Identification of Chemotaxis Sensory Proteins for Amino Acids in Pseudomonas fluorescens Pf0-1 and Their Involvement in Chemotaxis to Tomato Root Exudate and Root Colonization

Pseudomonas fluorescens Pf0-1 showed positive chemotactic responses toward 20 commonly-occurring l-amino acids. Genomic analysis revealed that P. fluorescens Pf0-1 possesses three genes (Pfl01_0124, Pfl01_0354, and Pfl01_4431) homologous to the Pseudomonas aeruginosa PAO1 pctA gene, which has been identified as a chemotaxis sensory protein for amino acids. When Pf01_4431, Pfl01_0124, and Pfl01_0354 were introduced into the pctA pctB pctC triple mutant of P. aeruginosa PAO1, a mutant defective in chemotaxis to amino acids, its transformants showed chemotactic responses to 18, 16, and one amino acid, respectively. This result suggests that Pf01_4431, Pfl01_0124, and Pfl01_0354 are chemotaxis sensory proteins for amino acids and their genes were designated ctaA, ctaB, and ctaC, respectively. The ctaA ctaB ctaC triple mutant of P. fluorescens Pf0-1 showed only weak responses to Cys and Pro but no responses to the other 18 amino acids, indicating that CtaA, CtaB, and CtaC are major chemotaxis sensory proteins in P. fluorescens Pf0-1. Tomato root colonization by P. fluorescens strains was analyzed by gnotobiotic competitive root colonization assay. It was found that ctaA ctaB ctaC mutant was less competitive than the wild-type strain, suggesting that chemotaxis to amino acids, major components of root exudate, has an important role in root colonization by P. fluorescens Pf0-1. The ctaA ctaB ctaC triple mutant was more competitive than the cheA mutant of P. fluorescens Pf0-1, which is non-chemotactic, but motile. This result suggests that chemoattractants other than amino acids are also involved in root colonization by P. fluorescens Pf0-1.

Certain strains of Pseudomonas fluorescens exert beneficial effects on plants (3,13,17). These strains are able to enhance plant growth indirectly by preventing the growth or action of plant-pathogenic microorganisms. P. fluorescens F113 protects sugar beet seedlings from damping-off disease caused by the fungal pathogen Pythium ultimum by producing the antifungal metabolite 2,4-diacetylphloroglucinol, hydrogen cyanide, and extracellular protease (1,22). P. fluorescens strain 2-79 is able to suppress wheat take-all produced by Gaeumannomyces graminis var. tritici by means of producing antibiotics phenazine-1-carboxylic acid (32,33). P. fluorescens WCS365 is a biocontrol agent against Fusarium oxysporum, which causes tomato foot and root rot (4). This strain is an efficient root colonizer and it is assumed that it can favorably compete for available habitable niches on a root surface with the pathogenic fungus.
Efficient root colonization by plant-beneficial rhizobacteria is assumed to be essential for the biocontrol of root pathogens (30). In previous studies, it was demonstrated that motility and chemotaxis are important traits for root colonization by P. fluorescens. Simons et al. reported that a mutant of P. fluorescens WCS365, lacking flagella was outcompeted in the root-tip colonization assay when inoculated in competition with the parental WCS365 strain (24). Conversely, Barahona et al. showed that a hypermotile mutant of P. fluorescens F113, was more competitive for rhizosphere colonization than the wild-type strain and exhibited improved biocontrol activity against the pathogenic fungus F. oxysporum and the pathogenic oomycete Phytophthora cactorum compared with the wild type strain (3). de Weert et al. demonstrated that cheA mutant of P. fluorescens WCS365, which is nonchemotactic but motile, colonized the tomato root tip less efficiently than the wild-type strain in the competitive root colonization assay (6).
Plant root exudates contain various organic compounds. Major components of tomato root exudate are amino acids (glutamic acid, aspartic acid, leucine, isoleucine, and lysine as major components [25]), organic acids (especially citric acid, malic acid and succinic acid [10]), and sugars (glucose and xylose as major components [14]). Previous studies demonstrated that P. fluorescens strains exhibit chemotactic responses toward plant seed and root exudates and their components (6,20,26,30); therefore, it is supposed that chemotaxis to components of plant root exudates is involved in effective root colonization.
Methyl-accepting chemotaxis proteins (MCPs) are chemotaxis sensory proteins responsible for the detection of chemotactic ligands (11). MCPs are membrane-spanning homodimers and typical features of MCPs are as follows: a positively charged N terminus followed by a hydrophobic membrane-spanning region, a hydrophilic periplasmic domain, a second hydrophobic membrane-spanning region and a hydrophilic cytoplasmic domain (7). Chemotactic ligands bind to periplasmic domains of MCPs and their binding initiates chemotaxis signaling. The diverse ligand specificities among MCPs reflect amino acid sequence diversities of periplasmic domains of MCPs. The C-terminal cytoplasmic domains of MCPs are relatively conserved. A 44-amino-acid highly conserved domain (HCD) is located in the cytoplasmic domain. MCPs from phylogenetically diverse bacteria have been shown to possess HCD (34). Blastp analysis found that in P. fluorescens Pf0-1 (accession number: CP000094), Pf-5 (accession number: CP000076), and SBW25 (accession number: AM181176) genomes, there are 37, 34, and 46 gene products possessing HCD, respectively, and all have been annotated as MCPs; however, it is still unknown which exudate components and MCPs are involved in efficient root colonization by P. fluorescens. In the present study, we identified and characterized MCPs for amino acids, major components of plant root exudates, in P. fluorescens Pf0-1 and investigated their involvement in chemotaxis toward tomato root exudate and root colonization.

Bacterial strains, plasmids, and growth conditions
Bacterial strains and plasmids used in this study are listed in Table 1. Escherichia coli JM109 (18) and S17-1 (23) were used for plasmid construction and transconjugation, respectively. P. fluorescens, Pseudomonas aeruginosa and E. coli strains were grown with shaking in 2× YT medium (18) supplemented with appropriate antibiotics. P. aeruginosa and E. coli strains were cultivated at 37°C, while P. fluorescens strains were grown at 28°C.

Chemotaxis assay
The computer-assisted capillary assay method was carried out as described previously (16). Cells in a 10 µL suspension were placed on a coverslip, and the assay was started by placing the coverslip upside down on the U-shaped spacer to fill the chemotaxis chamber with the cell suspension. Cells were videotaped over 3 min. Digital image processing was used to count the number of bacteria accumulating toward the mouth of a capillary containing a known concentration of an attractant plus 1% (w/v) agarose. The strength of the chemotactic response was determined by the number of bacterial cell per frame. The chemotaxis buffer was 10 mM HEPES (N-2-hydroxyethylpiperazine-N'-ethanesulfonic acid) buffer (pH 7.0).

DNA manipulation
Standard procedures were used for plasmid DNA preparations, restriction enzyme digestions, ligations, transformations, and agarose gel electrophoresis (18). PCR reactions were carried out using KOD Plus DNA polymerase (Toyobo, Tokyo, Japan) according to the manufacturer's instructions. Oligonucleotides used for PCR are listed in Table 2. P. aeruginosa was transformed by electroporation as described previously (15). Plasmids were introduced to P. fluorescens strains by transconjugation using E. coli S17-1 (23).
Plasmid construction and construction of deletion mutants of P. fluorescens Pf0-1 The Pfl01_0124, Pfl01_0354, and Pfl01_4431 genes were amplified from P. fluorescens Pf0-1 genome by PCR using PFL01f/ PFL01r, PFL02f/PFL02r, and PFL03f/PFL03r primer sets, and cloned into broad-host-range plasmid pUCP18 (21) to construct pFLCP01, pFLCP02, and pFLCP03, respectively. Suicide plasmids pUGMF01, pUGMF02, and pUGMF03 were constructed for unmarked gene deletion in P. fluorescens Pf0-1. PCR using primer sets DPFL01Uf/DPFL01Ur and DPFL01Df/DPFL01Dr was conducted to amplify 1.2-kb regions upstream and downstream of Pfl01_0124 from the P. fluorescens Pf0-1 genome, respectively. Amplified upstream and downstream regions were digested with PstI-BamHI and BamHI-EcoRI, respectively, and ligated with the backbone of PstI-EcoRI-digested pK18mobsacB (19) to obtain pUGMF01. PCR using primer sets DPFL02Uf/DPFL02Ur and DPFL02Df/DPFL02Dr was conducted to amplify a 1.4-kb upstream region and a 1.3-kb downstream region of Pfl01_0354 from the P. fluorescens Pf0-1 genome, respectively. Amplified upstream and downstream regions were digested with SalI-BamHI and SphI-SalI, respectively, and ligated with the backbone of SphI-BamHI-digested pK18mobsacB to obtain pUGMF02. PCR using primer sets DPFL03Uf/DPFL03Ur and DPFL03Df/DPFL03Dr was conducted to amplify a 1.3-kb upstream region and a 1.2-kb downstream region of Pfl01_0354 from the P. fluorescens Pf0-1 genome, respectively. Amplified upstream and downstream regions were digested with HindIII-XhoI and XhoI-EcoRI, respectively, and ligated with the backbone of HindIII-EcoRI-digested pK18mobsacB to obtain pUGMF03. The chromosomal Pfl01_0124, Pfl01_0354, and Pfl01_4431 genes were deleted by the unmarked gene deletion technique (19) using suicide plasmids pUGMF01, pUGMF02, and pUGMF03, respectively. Unmarked gene deletion was confirmed by PCR using PCR primers specific to upstream and downstream sites of each gene.

Preparation of tomato root exudate
Exudate was prepared from a plant species, tomato (Solanum lycopersicum cv. Oogatahukuju). Tomato seeds were sterilized by gentle shaking for 10 min in a solution of 8.75% (v/v) sodium hypochloride supplemented with 0.1% (v/v) Tween 20. The sterilized seeds were soaked six times for 15 min in sterile demineralized water. Nine sterile seeds were placed in 3 mL SSE medium (2), consisting of 5 µM KH2PO4, 4 mM CaSO4, 2 mM MgCl2, 2.5 mM NH4NO3, 0.5 mM KOH, 2.5 mM NaOH, and 0.02 mM Fe (as FeEDTA), and were allowed to grow in a climate-controlled growth chamber (NK System, Osaka, Japan) at 28°C, and 16 h of daylight. After 18 days, root exudates were collected and evaporated to dryness at 45°C under a vacuum, dissolved in 1 mL water, and sterilized by membrane filtration (0.45-µm pore size).

Selection for rifampicin resistance mutants
Spontaneous rifampicin-resistant mutants of P. fluorescens were generated by spreading bacterial cells, grown overnight in 2× YT, onto 2× YT agar plates containing 20 µg mL −1 rifampicin. The plates were incubated at 28°C for 20 h to form colonies. The resulting rifampicin-resistant colonies were then streaked on 2× YT agar containing 50 µg mL −1 rifampicin, and Rif r strains were subsequently maintained on this medium. One mutant showing a growth rate similar to that of the wild-type strain was selected and designated Pf0-1Rif. Similarly, a rifampicin-resistant mutant of P. fluorescens FLD3 was obtained and designated FLD3Rif. Oogatahukuju) were sterilized as described at "Preparation of tomato root exudate" section. To synchronize germination, seeds were placed on Petri dishes containing PNS solidified with 1.5% (w/v) Bacto Agar (Becton, Dickinson and Company, Franklin Lakes, NJ, USA) and incubated overnight in the dark at 4°C, followed by incubation at 28°C for 2 days. A germinated seed was aseptically placed at the center of a growth tube and 5 mm below the surface of quartz sand. Bacterial cells were grown for 14 h in 2× YT medium, centrifuged (3,300×g, 2 min), washed three times with PNS, and adjusted to 10 7 CFU mL −1 in PNS. For the colonization assay, 100 µL bacterial cell suspensions were added to the edge of a plant growth tube. For the competitive colonization assay, 100 µL of 1:1 (v/v) mixture of the tested strain and the competitor were mixed and inoculated at the edge of a plant growth tube. The plant growth tubes were incubated in a climate-controlled growth chamber (28°C, 16 h daylight) to allow the plantlets to grow. After 7 days of growth in the plant growth tubes, the root systems of tomato were mostly unbranched. The root tip (1 to 2 cm length) was removed and shaken vigorously in the presence of adhering sand particles in 0.5 mL of PNS to remove bacteria. The bacterial suspension was diluted and 100 µL was plated on 2× YT agar plates. For the competitive colonization assay, the bacterial suspension was spread on 2× YT agar plates with and without rifampicin. For statistical analysis, the nonparametric Wilcoxon-Mann-Whitney test was used (27).

Results
Chemotactic responses toward amino acids by P. fluorescens Pf0-1 Amino acids are one of the major tomato exudate components (25) and strong chemoattractants of P. fluorescens strains (6,26,28); therefore, we focused the present study on the identification of a chemotaxis sensory protein(s) for amino acids in P. fluorescens Pf0-1. We first measured chemotactic responses of P. fluorescens Pf0-1 toward each of twenty commonly-occurring L-amino acids by the computer-assisted capillary assay (16). P. fluorescens Pf0-1 exhibited chemotactic responses toward all twenty amino acids (Table 3). In particular, Cys, Gln, Gly, Ile, Lys, Met, Phe, Pro, and Ser were strong chemoattractants of P. fluorescens Pf0-1, while it showed weak chemotactic responses toward Glu and Trp.

Identification of chemotaxis sensory proteins for amino acids
In Pseudomonas aeruginosa PAO1, PctA, PctB, and PctC have been identified as MCPs for amino acids (12,31). There is 46-70% identity among amino acid sequences of periplasmic domains of these MCPs. To search for the PctA homologue of P. fluorescens Pf0-1, BLASTP analysis was performed on a protein database of P. fluorescens Pf0-1 by using the 250 amino acid sequence of the putative periplasmic domain of PctA (residues 28 to 277 of PctA [accession number: NP 252999.1]) as a query sequence. BLASTP search found that three proteins, Pfl01_4431 (accession number: ABA76168), Pfl01_0124 (ABA71868), and Pfl01_0354 (ABA72098), showed the highest similarity to the query sequence (65%, 58%, and 46% identity, respectively). They possessed HCD in the C-terminal regions and have been annotated as MCPs on the basis of homology. Pfl01_4431, Pfl01_0124, and Pfl01_0354 are 59, 56, and 44% identical to the periplasmic domain of PctB (residues 27 to 277), and 49, 45, and 51% identical to that of PctC (residues 27 to 280).
To investigate whether Pfl01_4431, Pfl01_0124, and Pfl01_0354 act as MCPs for amino acids, their genes were cloned into broad-host-range plasmid pUCP18 (21), the resulting recombinant plasmids were introduced into the pctA pctB pctC triple mutant of P. aeruginosa PAO1 (P. aeruginosa PCT2) ( Table 3. respectively, while PCT2 (pFLCP02) exhibited moderate responses only to Met. No common structural features are found among side chains of amino acids to which both PCT (pFLCP01) and PCT2 (pFLCP03) responded. Based on these results, the Pfl01_4431, Pfl01_0124, and Pfl01_0354 genes were designated ctaA, ctaB, and ctaC (cta: chemotactic transducer of amino acids). We constructed a ctaA ctaB ctaC triple mutant of P. fluorescens Pf0-1 (P. fluorescens FLD3) to assess the possibility of chemotaxis sensory protein(s) other than CtaA, CtaB, and CtaC. The triple mutant FLD3 showed moderate responses only toward Cys and Pro (Table 3). We then examined ctaB ctaC, ctaA ctaC, and ctaA ctaB double mutants of P. fluorescens Pf0-1 (P. fluorescens FL4431, FL0124, and FL0354, respectively) for their chemotaxis to amino acids to investigate the role of each MCP in amino acid chemotaxis in P. fluorescens Pf0-1. The ctaA ctaB double mutant showed a strong response to Met and Cys and weak or moderate responses to Arg, Gly, Pro, and Thr (Table 3). Since P. fluoresces FLD3 showed moderate responses to Cys and Pro, Met is the main chemoattractant of CtaC. The ctaB ctaC and ctaA ctaC double mutants showed strong responses to several amino acids and their response patterns were similar to those of PCT2 (pFLCP03) and PCT2 (pFLCP01), respectively. These results suggest that CtaA and CtaB play the major roles in amino acid chemotaxis in P. fluorescens Pf0-1 (Fig.  1).
Of the major root exudate components other than amino acids, malic acid and succinic acid were strong attractants to P. fluorescens Pf0-1 (data not shown); therefore, we examined P. fluorescens FLD3 for its ability to respond to malic acid and succinic acid. It showed chemotactic responses to malic acid and succinic acid comparable to those by the parental strain (data not shown), suggesting that CtaA, CtaB, and CtaC are not involved in the detection of malic acid and succinic acid.
Chemotaxis of P. fluorescens strains to tomato root exudate P. fluorescens Pf0-1 wild-type and mutant strains were tested for chemotaxis to tomato root exudate to assess the involvement of MCPs for amino acids in chemotaxis to the root exudates. P. fluorescens Pf0-1 wild-type strain was strongly attracted by tomato root exudate, while ctaA ctaB ctaC triple mutant showed much decreased responses (Fig.  2). The double mutants showed stronger responses to root exudate than the triple mutant, but weaker responses than the P. fluorescens Pf0-1 wild-type strain. In particular, ctaA ctaB double mutant showed the weakest responses among the double mutants. This result suggests that amino acids are the major chemoattractants of P. fluorescens Pf0-1 in tomato root exudate. It also suggests that CtaA, CtaB, and CtaC are responsible for chemotaxis to root exudate to various degrees and that CtaC is less responsible than CtaA and CtaB.

Root colonization analysis
In order to investigate the importance of chemotaxis to amino acids in the root colonization process, we examined P. fluorescens Pf0-1, spontaneous rifampicin resistant mutant of Pf0-1 (Pf0-1Rif), the ctaA ctaB ctaC triple mutant, and its spontaneous rifampicin-resistant mutant (FLD3Rif) as well as double mutants for their root-colonizing ability by the gnotobiotic root colonization system. We also tested the cheA mutant of Pf0-1, which is a general non-chemotactic mutant, with the root colonization assay to confirm the report by de Weert et al. that flagella-driven chemotaxis is an important trait for tomato root colonization by P. fluorescens (6). We confirmed that there were no significant differences in growth in LB medium between mutants and the wild type Pf0-1. When germinated tomato seedlings were inoculated with single strains, all mutants colonized the tomato root to the same extent as the wild-type strain (Fig. 3A). We then carried out competitiveness assays between chemotaxis mutants and the wild-type strain by inoculating seedlings with a 1:1 mixture. Because Pf0-1Rif and FLD3Rif competed fully with Pf0-1 and FLD3, respectively (data not shown), we used Pf0-1Rif and FLD3Rif as competitor strains in competitive colonization assays to distinguish the competing strains from  (Table 3).   Fig. 3B were the same level as in Fig. 3A. There were significant (P<0.05) differences in colonization between Pf0-1Rif and ΔcheA, Pf0-1Rif and FLD3, FLD3Rif and ΔcheA, FLD3Rif and Pf0-1, and FLD3Rif and FL0124. tested strains. As previously shown by de Weert et al. (6), the non-chemotactic cheA mutant was a very poor competitor and showed more than 10-fold reduced ability to colonize tomato roots (Fig. 3B). The ctaA ctaB ctaC triple mutant exhibited higher competitive colonization ability than the cheA mutant, but it still showed an approximately 2-fold impaired colonization ability in the competitive colonization assay. This result indicates that chemotaxis to amino acids plays a role in the root colonization process by P. fluorescens. FL0124 (ctaA ctaC double mutant) and FL4431 (ctaB ctaC double mutant) showed almost 2-fold superior colonization ability than FLD3Rif, while FL0354 (ctaA ctaB) showed only 1.3-fold superior colonization ability than FLD3Rif. We then examined competitiveness between ΔcheA (cheA mutant) and FLD3Rif (ctaA ctaB ctaC triple mutant). As shown in Fig.  3B, FLD3Rif is more competitive than cheA mutant.

Discussion
There are two classes of MCPs for amino acids in bacteria. One class includes Tar and Tsr of E. coli and Salmonella enterica serovar Thyphimurium, and the other includes P. aeruginosa PAO1 PctA. Tsr is an MCP for the attractants Ser, Ala and Gly, while Tar is an MCP for attractants Asp and Glu (29). These MCPs possess short periplasmic domains (ca. 150 amino acid residues), and their ligand specificity is relatively narrow. P. aeruginosa PAO1 PctA, which detects 18 commonly-occurring L-amino acids, shows broader ligand specificity, and its periplasmic domain (ca. 240 amino acid residues) is longer than those of Tsr and Tar (12). CtaA and CtaB, the main MCPs for amino acids in P. fluorescens Pf0-1, detect 16 amino acids (Fig. 1, Table 3), and possess long periplasmic domains, suggesting that CtaA and CtaB belong to the class of PctA-type MCPs. BLASTP search of genome databases revealed that like P. aeruginosa PAO1 and P. fluorescens Pf0-1, other Pseudomonas bacteria possess 2-4 PctA-type MCPs, but not Tar/Tsr homologues. Thus, amino acids are supposed to be major chemoattractants of Pseudomonas bacteria.
de Weert et al. reported that the cheA mutant of P. fluorescens WCS365 was much less competitive for tomato root colonization than the wild-type strain and concluded that flagella-driven chemotaxis toward root exudate is an important trait for competitive root colonization (6); however, it has not been identified which chemotaxis ligands in root exudate are involved in effective root colonization. In this study, we demonstrated that ctaA ctaB ctaC mutant impaired in chemotaxis to amino acids showed a significant reduced ability to colonize tomato root in competitive root colonization assays using the wild-type strain as the competitor strain (Fig. 3B). This triple mutant was impaired in chemotaxis to amino acids (Table 3), but showed parental responses to malic acid and succinic acid, another major components of root exudate. These results suggest that amino acids play a role as chemoattractants for effective root colonization. Data in Table 3 and Fig. 1 suggest that CtaA and CtaB are the major MCPs for amino acids and root exudate and that CtaC contributes less to chemotaxis to amino acids and root exudate. Consistent with these data, CtaA and CtaB also make a greater contribution to the root colonization process than CtaC (Fig. 3B). In this study, we demonstrate that chemotaxis to amino acids is involved in effective root colonization by P. fluorescens, but it remains to be elucidated which amino acids are involved in this process.
Additionally, ctaABC mutant was more competitive for root colonization than cheA mutant (Fig. 3B), suggesting that chemoattractants other than amino acids are involved in root colonization. Since P. fluorescens Pf0-1 exhibits marked responses to organic acids (especially succinic acid and malic acid) (data not shown), we suppose that chemotaxis to organic acids such as succinic acid and malic acid is also involved in root colonization by P. fluorescens. We are now searching MCPs for organic acids in P. fluorescens Pf0-1 to investigate the involvement of chemotaxis to organic acids in root colonization.