Sensory Repertoire of Bacterial Chemoreceptors

SUMMARY

Chemoreceptors in bacteria detect a variety of signals and feed this information into chemosensory pathways that represent a major mode of signal transduction. The five chemoreceptors from Escherichia coli have served as traditional models in the study of this protein family. Genome analyses revealed that many bacteria contain much larger numbers of chemoreceptors with broader sensory capabilities. Chemoreceptors differ in topology, sensing mode, cellular location, and, above all, the type of ligand binding domain (LBD). Here, we highlight LBD diversity using well-established and emerging model organisms as well as genomic surveys. Nearly a hundred different types of protein domains that are found in chemoreceptor sequences are known or predicted LBDs, but only a few of them are ubiquitous. LBDs of the same class recognize different ligands, and conversely, the same ligand can be recognized by structurally different LBDs; however, recent studies began to reveal common characteristics in signal-LBD relationships. Although signals can stimulate chemoreceptors in a variety of different ways, diverse LBDs appear to employ a universal transmembrane signaling mechanism. Current and future studies aim to establish relationships between LBD types, the nature of signals that they recognize, and the mechanisms of signal recognition and transduction.

INTRODUCTION

Bacteria need to constantly adapt to changing environmental conditions to ensure survival. This is achieved through a variety of signal transduction pathways that most commonly include one-component systems (1), two-component systems (2), and chemoreceptor-based signaling cascades, also referred to as chemotaxis or chemosensory systems (3, 4). Chemosensory pathways mediate chemotaxis and type IV pilus-based motility and also regulate other cellular processes (5–7). Central to these systems is the ternary complex formed by chemoreceptors (also known as methyl-accepting chemotaxis proteins [MCPs] [8] and transducer-like proteins or Tlps [9]), the CheA histidine kinase, and the CheW coupling protein. Chemoreceptors recognize various signals and transmit this information to CheA, which ultimately controls the direction of flagellar motor rotation (reviewed in references 3, 8, and 10). Chemoreceptors contain two principal modules: input and output. The input module is usually composed of a single global domain, although in some chemoreceptors, two or more domains comprise the input (11). The output module is a conserved structure, a long dimeric four-helix bundle (4HB) composed of two symmetric antiparallel coiled coils, which comprises the cytoplasmic signaling domain (annotated the MA domain in the SMART database and the MCPsignal domain in the Pfam database) (12, 13). In transmembrane chemoreceptors that form the most common membrane topology class (11), input-output signaling is mediated by the HAMP (found in histidine kinases, adenylate cyclases, methyl-accepting chemotaxis proteins, and phosphatases) domain (14). Chemoreceptors can recognize the signal via their input module in several ways. One common mode of sensing is by the direct binding of chemoeffectors to the ligand binding input domain (LBD) (15, 16). Some input domains contain cofactors, such as heme or flavin adenine dinucleotide (FAD), which enable chemoreceptors to recognize oxygen and changes in redox status (17–19). Chemoreceptor input domains can also bind chemoeffector-loaded periplasmic binding proteins (20–22). Collectively, input domains are often referred to as sensory domains, although the term LBD is also used in a broader sense to depict the input module of chemoreceptors.

Escherichia coli and Salmonella enterica serovar Typhimurium are the classic model organisms to study chemoreceptor-based signaling, but a variety of other species were studied in this respect, primarily during the last decade (23). E. coli has five chemoreceptors that funnel stimuli into a single chemotaxis signaling cascade (8). Four of these receptors, Tar, Tsr, Trg, and Tap, contain a periplasmic 4HB LBD, whereas the signal input of the fifth receptor, Aer, occurs via a PAS (found in Per-Arnt-Sim proteins) sensory domain (24) located in the cytoplasm. Genome analyses of other bacteria have shown that many of them possess more complex chemosensory systems (7, 25, 26). First, many other bacteria possess a much larger number of chemoreceptors: on average, there are 14 chemoreceptor genes per genome (25), but in some species, more than 80 chemoreceptor genes were identified (27). The number of chemoreceptors was found to depend on bacterial lifestyle (28). For example, bacteria that are able to inhabit multiple and variable environments encode approximately five times more chemoreceptors than do those that live in a specific ecological niche (25). Second, many bacterial genomes encode multiple chemosensory pathways (7, 26), which is in marked contrast to the single pathway in E. coli. In this respect, Rhodobacter sphaeroides and Pseudomonas aeruginosa, with three (4) and four (5, 29, 30) chemosensory pathways, respectively, emerged as model organisms to study complex chemosensory networks.

The availability of sequenced genomes from many different chemotactic species revealed that the chemoreceptor family exhibits substantial divergence. While the cytoplasmic signaling domain is conserved in all chemoreceptors (13), there are large differences at the levels of protein topology and LBD type. Although a transmembrane topology with an extracellular LBD is predominant, some chemoreceptors either are entirely cytosolic, lack a LBD, or are membrane anchored and possess an LBD with a cytosolic location (11, 31). A computational analysis performed in 2010 showed that the large majority of chemoreceptors possess an LBD, but its type could be predicted for only a small fraction of them (25).

A central question in the field is how to identify chemoreceptor function, which ultimately depends on which chemoeffector binds to its LBD. However, the different families of LBDs are characterized by enormous sequence divergence (32, 33). This and the fact that chemoeffector specificity can be different in LBDs with similar sequences (34, 35) have largely hampered the functional annotation of chemoreceptors in genomic data sets by extrapolation from experimentally studied homologs in model organisms. Experimental approaches were thus necessary to elucidate chemoreceptor function. In this context, high-throughput ligand screening assays using individual, recombinantly produced LBDs (36–38) become a highly efficient strategy.

Over the last several years, tremendous progress has been made in the field, resulting in (i) the functional characterization of different chemoreceptors in many bacterial species, (ii) the availability of several chemoreceptor LBD structures, and (iii) the development of improved bioinformatics tools enabling computational identification of the LBD type. Here, we review these new developments that allowed us to assess the structural and functional diversity of the chemoreceptor sensory repertoire in greater depth and detail.

CHEMORECEPTORS IN MODEL ORGANISMS

E. coli and S. Typhimurium are classical model organisms to study chemoreceptors. Here, we briefly review the knowledge on these models and then focus on a few other selected organisms for which information about chemoreceptors and their functions became available in recent years, namely, Bacillus subtilis, Helicobacter pylori, Campylobacter jejuni, Pseudomonas aeruginosa, Pseudomonas putida, and Pseudomonas fluorescens. Due to space constraints, it is impossible to review all the information about chemoreceptors in many other species that are subjects of experimental investigation, such as Azospirillum brasilense (39), Comamonas testosteroni (40), Myxococcus xanthus (41), Rhodobacter sphaeroides (42), Sinorhizobium meliloti (43), Vibrio cholerae (44), and others.

E. coli and S. Typhimurium

The enteric bacteria E. coli and S. Typhimurium are ubiquitous colonizers of the intestines of animals. Motility is important for colonization by both commensal and pathogenic E. coli and S. Typhimurium strains (45, 46). Fundamental mechanisms of chemosensory signaling were determined by using these models (reviewed in references 3 and 8). E. coli exhibits chemotaxis toward and away from many chemoeffectors (attractants and repellents) (47–49), and responses to amino acids, sugars, dipeptides, pyrimidine bases, neurotransmitters, phenol, quorum sensing signals, substrates of the phosphotransferase system (PTS), pH, and temperature have been studied in detail (50–60).

E. coli has five chemoreceptors, four of which (Tar, Tsr, Trg, and Tap) have similar domain architectures and contain the 4HB domain as a sensory module (Fig. 1 and Table 1). Tsr mediates attractant responses to serine (61) and quorum autoinducer 2 (AI-2) (50) as well as responses to oxygen, redox, and oxidizable substrates (18, 19, 62). Serine binds directly to the Tsr LBD (16), whereas AI-2 taxis requires the periplasmic AI-2 binding protein LsrB (50). Tsr was also shown to govern taxis to 3,4-dihydroxymandelic acid, a metabolite of norepinephrine that is produced by human cells (51). Tar mediates attractant responses to aspartate (61) through direct binding to its LBD (63) and to maltose via binding to a maltose binding protein (57, 64). Tar also governs negative responses to metal ions by an as-yet-unidentified mechanism (65). Trg is a chemoreceptor for attractant responses to ribose and galactose (66) that are mediated via its interaction with ribose and galactose binding proteins, respectively. Tap is stimulated by a dipeptide-loaded periplasmic protein (54) and by pyrimidines (52), causing attractant responses in both cases. In contrast to these four receptors, the Aer chemoreceptor contains a cytosolic, redox-sensitive, FAD-containing PAS domain (Fig. 1). The Aer chemoreceptor mediates responses to oxygen as well as energy taxis (18, 19, 62). Many pathogenic E. coli strains lack either Trg, Tap, or both (67).

FIG 1.

The chemoreceptor repertoires of E. coli, S. Typhimurium, and B. subtilis. Receptor topology, names, and locus tags of E. coli K-12, S. Typhimurium strain LT2, and Bacillus subtilis subsp. subtilis strain 168 are shown. The LBDs are colored and their type is annotated according to data in the Pfam database. Orthologous groups of chemoreceptors (including paralogs) are highlighted by blue shading.

TABLE 1.

Functionally characterized chemoreceptors of E. coli, S. Typhimurium, and B. subtilis

Locus tag (chemoreceptor name)	LBD type	Effector(s)	Binding mode(s)	Reference(s)
E. coli
b4355 (Tsr)	4HB	Serine	Direct	50, 61
		AI-2	Indirect	50
		Redox substrates, aerotaxis	Unknown	19, 62
		3,4-Dihydroxymandelic acid	Direct	51
b1886 (Tar)	4HB	Aspartate	Direct	61, 63
		Maltose	Indirect	57, 64
		Metal ions	Unknown	65
b1421 (Trg)	4HB	Ribose, galactose	Indirect	66
b1885 (Tap)	4HB	Dipeptides	Indirect	54
		Pyrimidines	Unknown	52
b3072 (Aer)	PAS	Aerotaxis, energy	Unknown	18, 19, 62
S. Typhimurium
STM1919 (Tar)	4HB	Cysteine, aspartate	Direct	70, 208
STM4533 (Tsr)	4HB	Cysteine	Unknown	70
STM3577 (Tcp)	4HB	Citrate	Direct	209
		Phenol	Unknown
STM3152 (McpB)	4HB	Cystine	Unknown	70
STM3216 (McpC)	4HB	Cystine	Unknown	70
B. subtilis
BSU10380 (HemAT)	Protoglobin	Aerotaxis	Unknown	17
BSU13950 (McpC)	dCache	12 amino acids	Direct	88, 89
		PTS substrates	Indirect	194, 195
BSU31240 (McpA)	dCache	Glucose, α-methylglucoside	Unknown	9
BSU31260 (McpB)	dCache	4 amino acids	Direct	9

S. Typhimurium LT2 has nine chemoreceptors (Fig. 1 and Table 1), four of which (Tar, Tsr, Trg, and Aer) are orthologs of the corresponding E. coli proteins (68). It lacks the Tap chemoreceptor but has five chemoreceptors that are not present in E. coli. Tcp mediates attractant responses to citrate and repellent responses to phenol (69). McpB and McpC appear to serve as chemoreceptors for repellent responses to cysteine (70). The function of two other chemoreceptors, Tip (71) and McpA (72), is still unknown. At the C-terminal part, McpA contains a CZB (chemoreceptor zinc binding) domain. This domain was first identified in the cytoplasmic chemoreceptor TlpD from Helicobacter pylori (see below) (73). TlpD mediates energy taxis (74) as well as chemotaxis to reactive oxygen species (ROS) (75) and pH (76), and the role of zinc binding in these processes remains to be established. As in the case of E. coli, a majority of chemoreceptors in S. Typhimurium contain a periplasmic 4HB domain (Fig. 1). More detailed information on the chemosensory mechanisms of E. coli and S. Typhimurium was reported previously (8, 10, 77).

Bacillus subtilis

B. subtilis is a member of the phylum Firmicutes, which is often found in soil but is also present in water or associated with plants or inhabits the gastrointestinal tract of ruminants and humans (78). It is an obligate aerobe, which, however, is capable of surviving under anaerobic conditions in the presence of nitrates or glucose. It is the best-characterized Gram-positive bacterium and the first nonenterobacterial model used to study chemotaxis (79). Initial studies of B. subtilis demonstrated that a galactose binding protein was essential for chemotaxis toward this sugar (80, 81), and subsequent studies revealed chemotaxis to many other sugars (82). B. subtilis shows repellent responses to different compounds, including uncouplers of oxidative phosphorylation (79, 83). Chemoattraction was observed for the 20 proteinogenic amino acids (84–86) and oxygen (17, 87).

B. subtilis has 10 chemoreceptors (Fig. 1 and Table 1). In contrast to E. coli and S. Typhimurium, it contains only one chemoreceptor with a 4HB domain, whereas five receptors contain a dCache_1 (double Cache 1) domain. This domain is composed of two α/β subdomains, each homologous to the PAS domain, and a long N-terminal helix, and it was recently shown to be the most ubiquitous extracellular LBD in prokaryotes (33). Two of the dCache_1 domain-containing chemoreceptors, McpB and McpC, mediate chemotaxis toward amino acids. McpC is a broad-range receptor that mediates chemotaxis to all amino acids except l-asparagine (88). Interestingly, McpB was identified as a chemoreceptor for this amino acid (9). McpB has thus evolved to recognize, with high specificity, the only l-amino acid that is not recognized by the broad-range McpC receptor.

Although McpC mediates chemotaxis toward 19 amino acids, only 12 of them bind to its recombinant LBD (89). Using pulldown experiments with the immobilized McpC-LBD, four amino acid binding proteins were identified, suggesting that, similarly to the Tar chemoreceptor in E. coli, McpC recognizes signals both directly by ligand binding and via interactions with ligand binding proteins (89).

Three genes that encode McpB paralogs, namely, TlpA, McpA, and TlpB, are located adjacent to the mcpB gene on the B. subtilis chromosome (9). McpA was shown to mediate chemotaxis toward glucose and α-methylglucoside, whereas chemoeffector screening of TlpA and TlpB mutants failed to provide any information as to receptor function (9). The TlpC chemoreceptor contains the sCache (single Cache) domain, and similarly to TlpA and TlpB, its function remains unknown.

Genome analysis revealed the presence of three other chemoreceptors in B. subtilis, YfmS, YoaH, and YvaQ, that contain 4HB, CHASE (cyclase/histidine kinase-associated sensory extracellular) (90, 91), and PAS domains as their sensory LBDs, respectively (Fig. 1). However, no information as to their function is available. The soluble HemAT receptor (Fig. 1) mediates chemotaxis toward oxygen. However, in contrast to E. coli, where aerotaxis is mediated by the PAS domain-containing Aer chemoreceptor (via redox sensing) and by the 4HB domain-containing Tsr chemoreceptor (via proton motive force sensing), HemAT governs aerotaxis via direct oxygen sensing by a protoglobulin domain, which contains a bound heme (17, 92).

Helicobacter pylori

H. pylori belongs to the class Epsilonproteobacteria. The elevated motility of this pathogen in dilute and viscous media is based on the action of a bundle of unipolar sheathed flagella. It is a microaerophilic organism, and its main habitat is the upper gastrointestinal tract. Bacteria penetrate the gastric mucous layer and cause gastric ulcers, adenocarcinomas, and lymphomas. Its ability to colonize gastric mucosa depends largely on the capacity to produce urease and to neutralize the acidic pH of the stomach by breaking urea into ammonia and bicarbonate (93). Chemotaxis plays a fundamental role in the mechanism of infection (94).

Urea and bicarbonate emanate from the human gastric epithelium, and chemotaxis of H. pylori to both compounds was observed (95–97). Other detected attractants include mucin (98), various amino acids (96, 99), and cholesterol (100), whereas repellent responses were observed for salts (96), AI-2 (101), and metal ions (75, 102). In addition, other behavioral responses, for example, pH (102, 103), ROS (75), and energy taxis (74), were also described. All four chemoreceptors of H. pylori (Fig. 2 and Table 2) were found to play a role in the infection process: mutation of TlpA and TlpC reduced stomach colonization (104), whereas TlpD was required for the proliferation of H. pylori in the gastric antrum, the lowermost part of the stomach (105), or throughout the stomach in different animal models (105, 106). TlpB was shown to be important for persistent colonization in vivo (97, 103), and mutation of TlpB resulted in decreased inflammation (107, 108).

FIG 2.

The chemoreceptor repertoires of H. pylori and C. jejuni. Receptor topology and locus tags of H. pylori 26695 and C. jejuni subsp. jejuni NCTC 11168 are shown. Pfam names for all LBDs are shown, except for 4HB, which is predicted to be a divergent four-helix bundle domain. Domains that are not detected by the Pfam tool HMMER but are recognized by the more sensitive HHpred tool are indicated as empty rectangles. Orthologous groups of chemoreceptors (including paralogs) are highlighted by blue shading.

TABLE 2.

Functionally characterized chemoreceptors of H. pylori and C. jejuni

Locus tag (chemoreceptor name)	LBD type	Effector(s)	Binding mode	Reference(s)
H. pylori
HP0099 (TlpA)	dCache	Arginine, bicarbonate	Unknown	99
HP0103 (TlpB)	sCache	pH	Unknown	103
		AI-2	Indirect	110
		Urea, hydroxyurea, formamide, acetamide	Direct	97, 109
HP0599 (TlpD)	CZB	Energy	Unknown	74
		Hydrogen peroxide, paraquat	Unknown	75
C. jejuni
Cj0262c (DocC)	dCache	Deoxycholate	Unknown	117
Cj1506c (CcaA)	dCache	Aspartate	Direct	116
Cj1564 (CcmL)	dCache	3 amino acids, succinate	Direct	38
		Malate, fumarate, purine, thiamine, Ile		179
Cj1189c (CetB)	PAS	Energy	Unknown	121
Cj1190c (CetA)	None	Energy	Unknown	121
Cj1110c (CetZ)	PAS	Energy	Unknown	125
Cj0951c (Tlp7)	None	Formate	Unknown	124
Cj0952c	Unknown	Formate	Unknown	124
Tlp11a	dCache	Galactose	Direct	119

^aFound in only a few isolates.

TlpB is an intriguing chemoreceptor (Fig. 2), because it mediates repellent responses to acidic pH (103, 109) and to the AI-2 quorum sensing signal (101) as well as attractant responses to compounds released by epithelial tissues, such as urea (97). Urea binds to TlpB with high specificity and affinity (97, 109). Strains containing TlpB with single-amino-acid substitutions that prevented urea binding showed responses to AI-2 but lost the pH taxis phenotype (109). A model in which bound urea, acting as a cofactor, modulates the properties of a pH-sensing aspartate residue was proposed. The structure of the urea-containing TlpB LBD has been solved, and it was initially classified as a PAS domain (109). However, a recent study revealed that all extracellular “PAS-like” domains belong to the Cache domain superfamily (33). Accordingly, the TlpB LBD is now classified as an sCache domain. The TlpB LBD does not bind AI-2 directly: AI-2 chemotaxis depends on two periplasmic ligand binding proteins, and consequently, an indirect binding mechanism was proposed (110).

Two other chemoreceptors, TlpA and TlpC, possess dCache domains (Fig. 2). TlpA was identified as a chemoreceptor for arginine and bicarbonate because a corresponding mutant lost chemotaxis to these compounds (111). However, the mode by which these chemoeffectors stimulate the receptor remains unknown. The signals that are recognized by the TlpA paralog TlpC are currently unknown; however, indirect evidence suggests that TlpC can modulate the TlpB-mediated response to acid by an as-yet-unknown mechanism (102).

The fourth receptor in H. pylori, TlpD, is soluble and is composed of an MCP signaling domain and a CZB domain (73) (Fig. 2). TlpD was proposed to serve as an energy taxis receptor that mediates repellent responses away from conditions of reduced electron transport and away from electron transport inhibitors (74). A recent report showed that TlpD mediates a repellent response to agents that promote oxidative stress, such as hydrogen peroxide and paraquat (75). This chemoreceptor appears to function independently of the three transmembrane chemoreceptors, because it forms an autonomous polar signaling complex (75). TlpD localization is affected by metabolic activity (catalase deficiency and lower-energy conditions), which provides clues to its potential mechanism of action (112).

Campylobacter jejuni

Campylobacter species are commensal microorganisms present in the gastrointestinal tracts of different animals and are transmitted to humans primarily by the consumption of infected food. C. jejuni is one of the main causative agents of gastroenteritis but can also lead to more serious diseases such as Guillain-Barre syndrome, Miller Fisher syndrome, and arthritis (113). Mutants with deletions of various chemoreceptor genes showed up to a 10-fold decrease in the ability of bacteria to invade host cells (114). C. jejuni shows attractant responses to amino and organic acids, sugars, and nucleotides (38, 114–119), whereas lysine, arginine, glucosamine, thiamine, and bile constituents act as repellents (38, 115).

The majority of completely sequenced C. jejuni genomes contain 10 chemoreceptor genes, as exemplified by the widely used C. jejuni subsp. jejuni laboratory strain NCTC 11168 (Fig. 2 and Table 2). However, the number of chemoreceptor genes per C. jejuni genome can vary from 5 to 11. Of the 10 typical C. jejuni chemoreceptors (named consecutively Tlp1 to Tlp10 [120]), 5 appear to detect signals in the periplasm, and another 5 appear to detect signals in the cytoplasm. Four of the five transmembrane receptors contain a dCache_1 domain as a single sensory module (Fig. 2). Signal specificity was established for two of them: Tlp1 (renamed CcaA) was identified as an aspartate chemoreceptor (116), and Tlp3 (renamed CcmL) was identified as a chemoreceptor for multiple ligands, which directly binds isoleucine, lysine, arginine, glucosamine, succinate, malate, fumarate, purine, and thiamine (38). CcmL and another chemoreceptor with a similar domain architecture, Tlp4 (renamed DocC), appear to mediate chemotaxis to bile and its major component deoxycholate (117).

C. jejuni also has an unusual bipartite chemoreceptor encoded by two separate genes, cetA and cetB, that are located adjacent to each other (121). Two proteins encoded by these genes work in tandem by forming a fully functional chemoreceptor. The CetA/CetB pair is the first studied example of a “split-gene” chemoreceptor. CetA (or Tlp9) is composed of a chemoreceptor signaling domain and a HAMP domain and is anchored to the membrane by two transmembrane helices (Fig. 2), whereas CetB is a stand-alone PAS domain. The cetA and cetB genes are cotranscribed, and both proteins are membrane associated and present in protein complexes (122). This bipartite chemoreceptor system was shown to govern energy taxis (121), and follow-up studies showed that both components are homologous to energy taxis chemoreceptors in other species, where signal input is achieved through FAD-containing PAS domains, similar to a canonical Aer chemoreceptor in E. coli (39, 123).

More recently, a second bipartite chemoreceptor system was described for C. jejuni (124). Similar to the CetA/CetB pair, Tlp7 is a stand-alone signaling domain that interacts with a protein encoded by the cj0952c gene, which consists of a periplasmic domain flanked by two transmembrane regions and a HAMP domain. This bipartite chemoreceptor system appears to mediate the attractant response to formate and is important for invasion of host cells (124). The soluble CetZ chemoreceptor (Tlp8) has dual PAS domains and was recently shown to counteract the CetA/B system in guiding C. jejuni cells toward optimal energy resources (125).

Roles of other chemoreceptors in C. jejuni have not been established yet; however, their domain composition provides hints as to their putative functions. For example, the soluble Tlp5 chemoreceptor contains a FIST (found in F-box and intracellular signal transduction proteins) domain, a widely distributed sensory module which was suggested to bind primarily amino acids (126). Another soluble chemoreceptor, Tlp6, has a C-terminal CZB domain and is orthologous to the H. pylori TlpD chemoreceptor (73) that mediates a repellent response to conditions that promote oxidative stress (75).

In a recent study, the chemoreceptor Tlp11 was identified, which is present in only a few genomes of invasive C. jejuni species. This receptor binds galactose via its periplasmic dCache_1 domain and mediates positive chemotaxis toward this sugar (119).

Pseudomonas

Figure 3 summarizes the chemoreceptor repertoire of three different species that belong to the genus Pseudomonas, a member of the Gammaproteobacteria. P. aeruginosa is an opportunistic pathogen of humans, animals, and plants and is also present in soil and water. Strain PAO1 is the common reference strain and corresponds to a spontaneous chloramphenicol-resistant mutant of the original PAO strain, which was isolated from a human wound (127). P. putida KT2440 is a Tol plasmid-cured derivative of the mt-2 strain. This strain was isolated from soil, is nutritionally versatile, and has a saprophytic lifestyle (128). P. fluorescens strain Pf0-1 was isolated during a study of bacterial fitness in soil (129). This strain efficiently colonizes plant roots, and chemotaxis was found to be essential for this process (130, 131). In general, Pseudomonas strains have extraordinary metabolic versatility and a large number of chemoreceptors (25). The three model strains analyzed here contain 26 chemoreceptors (PAO1), 27 chemoreceptors (KT2440), and 37 chemoreceptors (Pfl0-1). The chemoeffector repertoire of pseudomonads is very diverse and includes organic and amino acids, aromatic hydrocarbons, sugars, fatty acids, peptides, bivalent metal ions, inorganic anions, herbicides, morphine, as well as purine and pyrimidine bases (132).

FIG 3.

The chemoreceptor repertoires of three different Pseudomonas strains. Receptor topology and locus tags of P. aeruginosa PAO1, P. putida KT2440, and P. fluorescens Pf0-1 are shown. The LBDs are colored, and their type is annotated. Orthologous groups of chemoreceptors (including paralogs) are highlighted by blue shading. The question mark indicates a domain of an unknown type.

P. aeruginosa.

Many bacteria possess multiple chemosensory pathways, and P. aeruginosa emerged as an important model to study this property. Two of the P. aeruginosa pathways, Che1 and Che2, control flagellum-mediated taxis (30, 133). The Wsp pathway modulates c-di-GMP levels (5), and the Chp pathway is responsible for type IV pilus-mediated motility (29, 134) and for regulating cAMP levels (135).

The P. aeruginosa PAO1 chemoreceptors that mediate chemotaxis to amino acids and inorganic phosphate (P_i) are the best-studied chemoreceptors (Table 3). This strain responds to all 20 proteinogenic amino acids (136), and no amino acid taxis was observed in the PctA/PctB/PctC (PA4309/PA4310/PA4307) triple mutant (Fig. 3) (137). Each of these three paralogous chemoreceptors contains a dCache domain as a single sensory module. Similar to B. subtilis (see above), P. aeruginosa has one broad-spectrum receptor, PctA, which binds most of the amino acids (137, 138). PctB binds several of the PctA ligands but has a strong preference for glutamine, which is one of the two amino acids not recognized by PctA. The signal input in PctA and PctB (represented by the binding constants of individual ligands) correlates with the magnitude of its output, as determined by fluorescence resonance energy transfer (FRET) analyses of a PctA-Tar chimera (139). PctC binds with modest affinity to two of the amino acids that are also recognized by PctA. This receptor, however, has evolved to bind with very high affinity to gamma-aminobutyrate (GABA) and mediates chemotactic responses to low GABA concentrations (34, 138). PctA, PctB, and PctC were also identified as chemoreceptors for trichloroethylene (TCE) (140), but this compound is not recognized directly by their LBDs (138).

TABLE 3.

Functionally characterized chemoreceptors of different pseudomonads

Locus tag (chemoreceptor name[s])	LBD type	Effector(s)	Binding mode	Reference(s)
P. aeruginosa
PA0176 (Aer2, McpB)	PAS	O2, NO, CO, cyanide	Direct	148, 153, 154, 210
PA0180 (CttP, McpA)	None	Chloroethylenes	Unknown	140
PA0411 (PilJ)	PilJ	Phosphatidylethanolamine	Unknown	158
PA1561 (Aer, TlpC)	PAS	Aerotaxis	Unknown	148, 210
PA2561 (CtpH)	4HB	Inorganic phosphate	Direct	141
PA2652 (CtpM)	sCache	Malate	Unknown	145
PA2654 (TlpQ)	dCache	Ethylene	Unknown	146
PA3708 (WspA)	4HB	Growth on solid surfaces	Unknown	211
		Ethanol	Unknown	212
PA4307 (PctC)	dCache	GABA, histidine, proline	Direct	34, 136–138
		Chloroethylenes	Unknown	213
PA4309 (PctA)	dCache	17 amino acids	Direct	136–138
		Chloroethylenes	Unknown	213
		Chloroform	Unknown
		Methylthiocyanate	Unknown
PA4310 (PctB)	dCache	5 amino acids	Direct	136–138
		Chloroethylenes	Unknown	213
PA4844 (CtpL)	HBM	Inorganic phosphate	Indirect	141, 214
		Chloroaniline, catechol	Unknown
PA5072 (McpK)	HBM	α-Ketoglutarate	Direct	147
P. putida
PP0320 (McpH)	dCache	Metabolizable purines	Direct	162
PP1228 (McpU)	dCache	Polyamines	Direct	163
PP1371 (McpG)	dCache	GABA	Direct	34
PP2111 (Aer2)	PAS	Aerotaxis, methylphenols	Unknown	169
		Phenylacetic acid	Unknown	215
PP2249 (McpA)	dCache	12 amino acids	Direct	163
PP2257 (Aer1)	PAS	Aerotaxis	Unknown	216
PP2861 (McpP)	sCache	Acetate, pyruvate, propionate, l-lactate	Direct	168
PP4658 (McpS)	HBM	TCA intermediates, acetate, butyrate	Direct	164
PP5020 (McpQ)	HBM	Citrate, citrate/Mg2+	Direct	167
Pput_0623 (McpC)a	dCache	Nicotinic acid, cytosine	Unknown	172, 173
Pput_2149 (PcaY)a	4HB	Cyclic carboxylic acids	Unknown	174
Pput_4520 (McfS)a	HBM	Malate	Unknown	175
Pput_4894 (McfQ)a	HBM	Citrate, fumarate	Unknown	175
Pput_0339 (McfR)a	4HB	Succinate, malate, fumarate	Unknown	175
P. fluorescens
Pfl01_0124 (CtaB)	dCache	Amino acids	Unknown	131
Pfl01_0354 (CtaC)	dCache	Amino acids	Unknown	131
Pfl01_0728 (McpS)	HBM	Malate, succinate, fumarate	Unknown	130
Pfl01_3768 (McpT)	sCache	Malate, succinate	Unknown	130
Pfl01_4431 (CtaA)	dCache	Amino acids	Unknown	131
NbaYb	sCache	Nitrobenzoate	Unknown	176

^aP. putida F1.

^bP. fluorescens KU-7.

Two chemoreceptors in P. aeruginosa respond to changes in P_i concentrations: CtpL (PA4844) detects low P_i concentrations, and CtpH (PA2561) detects high P_i concentrations (141). However, these chemoreceptors differ in their sensory modules: CtpL is predicted to possess an HBM (helical bimodular) domain (142), whereas CtpH has a 4HB domain (Fig. 3). A recent study showed that P_i binds directly to CtpH but not to CtpL (22). CtpL is activated by binding the P_i-loaded PstS periplasmic binding protein that also provides P_i to the Pst transporter (143). Indirect chemoreceptor activation by periplasmic binding proteins was previously shown to mediate chemotaxis to sugars, dipeptides, and AI-2 in E. coli (50, 54, 57, 64, 144). The discovery of a similar mechanism in a different bacterial family, involving a different chemoreceptor type and chemoeffector, suggests that such systems are widespread.

Other chemoreceptors with identified specificity include the malate-specific sCache domain-containing PA2652 chemoreceptor (145), the dCache domain-containing TlpQ (PA2654) receptor for the plant hormone ethylene (146), and the α-ketoglutarate-specific McpK (PA5072) chemoreceptor that harbors an HBM domain (147). The membrane-bound McpA (CttP or PA0180) chemoreceptor mediates positive chemotaxis to TCE. McpA has no LBD (Fig. 3), and the mechanism of its action is unknown.

P. aeruginosa has two chemoreceptors, namely, PA1561 (Aer/TlpC) and PA0176 (Aer-2/McpB/TlpG), which were suggested to mediate aerotaxis (148). While Aer is homologous to that of E. coli, Aer-2/McpB has a very unusual domain architecture. This soluble receptor is composed of three HAMP domains followed by a PAS domain, two HAMP domains, and the signaling domain (149). Similar to the high-abundance chemoreceptors Tar and Tsr in E. coli, Aer-2/McpB is one of the two P. aeruginosa chemoreceptors that contains a pentapeptide tethered via a linker to its C-terminal end. It has been shown that CheR2, and not any other CheR homolog, binds exclusively to this pentapeptide and methylates Aer-2/McpB (150). It was concluded that the presence of this pentapeptide permits the targeting of a specific chemoreceptor with a dedicated CheR homolog, which may be one of the molecular mechanisms to ensure the specificity of the interaction of signaling proteins within a given pathway. This receptor may be associated with pathogenicity, because the mutation of its cognate CheB2 methylesterase causes a dramatic reduction in virulence (151). The Aer-2/McpB PAS domain contains heme, which enables this chemoreceptor to recognize O₂, NO, CO, and cyanide. A recent study revealed that oxygen binds to the heme-containing PAS domain with a K_D (equilibrium dissociation constant) of 16 μM, which is comparable to the affinities of other PAS-heme O₂ sensors (152). This and the fact that amino acid substitutions in the gas binding site affected primarily the binding of O₂ and not of other gases led to the proposition that O₂ is the natural ligand for this chemoreceptor (152). Comparison of the structures containing ferric, ligand-free heme (153) and cyanide-bound ferric heme (154) also led to a model of receptor function (153). Changes in the heme ligation state cause conformational changes in the PAS domain that shift the PAS monomer-dimer equilibrium or alternatively cause a rearrangement of the PAS dimer. The exact role of Aer-2/McpB remains unclear: the aer-2 (mcpB) mutant shows an aerotaxis phenotype comparable to that of the wild type (133).

BdlA (PA1423) is a soluble chemoreceptor, which is composed of two PAS domains followed by the signaling domain (Fig. 3). This specific domain architecture defines a widely spread chemoreceptor subfamily, several members of which were shown to monitor changes in the redox status of the electron transport system (39). Initial studies showed that the BdlA mutant was deficient in biofilm dispersion and had increased adherence properties and increased intracellular levels of cyclic di-GMP (155). BdlA employs an unorthodox mechanism. The full-length protein appears to be inactive. However, during growth in biofilms, but not during planktonic growth, an amino acid in the segment linking the PAS domains with the signaling domain is phosphorylated by an as-yet-unidentified kinase, and this step permits the proteolytic cleavage of the N-terminal PAS domain. It was proposed that the resulting truncated BdlA chemoreceptor forms efficient signaling complexes (156).

The PilJ chemoreceptor (PA0411) contains dual PilJ domains in its N terminus, and it feeds into the Chp pathway controlling type IV pilus-mediated motility and cAMP levels (29, 135, 157). The pilJ mutant showed a defect in twitching motility toward phosphatidylethanolamine (PE), and it was proposed that PE may be sensed by this receptor (158). PilJ mutant cells have shortened pili, suggesting that PilJ is required for full pilus assembly and/or extension (159). This agrees with data from another study showing that the Chp pathway modulates cAMP levels, which in turn impacts type IV pilus production (135). A more recent study showed that PilA, a subunit that assembles to form type IV pili, directly interacts with PilJ, most likely via its periplasmic LBD (160). The authors of that study suggested that PilJ may act as a mechanosensor that can detect conformational changes induced in stretched type IV pili.

P. putida.

P. putida KT2440 has three chemosensory pathways (161) orthologous to the Che1, Wsp, and Chp pathways in P. aeruginosa; it lacks the Che2 pathway, which is associated with virulence in P. aeruginosa. Eight of its 27 chemoreceptors possess a dCache LBD (Fig. 3). Four of these chemoreceptors, all belonging to the same paralogous group, have been functionally annotated (Table 3). PP0320 (McpH) mediates chemotaxis toward metabolizable purine derivatives such as adenine and guanine, but it does not recognize nonmetabolizable derivatives such as theobromine and theophylline (162). PP1371 (McpG) is a GABA-specific chemoreceptor, and its dissociation constant for this ligand (K_D = 175 nM) is one of the highest ever observed for a chemoreceptor. GABA is present in plant root exudates, and bacterial root colonization is delayed in an McpG mutant (34). PP1228 (McpU) was identified as a chemoreceptor that specifically binds three polyamines, putrescine, cadaverine, and spermidine; PP2249 (McpA) responds to 12 different proteinogenic amino acids (163). The common feature of the above-mentioned chemoreceptors is that they recognize their ligands directly and that all attractants are amines that support bacterial growth as sole N sources.

McpS (PP4658) mediates chemotaxis to six tricarboxylic acid (TCA) cycle intermediates, butyrate, and acetate (164, 165), and it is the first characterized chemoreceptor with an HBM domain (142). This domain is composed of 6 helices that are arranged into two stacked helical modules (166). Each module contains a ligand binding site; malate and succinate bind to the membrane-proximal bundle, whereas acetate binds to the membrane-distal bundle. It has been demonstrated that ligand binding to each module causes chemotaxis and that signals are additive (166). In a superimposition of both modules, by translation and 180° rotation, both ligand binding sites overlap. In addition, the Tar LBD can be closely superimposed onto the individual McpS LBD modules, and the Tar aspartate binding site overlaps the binding sites at the McpS modules. McpS thus has a bimodular LBD, and the binding of different ligands to the individual modules was proposed to be a mechanism to fine-tune chemotactic responses (166). The dCache domain is also characterized by a bimodular arrangement, but it remains to be elucidated whether ligands also bind to both modules.

Despite the facts that citrate is abundantly present in plant tissues and root exudates and that it serves as a preferred carbon source, McpS binds citrate with low affinity and does not bind citrate-metal ion complexes (164). The McpS paralog PP5020 (McpQ) was found to specifically bind citrate, either in its free form or in complex with metal ions (167). PP2861 (McpP) is a receptor specific for C-2 and C-3 carboxylic acids and recognizes its ligands directly via the sCache domain (168). In contrast to P. aeruginosa, aerotaxis in P. putida is likely mediated by three paralogous chemoreceptors that all have the exact domain architecture of E. coli Aer, namely, PP2257 (Aer-1), PP2111 (Aer-2), and PP4521 (Aer-3) (169) (Fig. 3).

A characteristic feature of many P. putida strains is that they are able to degrade various aromatic compounds, and many of them serve as chemoattractants (132). So far, two chemoreceptors have been unambiguously identified as mediators of aromatic hydrocarbon taxis: McpT of P. putida DOT-T1E, for taxis to many different mono- and biaromatic hydrocarbons (170), and NahY of P. putida G7, for taxis to naphthalene (171). In contrast to the very large majority of chemoreceptors, McpT and NahY are encoded on plasmids. Another common feature is that both receptors possess 4HB LBDs.

Sensory specificity of several other chemoreceptors was identified in P. putida strain F1. McpC (orthologous to PP0584) (Fig. 3) was identified as a chemoreceptor for cytosine (172) and nicotinic acid (173). Another study led to the identification of a chemoreceptor, termed PcaY, which is an ortholog of PP2643 (Fig. 3); it responds to different aromatic acids such as benzoate and its derivatives (174). McpS of P. putida KT2440 is the dominant chemoreceptor for TCA cycle intermediates, and only very minor responses are observed for the mcpS mutant (164). In contrast, mutation of the McpS ortholog McfS in P. putida F1 (99.5% sequence identity) had almost no effect on chemotaxis toward TCA cycle intermediates (175). Two other chemoreceptors for TCA cycle intermediates, McfQ and McfR, were identified in P. putida F1, and there is evidence for additional, as-yet-unidentified chemoreceptors for TCA cycle intermediates (175). These studies highlight unique evolutionary paths for chemoreceptors, where, in contrast to the general rule, orthologs can respond to different spectra of chemoeffectors, while the same chemoeffector might be recognized by nonhomologous LBDs (e.g., the HBM domain in McfS/McfQ and the 4HB domain in McfR).

P. fluorescens.

Five of the 37 chemoreceptors in P. fluorescens Pf0-1 have been functionally characterized. The response to amino acids was modulated primarily by three chemoreceptors that are orthologous to P. aeruginosa PctA, PctB, and PctC. Consequently, as their P. aeruginosa counterparts, these receptors, Pfl01_4431 (CtaA), Pfl01_0124 (CtaB), and Pfl01_0354 (CtaC) (Fig. 3), contain a dCache domain as a single dedicated sensory module. While CtaA and CtaB are broad-range receptors (each one responds to 16 amino acids), CtaC has a relatively narrow ligand profile and mediates taxis to only 5 amino acids (Cys, Arg, Gly, Met, and Thr) (131).

Chemoreceptors that mediate a strong attractant response to the TCA cycle intermediates succinate, malate, and fumarate in P. fluorescens Pf0-1 have been identified. Both Pfl01_0728 (McpS) and Pfl01_3768 (McpT) respond to malate and succinate, whereas McpS also responds to fumarate in addition to these two compounds (130). The sCache domain-containing chemoreceptor McpT is orthologous to the malate-specific chemoreceptor PA2652 of P. aeruginosa, whereas the HBM-containing chemoreceptor McpS is an ortholog of P. aeruginosa PA5072. Thus, as in the case of P. putida F1, chemotaxis of P. fluorescens to organic acids is mediated by nonhomologous chemoreceptors with different LBDs.

A study of P. fluorescens strain KU-7, which is able to metabolize 2-nitrobenzoate, led to the identification of the NbaY chemoreceptor (176). The nbaY gene was found on the plasmid next to the genes encoding the two initial enzymes of the 2-nitrobenzoate degradation pathway. The proximity of genes involved in degradation and chemotaxis strongly suggests that there is a functional link; indeed, the wild-type strain shows positive chemotaxis to 2-nitrobenzoate, and NahY was identified as a receptor for this response.

The majority of chemoreceptors in Pseudomonas remain functionally uncharacterized. Furthermore, no known LBD type was detected (even when using the most sensitive profile-profile similarity searches) in the P. aeruginosa chemoreceptor PA2867 and its orthologs in two other species (Fig. 3). However, transcript levels of the P. putida KT2440 ortholog (PP2310) were highest among all chemoreceptors in this strain (177), and mutation of the corresponding gene increased biofilm formation (163).

CHEMORECEPTOR LBD DIVERSITY AND ABUNDANCE

Following the analysis of the chemoreceptor repertoire of model organisms, we revisited the issue of chemoreceptor LBD diversity and abundance on the genomic scale. The latest analysis was reported in 2010 (7), and the amount of available genomic information has grown dramatically in recent years. To address this issue, we analyzed the domain architectures of all sequences matching the chemoreceptor signaling domain (MCPsignal; Pfam accession number PF00015), which is the accepted criterion for the chemoreceptor definition (13), that are currently available in the Pfam 31.0 database and its underlying UniProt reference proteome database (178) as of March 2017. This search retrieved 26,530 sequences, 10,658 of which contained either no LBD or a putative LBD that cannot be matched to any known domain model in Pfam. The remaining 15,872 sequences contain at least one known LBD. This set of sequences was used to determine LBD diversity and abundance (Table 4 and Fig. 4).

TABLE 4.

Genomic diversity and abundance of chemoreceptor ligand binding domains^a

Domain superfamilyb	LBD type	Domain model	No. of domains
4HB_MCP	Single 4HB	4HB_MCP_1	3,998
		TarH	861
		CHASE3	392
	Double 4HB	HBM	231
Cache-like	Double Cache	dCache_1	3,515
		Cache_3-Cache_2	307
		dCache_3	202
		dCache_2	183
		Ykul_C	2
	Single Cache	sCache_2	1,165
		sCache_3_2	60
		sCache_3_3	368
		Diacid_rec	40
		CHASE4	19
		CHASE8	7
PAS	PAS	PAS_3	1,930
		PAS_9	682
		PAS_4	459
		PAS_8	97
		PAS	111
		PAS_7	8
		PAS_6	1
		PAS_10	1
GAF	GAF	GAF	463
		GAF_2	74
Protoglobin	Protoglobin	Protoglobin	552
		Globin	8
		Bac_globin	2
PBP	Periplasmic binding protein	Phosphonate-bd	30
		SBP_bac_5	29
		SBP_bac_3	23
		SBP_bac_8	7
		Peripla_BP_4	16
		Peripla_BP_5	2
		Peripla_BP_3	1
		Peripla_BP_6	1
		OpuAC	5
		DctP	2
		NMT1	2
HNOX-like	Heme and NO binding	HNOB	75
4Fe-4S	Iron-sulfur cluster binding	Fer4	14
		Fe4_10	20
		Fer4_7	6
		Fer4_9	4
		Fer4_6	1
		FeS	33
GPCR_A	GPCR-like	7TMR-DISM_7TM	16
NADP_Rossman	NAD binding	GFO_IDH_MocA	9
		Semialdhyde_dh	1
Gx_transp	Transporter	5TM-5TMR_LYT	9
Beta_propeller	Extracellular binding	Reg_prop	8
GBD	Galactose binding	7TMR-DISMED2	6
		CBM_4_9	2
TPR	Protein-protein interactions	TPR_19	4
		ANAPC3	1
Cupin	Nucleotide binding	cNMP_binding	3
HHH	Helix turn helix	HHH_5	2
E-set	Sugar binding	Y_Y_Y	2
Phosphatase	Cyanide binding	Rhodanese	2
P-loop_NTPase	Nucleotide binding	ABC_tran	1
2heme_cytochrom	Heme binding	Ni_hydr_CYTB	1
ATP-grasp	Glutathione binding	GSH-S_ATP	1
Flavodoxin	FMN/FAD binding	Falvodoxin_2	1
No assigned superfamily	Zinc binding	CZB	582
	PilJ	PilJ	326
	Nitrate and nitrite binding	NIT	318
	c-di-GMP binding	PilZ	137
	Oxygen binding	Hemerythrin	98
		PocR	91
	Domain of unknown function	DUF3365	83
	Amino acid binding	FIST	62
	Hydrogen binding	Fe_hyd_lg_C	45
	Integral membrane sensor	MHYT	28
	Adenosyl group binding	CBS	21
	Ammonium transporter	Ammonium_trans	16
	Ligand binding	PrpR_N	9
	Domain of unknown function	DUF4077	5
	Integral membrane domain	MASE3	5
	Quorum sensing	AI-2E_transport	1
	Nucleoside binding	Gate	1
	Phosphate/sugar binding	PTS_EIIC	1

^aData from Pfam 31.0 and the underlying UniProt reference proteome databases (178) as of March 2017. A total of 26,530 sequences were retrieved by using the MCPsignal domain model. A total of 17,907 LBDs matching Pfam domain models were identified. GPCR, G-protein-coupled receptor; GBD, galactose-binding domain.

^bPfam clan.

FIG 4.

Relative abundances of different LBD types in chemoreceptors. The analysis includes sequences matching the chemoreceptor signaling domain (MCPsignal; Pfam accession number PF00015) that were available in the Pfam 31.0 database and its underlying UniProt reference proteome database (178) as of March 2017. See Table 4 for details.

More than 80 different LBD types were found in this data set (Table 4), a number which is at least three times larger than the previously reported number (7); however, their distribution is uneven. Only a few LBD types are abundant, whereas the majority of them are found infrequently. The 4HB domain (defined by Pfam models 4HB_MCP_1 and TarH) is the most abundant LBD in chemoreceptors (Table 4 and Fig. 4 and 5). This domain was previously identified as a ubiquitous extracytoplasmic sensory module found in all major signal transduction families of bacteria and archaea (32). Ligands that are directly recognized by representatives of this domain family are shown in Fig. 6. This domain is best studied for the E. coli chemoreceptors Tar and Tsr, and the mechanism of ligand binding and signal transduction is well understood (reviewed in reference 8).

FIG 5.

Structural diversity of LBDs. Structures of different LBD types that are found in chemoreceptors are shown. Bound ligands are shown in red. Different chain colors indicate that the domain was experimentally shown to be dimeric. Domain definitions were obtained from Pfam. PDB accession numbers are shown in parentheses.

FIG 6.

Diversity of ligands recognized by the major classes of chemoreceptor LBDs. The following ligands for which direct binding was observed are shown (along with references for the corresponding evidence): E. coli Tar (Ec-Tar) (63), Ec-Tsr (50), C. testosteroni MCP2901 (Ct-MCP2901) (203), Ct-MCP2983 (204), Ct-MCP2201 (40), P. aeruginosa CtpH (Pa-CtpH) (22, 141), P. putida McpS (Pp-McpS) (164), Pp-McpQ (167), H. pylori TlpB (Hp-TlpB) (97, 109), Pp-McpP (168), B. subtilis McpC (Bs-McpC) (89), Bs-McpB (9, 180), C. jejuni CcaA (Cj-CcaA) (116), Cj-CcmL (38, 179), Pa-PctA (138), Pa-PctB (138), Pp-McpA (163), V. cholerae Mlp37 (Vc-Mlp37) (44), Vc-Mlp24 (205), S. meliloti McpU (Sm-McpU) (206), Cj-Tlp11 (119), Pa-PctC (138), Pp-McpG (34), Sm-McpX (207), Pp-McpU (163), and Pp-McpH (162).

The dCache_1 domain is the second most abundant LBD (Table 4). The dCache domains are composed of two PAS-like modules and a long N-terminal α-helix (Fig. 5). The different subfamilies of dCache domains likely originated from sCache domains that contain only one PAS-like module (33). dCache is one of the two characterized LBDs that have a bimodular arrangement. In all structural and functional studies of dCache domains conducted to date, chemoeffectors were shown to bind to the membrane-distal module (44, 89, 138, 179, 180), and no physiologically relevant ligand interacting with the membrane-proximal module has been identified. The latter module might be involved in signal transmission to the membrane. Different ligands, primarily amines, bind to dCache domains directly (Fig. 6). In contrast to dCache, chemoeffectors bind to both modules of another characterized LBD with a bimodular arrangement, the HBM domain (142). For example, acetate binds to the distal module of the HBM domain in the McpS chemoreceptor of P. putida, whereas malate and succinate bind to the membrane-proximal module (166). Furthermore, ligand binding at both modules is agonistic (166).

PAS domain-containing chemoreceptors represent the third most abundant subfamily (Fig. 4). Although PAS and sCache domains share a similar fold, they are defined and recognized by different sequence signatures. Similar to 4HB and Cache domains, PAS domains are also omnipresent in bacterial signal transduction systems, but in contrast to the 4HB and Cache domains, they are also widely distributed in eukaryotes (24, 181). This domain is found almost exclusively in the cytosol (24, 33), and many family members contain bound heme, FAD, or flavin mononucleotide (FMN) and are involved in oxygen and redox sensing, which govern aerotaxis and related behavioral responses.

The three most abundant LBD superfamilies (4HB, Cache, and PAS) account for more than 80% of chemoreceptors with known LBDs, whereas the CZB, GAF (found in cGMP-specific phosphodiesterases, adenylyl cyclases, and FhlA), protoglobin, NIT (nitrate and nitrite sensing), and PilJ LBDs account for 2 to 3% each (Fig. 4 and Table 4). Representatives of more than 30 other LBD superfamilies and families account for only 4% of all chemoreceptors with known LBDs (Fig. 4 and Table 4). The CZB domain-containing chemoreceptor TlpD mediates energy, ROS, and pH taxis in H. pylori (73, 75, 76), whereas the protoglobin domain of the HemAT chemoreceptor in the archaeon Halobacterium salinarum and in B. subtilis contains a bound heme and mediates taxis to oxygen (17). While chemoreceptors containing representatives of other LBD families have never been studied, putative functions for many of them can be predicted by using comparative genomics. For example, known ligands for GAF domains include cyclic nucleotides (182), whereas the NIT domain was predicted to bind nitrates and nitrites (183). The FIST domain is hypothesized to be associated with amino acid sensing (126), the PilZ domain binds the second messenger c-di-GMP (184), the PocR domain was proposed to be involved in sensing simple hydrocarbon derivatives (185), the HNOB (heme-NO-binding) domain is predicted to function as a heme-dependent sensor for gaseous ligands (186), and TPR (tetratricopeptide repeat) domains typically bind other proteins (187). Many putative LBDs appear to be derivatives of enzymes and transporters with a known or predicted specificity (Table 4). Their lower abundance suggests that many such chemoreceptors evolved recently in some lineages, helping bacteria to adapt to specific environments.

The vast majority of LBDs that have been studied are located in the N-terminal region of chemoreceptors; however, some chemoreceptors, including those from model organisms (Fig. 1 to 3), contain LBDs as their C-terminal domains. Genomic analysis suggests that approximately 5% of LBDs in microbial chemoreceptors are located at the C terminus, and the CZB domain is the most abundant among them. Some of the chemoreceptor-associated LBDs are always located at the very C terminus, for example, the PilZ and hemerythrin domains. Some LBD types are predominantly located at the C terminus but occasionally can be found in the N terminus, for example, domains that belong to the periplasmic binding protein (PBP) superfamily. Finally, many LBD types that are usually located at the N terminus occasionally can be found in the C terminus, for example, domains from the Cache and PAS superfamilies. The molecular mechanism of signal transduction by C-terminally located LBDs remains to be established, but it likely involves a direct interaction with the HAMP domain and/or the signaling domain.

Approximately 14% of bacterial chemoreceptors lack transmembrane regions and are predicted to be located in the cytosol (31). A characteristic feature of this group is the predominant presence of PAS domains that are found in nearly one-half of the cytosolic chemoreceptors (31). In contrast, transmembrane chemoreceptors predominantly contain 4HB and Cache domains as extracellular LBDs.

Current structural information about different LBD types is summarized in Fig. 5, revealing two dominant folds: antiparallel α-helix bundles (4HB, HBM, PilJ, CHASE, and NIT domains) and PAS-like α/β-folds (domains that belong to the PAS, GAF, and Cache superfamilies).

DIVERSITY OF SENSING MECHANISMS AND WAYS TO STIMULATE A CHEMORECEPTOR

The chemoreceptor structural unit is a homodimer. Chemoreceptor homodimers interact to form trimers of dimers that are functional units of signal transduction, and oligomerization appears to be mediated primarily by the signaling domain of chemoreceptors (188). Comprehensive studies of the sensory mechanism were carried out on E. coli chemoreceptors that contain the 4HB LBD. It was shown that aspartate binds only to the dimeric state of the recombinant Tar LBD with a stoichiometry of 1 molecule per dimer and very strong negative cooperativity. Ligand binding was found to stabilize the dimer and to shift the monomer-dimer equilibrium to the dimeric state (189, 190). The analysis of the Tar LBD structure provided the basis for this binding mechanism. The aspartate binding sites are located at the dimer interface, and amino acids from both monomers of the dimer establish contacts with the bound ligand (Fig. 7A), causing dimer stabilization (63).

FIG 7.

Different sensing mechanisms. (A to C) Sensing mechanism in which ligand binding induces LBD dimerization. (D and E) Sensing mechanism in which ligand binding does not induce LBD dimerization. (A and B) Zoomed-in images of the binding pockets of the Tar LBD (4HB) with bound Asp (PDB accession number 4Z9H) (A) and the McpS LBD (HBM) with malate (PDB accession number 2YFA) (B). The two monomers of the dimer are colored differently. In both cases, the binding site is at the dimer interface, and amino acids from both monomers are involved in ligand binding. (C) Analytical ultracentrifugation studies of the LBD of the McpS homolog McpQ (HBM) in the absence and presence of its ligand citrate. (D) Binding pocket of the dCache_1 LBD of the CcmL receptor (PDB accession number 4XMR) containing bound Ile. Amino acids involved in Ile binding are from the same monomer. (E) Analytical ultracentrifugation data for the PctA LBD (dCache_1) in the absence and presence of Ala. Data were reported previously (63, 138, 166, 167, 179). c(s), sedimentation coefficient distribution; AU, absorbance units; S, Svedberg units.

More recent studies of bimodular HBM domains produced results that are very similar to what was observed for the 4HB domain. In the P. putida McpS LBD, malate binds with a stoichiometry of 1 molecule per dimer, and similar to the Tar LBD structure, the malate binding site in McpS LBD is at the dimer interface; residues from both monomers participate in binding (Fig. 7B). Analytical ultracentrifugation studies showed that the individual LBD of McpS is present in a monomer-dimer equilibrium and that ligand binding shifts this equilibrium entirely to the dimeric state (Fig. 7C) (164). Similar observations were made in studies of the individual LBDs of the McpQ (167) and McpK (147) chemoreceptors. However, new structural information that became available for other LBD types suggests that this binding mode may not be universal. Structures of sCache and dCache domains show that the ligand is buried within the binding cavity of a single protein monomer (Fig. 7D) (44, 109, 179). In agreement with this structural information, the dCache LBDs of the PctA (138) and TlpC (191) chemoreceptors were found to be monomeric, and in the case of the PctA LBD, a saturating ligand concentration did not alter its oligomeric state (Fig. 7E) (138). The data therefore suggest that there are different sensing mechanisms with regard to ligand-induced LBD dimerization.

Newly available data also show that there are different ways to stimulate chemoreceptors (Fig. 8). A canonical chemoreceptor such as E. coli Tar can be stimulated either by direct ligand binding or by complexing with a ligand binding protein. The example of the PilJ chemoreceptor suggests that stimulation can also be achieved by binding to other proteins such as PilA, the subunit that forms type IV pili (160). Studies of the mechanism underlying phenol taxis in E. coli, where Tar senses phenol as an attractant and Tsr as a repellent, resulted in the proposition of an alternative mechanism. Diffusible chemoeffectors such as phenol may stimulate a chemoreceptor by perturbing the structural stability or position of the transmembrane bundle helices or other elements, which ultimately causes a modulation of kinase activity (53). Those authors suggested that behavioral responses to cytoplasmic pH and temperature may also involve the detection of alterations in the transmembrane regions, which is in agreement with a significant body of evidence showing that the temperature-sensing histidine kinases employ similar mechanisms (192, 193). However, there is evidence that chemotaxis to diffusible chemoeffectors is not mediated exclusively by the direct action of the effector on the transmembrane helices. One such example is McpT of P. putida DOT-T1E, which mediates chemotaxis to various hydrocarbons (170). Several lines of evidence suggest that this chemoreceptor operates by a canonical mechanism involving specific ligand recognition at the LBD. First, McpT has a broad and well-defined ligand profile and does not mediate responses to many highly hydrophobic and diffusible ligands, and second, a single-amino-acid substitution in its 4HB LBD abolished receptor function. Chemoreceptor stimulation by PTS substrates appears to occur via yet another mechanism (Fig. 8): a model in which the stimuli generated by PTS substrates are transmitted to the cytosolic fragment of the McpC chemoreceptor of B. subtilis was proposed (194, 195). Changes in the redox state of the FAD cofactor is yet another mode of stimulation, which is seen in chemoreceptors that modulate redox and energy taxis (18, 19, 196).

FIG 8.

Diversity in signal recognition and modes of action. LBDs are shown in green.

Approximately 18% of chemoreceptors lack sensory domains (25). Around one-half of these receptors are composed exclusively of a signaling domain, whereas the other half also contains transmembrane regions (25). The function of this receptor family is poorly understood. However, the discovery of bipartite chemoreceptors (122–124) comprised of an individual sensory domain, such as the PAS domain, and a separate sensorless chemoreceptor (Fig. 8) may provide import clues to their potential mechanism of action.

The example of the BdlA chemoreceptor in P. aeruginosa, which requires phosphorylation and, subsequently, the proteolytic cleavage of one of the sensor domains (Fig. 8C) (156), demonstrates that there might be other unorthodox ways of chemoreceptor stimulation. Future research will show whether other receptors employ similar mechanisms and likely reveal other ways to stimulate chemoreceptors.

COMMON MECHANISM FOR TRANSMEMBRANE SIGNALING

Despite the structural diversity of LBDs, there is evidence for a common mechanism of transmembrane signaling. This mechanism has been identified for the E. coli chemoreceptors and consists of chemoeffector-induced, piston-like, and rotational motions of the last α-helix of the 4HB domain, which extends into the second transmembrane helix (197–200). A recent report identified similar types of movements in the sensory domain on a histidine kinase, further strengthening the argument for a common mechanism for transmembrane signaling in bacteria (201). As shown in Fig. 4 and Table 5, chemoreceptors are equipped with a range of structurally diverse LBDs, which in turn raises the question of whether there are different transmembrane signaling mechanisms. Recent studies reported the construction of chimeric receptors in which NIT, HBM, sCache, and dCache LBDs were fused to the cytosolic fragment of the E. coli Tar chemoreceptor (34, 139, 202). All of these constructs were functional and mediated efficient and specific responses to the corresponding chemoeffectors, suggesting that structurally and functionally diverse LBDs in chemoreceptors employ a single universal transmembrane signaling mechanism.

TABLE 5.

Chemoreceptor ligand binding domains with known three-dimensional structures

Species	Protein	LBD type	Ligand	PDB accession no.	Reference(s)
E. coli	Tar	4HB	Aspartate	4Z9H	217
V. cholerae	Mlp37	dCache	Taurine	5AVF	44
			Serine	5AVE
C. jejuni	Tlp1	dCache	None	4WY9	218
C. jejuni	CcmL	dCache	Isoleucine	4XMR	179
H. pylori	TlpB	sCache	Urea	3UB6	109
			Acetamide	3UB7
			Formamide	3UB8
			Hydroxyurea	3UB9
Vibrio parahaemolyticus	Q87T87_VIBPA	sCache	Pyruvate	4EXO, 2QHK	109
P. putida	McpS	HBM	Malate	2YFA	166
			Succinate	2YFB
E. coli	Tsr	4HB	Serine	2D4U, 3ATP	35
S. Typhimurium	Tar	4HB	Aspartate	2LIG, 1WAT, 1JMW	63, 219, 220
B. subtilis	HemAT	Protolobin	Cyano	1OR4	221
P. aeruginosa	McpB (Aer2)	PAS	Ferric heme	4HI4	153
			Cyano	3VOL	154

CONCLUSIONS

As key components of navigation systems in bacteria and archaea, chemoreceptors enable motile cells to recognize numerous environmental and internal signals that ultimately affect cell behavior and other functions. The types of signals and the mechanisms of signal recognition by chemoreceptors were studied primarily in model organisms such as E. coli and S. enterica as well as B. subtilis, but emerging new models of chemotaxis and the availability of genomic data are changing paradigms and revealing new information about the sensory repertoire of chemoreceptors. Nearly a hundred different types of protein domains are now found in chemoreceptor sequences as known or predicted LBDs, although only a few of them appear to be ubiquitous. Bacteria seem to rely primarily on evolutionarily “old” specialized sensory domains, such as 4HB, Cache, and PAS, that provide countless opportunities to recognize various signals. On the other hand, many bacterial species and strains are “experimenting” with newly evolved versions of chemoreceptors that recruit other domain types as their sensory modules. Consequently, in some species, the chemoreceptor sensory repertoire is extremely limited, whereas in other species, it is very broad. LBDs of the same type that are very similar in sequence might recognize very different ligands, whereas the same ligand can be recognized by sensory domains that belong to fundamentally different protein folds. Although these extremes present a serious challenge to our efforts toward predictive biology, recent experimental and computational advances have begun to reveal certain tendencies. We now know that dCache domains primarily recognize amines and that HBM domains primarily bind TCA cycle intermediates, whereas PAS domains tend to contain redox-sensitive cofactors. The identification of bimodular LBDs, such as dCache_1 and HBM domains, exposed new ways for recognizing multiple signals. On the other hand, signals recognized by the membrane-proximal subdomain of dCache-type domains remain to be identified. The indirect binding of many different ligands emerges as a common mode of signal recognition by chemoreceptors, which might enable the coordination of chemotaxis and uptake, and warrants the development of new strategies and lines of inquiry. Unconventional chemoreceptors that lack LBDs and chemoreceptors with cytoplasmic C-terminal LBDs, for which no mechanistic understanding of signaling is available, add to the overall complexity of chemosensing in bacteria. Future genomics-driven studies will undoubtedly reveal many more relationships between signals and signal-recognizing domains and will broaden and deepen our understanding of the sensory repertoire of chemoreceptors in particular and bacterial signal transduction in general.