The paralogous receptors PctA, PctB and PctC of Pseudomonas aeruginosa were reported to mediate chemotaxis to amino acids, intermediates of amino acid metabolism and chlorinated hydrocarbons. We show that the recombinant ligand binding regions (LBRs) of PctA, PctB and PctC bind 17, 5 and 2 l-amino acids respectively. In addition, PctC-LBR recognized GABA but not any other structurally related compound. l-Gln, one of the three amino acids that is not recognized by PctA-LBR, was the most tightly binding ligand to PctB suggesting that PctB has evolved to mediate chemotaxis primarily towards l-Gln. Bacteria were efficiently attracted to l-Gln and GABA, but mutation of pctB and pctC, respectively, abolished chemoattraction. The physiological relevance of taxis towards GABA is proposed to reside in an interaction with plants. LBRs were predicted to adopt double PDC (PhoQ/DcuS/CitA) like structures and site-directed mutagenesis studies showed that ligands bind to the membrane-distal module. Analytical ultracentrifugation studies have shown that PctA-LBR and PctB-LBR are monomeric in the absence and presence of ligands, which is in contrast to the enterobacterial receptors that require sensor domain dimers for ligand recognition.
Bacteria adapt to changing environmental conditions through various signal transduction mechanisms. Genome analyses suggest that bacterial signal transduction is primarily achieved through the action of one-component systems, two-component systems and chemoreceptor-based signalling (Galperin, 2005). Typically, chemoreceptors are composed of a ligand binding region (LBR) and a signalling domain. The latter domain mediates the interaction of the chemoreceptor with the CheA sensor kinase and the CheW adaptor protein. Signal binding at the LBR triggers a molecular stimulus that is transmitted to the signalling domain, which in turn modulates CheA autophosphorylation and consequently transphosphorylation of the response regulator (Hazelbauer et al., 2008). Chemosensory systems mediate flagellum-mediated chemotaxis, type IV pili mediated taxis or are involved in the regulation of alternative cellular processes (Hickman et al., 2005; Zusman et al., 2007; Wuichet and Zhulin, 2010).
Enterobacteria are the primary model organisms in the study of chemosensing and in particular their response towards amino acids has been studied (Baker et al., 2006). In enterobacteria responses towards amino acids are mediated by the Tar and Tsr chemoreceptors that both possess a 4-helix bundle LBR (Milburn et al., 1991; Tajima et al., 2011). Tar and Tsr recognize l-aspartate and l-serine, respectively, with affinities of around 5 μM (Clarke and Koshland, 1979). The molecular determinants for ligand recognition at both receptors reside in their respective LBRs, since ligands bind with similar affinities to the recombinantly produced LBRs (Milburn et al., 1991; Milligan and Koshland, 1993; Tajima et al., 2011). Tar-LBR is present in a monomer–dimer equilibrium (Milligan and Koshland, 1993) and aspartate was found to bind exclusively to the dimeric form of the LBR (Milburn et al., 1991). Bound aspartate establishes contacts with amino acids from both monomers of the dimer and as a consequence ligand binding stabilizes the dimeric form of the protein (Milburn et al., 1991; Milligan and Koshland, 1993). Aspartate binding triggers a piston-like movement of the final helix of the 4-helix bundle, which corresponds to the molecular stimulus that is transmitted across the membrane (Ottemann et al., 1999). Escherichia coli is attracted to 7 and repelled by another 7 amino acids (Hedblom and Adler, 1983).
In contrast, the human pathogen Pseudomonas aeruginosa is strongly attracted to all 20 l-amino acids (Kuroda et al., 1995). Further research has led to the identification of three paralogous chemoreceptors, PctA, PctB and PctC, that mediate the response towards 18, 7 and 2 l-amino acids respectively (Taguchi et al., 1997). Infections with P. aeruginosa in cystic fibrosis patients causes chronic lung infections, which lead to a high degree of morbidity and mortality (Brugha and Davies, 2011). In the course of respiratory infections of cystic fibrosis patients, P. aeruginosa undergoes phenotypic changes to better adapt to this habitat. One of these changes is the transition from the prototrophic (wild-type) type to a phenotype characterized by an auxotrophism for l-amino acids (Barth and Pitt, 1995; 1996). The clinical significance of this is unclear, but the chemotactic movement of auxotroph P. aeruginosa strains towards l-amino acids is likely to contribute to virulence.
In addition to amino acid chemotaxis Taguchi et al. reported that P. aeruginosa also responded to intermediates of the amino acid metabolism, like putrescine, cadaverine and GABA (Taguchi et al., 1997). Since this chemotactic behaviour was not observed in the triple pctABC mutant, the authors suggested that one or several of these receptors are involved in taxis towards these metabolites (Taguchi et al., 1997). P. aeruginosa also exhibited repellent responses to different toxic chlorinated compounds like trichloroethylene and chloroform (Shitashiro et al., 2003). This taxis was significantly decreased in the pctA mutant, which has led to the conclusion that PctA is the primary receptor for the repellent response towards chlorinated contaminants (Shitashiro et al., 2005). In summary, data currently available show that PctA, PctB and PctC have multiple functions, namely in mediating chemoattraction to amino acids and intermediates of the amino acid metabolism as well as mediating chemorepellent responses towards chlorinated compounds (Shitashiro et al., 2005).
There are significant differences in the chemosensory systems of enterobacteria and environmental bacteria (Krell et al., 2011). An analysis of all available chemoreceptor sequences showed that one such difference is the size of the chemoreceptor LBRs that can be classified into cluster I (around 150 amino acids) and cluster II regions (around 250 amino acids) (Lacal et al., 2010a). The well-studied enterobacterial receptors possess cluster I receptors with 4-helix bundle type LBRs (Milburn et al., 1991). However, an estimated 40% of all chemoreceptors possess cluster II domains. Although abundant, cluster II receptors are poorly characterized. It appears that the family of cluster II ligand-binding domains is composed of double PDC (PhoQ/DcuS/CitA) like domains and McpS-LBR like domains (Pineda-Molina et al., 2012) (Fig. 1). Both sensor domains show a bimodular arrangement. Double PDC domains are formed by a long N-terminal helix and two stacked PAS-like domains (Fig. 1A), similar to those seen in PhoQ, DcuS and CitA (Zhang and Hendrickson, 2010). PAS domains are ubiquitous bacterial sensor domains. Double PDC domains are frequently found in sensor kinases but were also identified in chemoreceptors such as McpB and McpC of Bacillus subtilis (Glekas et al., 2010; 2012). In contrast, the McpS-LBR consists of two stacked 4-helix bundles (Fig. 1B) (Pineda-Molina et al., 2012). We have shown recently that each of these bundles binds different chemoattractants and that this binding causes a chemotactic response (Fig. 1) (Pineda-Molina et al., 2012). In the case of double PDC domains it has not yet been established whether ligands bind to each of the two structural modules. The LBR of PctA, PctB and PctC are of approximately 250 amino acids indicating that they belong to the cluster II family.
Here we report a characterization of these three paralogous receptors. The recombinant LBRs are analysed by a set of biophysical techniques including isothermal titration calorimetry, analytical ultracentrifugation, differential scanning calorimetry and circular dichroism spectroscopy. Homology modelling and site-directed mutagenesis experiments led to the identification of the signal binding site. Experiments have led to the identification of PctC as chemoreceptor for the non-protein amino acid GABA.
The homology models of PctA-LBR, PctB-LBR and PctC-LBR reveal a double PDC fold
PctA, PctB and PctC are paralogous proteins and share an overall sequence identity of 69%. Using the Dense Alignment Surface (DAS) method (Cserzo et al., 1997) two transmembrane regions were predicted for each paralogues flanking a LBR of 252–255 amino acids (Fig. S1). Sequence diversity resides primarily in the LBRs whereas the cytosolic signalling domains are highly conserved (Fig. S1). Pairwise sequence alignments showed that PctA-LBR and PctB-LBR are closely related (70% identity), whereas similarities of both sequences with PctC-LBR are much lower (46% and 54% identity respectively). The LBRs are un-annotated in the PROSITE database. However, a blast search in the protein data bank revealed that PctA-LBR shares 30% sequence identity with a LBR of an uncharacterized Vibrio cholera chemoreceptor (pdb entry 3c8C). This structure was used to generate homology models of the three LBRs (Fig. 1A, Fig. S2). Ramachandran plots show that the models are of acceptable geometry (Fig. S3). The three models were very similar and are composed of a long N-terminal helix, followed by two modules each representing a globular α/β domain. This protein fold, termed double PDC domain, is the predominant sensor domain in histidine kinases (Zhang and Hendrickson, 2010) and also abundant in chemoreceptors (Pineda-Molina et al., 2012).
PctA-LBR, PctB-LBR and PctC-LBR bind respectively 17, 5 and 2 l-amino acids
PctA, PctB and PctC were found to mediate taxis towards three groups of compounds, namely amino acids, intermediates of amino acid metabolism and chlorinated compounds. To establish whether these receptors recognize ligands directly, PctA-LBR, PctB-LBR and PctC-LBR were obtained as recombinant proteins (Fig. S4) and submitted to microcalorimetric titrations (Krell, 2008). Initial experiments involved titrations with l-alanine and were conducted in polybuffer, pH 7.0, 150 mM NaCl, 10% glycerol. As shown in Fig. 2A the titration of PctA-LBR resulted in exothermic heat changes that diminished as protein saturation with ligands advanced. Data analysis revealed that l-Ala binding was driven by favourable enthalpy and entropy changes. Binding was tight and a KD value of 0.7 μM was obtained. This experiment was then repeated with the same buffer at pH values between 4 and 9. Binding was observed for all pH values and the pH optimum was at pH 7.0 (Fig. S5). The optimal salt concentration was 150 mM NaCl and a reduction of affinity was observed in the presence of either 50 or 300 mM NaCl (Fig. S5).
Subsequent experiments involved the titration of PctA-LBR with the remaining l-amino acids (Fig. 2A, Table 1). With the exception of l-Asp, l-Glu and l-Gln all other l-amino acids bound. Amino acids with the lowest dissociation constants were l-Thr (KD = 0.28 μM), l-Val (0.34 μM) and l-Pro (0.6 μM), whereas amino acids with highest dissociation constants were l-His (KD = 28 μM) and l-Leu (116 μM). This implies that the affinities of amino acids for PctA-LBR binding differ by a factor of around 400. It should be noted that tightest binding amino acids, l-Thr and l-Val, are similar in their structures. PctA-mediated chemotaxis to l-Glu was described (Taguchi et al., 1997) but under the experimental conditions used we were unable to detect l-Glu binding (note: in the absence of binding the experiment was repeated at 15°C, to exclude the possibility that enthalpic and entropic contributions cancel out each other at 30°C, the analysis temperature used for the experiments reported).
Table 1. Thermodynamic parameters of ligand binding derived from the microcalorimetric titration of PctA-LBR, PctB-LBR and PctC-LBR with different ligands
PctA-LBR, PctB-LBR and PctC-LBR were titrated with the following compounds, which however did not bind: d-Ala, d-Gln, d-Glu, the dipeptides l-Ala-l-Ala, l-Ala-l-Gln, l-Ala-l-Gly, cadaverine, putrescine, trichloroethylene, chloroform and methylthiocyanate. In addition PctC-LBR was titrated with succinate semialdehyde, methyl-gammabutyrate and butyrate and an absence of binding was observed in all cases. Data are means and standard deviations from three experiments.
(5.61 ± 1.1) 105
1.8 ± 0.4
−7.4 ± 1
(1.57 ± 0.1) 104
64 ± 4
1.52 ± 0.4
(6.08 ± 0.5) 105
1.6 ± 0.1
−7.5 ± 0.3
912 ± 73
1096 ± 88
0.95 ± 0.7
(1.94 ± 0.6) 105
5.2 ± 1.7
−6.8 ± 3.6
(4.27 ± 1.6) 105
2.3 ± 0.9
−3.5 ± 0.9
(9.27 ± 2) 105
1.1 ± 0.2
−3.4 ± 0.3
(1.38 ± 0.2) 106
0.72 ± 0.1
−5.7 ± 0.4
(1.56 ± 0.1) 103
641 ± 50
0.92 ± 0.8
(2.97 ± 1.5) 106
0.34 ± 0.2
−2.4 ± 0.3
(4.05± 0.9) 104
24 ± 5.6
3 ± 1.5
(8.58 ± 7.8) 103
116 ± 10
9.7 ± 0.9
(1.10 ± 0.2) 106
0.91 ± 0.2
−4.3 ± 0.3
(2.18 ± 0.1) 104
46 ± 2
0.55 ± 0.1
(4.92 ± 0.6) 105
2.0 ± 0.2
−5.0 ± 0.5
(8.54± 0.8) 105
1.2 ± 0.1
−5.49 ± 0.1
(8.05 ± 1.03) 105
1.2 ± 0.2
−8.5 ± 0.5
(1.26 ± 0.1) 106
0.79 ± 0.1
−9.8 ± 0.4
(3.51 ± 0.7) 106
0.28 ± 0.1
−3.4 ± 0.2
(3.55 ± 0.7) 104
28 ± 5
0.97 ± 0.2
(5.6 ± 0.9) 104
17 ± 3
0.6 ± 0.03
(1.67 ± 0.5) 106
0.60 ± 0.2
−6.5 ± 0.9
(1.25 ± 0.1) 104
80 ± 6
−0.33 ± 0.05
(4.70 ± 1.7) 104
21 ± 7.7
−3.4 ± 8.4
(8.0 ± 0.3) 105
1.2 ± 0.3
−13.2 ± 0.1
PctB was reported to mediate taxis towards 7 different l-amino acids (Taguchi et al., 1997) and we have titrated PctB-LBR with all natural l-amino acids. Interestingly, PctB-LBR bound tightly l-Gln (KD = 1.2 μM, Fig. 2B), which is one of the 3 amino acids that is not recognized by PctA-LBR. In addition PctB-LBR bound l-Arg, l-Met, l-Ala and l-Lys with significantly lower affinities (KD = 46 to 1096 μM, Fig. 2B, Table 1). The thermodynamic binding mode of l-Gln was different from that of the remaining ligands. l-Gln binding was exothermic, indicating favourable enthalpy changes, whereas the binding of the other amino acids was endothermic and thus driven by favourable entropy changes (Fig. 2B, Table 1). PctB-LBR is composed of two modules (Fig. S2) and the different thermodynamic binding modes may indicate that ligands interact with different modules. To assess this possibility we have conducted competition experiments. PctB-LBR was saturated with 75 μM l-Gln and then titrated with aliquots of 2 mM l-Arg, which did not reveal any binding. In a second experiment PctB-LBR was saturated with 650 μM l-Arg and the resulting complexes were titrated with l-Gln. The binding of l-Gln was characterized by a significantly lower affinity (KD = 10.9 μM) as compared with the binding to the unliganded protein. These experiments suggest that amino acids compete for binding at the same site at PctB-LBR, but it cannot be ruled out that the reduction in affinity may be the result of an allosteric effect. Since PctA-LBR recognized l-Arg, l-Met, l-Ala and l-Lys with significantly higher affinity (KD of around 1 μM) it can be suggested that PctA is the principal receptor for these amino acids. It was also reported that PctB mediates taxis to l-Glu and l-Tyr (Taguchi et al., 1997). As in the case of l-Glu binding to PctA-LBR, we were unable to observe binding of these amino acids.
The titration of PctC-LBR with the 20 amino acids resulted in the detection of binding for l-Pro and l-His (Table 1), which is in agreement with the chemotaxis data published (Taguchi et al., 1997). As reported above PctA-LBR recognized both amino acids with higher or similar affinities (Table 1). Ligands so far analysed are l-amino acids. To assess the specificity of recognition, the three proteins were titrated with a number of d-amino acids and dipeptides (compounds are listed in the legend to Table 1). In all cases an absence of binding was noted indicating that proteins are specific for l-amino acids.
PctC-LBR recognizes GABA
Pseudomonas aeruginosa but not the triple pctA/pctB/pctC mutant showed chemoattraction towards intermediates of the amino acid metabolism putrescine, cadaverine and GABA (Taguchi et al., 1997). To identify the chemoreceptor(s) involved the three recombinant proteins were titrated with these compounds. PctA-LBR and PctB-LBR were devoid of binding, whereas PctC-LBR bound specifically GABA but not putrescine and cadaverine (Fig. 2C). The affinity (KD = 1.2 μM) of PctC-LBR for GABA is significantly higher than those for l-Pro and l-His (Table 1). To explore the specificity of GABA recognition, McpC-LBR was titrated with structurally similar compounds like butyrate, succinate semialdehyde and methyl gamma-aminobutyrate. As shown in Fig. 2C no binding was observed, consistent with a high specificity for GABA.
PctA-LBR does not recognize chlorinated compounds
PctA was identified as the primary receptor responsible for the repellent responses from chlorinated toxic compounds (Shitashiro et al., 2003). To determine whether these compounds are recognized by PctA its recombinant LBRs was titrated with 2–5 mM solutions of trichloroethylene, chloroform and methylthiocyanate. In all three cases an absence of binding was noted. After the completion of each titration, the resulting protein/chlorinated compound mixtures were titrated with l-Ala as a positive control. The repetition of the experiments at 15°C also failed to obtain binding. The binding constants of ligands for the three receptor proteins are summarized in Fig. 3. The paralogues recognized together 18 l-amino acids with affinities ranging from 0.28 to 1096 μM and in addition GABA. l-Asp and l-Glu were not recognized by any of these receptors. Six amino acids (l-Arg, l-Lys, l-Ala, l-Met, l-His and Pro) were recognized by two chemoreceptors (Fig. 3). l-Gln and GABA bound with identical affinities to PctB-LBR and PctC-LBR respectively.
PctB and PctC are the sole chemoreceptors for chemoattraction to l-glutamine and GABA respectively
PctB and PctC may have evolved to assure attraction to l-glutamine and GABA respectively. To assess the contribution of both receptors in chemotaxis, quantitative capillary assays of wild type and mutant strains towards different concentrations of l-Gln and GABA were carried out (Fig. 4). The wild type strain showed a clear dose-dependent chemotactic response to l-Gln and GABA. Mutation of pctB and pctC abolished chemoattraction to l-Gln and GABA respectively (Fig. 4). In the case of the pctB mutant slight repellent response was observed, which may suggest the existence of another chemoreceptor that mediates this response. Repellent responses were not due to pH effects since the pH of the l-Gln solution was identical to that of the chemotaxis buffer. The complementation of both mutants with plasmids harbouring the pctB and pctC gene led to a partial recovery of chemotaxis (Fig. 4). To determine whether P. aeruginosa can use l-Gln and GABA as sole carbon source we have recorded growth curves in minimal medium supplemented with l-Gln, GABA, succinate and glucose. Efficient growth was observed in all four cultures (Fig. S6).
As shown above the Pct receptors recognize specifically the l-isomer of amino acids. To evaluate whether P. aeruginosa may respond to d-amino acids we have conducted chemotaxis drop and quantitative capillary assays with both isomers of Ala, Glu and Gln. d-Ala and d-Glu are components of the peptidoglycan layer of the cell wall (Typas et al., 2012) and in this context chemotaxis to these compounds may be beneficial. As shown in Fig. S7 clear responses using both assays were observed for l-Ala, l-Glu and l-Gln, but no taxis to the d-isomers of both amino acids could be detected.
Identification of the ligand binding site for l-amino acids
The homology model of PctA-LBR (Fig. 1A) shows a bimodular arrangement in which each module represents a potential ligand binding site. The sensor domain of the DctB kinase has a similar structure and its 3D structure shows succinate bound to the membrane distal module (Zhou et al., 2008). To determine whether ligand recognition at PctA also occurs on the membrane-distal module, amino acids present in the cavity of the distal module of the PctA-LBR were identified and the corresponding alanine substitution mutants prepared. These amino acids were R126, W128, Y144, D146 and D173 (Fig. 1A). In the sequence alignment of PctA, PctB and PctC (Fig. S1) R126 and W128 are fully conserved, whereas the remaining amino acids are only partially conserved. The resulting mutants were submitted to microcalorimetric titration with l-Ala, l-Arg, l-Thr, l-Trp and l-Pro (Table 2). Mutants R126A and D146A failed to recognize any of the 5 amino acids, suggesting a role of these amino acids in the recognition of main chain atoms. Mutant W128A recognized l-Ala, l-Arg, l-Thr and l-Pro with affinities 32- to 65-fold inferior to the native protein (Table 2) but failed to recognize l-Trp (Fig. 5, Table 2). This suggests that W128 plays a central role in the binding of aromatic amino acid side-chains. Two other mutants, D173A and Y144A, bound l-Trp but not any other amino acid tested (Fig. 5, Table 2). Data show that amino acids are recognized by the membrane-distal module.
Table 2. Dissociation constants (μM) derived from the microcalorimetric titration of native and mutant PctA-LBR with different amino acids
Data are means and standard deviations from three experiments.
0.3 ± 0.1
9.8 ± 2.1
0.7 ± 0.1
45.7 ± 6.1
0.6 ± 0.2
34.5 ± 4.6
1.8 ± 0.4
67.6 ± 10.5
2.3 ± 0.9
10.7 ± 0.6
7.1 ± 1.5
To verify whether these mutations abolish a response, quantitative capillary chemotaxis assays were conducted. To this end the mutant strain P. aeruginosa PCT2 (Table S1), deficient in the pctA, pctB and pctC genes, was complemented with plasmid pMAI18-1 (Table S1) harbouring pctA. Initial experiments involved the quantification of chemotaxis towards different concentration of l-Ala (Fig. S8A). An optimal response was obtained at 100 μM l-Ala and this concentration was used for further studies. Subsequently, five mutant derivatives of pMAI18-1 were generated encoding PctA with each of the above mentioned single amino acid substitutions. Assays of P. aeruginosa PCT2 complemented with any of these five plasmids showed either little or no chemotaxis towards l-Ala (Fig. S8B), confirming that these mutations in the PctA-LBR either strongly reduce or abolish taxis.
Different ligand binding modes are reflected in the unfolding properties of the PctB-LBR/ligand complexes
To assess whether there may be a functional communication between both modules of the double PDC structure we have conducted thermal unfolding studies. Using the two-domain transcriptional regulator TtgV we have shown that functional interaction between domains is reflected in the unfolding thermodynamics (Fillet et al., 2011). TtgV, which operates by a mechanism involving domain communication (Guazzaroni et al., 2007), unfolded in a single transition indicative of cooperative unfolding of both domains. In contrast a mutant with impaired inter-domain communication unfolded in two transitions (Fillet et al., 2011) indicative of a sequential unfolding of the individual domains.
The far UV CD spectra of PctB-LBR are shown in Fig. 6A. From the spectrum an α-helix content of 26% was calculated (Luo and Baldwin, 1997). This value is similar to the α-helix content of 28% derived from the PctB-LBR model (Fig. S2). Addition of ligands produced only minor alterations to the secondary structure (Fig. 6A). We have conducted thermal unfolding studies of PctB-LBR monitored by CD and Differential scanning calorimetry (DSC). In both techniques a temperature gradient is applied to the protein and unfolding is monitored. Unfolding studies using the CD signal at 220 nm revealed a cooperative unfolding of PctB-LBR and the midpoint of protein unfolding was at 58°C. When this experiment was repeated in the presence of l-Met and l-Arg, ligands that showed endothermic binding (Fig. 2B), a modest increase in the midpoint of protein unfolding of around 1.5°C was noted (Fig. 6B). However, an increase in the midpoint of unfolding by approximately 7.5°C was noted in the presence of l-Gln (Fig. 6B), the ligand that caused exothermic binding (Fig. 2B).
Analogous samples were then analysed by DSC. Unliganded PctB-LBR unfolded in a single event with a midpoint of unfolding (Tm) at 58.14°C. Consistent with the CD data, the addition of l-Met and l-Arg caused an increase in Tm by 1.28 and 1.64°C (Table 3, Fig. 6C), respectively, whereas the addition of l-Gln or l-Gln + l-Arg increased the Tm by around 7.6°C (Table 3, Fig. 6C). In the presence of ligand an increase in the unfolding cooperativity is noted as witnessed by narrower peaks. Taken together, the data indicate that the two structural modules of PctB-LBR unfold cooperatively. Data also suggest that the different binding modes for l-Gln (exothermic binding) and l-Met/l-Arg (endothermic binding) are reflected in the unfolding thermodynamics, since the increase in Tm in the presence of l-Gln was around fivefold superior to the corresponding values obtained for l-Met and l-Arg.
Table 3. Thermodynamic parameters for the thermal unfolding of PctB-LBR in the absence and presence of different l-amino acids as derived from differential scanning calorimetry studies
ΔH (kcal mol−1)
PtcB-LBR + l-Gln
PtcB-LBR + l-Arg
PtcB-LBR + l-Met
PtcB-LBR + l-Arg + l-Gln
PctA-LBR and PctB-LBR are monomeric in the absence and presence of ligands
Tar-LBR and McpS-LBR consist of a single and respectively a double 4-helix bundle (Milburn et al., 1991; Pineda-Molina et al., 2012). Both proteins are present in a monomer–dimer equilibrium and in both cases ligands bind only to the dimeric state (Milligan and Koshland, 1993; Lacal et al., 2010b). Since no information on the oligomeric state of double PDC like domains is available, PctA-LBR and PctB-LBR were analysed by analytical ultracentrifugation. Sedimentation velocity studies were conducted over a concentration range of 0.1–6 mg ml−1 (Fig. 7). The analysis of the profiles (Fig. S9) showed that neither protein presents a concentration-dependence of the sedimentation coefficient, indicating a lack of auto-association. The sedimentation coefficient distributions obtained for PctA-LBR and PctB-LBR are practically identical over the concentration range studied (Fig. 7) and reveal a neat peak corresponding to a single species with standard sedimentation coefficients of sw,20 = 2.3 S and sw,20 = 2.5 S for PctA-LBR and PctB-LBR respectively.
The HYDROPRO algorithm (Ortega et al., 2011) allows a prediction of the sedimentation coefficients of a protein from its atomic co-ordinates. Using the homology models of both protein monomers (Fig. 1 and Fig. S2) a standard s value (sw,20) of 2.5 S was predicted for both monomers. The fact that the predicted and measured S values are in close agreement is evidence that the single species observed in the sedimentation velocity experiments correspond to monomers. The above experiments were repeated in the presence of saturating concentrations of l-Ala (PctA-LBR) and l-Gln (PctB-LBR). The sedimentation coefficient distributions (Fig. 7) are practically identical to those corresponding to the sample in the absence of ligands, indicating that the ligand binding does not induce self-association.
To verify the above conclusions sedimentation equilibrium experiments were carried out (Figs S10 and S11) for PctA-LBR (0.7 mg ml−1) and PctB-LBR (0.6 mg ml−1) in the absence and presence of 2 mM l-Ala or 2 mM l-Gln respectively. Fitting of the gradient profiles revealed in all cases the presence of a single species. The average molecular weight derived from the analysis at three different speeds was 27 100 ± 100 Da (the error is the error of curve fitting) for PctA-LBR and 30 000 ± 100 Da for PctB that are close values the sequence derived masses of 29 800 Da and 31 800 Da respectively. In analogy to the sedimentation velocity studies the addition of l-Ala or l-Gln to PctA-LBR and PctB-LBR did not alter the oligomeric state.
Much of what we know in the field of bacterial taxis is due to the study of enterobacterial chemotaxis towards amino acids, which has been the object of extensive investigations. As a consequence enterobacterial taxis to amino acids can be regarded as a paradigm and reference system in the field.
A first major conclusion of this work resides in the fact that the bases of amino acid chemotaxis in P. aeruginosa are fundamentally different from the enterobacterial system. A first major difference resides in the receptor architecture. P. aeruginosa receptors are of cluster II and are predicted to contain double PDC ligand binding regions whereas enterobacterial receptors are of cluster I and contain 4-helix bundle sensing domains. In this context P. aeruginosa amino acid chemoreceptors are similar to the two amino acid chemoreceptors of B. subtilis, McpB and McpC that were also predicted to possess double PDC sensor domains (Glekas et al., 2010, Glekas et al., 2012). Using site-directed mutagenesis we show that ligands bind to the membrane distal module of PctA-LBR. This is consistent with studies of B. subtilis McpB, where it was demonstrated that mutations in the membrane distal module had a significant impact on ligand binding and chemotaxis whereas mutations of the membrane proximal module had little or no effect (Glekas et al., 2010).
A second major difference is the ligand profile of amino acid chemoreceptors in enterobacteria. Tar and Tsr have a relatively high ligand specificity for aspartate and serine respectively (Clarke and Koshland, 1979; Hedblom and Adler, 1983). P. aeruginosa has the broad range amino acid receptor PctA that was found to support chemotaxis to 19 amino acids (Taguchi et al., 1997) and of which its ligand binding region recognizes directly 17 amino acids (Fig. 3). Again strong parallels exist to the B. subtilis receptor McpC that also has a broad ligand spectrum since it supports taxis to 17 amino acids of which 11 bind directly in vitro. Apart from the broad range amino acid receptors PctA and McpC, P. aeruginosa and B. subtilis possess chemoreceptors that appear to have evolved to mediate taxis towards either l-Gln or l-Asn. McpB of B. subtilis mediates taxis primarily to l-Asn and is referred to as asparagine receptor (Hanlon and Ordal, 1994; Glekas et al., 2012), Our results show that PctB-LBR recognizes l-Gln with very high affinity, whereas the remaining four ligands are recognized with lower affinity. This thus suggests that PctB may have evolved to specifically respond to l-Gln. To determine whether the functional similarities between PctA and McpC as well as between PctB and McpB are reflected in sequence similarities, a cluster analysis of the sensor domains was conducted. However, the sequences with similar effector profiles did not cluster together indicating that subtle amino acid changes in the binding site may be responsible for the differences in the ligand profile.
Another major difference between enterobacterial and P. aeruginosa receptors is related to the mode of signal recognition. In the absence of bound ligand Tar-LBR is present in a dynamic monomer–dimer equilibrium and ligand was found to bind exclusively to the dimeric state of the protein, which in turn stabilized the dimer (Milligan and Koshland, 1993). The molecular reason for the requirement of the dimer for ligand binding resides in the fact that the ligand binding site is at the dimer interface and that amino acids from both monomers of the dimer inter act with bound aspartate (Milburn et al., 1991). Here we show that PctA-LBR and PctB-LBR are monomeric in solution and that the monomeric forms of both proteins bind ligands. It can therefore be concluded that the binding mode of these proteins does not require dimeric sensor domains. This is also supported by the 3D models of the Pct domains and of McpB-LBR that show a deep ligand binding cavity in the membrane distal module (Glekas et al., 2012). In both cases mutation of amino acids within this cavity resulted in a reduction or loss of ligand binding.
The common feature of enterobacterial, P. aeruginosa and B. subtilis amino acid receptors is their direct interaction with amino acids. However, for P. aeruginosa and B. subtilis receptors the number of amino acids that bind directly is inferior to the number of amino acids to which the receptor mediates a tactic response. McpC supports taxis to 17 amino acids but binds only 11 of them directly. Similar observations have been made for PctA and PctB that mediate taxis to 19 and 7 amino acids, respectively, but bind only 17 and 5 (Table 1). Glekas et al. (2012) have provided initial evidence that the remaining amino acids bind to the receptor in complex with periplasmic binding proteins. Further research will show whether this is also the case for the receptors studied here.
As a whole it can be concluded that the mechanism of amino acid chemotactic signalling of P. aeruginosa is much more similar to the system in B. subtilis than to the enterobacterial system. This is rather astonishing since P. aeruginosa is phylogenetically much closer to enterobacteria (both belong to class of Gammaproteobacteria of the phylum Proteobacteria) than to B. subtilis (that belong to the phylum Firmicutes). The observation of similar chemotactic systems in phylogenetically very distant bacteria may suggest that the predominant amino acid chemotaxis system in bacteria is that observed in B. subtilis and P. aeruginosa.
Another major finding is the identification of PctC as GABA chemoreceptor. The titration of PctC-LBR with closely related GABA homologues did not reveal any binding indicative of a high specificity in the molecular recognition of GABA. P. aeruginosa is ubiquitously present in the environment (Green et al., 1974) and also a ubiquitous opportunistic pathogen able to infect different animals and plants (Rahme et al., 1995; Cao et al., 2001). GABA is a ubiquitous, non-protein amino acid that has a multitude of different functions in all kingdoms of life. GABA is synthesized via the GABA shunt, which exists in animals, plants, fungi and bacteria (Shelp et al., 1999). GABA acts as an important inhibitory neurotransmitter in animals (Bormann, 2000) but is also abundantly present in plants where it was found to be the most abundant amino acid in tomato apoplastic liquid (0.5 mM) (Rico and Preston, 2008). Based on the observation that plant exposure to different types of stress, including bacterial infection, causes an increase in intracellular and extracellular GABA concentrations (Shelp et al., 1999; Allan et al., 2008), it was proposed that GABA acts as a signal molecule to mediate interactions with other organisms such as bacteria (Shelp et al., 2006). In addition, wounding of plants was found to stimulate a process which increased the GABA release from the wound into the external medium (Bown et al., 2006). Pseudomonas can use GABA as sole carbon and energy source (Rico and Preston, 2008). Based on the importance of GABA in plant physiology it appears plausible that the physiological reason for a GABA chemoreceptor is the capacity to colonize and infect plants. However, GABA is also an important neurotransmitter and it remains to be assess whether GABA taxis contributes to pathogenicity.
It was shown that PctA, PctB and PctC mediate taxis towards three groups of compounds, namely l-amino acids, different derivatives of amino acid metabolism (Kuroda et al., 1995; Taguchi et al., 1997) and chlorinated toxic compounds (Shitashiro et al., 2005). We show here that taxis towards 18 amino acids and GABA is mediated by direct binding to these receptors. No binding of cadaverine, putrescine and chlorinated compounds for any of the three proteins, which suggests that taxis towards these compounds is based on a mechanism involving binding proteins.
Strains and plasmids
The strains and plasmids used are provided in Table S1.
Cloning of PctA-LBR, PctB-LBR and PctC-LBR into expression plasmids, protein expression and purification
The DNA fragments encoding amino acids 30–278, 30–277 and 30–281 of PctA, PctB and PctC, respectively, were amplified using the primers listed in Table S2. These primers contained restriction sites for NdeI and BamHI. Resulting PCR products were digested with these enzymes and cloned into the expression plasmid pET28b(+). The resulting plasmids were termed pET28-PctA-LBR, pET28-PctB-LBR and pET28-PctC-LBR and were verified by DNA sequencing of the insert and flanking regions.
Escherichia coli BL21 (DE3) was transformed with either pET28-PctA-LBR and pET28-PctB-LBR and E. coli C43 with pET28-PctC-LBR. Cultures were grown in 2 l Erlenmeyer flasks containing 500 ml LB medium supplemented with 50 μg ml−1 kanamycin at 30°C until an OD660 of 0.6, at which point protein production was induced by adding 0.1 mM IPTG. Growth was continued at 18°C overnight prior to cell harvest by centrifugation at 10 000 g for 30 min. Cell pellets were resuspended in buffer A [30 mM Tris, 300 mM NaCl, 10 mM imidazole and 5% (vol/vol) glycerol, pH 7.0] and broken by French press at 1000 psi. After centrifugation at 20 000 g for 1 h, the supernatant was loaded onto a 5 ml HisTrap column (Amersham Bioscience), washed with five column volumes of buffer A and eluted with an imidazole gradient of 45–1000 mM in buffer A.
Isothermal titration calorimetry
Experiments were conducted on a VP-microcalorimeter (Microcal, Amherst, MA) at 30°C. Proteins were dialysed against polybuffer (5 mM Tris, 5 mM Pipes, 5 mM Mes), 10% glycerol (vol/vol), 150 mM NaCl, pH 7.0 and placed into the sample cell. Typically, 15–35 μM of protein was titrated with 1–5 mM effector solutions that were prepared in dialysis buffer immediately before use. The mean enthalpies measured from the injection of effectors into the buffer were subtracted from raw titration data prior to data analysis with the MicroCal version of ORIGIN. Data were fitted with the ‘One binding site model’ of ORIGIN.
Site-directed mutagenesis of amino acids in PctA-LBR
A modified version of the Hemsley method (Hemsley et al., 1989) was used to generate PctA-LBR mutants R126A, W126A, Y144A, D146A and D173. Pairs of partially overlapping mutagenic primers were used (Table S2) to amplify the entire plasmid with a high-fidelity DNA polymerase, which generated nicked circular DNA. After PCR amplification with Pfu Turbo DNA polymerase (Agilent Technologies) the methylated template DNA was eliminated by digestion with DpnI (an enzyme specific for methylated DNA). The resulting PCR products were electrotransformed into E. coli DH5 and colonies were selected on LB supplemented with kanamycin (50 μg ml−1). Plasmid DNA from resulting clones were isolated and inserts and flanking regions were sequenced for verification.
Derivatives of plasmid pMAI18-1 encoding PctA were constructed harbouring the same amino acid substitutions. To this end a 0.69 kb fragment containing the desired mutations were excised from the above constructed pET28 derivatives by a restriction of the five mutant plasmids with PstI and BstEII (Roche). The five mutant fragments were cloned into pMAI18-1 linearized with the same enzymes. The presence of the desired mutation was verified by DNA sequencing.
Bacteria were grown to early stationary phase in 2× YT medium at 37°C. The cultures were then gently washed twice with 10 mM HEPES, pH 7.0 and subsequently diluted in HEPES buffer to an OD600 of 0.06–0.08. Capillaries (Microcaps, Drummond Scientific, USA) were sealed at one end, warmed over the flame and the open end inserted into the chemoattractant solution, of which the pH had been adjusted to that measured for the bacterial suspension. The HEPES buffer lacking effectors was used as a control. The bacterial suspension was placed into a small chamber formed by placing a v-shaped needle onto a microscope slide. The system was then closed with a glass coverslip. The capillary was immersed into the cell suspension at its open end. After incubation for 10 min, the open end of the capillary was rinsed with water and placed into a microfuge tube containing 1 ml minimal medium M9. The sealed end was broken and the contents emptied into the tube. A volume of 20 μl of the resulting cell suspension was plated out agar plates containing M9 minimal medium supplemented with 15 mM succinate and incubated at 30°C. Colonies were counted after growth for 36 h.
Differential scanning calorimetry
Thermal denaturation experiments were carried out on a VP-DSC capillary-cell microcalorimeter from MicroCal (Northampton, MA, USA) at a scan rate of 60°C h−1. PctB-LBR solutions were prepared by dialysis against 5 mM Tris, 5 mM Pipes, 5 mM Mes, pH 7.0, 150 mM NaCl, and each ligand compound was subsequently added to the protein samples and the buffer. Before each experiment several buffer–buffer baselines were obtained to equilibrate the instrument. The experimental thermograms were baseline-subtracted, corrected from the instrument's response and normalized by the protein concentration. The calorimetric enthalpies were estimated by integration of the transition peaks.
Circular dichroism spectroscopy
CD experiments were performed on a Jasco J-715 (Tokyo, Japan) spectropolarimeter equipped with a thermostated cell holder. Measurements of the far-UV CD spectra (260–202 nm) were made with 1 mm path length quartz cuvettes using 15 μM protein in the absence and presence of ligands. The concentrations of the ligands were 37.5 μM l-Gln, 325 μM l-Arg, 235 μM l-Met, 325 μM l-Arg + 37.5 μM l-Gln. In thermal melting experiments, the CD signal was monitored as a function of temperature at 220 nm.
Experiments were performed on a Beckman Coulter Optima XL-I analytical ultracentrifuge (Beckman Coulter, Palo Alto, CA, USA) with an An-50 Ti 8-hole rotor with 12 mm path-length charcoal-filled epon double-sector centrepieces. Sedimentation velocity (SV) experiments were performed at 45 000 r.p.m. at 20°C. Samples of 390 μl at concentrations between 0.1 and 6 mg ml−1 in polybuffer, 10% glycerol (vol/vol), pH 7.0 were analysed. A series of 300 absorbance scans without time intervals between successive scans were taken at a wavelength of 280 nm for those samples with concentrations equal and below 1 mg ml−1 and at 300 nm for those above 1 mg ml−1. The data were analysed using size-distribution c(s) analyses in SEDFIT v12.1 software (Schuck, 2000). Buffer density (ρ = 1.0347 g ml−1) and viscosity (η = 0.0138 Poise) at 20°C, measured with a picnometer and a Ubehlode viscometer were used to correct the sedimentation coefficient values to standard conditions (s20,w, viscosity and density of the water, and 20°C). SEDNTERP software (Lauer and Firnhaber, 1992) was used to estimate the partial specific volume of PctA-LBR and PctB-LBR (0.737 and 0.731 ml g−1 respectively) from the amino acid sequence. Sedimentation equilibrium (SE) data were acquired for 150 μl samples, using 150 μl polybuffer as reference, at three different speeds (13 500, 22 000 and 27 000 r.p.m.) and the scans measured with absorbance optics at a wavelength of 280 nm. The average molecular weights were calculated using SEDPHAT v9.4 software (Vistica et al., 2004), fitting the concentration gradients observed by absorbance with the ‘single species of an interacting system’ model and corrected with the buoyancy factor by the partial specific volume calculated by SEDNTERP.
We are grateful to Dr Junichi Kato for his generosity in providing the P. aeruginosa strains and for his unconditional help. We acknowledge financial support from FEDER funds and Fondo Social Europeo through grants from the Junta de Andalucía (grants P09-RNM-4509 and CVI-7335 to T.K.), the Spanish Ministry for Economy and Competitiveness (Grant Bio2010-16937 to T.K. and BIO2010-17227 to J.L.R.) and from the BBVA Foundation (Grant BIOCON08 185/09 to T.K.). A.O. is recipient of a postdoctoral fellowship from Fundación Cajamurcia.