springer and Adler, 1977). The results presented here establish a firm relationship between the methylation reaction and sensory adaptation.     

In vivo and in vitro chemotactic methylation in Bacillus subtilis

Two doublets of Bacillus subtilis membrane proteins with molecular weights of 69,000 and 71,000 and of 30,000 and 30,800, were labeled by C3H3 transfer in the absence of protein synthesis. In addition, there was intense methylation of several low-molecular-weight substances. Both doublets were missing in a chemotaxis mutant. The equivalent proteins in Escherichia coli and Salmonella typhimurium are believed to be the methyl-accepting chemotaxis proteins. The higher-molecular-weight doublet bands were increased in degree of methylation upon addition of attractant to the bacteria. A methyltransferase from B. subtilis that methylates the wild-type membrane significantly better than the mutant membrane, using S-adenosylmethionine, has been partly purified. The methylated product was alkali labile and is probably a gamma-glutamyl methyl ester, as in E. coli and S. typhimurium. Ca2+ ion inhibited the methyltransferase, with a Ki of about 80 nM. Analysis of the in vitro methylation product showed labeling of the 69,000-dalton methyl-accepting chemotaxis protein and a low-molecular-weight protein, using wild-type membrane. Labeling of the low-molecular-weight protein but not of the 69,000 dalton protein was observed when the mutant membrane was used. The chemotaxis mutant tumbled much longer than the wild type when diluted away from attractant.     

Evidence for an intermediate methyl-acceptor for chemotaxis in Bacillus subtilis

Bacillus subtilis responds to chemotactic attractants by demethylating certain membrane-bound proteins, termed methyl-accepting chemotaxis proteins (MCPs) and by augmenting the evolution of methanol. We propose that the methanol comes from a methylated intermediate rather than directly from the MCPs themselves. First, repellent blocks attractant-induced smooth swimming and methanol formation, but not MCP demethylation. Second, prior treatment of cells with much attractant to reduce radiolabeling of MCPs and increase that of the putative intermediate caused increased, rather than decreased, production of methanol upon addition and then removal of the repellent. Third, such cells also produced much, rather than little, methanol upon addition of less attractant than during the pretreatment. We speculate that unmethylated intermediate causes tumbling; attractant causes its methylation and hence absence of tumbling (smooth swimming). Its demethylation during the period of smooth swimming affords adaptation.     

CheB is required for behavioural responses to negative stimuli during chemotaxis in Bacillus subtilis

The methyl-accepting chemotaxis protein, McpB, is the sole receptor mediating asparagine chemotaxis in Bacillus subtilis. In this study, we show that wild-type B. subtilis cells contain approximately 2,000 copies of McpB per cell, that these receptors are localized polarly, and that titration of only a few receptors is sufficient to generate a detectable behavioural response. In contrast to the wild type, a cheB mutant was incapable of tumbling in response to decreasing concentrations of asparagine, but the cheB mutant was able to accumulate to low concentrations of asparagine in the capillary assay, as observed previously in response to azetidine-2-carboxylate. Furthermore, net demethylation of McpB is logarithmically dependent on asparagine concentration, with half-maximal demethylation of McpB occurring when only 3% of the receptors are titrated. Because the corresponding methanol production is exponentially dependent on attractant concentration, net methylation changes and increased turnover of methyl groups must occur on McpB at high concentrations of asparagine. Together, the data support the hypothesis that methylation changes occur on asparagine-bound McpB to enhance the dynamic range of the receptor complex and to enable the cell to respond to a negative stimulus, such as removal of asparagine.     

Receptor conformational changes enhance methylesterase activity during chemotaxis by Bacillus subtilis

Addition and removal of the attractant asparagine causes methanol formation as a consequence of methylation and demethylation of conserved glutamate residues in the Bacillus subtilis chemotaxis receptor McpB C-terminal domain. We found that methanol was released on both addition and removal of asparagine even when the response regulator domain of CheB was removed (to produce CheB(141-357)). Thus, in undergoing the transition from unbound receptor to ligand-bound adapted receptor, the receptor must pass through a state of heightened susceptibility to demethylation by CheB that is independent of phosphorylation. The same result occurred when the aspartate phosphorylation site of CheB, Asp54, had been mutated to an asparagine residue, provided the enzyme was sufficiently induced. However, no methanol release was observed for an active site point mutant, cheB(S173C), in response to addition or removal of asparagine even when induced. Finally, methanol release was observed only for attractant addition in a mutant background lacking the coupling proteins, CheW and CheV, provided CheB(141-357) was present. Thus, on attractant addition, methanol must arise from a transient conformation of the receptor C-terminal domain that is an intrinsic property of the receptor; on attractant removal, however, methanol must arise from a different transient conformation, one dependent on the presence of coupling proteins.     

Chemotactic Adaptation Is Altered by Changes in the Carboxy-Terminal Sequence Conserved among the Major Methyl-Accepting Chemoreceptors

In Escherichia coli and Salmonella typhimurium, methylation and demethylation of receptors are responsible for chemotactic adaptation and are catalyzed by the methyltransferase CheR and the methylesterase CheB, respectively. Among the chemoreceptors of these species, Tsr, Tar, and Tcp have a well-conserved carboxy-terminal motif (NWET/SF) that is absent in Trg and Tap. When they are expressed as sole chemoreceptors, Tsr, Tar, and Tcp support good adaptation, but Trg and Tap are poorly methylated and supported only weak adaptation. It was recently discovered that CheR binds to the NWETF sequence of Tsr in vitro. To examine the physiological significance of this binding, we characterized mutant receptors in which this pentapeptide sequence was altered. C-terminally-mutated Tar and Tcp expressed in a receptorless E. coli strain mediated responses to aspartate and citrate, respectively, but their adaptation abilities were severely impaired. Their expression levels and attractant-sensing abilities were similar to those of the wild-type receptors, but the methylation levels of the mutant receptors increased only slightly upon addition of attractants. When CheR was overproduced, both the adaptation and methylation profiles of the mutant Tar receptor became comparable to those of wild-type Tar. Furthermore, overproduction of CheR also enhanced adaptive methylation of wild-type Trg, which lacks the NWETF sequence, in the absence of any other chemoreceptor. These results suggest that the pentapeptide sequence facilitates effective adaptation and methylation by recruiting CheR.     

Tar-dependent and -independent pattern formation by Salmonella typhimurium.

When Salmonella typhimurium cells were allowed to swarm on either a minimal or complex semisolid medium, patterns of cell aggregates were formed (depending on the thickness of the medium). No patterns were observed with nonchemotactic mutants. The patterns in a minimal medium were not formed by a mutant in the aspartate receptor for chemotaxis (Tar) or by wild-type cells in the presence of alpha-methyl-D,L-aspartate (an aspartate analog), thus resembling the patterns observed earlier in Escherichia coli (E. O. Budrene and H. C. Berg, Nature [London] 349:630-633, 1991) and S. typhimurium (E. O. Budrene and H. C. Berg, Abstracts of Conference II on Bacterial Locomotion and Signal Transduction, 1993). Distinctively, the patterns in a complex medium had a different morphology and, more importantly, were Tar independent. Furthermore, mutations in any one of the genes encoding the methyl-accepting chemotaxis receptors (tsr, tar, trg, or tcp) did not prevent the pattern formation. Addition of saturating concentrations of the ligands of these receptors to wild-type cells did not prevent the pattern formation as well. A tar tsr tcp triple mutant also formed the patterns. Similar results (no negative effect on pattern formation) were obtained with a ptsI mutant (defective in chemotaxis mediated by the phosphoenolpyruvate-dependent carbohydrate:phosphotransferase system [PTS]) and with addition of mannitol (a PTS ligand) to wild-type cells. It therefore appears that at least two different pathways are involved in the patterns formed by S. typhimurium: Tar dependent and Tar independent. Like the Tar-dependent patterns observed by Budrene and Berg, the Tar-independent patterns could be triggered by H(2)O(2), suggesting that both pathways of pattern formation may be triggered by oxidative stress.     

Excitatory signaling in bacterial probed by caged chemoeffectors.

Chemotactic excitation responses to caged ligand photorelease of rapidly swimming bacteria that reverse (Vibrio alginolyticus) or tumble (Escherichia coli and Salmonella typhimurium) have been measured by computer. Mutants were used to assess the effects of abnormal motility behavior upon signal processing times and test feasibility of kinetic analyses of the signaling pathway in intact bacteria. N-1-(2-Nitrophenyl)ethoxycarbonyl-L-serine and 2-hydroxyphenyl 1-(2-nitrophenyl) ethyl phosphate were synthesized. These compounds are a 'caged' serine and a 'caged' proton and on flash photolysis release serine and protons and attractant and repellent ligands, respectively, for Tsr, the serine receptor. The product quantum yield for serine was 0.65 (+/- 0.05) and the rate of serine release was proportional to [H+] near-neutrality with a rate constant of 17 s-1 at pH 7.0 and 21 degrees C. The product quantum yield for protons was calculated to be 0.095 on 308-nm irradiation but 0.29 (+/- 0.02) on 300-350-nm irradiation, with proton release occurring at > 10(5) s-1. The pH jumps produced were estimated using pH indicators, the pH-dependent decay of the chromophoric aci-nitro intermediate and bioassays. Receptor deletion mutants did not respond to photorelease of the caged ligands. Population responses occurred without measurable latency. Response times increased with decreased stimulus strength. Physiological or genetic perturbation of motor rotation bias leading to increased tumbling reduced response sensitivity but did not affect response times. Exceptions were found. A CheR-CheB mutant strain had normal motility, but reduced response. A CheZ mutant had tumbly motility, reduced sensitivity, and increased response time to attractant, but a normal repellent response. These observations are consistent with current ideas that motor interactions with a single parameter, namely phosphorylated CheY protein, dictate motor response to both attractant and repellent stimuli. Inverse motility motor mutants with extreme rotation bias exhibited the greatest reduction in response sensitivity but, nevertheless, had normal attractant response times. This implies that control of CheY phosphate concentration rather than motor reactions limits responses to attractants.