The two-component colR/S system ofPseudomonas£uorescens WCS365 plays a role in rhizosphere competence through maintaining the structure and function of the outer membrane

Pseudomonas ﬂuorescens strain PCL1210, a competitive tomato root tip colonization mutant of the efﬁcient root colonizing wild type strain WCS365, is impaired in the two-component sensor-response regulator system ColR/ColS. Here we show that a putative methyltransferase/ wapQ operon is located downstream of colR/colS and that this operon is regulated by ColR/ColS. Since wapQ encodes a putative lipopolysaccharide (LPS) phosphatase, the possibility was studied that the integrity of the outer membrane of PCL1210 was altered. Indeed, it was shown that mutant PCL1210 is more resistant to various chemically unrelated antibiotics which have to pass the outer membrane for their action. In contrast, the mutant is more sensitive to the LPS-binding antibiotic polymyxin B. Mutant PCL1210 loses growth in competition with its wild type when grown in tomato root exudate. Mutants in the methyltransferase /wapQ operon are also altered in their outer membrane permeability and are defective in competitive tomato root tip colonization. A model for the altered outer membrane of PCL1210 is discussed.


Introduction
Decreases in crop yields are a world-wide problem. To solve this problem is not only of economical importance but it is also important for the environment. An alternative for chemical pesticides can be the use of beneficial microorganisms. To understand the molecular basis of the control of pathogens by beneficial bacteria it is necessary to study traits essential for biocontrol. Efficient colonization of root systems by beneficial microorganisms is one of the most important traits necessary to establish microbiological control . After transposon mutagenesis some genes and traits involved in root colonization have been identified . For example, mutant PCL1210, a transposon mutant of the efficient root colonizing biocontrol strain Pseudomonas fluorescens WCS365 , was initially isolated as a competitive potato root tip colonization mutant (Dekkers et al., 1998). Similarly, the mutant is also impaired in root colonization on various other plants, indicating a broad host range mutation (Dekkers et al., 1998). Three ORFs (open reading frames) are present on the fragment flanking the transposon. The first two ORFs of the complementing region, colR and colS, show strong homology with known members of two-component regulatory systems (Dekkers et al., 1998). The third one encodes a methyltransferase. Two-component systems, of which more than 50 different sets are presumed to be present in all bacterial species (Rodrigue et al., 2000), enable bacterial cells to respond adequately to a large variety of environmental stimuli. These adaptive responses range from very fast changes necessary for a chemotactic response (Swanson et al., 1994) to changes that in the end lead to the production of secondary metabolites (Laville et al., 1992). The mutation in mutant PCL1210 is thought to lead to a decreased responsiveness towards the rhizosphere environment that eventually leads to impaired competitive colonization fitness (Dekkers et al., 1998). Here we report that further sequencing of DNA downstream colR and colS revealed the methyltransferase/wapQ operon which appears to be regulated by the colR/S system. Further studies on the methyltransferase/wapQ operon suggest involvement in the fine-tuning of the permeability function of the outer membrane. A possible model is proposed and discussed in this paper.

Materials and methods
Bacterial strains and growth conditions Overnight cultures of P. fluorescens WCS365 and mutant derivatives were grown at 28 1C in liquid King's medium B (KB;King et al., 1954) under vigorous aeration or on solidified medium containing 1.8% agar. When appropriate, the media were supplemented with kanamycin or tetracycline in final concentrations of 50 and 80 mg mL À1 , respectively. To distinguish between P. fluorescens WCS365 (white/ yellow colonies) and mutant PCL1210 (blue colonies), 5-bromo-4-chloro-3-indolyl-b-D-galactoside (X-gal) was used in a final concentration of 40 mg mL À1 as an indicator for b-galactosidase activity.

Antibiotic sensitivity
For experiments in which antibiotic resistance was measured, 0.1 mL of an overnight culture of PCL1210 or WCS365 in KB was diluted into 3 mL of KB top agar (0.6%). The mixture was poured onto the surface of KB plates and was allowed to solidify. Filter paper discs with a diameter of 6 mm (Filter paper GB002, Schleicher and Schuell, Dassel, Germany) were soaked in solutions of various antibiotics and placed on the agar plates. Antibiotic concentrations in the solutions were spectomycin, 100 mg mL À1 ; tetracycline, 8 mg mL À1 ; rifampicin, 2 mg mL À1 ; chloramphenicol, 10 mg mL À1 ; and streptomycin, 500 mg mL À1 . After overnight growth at 28 1C, sensitivity towards the antibiotics was measured by determining the mean (n = 8) inhibition zone. The obtained data were analyzed for significance with Student's t-test (P = 0.05), using SPSS software (SPSS Inc., Chicago, IL).
A second method used to determine and quantify sensitivity towards antibiotics was performed by growing bacteria overnight and plating these subsequently on increasing concentrations of the appropriate antibiotic.
Sensitivity towards paraquat, a superoxide-generating agent, was measured by inoculating 50 mL of an overnight culture into 5 mL of liquid KB supplemented with different concentrations of paraquat, followed by overnight growth at 28 1C and measurement of the OD 620 nm .
Polymyxin B sensitivity was determined using overnight cells grown in BM2 succinate medium (Gilleland Jr. et al., 1974) supplemented with 0.5 mM MgSO 4 . Cells were diluted 100-fold in the same medium as was used for growth in overnight culture except that it was supplemented with different concentrations of polymyxin B. After growth overnight at 28 1C the OD 620 nm was measured.

Growth in competition
Competitive growth on root exudate or putrescine (5 mM) as the sole carbon source was performed using overnight grown cells of mutant PCL1210 and wild type WCS365 in liquid basic medium (BM; Lugtenberg et al., 1999). Cells were washed with sterile phosphate-buffered saline (PBS) (Hoffland et al., 1989) and subsequently diluted in BM containing root exudate or putrescine to a final OD 620 nm of 0.1. Equal volumes of these cultures were mixed. The mixture was allowed to grow overnight at 28 1C and subsequently diluted again 1000-fold in fresh root exudate or putrescine. This procedure was repeated twice. The ratio between the wild type and the mutant cells was determined at various time intervals by plating diluted samples on solid KB medium supplemented with or without kanamycin.

General DNA techniques
All general DNA techniques, including restriction enzyme digestion, ligations, plasmid isolations, transformation of Escherichia coli and nucleotide sequencing were performed as described by Sambrook & Russel (2001).

Analysis of membrane protein and lipopolysaccharide patterns
Analysis of cell envelope protein patterns (Lugtenberg et al., 1975), and analysis of lipopolysaccharide (LPS) ladder patterns (de Weger et al., 1987) by SDS-polyacrylamide gel electrophoresis was performed as described previously (Goldman & Leive, 1980;Palva & Makela, 1980). Cell envelope protein patterns were analyzed after growth of cells in KB and of cells in BM2 medium supplemented with low (0.02 mM) and high (0.5 mM) Mg 21 concentrations. Proteins that were tightly associated with the peptidoglycan were isolated as described by Lugtenberg et al. (1977). Cell envelopes that were treated with trypsin were prepared as described previously (Overbeeke & Lugtenberg, 1980).

Rate of permeation of ampicillin through the outer membrane
The rate of permeation of the b-lactam antibiotic ampicillin through the outer membrane of whole cells was measured using a method originally described by Zimmermann & Rosselet (1977) and modified by Overbeeke & Lugtenberg (1982). Briefly, after overnight growth in KB medium, cells were spun down and washed with buffer A (10 mM HEPES-5 mM MgCl 2 , pH 7.0) and set to an OD 660 nm value of 0.1. Subsequently the assay was started by the addition of different b-lactam antibiotics at a concentration of 8 mM. After 15 min the reaction was stopped by the addition of freshly prepared stop-mix [containing 8 mM I 2 , 0.32 M KI, 2 M HAc, 1 M Na 2 WO 4% and 2% (w/v) starch]. After an additional incubation of 20 min for the colour development, the absorbance of the solution was measured at an optical density of 623 nm.
Competitive tomato root tip colonization assays in a gnotobiotic sand system Competitive root tip colonization assays were performed as described by Simons et al. (1996) using the gnotobiotic system containing sterile quartz sand to which 10% (v/w) PNS (plant nutrient solution) (Hoffland et al., 1989) was added to moisten the sand. For colonization experiments, sterile germinated tomato (Lycopersicon esculentum Mill. cv. Carmello; Syngenta, Enkhuizen, the Netherlands) seedlings were inoculated with a mixture of parental strain P. fluorescens WCS365 and one of its mutants in a 1 : 1 ratio. The seedlings were placed in the tubes and allowed to grow in a climate-controlled growth chamber at 18 1C, 70% relative humidity, and 16 h of daylight. After 7 days, root tips (1 cm in length) were isolated with adhering sand, and bacteria were removed from the root parts by shaking in 1.0 mL PBS. The CFU of parental and mutant cells were determined by plating dilutions on solidified KB medium and on the same medium supplemented with kanamycin. The detection limit is 100 CFU mL À1 . All results were statistically analyzed using the nonparametric Wilcoxon-Mann-Whitney test (Sokal & Rohlf, 1981). To avoid log 0 cases, calculations were carried out using log (CFU11) cm À1 root tip.

b-galactosidase assay
Cells were grown in root exudate or BM medium with 5 mM putrescine as the sole carbon source at 28 1C. b-galactosidase assays were performed as described by Miller (1972).

Characterization of genes downstream colR/colS
Mutant PCL1210 was obtained as a result of the screening of a Tn5lacZ mutant bank of the good root colonizer P. fluorescens WCS365 (Dekkers et al., 1998). The transposon was found to be inserted in the colS gene of a colR/colS twocomponent regulatory system. Further sequencing of the DNA located downstream of colR/S revealed two open reading frames (genbank accession no. Y09798). The first open reading frame, orf222, shows 80% identity and 87% similarity at the amino acid level to a SAM-dependent methyltransferase of P. fluorescens PfO1 (Genebank accession no. ZP0264083). The predicted amino acid sequence of the second ORF shows significant homology with the KDO, lipopolysaccharide kinase family (pfam06 293), such as inaA of E. coli (Carlson et al., 1984;White et al., 1992) and the putative heptose kinase wapQ of Pseudomonas aeruginosa (Walsh et al., 2000). Upstream of the methyltransferase two putative s70 promoters could be identified, of which the first one is located at bp 2936-18 bp-2912 (-10--35) and the second one at bp 2926-16 bp-2940 (-10--35). Both ORFs are preceded by a Shine Dalgarno sequence that is correctly spaced from both start codons. As no putative promoter sequence could be identified between the methyltransferase and wapQ, they probably comprise one transcriptional unit in P. fluorescens WCS365 (Fig. 1). No significant homology could be identified further downstream of wapQ. In P. aeruginosa PAO1 a similar highly conserved gene cluster containing five genes can be identified (PA4377-PA4381). In contrast to the four genes we found that a very small gene, PA4377, is present in P. aeruginosa PAO1 (http://www.pseudomonas.com/) downstream of WapQ.
ColR/colS regulates the methyltransferase/ wapQ operon It could be possible that the ColR/S system is regulating the downstream orf222-inaA/wapQ operon. To test this possibility, the promoter of the methyltransferase gene was cloned in front of a promoterless lacZ gene in vector pML103 (Labes et al., 1990), resulting in pMP5579. As a control, a plasmid in which the promoter of the methyltransferase is present in the wrong orientation was constructed (pMP5580 , Table 1). Subsequent b-galactosidase activity assays showed that pMP5579 indeed showed promoter activity, whereas the control construct did not (Fig. 2). Moreover, when cells were grown in tomato root exudate or BM containing 5 mM of putrescine, promoter expression in the colR/colS mutant, PCL1251, was significantly decreased, relative to that in the wild type (Fig. 2). This result shows that the methyltransferase/wapQ operon belongs to the regulon of the ColR/ColS two-component system.

Outer membrane integrity
The in silico analysis of wapQ suggests that the encoded protein could play a role in LPS modification (phosphorylation) and therefore in outer membrane functioning. Since the ColR/ColS proteins (co)regulate wapQ, we tested the effect of the colR/colS mutation on a number of outer membrane processes. No differences were observed between the parental strain and mutant, PCL1210, concerning the LPS ladder patterns or cell envelope protein patterns (data not shown), however, phosphorylation and/or methylation differences are also not expected to change the electrophoretic mobility of a large molecule such as LPS. Subsequently, the sensitivity towards a wide range of antibiotics was tested on plates on which filter paper discs soaked in various solutions of these compounds were placed. This analysis showed that mutant PCL1210 was slightly but significantly more resistant towards the antibiotics streptomycin (Fig. 3), rifampicin, spectinomycin and tetracycline (data not shown). Since polymyxin B binds to LPS (Nikaido & Vaara, 1987), the colS/colR mutant could show changed sensitivity towards this compound. The observation that the growth of this mutant is more sensitive to polymyxin B than the growth of parental strain is (Fig. 4) supports this notion. To study whether the increased sensitivity of the PCL1210 was due to the effect on the methyltransferase/wapQ expression, we also constructed mutants in both these genes. For the construction of a mutant in the wapQ homologue of WCS365, the forward primer was designed based on WCS365 sequence while the reverse primer was designed based on an alignment of wapQ homologues of P. aeruginosa and Pseudomonas putida. The resulting mutant is referred to as PCL1823 (Table 1). The mutants were not impaired in growth rate when grown in KB medium (data not shown). Indeed, both these mutants showed similar sensitivity results (data not shown). In addition, mutant PCL1210 appeared to be more resistant towards paraquat (Fig. 5).

Rate of permeation of ampicillin through the outer membrane
To analyze directly whether the outer membrane of the mutant is altered, the rate of permeation of ampicillin through the outer membrane was measured. The b-lactamase present in the periplasmic space will open the b-lactam ring of those ampicillin molecules that have diffused through the outer membrane protein pores after exogenous application of the antibiotic (Pugsley et al., 1987). The bla gene of pMMB67HE (Furste et al., 1986), a plasmid which appeared not to be stable in P. fluorescens, was cloned as a HindIII fragment into the broad host range vector pBBR5MCS5 (Kovach et al., 1995), selecting for gentamycin resistance. The resulting plasmid pMP4068 (Table 1) was electroporated into strains WCS365 and PCL1210. The respective resulting strains PCL1830 and PCL1831 (Table  1) were used to determine the rate of permeation of ampicillin through the outer membrane. The results of three experiments showed that the b-lactam antibiotic ampicillin is taken up efficiently by the wild type strain, with an average rate of permeation of 6.2 nmol À1 min À1 10 8 cells. In contrast, the relative rate of b-lactam uptake of mutant PCL1210 was only 18% of that of the wild type. This result indicates that the transport of ampicillin through the general porin(s) was severely affected in the mutant. colS mutant PCL1210 loses growth in competition on root exudate and its main component putrescine Because differences in outer membrane permeability, such as the uptake of putrescine, were observed, we reasoned that the colonization defect of the colR/colS mutant could be due to the slower uptake of root exudates components. There-fore, a growth competition was performed in BM medium containing 5 mM of the major root exudate component putrescine as the sole carbon source. The results (Fig. 6) show that the mutant loses slowly but gradually in competition from its parental strain. After approximately 30 generations four times more wild type cells are present than mutant cells (Fig. 6).

Competitive colonization of a methyltransferase-and wapQ mutant
To test whether the methyltransferase and wapQ genes downstream of colR/colS are indeed involved in root colonization, new mutants were constructed by homologous recombination. Mutant PCL1812 is mutated in a gene homologous to a methyltransferase gene ( Table 1). The competitive colonizing behaviour of P. fluorescens WCS365 and its mutants on tomato roots was analyzed in the gnotobiotic sand system as described by Simons et al. (1996). Both mutants PCL1812 (methyltransferase) and PCL1823 (wapQ) showed a fivefold decrease in root tip colonizing ability (Table 2).

Discussion
For biological control root colonization is an essential trait. If a bacterium is not able to colonize the root system efficiently, its biocontrol activity is decreased or completely absent (Chin-A-Woeng et al., 1998;Kamilova et al., 2005). When root colonization is studied, CFU are determined from the root tip. Distribution of inocula along the root system shows a gradient (Simons et al., 1996). The major differences between a wild type and a colonization mutant are found on the root tip since that part of the root is most difficult to reach (Chin-A-Woeng et al., 1997;de Weert et al., Fig. 6. Competitive growth of Pseudomonas fluorescens WCS365 (dark bars) and colS mutant PCL1210 (grey bars) in BM with root exudate (a) or putrescine (b) as the sole carbon source. At time point zero individual cultures of both strains were mixed and the number of CFU of each strain was determined. The mixture was diluted 1000-fold, incubated overnight after which the number of CFU of each strain was determined again. This procedure was repeated twice.

2002). A criterion for a competent root colonizer is that it
can efficiently reach the root tip. Unravelling the mechanism behind root colonization is necessary to be able to improve the biocontrol abilities of microorganisms. In this way we can work in an environmentally friendly way on controlling phytopathogens. In a previous study it has been shown that colS mutant PCL1210 is impaired in competitive root tip colonization (Dekkers et al., 1998). Here we show that a two-gene operon encoding a methyltransferase and a wapQ is present downstream of the colR/colS operon (Fig. 1).
In E. coli the gene inaA is induced by acidification of the external medium. Expression increases with decreasing intracellular pH and with decreasing temperature. In general, inaA expression correlated with a decrease in intracellular pH (White et al., 1992). In E. coli inaA is regulated by both the soxRS system, which regulates superoxide stress response, and the MAR system, which regulates multiple antibiotic resistance (Rosner & Slonczewski, 1994).
The wapQ gene of P. aeruginosa encodes a putative heptose kinase. LPS in the outer membrane contain several heptoses, which are generally phosphorylated. However, the extent of LPS modification is known to vary. For instance, P. fluorescens LPS is more heavily phosphorylated at higher temperatures and this modification is probably affecting the pore size of the main porin of P. fluorescens OprF (El Hamel et al., 2000). LPS is known to be strongly associated with bacterial porins of P. aeruginosa and E. coli. It is among others involved in the channel activity of some porins. The negatively charged phosphates on heptose are assumed to contribute to stabilization of the outer membrane structure (Lugtenberg & van Alphen, 1983). This notion, together with the finding that colS mutant PCL1210 is more resistant to various chemically unrelated antibiotics, suggested to us a model (Fig. 7) in which passive diffusion of these molecules through the outer membrane pore protein, which interacts with LPS (Lugtenberg & van Alphen, 1983) is less efficient in the mutant.
As two-component systems often regulate genes in their close proximity (Swanson et al., 1994), we tested whether colR/colS regulates the expression of the methyltransferase/ wapQ operon. This appeared to be the case although substantial expression still occurs in the ColR/ColS mutant (Fig.  2). Because inaA expression in E. coli is regulated by at least two regulatory systems, soxRS and MAR (Rosner & Slonczewski, 1994), a similar complex regulation may occur in Pseudomonas. This would explain the high residual activity of the methyltransferase/wapQ promoter in a colR/colS background (Fig. 2). The proposal that colR/colS is required for optimal expression of the methyltransferase/wapQ operon is further supported by the observation that like colS mutant PCL1210 (Dekkers et al., 1998) both mutants PCL1812 (methyltransferase) and wapQ mutant PCL1823 are impaired in competitive tomato root tip colonization ( Table 2).
The observation that the rate of degradation of ampicillin by periplasmic b-lactamase is 18% of wild type values in mutant PCL1210 shows that the colS mutation negatively affects the permeability of the outer membrane. Consistent with this conclusion is the observation that PCL1210 loses competition for root exudate and the major root exudate component (Kuiper et al., 2001;de Weert et al., 2006) putrescine (Fig. 6). The increased sensitivity of PCL1210 for the antibiotic polymyxin B (Fig. 4), which attacks the outer membrane from the outside by binding to LPS (Nikaido & Vaara, 1987), indicates that, in the mutant, LPS molecules have indeed changed (Ernst et al., 1999). This observation is consistent with a decrease of phosphorylation of heptose in the inner core region of LPS caused by a decreased activity of the WapQ heptose kinase as suggested by the model. It is essential that phosphate residues are present at two locations in the inner core-lipid A region per LPS molecule in wild  In every experiment 10 seedlings were inoculated and plants were individually processed after growth. When values in the same row are followed by a different letter, they are significantly different at P = 0.05 on nonparametric multiple comparisons using the Wilcoxon-Mann-Whitney test.
type cells (Wilkinson, 1981;Walsh et al., 2000), however the third phosphorylation site of heptose I residue is not essential in P. aeruginosa (Walsh et al., 2000). We propose the following working model. The ColS sensor of the ColR/ColS two-component system is activated by an unknown environmental stimulus. As a consequence ColS is autophosphorylated and subsequently transfers its phosphate residue to ColR. Phosphorylated ColR activates the methyltransferase/wapQ operon. The wapQ product is homologous to a P. aeruginosa heptose kinase responsible for phosphorylating heptose I in the LPS core region (Walsh et al., 2000) The resulting negatively charged phosphorylated LPS is thought to interact with a positively charged part of the pore protein, thereby contributing to keeping the pore open. Considering the possible function of this gene as a methyltransferase, it is possible that modified methylation of LPS also plays a role in the altered outer membrane properties of colS mutant PCL1210. Several antibiotics are able to diffuse through the pore of wild type cells. In cells of colS mutant PCL1210 the heptose kinase encoded by wapQ is less active. As a consequence, the phosphate residues that help to keep the pore open in wild type cells are absent in mutant cells which therefore have a narrower pore through which antibiotics diffuse with a slower rate. Moreover, the space between the pore protein and the LPS molecule at the outside of the cells becomes wider in the mutant which enables polymyxin B to bind to LPS more efficiently. The narrower pore opening in mutant cells makes these cells less efficient than those of the wild type in competing for the low nutrient levels in the rhizosphere. Consequently, the latter cells lose in competitive tomato root tip colonization.
In conclusion, the obtained results support the model described above and presented in Fig. 7 implying that the ColR/ColS regulatory system is required for optimal expression of the methyltransferase/wapQ operon and that wapQ encodes a heptose kinase. Under-phosphorylation of heptose causes a narrowing of the protein pores in the outer membrane which results in a higher resistance towards several antibiotics and a poorer uptake of the major root exudate component putrescine when presented in a low concentration. Considering the possible function of the downstream gene as a methyltransferase, it is possible that modified methylation of LPS also plays a role in the altered outer membrane properties of colS mutant PCL1210. All this leads on the tomato root to a decreased ability of the mutant, after mixed inoculation of the seed, to reach the root tip in competition with the wild type.