Metabolic Adaptation of Ralstonia solanacearum during Plant Infection: A Methionine Biosynthesis Case Study

MetE and MetH are two distinct enzymes that catalyze a similar biochemical reaction during the last step of methionine biosynthesis, MetH being a cobalamin-dependent enzyme whereas MetE activity is cobalamin-independent. In this work, we show that the last step of methionine synthesis in the plant pathogen Ralstonia solanacearum is under the transcriptional control of the master pathogenicity regulator HrpG. This control is exerted essentially on metE expression through the intermediate regulator MetR. Expression of metE is strongly and specifically induced in the presence of plant cells in a hrpG- and metR-dependent manner. metE and metR mutants are not auxotrophic for methionine and not affected for growth inside the plant but produce significantly reduced disease symptoms on tomato whereas disruption of metH has no impact on pathogenicity. The finding that the pathogen preferentially induces metE expression rather than metH in the presence of plant cells is indicative of a probable metabolic adaptation to physiological host conditions since this induction of metE occurs in an environment in which cobalamin, the required co-factor for MetH, is absent. It also shows that MetE and MetH are not functionally redundant and are deployed during specific stages of the bacteria lifecycle, the expression of metE and metH being controlled by multiple and distinct signals.


Introduction
Ralstonia solanacearum is a Gram-negative soil-borne b-proteobacterium which is the causal agent of bacterial wilt, one of the most devastating bacterial plant diseases in the world. The strong impact of this pathogen results from its worldwide geographical distribution and its wide host range, since it affects more than 200 plant species, including important food crops such as tomato, potato, banana and eggplant [1]. R. solanacearum infects plants via wounds, the root tip, or cracks at the sites of lateral root emergence. The bacteria subsequently colonize the root cortex, invade the xylem vessels and reach the stem and aerial parts of the plant through the vascular system [2,3]. R. solanacearum can rapidly multiply in the xylem up to very high cell densities, leading to wilting symptoms and plant death.
As soon as it enters the host root tissue, R. solanacearum has to face hostile environmental conditions due to plant defence reactions and to limited nutritional resources in the plant apoplasm. The pathogen relies on a complex regulation network that coordinates the expression of its various pathogenicity determinants according to the environmental stimuli [4]. A key player of this regulatory system is HrpG, which directly or indirectly controls the transcriptional induction of more than 350 genes in the presence of plant cells, including those directing the synthesis of the Type III Secretion System (T3SS) and effector substrates which is essential to pathogenesis [5,6,7]. HrpG integrates multiple environmental signals and controls, beyond the T3SS, many other bacterial functions that promote disease (reviewed in [8]).
In order to mobilize efficiently the host resources and to ensure its multiplication inside plant tissues, the bacterium has to deploy energy to set up the necessary elements required for the success of the infectious process. This implies that the pathogen has to be metabolically proficient, and little is known about the specific metabolic requirements once the pathogen invades its host. Some R. solanacearum mutants deficient for methionine, tryptophan or leucine [9,10] or folate [11] biosynthesis were found to be avirulent on plants. However, because these mutants were auxotrophic, the loss of virulence resulted presumably from nutritional deficiency, which is the case if the host fails to provide an adequate source of the required metabolite in the infection court.
Genetic robustness, which characterizes the constancy of the phenotype in face of heritable perturbations, has been attributed to two main causes. One is gene duplication and the other is the environmental specificity of some of the dispensable genes [12]. Since the host plant represents a specific environment for plant pathogens, we were interested to look for the possible existence of metabolic genes of R. solanacearum involved in the adaptation to the host plant. In this study, we studied the role of methionine biosynthesis in R. solanacearum pathogenicity, focusing our analysis on two enzymes, MetE and MetH, which perform a similar biochemical reaction and apparently look as functionally redundant. Methionine is a sulfur-containing amino acid which, apart from its function as a structural component of proteins plays also an important role in methyl transfer reactions via its derivative S-adenosylmethionine [13]. The methionine biosynthetic pathway and its regulation differ among various groups of organisms. Main differences are found in the nature of the acylhomoserine intermediate formed in the first step of the pathway and in the method of assimilation of the sulfur atom ( Figure 1). From what can be deduced from the genomic data, R. solanacearum strain GMI1000 synthesises homocysteine from homoserine through the consecutive action of a homoserine-O-actetyltranseferase (encoded by metX and metW, RSc0027-RSc0026) and a homocysteine synthase (encoded by metZ, RSc1562). The first part of this pathway is different from the E. coli biosynthesis pathway and resembles to the one described in P. aeruginosa or P. putida [14]. The last step involves the methylation of homocysteine to produce methionine: this reaction is catalyzed by a cobalamin (vitamin B12)dependent transmethylase, the metH gene product, or by a cobalamin-independent transmethylase, the metE (RSp0676) gene product. In GMI1000, MetH is encoded by two genes, metHa (RSc0295) and metHb (RSc0294), organized in an operon.
In E. coli (which does not synthesize vitamin B12), the availability of the vitamin cofactor determines which transmethylase is expressed since cobalamin represses the expression of metE [15]. Both the metE and metH genes, and metA, require an activator protein encoded by metR in order to be efficiently expressed [16]. Another regulatory protein, MetJ, acts as a transcriptional repressor of expression for all the genes of the methionine biosynthetic pathway [17]. R. solanacearum GMI1000 can synthesize cobalamin [18] and possesses a metR gene (RSp0677) lying just next to metE on the megaplasmid replicon, but does not harbour any gene related to metJ. Our data show that regulation of methionine biosynthesis is operated differently in R. solanacearum than in E. coli, and is controlled by the master pathogenicity regulator HrpG. We also present evidence that methionine biosynthesis significantly contributes to R. solanacearum pathogenicity, although methionine auxotrophy has limited or no impact on bacterial growth in plant tissues.

HrpG Regulates the Expression of Methionine Synthase Genes
The list of 112 genes positively regulated by HrpG in strain GMI1000, previously identified through transcriptomic analyses, comprised the methionine synthase cobalamin-independent gene metE [5]. To better investigate the role of HrpG in the regulation of methionine synthase genes, lacZ genomic fusions were created in the metE and metHab genes and expressed in different genetic backgrounds. The resulting strains were cultivated in minimal medium supplemented with glutamate which is known to induce HrpG activity [19]. Figure 2 shows the level of b-galactosidase activity monitored for each strain in this condition. The metE was more expressed in minimal medium than metHab genes since the bgalactosidase activity of the metE::lacZ fusion is three times superior to the one measured for the metH::lacZ fusion in the wild type strain. These data also confirmed that the expression of metE depends on the presence of HrpG but not on HrpB, the transcriptional activator of the T3SS, which is also under the control of HrpG ( Figure 2A). The metE::lacZ fusion was expressed as much in the wild type strain as in the hrpB mutant background but expression was decreased by a 10-fold factor in the hrpG mutant. Expression of metH was also modulated by HrpG as the bgalactosidase activity of the metH::lacZ fusion was decreased by a 2.5-fold factor in the hrpG mutant compared to its expression in the wild type ( Figure 2B). These results showed that HrpG controls expression of genes involved in the last step of methionine biosynthesis: this control is exerted independently from HrpB and appears tighter on metE than on metHab.

HrpG-dependent Regulation of metE is Exerted via the Regulator MetR
In E. coli, expression of metE and metH genes is known to be controlled by the MetR regulator [16]. To investigate the role of MetR on metE expression, the metE::lacZ fusion was introduced in a metR mutant background and the b-galactosidase activity was monitored after growth of the strain in minimal medium supplemented with glutamate. Expression of metE was abolished in a metR mutant. These results confirmed that expression of metE in R. solanacearum is controlled by MetR as almost no bgalactosidase activity was measured in the metR mutant genetic background ( Figure 2A). However, no significant decrease of metHab expression was measured in the metR mutant background (data not shown), showing that metE and metHab are differentially controlled by MetR.
Finally, a plasmid-borne metR::lacZ fusion was constructed and expression of this construct was reduced of a 2.5-fold factor in the hrpG mutant vs the wild-type strain, indicating that metR expression was dependent on HrpG ( Figure 2C). Taken together these results suggested that the HrpG-dependent regulation of metE is indirect and mediated by MetR whereas the expression of metHab escapes this control. These results were also supported by the observation that the metE promoter possesses an E. coli MetR-like consensus recognition sequence (TGAANNNNNTTCA) [20] but no similar sequence was found in the metH gene promoter region.
R. solanacearum metE, metH and metR Mutants are not Auxotrophic for Methionine As in other microorganisms possessing two methionine synthase genes, the metE and metH single mutants were not auxotrophic for methionine and showed similar growth rates as the wild type strain GMI1000 in minimal medium supplemented with glutamate ( Figure 3). Here, the obvious functional redundancy of these genes explains the prototrophy phenotypes. A main difference between the MetE and MetH synthases is whether they require for activity cobalamin as a co-factor. We confirmed that the GMI1000 MetE and MetH synthases indeed behave as cobalamin-independent and -dependent enzymes, respectively. Disruptions of the metE and metHab genes were introduced in the cobalamin (cobN) mutant strain GMI 1783 (see Material & Methods). The resulting strains GMI1784 (cobN metE) and GMI1785 (cobN metHab) were cultivated on minimal medium. Only the cobalamin, metH double mutant was able to grow on minimal medium proving that the presence of metE gene enables the strain to synthesise methionine in the absence of cobalamin. Inversely, the cobalamin, metE double mutant was only able to grow when minimal medium was supplemented with cobalamin, proving that MetH requires the presence of cobalamin to synthesise methionine ( Figure S1).
More surprisingly the R. solanacearum metR single mutant was not auxotrophic for methionine whereas the corresponding mutant is described to be auxotrophic in E. coli [16]. Indeed a metR mutant showed a growth rate in minimal medium similar to the metE and metH single mutants as well as to the wild type strain ( Figure 3). To better understand the prototrophy of the metR mutant, double mutants were created and tested for their auxotrophy for methionine. As expected, the metE metH double mutant was auxotrophic for methionine, thus proving that no other enzyme is able to perform the last step of methionine biosynthesis pathway in R. solanacearum. On the other hand, the metE metR double mutant was not auxotrophic for methionine and showed no differential growth rate in minimal medium compared to the wild type strain. The prototrophy of metE metR double mutant compared to the auxotrophy of the metE metH double mutant suggested that in the metE metR double mutant the expression of metH genes was sufficient to relieve the auxotrophy for methionine in this strain. This hypothesis was validated by creating a metR metH double mutant, which was subsequently found to be auxotrophic for methionine ( Figure 3). This result confirmed that the expression of  . Growth curves of each single and double mutant in the metE, metH or metR genes. Growth was measured and compared to the wild type strain in minimal medium supplemented with glutamate at a 20 mM final concentration using a microplate reader. six replicates were measured for the non auxotrophic strains and three replicates were used for the auxotrophic strains. Results represent the mean value for all replicates at each time point. The metE metH and metH metR double mutant strains are auxotrophic for methionine. doi:10.1371/journal.pone.0036877.g003 metH in R. solanacearum is not dependent on MetR. Furthermore the observation that a hrpG metH double mutant is not auxotrophic for methionine suggested that in this mutant a residual expression of metR and, in turn, of metE is sufficient to relieve methionine auxotrophy.

Expression of the Methionine Synthase Gene metE, but not metH, is Specifically Induced in the Presence of Plant Cells
Since expression of hrpG is induced in the presence of plant cells [19], we determined whether the expression of methionine synthase genes was also specifically induced in this condition. Strains carrying reporter lacZ fusions in metE and metH genes were grown in the presence of Arabidopsis thaliana plant cells and in Gamborg medium, the synthetic medium used for plant cell growth. Expression level of the metE and metHab genes in these conditions was compared to their level of expression monitored in complete medium (which does not induce the expression of hrpG) and in minimal medium (which induces HrpG activity) [19]. The results showed that the expression of metE follows the HrpG induction pattern ( Figure 4). The expression of the metE::lacZ fusion was strongly induced in the presence of plant cells compared to the expression in complete medium and this expression was decreased by a 4-fold factor in the hrpG mutant compared to the wild type strain. On the other hand, expression of metHab was not induced in the presence of plant cells compared to the non-inducing complete medium ( Figure 4B). These data confirm that the two methionine synthase genes metE and metH have different expression patterns in R. solanacearum and that metE expression is specifically induced in the presence of plant cells, similarly to hrpG [19].
The reduced expression of metE in the hrpG mutant appeared less pronounced in the presence of plant cells than the one observed in minimal medium, and thus suggested that plant cell co-culture conditions could induce a residual hrpG-independent expression of metE. The finding that metE expression is completely abolished in presence of plant cells in a metR mutant (4-fold reduction in hrpG versus 50-fold in metR, see Figure 4A) also suggested that this hrpG-independent inducer was acting on metR.

Homocysteine Induces the MetR-dependent Expression of metE
Since homocysteine is a co-activator of MetR regulatory protein in E. coli [21] we investigated whether homocysteine could also be an inducer of metE expression in R. solanacearum. Homocysteine was added to the Gamborg medium at a 100 mM final concentration and b-galactosidase activity was monitored for the different strains in this condition. Expression of metE was induced in the presence of homocysteine by a 30-fold factor in the wild type compared to the expression of metE in Gamborg medium without addition of homocysteine ( Figure 4A). This induction was independent from HrpG as the expression of metE was similar in Gamborg medium supplemented with homocysteine in the wild type genetic background and in the hrpG mutant ( Figure 4A). This induction of metE expression by homocysteine was only dependent on MetR as a 20-fold decrease in b-galactosidase activity was measured for metE::lacZ fusion between the wild type background and its metR mutant derivative. On the other hand, expression of metHab gene was independent from the presence of homocysteine as no induction of b-galactosidase activity is measured using the metH::lacZ fusion with or without homocysteine ( Figure 4B). metE and metR Mutants are not Impaired for in planta Growth but have Reduced Pathogenicity Because the methionine synthase genes belong to the HrpG virulence regulon, pathogenicity of the single disruption mutants for the metE, metHab and metR genes and the double mutant metEH was assayed on tomato and compared to the wild type strain. We first used an inoculation procedure by drenching the tomato plants with a bacterial suspension, which most closely reproduces the natural root infection process. In these conditions, the double mutant metE metH was completely avirulent and produced no disease symptoms ( Figure 5A). This probably results from the inability of this mutant strain to invade or multiply within its host since no bacteria were recovered from the stem of inoculated plants (data not shown). The metE and metR single mutants exhibited a significantly reduced aggressiveness with symptom rates delayed from 2 to 3 days compared to the wild type strain (p = 0,007 for metE mutant and p = 0,002 for metR mutant). The metH mutant was not significantly altered in pathogenicity since the disease progress curve is almost similar to that of the wild-type GMI1000 (p = 0,574).
To investigate whether the survival in the soil and the root invasion steps were involved in the decreased aggressiveness of the different mutants, another mode of inoculation was used by injecting the bacteria directly into the stem of tomato plants. Using this mode of inoculation, the metEmetH double mutant remained completely avirulent and the metE and metR mutants exhibited a reduced aggressivity phenotype comparable to the one observed with the soil drenching inoculation method ( Figure 5B) (p,0,001 for metE mutant and p = 0,002 for metR mutant). Again, no significant difference could be observed between the aggressiveness of the metH mutant strain compared to the wild type strain (p = 0,1). The pathogenic behaviour of these mutant strains was similar using both inoculation methods, thus suggesting that methionine biosynthesis is not critical in the first step of the infectious process but rather inside the plant, during the bacterial colonisation of host tissues.
In order to evaluate whether methionine concentration could be a limiting factor in the plant stem preventing normal growth of the methionine mutants, the in planta growth of the different mutants was quantified. After inoculation directly into the tomato stems, the bacteria were recovered from the aerial part of the plants at 0, 3 and 5 days post inoculation (dpi) and serial dilutions were plated on BG agar medium for bacterial enumeration. Methionine auxotrophy of recovered bacteria was checked on synthetic medium deprived from methionine to ensure that no reversion in the metEmetH mutant occurred in planta. The results showed that the single mutants in metE, metR and metH genes have a similar growth to the wild type strain and reach a bacterial density of 10 8 CFU gFW 21 at 5 dpi ( Figure 6). More surprisingly, the metEmetH double mutant was also able to multiply in the plant stem. Methionine concentration in the tomato xylem sap was quantified around 3.7 mM [22] and we observed that a GMI1000 metEmetH mutant is still able to grow on a synthetic minimal medium supplemented with 3.7 mM methionine, albeit less efficiently than the wild-type strain ( Figure S2). This probably explains why the bacterial population of this double mutant is statistically inferior to the one measured for the other strains at 3 and 5 dpi, but remains able to multiply inside the plant to reach a density of 10 7 CFU/gFW at 5 dpi. These results indicated that the last step of methionine biosynthesis is necessary for the full pathogenicity of R. solanacearum and revealed that the reduced aggressiveness of the metE and metR prototrophic mutants did not result from impaired growth in the plant xylem.

Part of the Methionine Biosynthesis Pathway is Under the Control of the Pathogenicity Regulon in R. solanacearum
The OmpR family regulatory protein HrpG plays a key role in coordinating different categories of pathogenicity genes required to overcome the plant defence responses and to survive in the intercellular spaces during the parasitic stage inside the plant [8]. HrpG was originally identified as a transcriptional regulator of the T3SS in R. solanacearum and further studies identified other regulatory targets more generally involved in the adaptation to life in the host [5]. This study uncovers a direct link between basal metabolism and pathogenicity in R. solanacearum, revealing that methionine metabolism is specifically activated once the bacteria are in the presence of plant cells and is genetically connected through the HrpG regulon to other essential pathogenicity determinants such as the T3SS. The observation that no other amino acid biosynthetic pathway than methionine appears to be controlled by HrpG from the transcriptomic data [5], indicates that HrpG does not stimulate a general activation of R. solanacearum metabolic pathways during infection but specifically the final step of the methionine biosynthesis pathway. Interestingly, a recent report on the transcriptomic characterization of the regulon controlled by the HrpG paralogue in Xanthomonas axonopodis pv. citri revealed that expression of metE is also hrpGdependent in this plant pathogen [23], thus reinforcing the hypothesis of a specific requirement of a proficient methionine biosynthesis pathway in these bacteria during plant infection.
In this study, we provide evidence that HrpG controls the expression of the major methionine synthase gene during infection, metE, through another regulatory gene, metR. Transactivation experiments in a heterologous E. coli system confirmed that metR, as hrpB, is a direct HrpG regulatory target (L.P. & S.G, unpublished results). metR is an ubiquitous regulator of the methionine biosynthetic pathway among Proteobacteria and this study provides the first example of the control of metR by a central pathogenicity regulator. Some of the regulatory mechanisms revealed in this study also differ from what is known in E. coli since in R. solanacearum the control exerted by the MetR regulator appears less essential to the biosynthesis: first, a GMI1000 metR mutant is not auxotrophic for methionine, contrary to what described for E. coli [16], and second, the genes involved in the first part of the biosynthesis pathway such as metX do not harbour any MetR recognition sequence in their promoter region, whereas the expression of this enzyme is controlled by MetR in E. coli [21,24]. In R. solanacearum, MetR curiously seems to modulate only the expression of metE in the methionine biosynthesis pathway in response to a plant signal. Disruption of metR affects pathogenicity, but not more than what observed for a metE mutant, suggesting that the delayed pathogenicity phenotype of the metR mutant is only due to the control exerted on metE.
MetR activity is enhanced by the metabolic inducer homocysteine, which results in increased expression of metE. Our data suggest that homocysteine is also present in plant cell co-culture conditions. Indeed, the expression of metE in the presence of plant cells is decreased by a 20-fold factor in the metR mutant compared to the hrpG deletion background (Figure 4), indicating that the MetR co-factor homocysteine induces a residual expression of metE in the hrpG mutant. On the other hand, expression of metH appears to be independent from homocysteine, in accordance to what is known in E. coli, where metH expression depends on MetR but it is not induced by the presence of homocysteine [25]. This difference for metR function in R. solanacearum and E. coli probably depends on the divergent environmental cue driving expression of methionine synthase genes (Figure 7). In E. coli a major constraint is to differentially control the expression of the two redundant enzymes depending on the availability of cobalamin in the milieu (since E. coli does not synthetize Vitamin B12) [26]. Conversely, R. solanacearum requires efficient methionine synthesis once inside the host environment, and metR appears to have been placed under hrpG control as a relay to specifically induce metE expression.

A Metabolic Adaptation to Host Environmental Conditions
It is noteworthy that the control of HrpG leads to a very high expression of metE, the cobalamin-independent enzyme, rather than metH, the cobalamin-dependent methionine synthase. Gene expression data confirm that in the presence of plant cells expression of metE is strongly induced and results in ten-fold more expression than metH (Figure 4). Cobalamin is not synthetized by plants and only the cobalamin-independent methionine synthase activity has been described in higher plants [27]. It is therefore tempting to speculate that R. solanacearum mobilizes preferentially metE during the infection process to synthesize methionine in a cobalamin-deficient environment since the ability to synthesize cobalamin, one of the largest known non-polymeric natural products, requires a substantial energetic investment, necessitating the activity of ,30 genes [28]. Accordingly, disruption of metH does not affect pathogenicity. The MetH enzyme was probably conserved through evolution because catalysis via cobalamin is mechanistically more advantageous, the k cat value for MetH being 50-100 times higher than for MetE [13]. Although they encode enzymes performing an identical biochemical reaction, the metE and metH genes are not functionally redundant in R. solanacearum since they are differentially mobilized during the parasitic or saprophytic phases. This picture provides an example of a tight metabolic adaptation of the pathogen to specific conditions encountered in its host.
It is also possible that the strong expression of metE in the presence of plant cells counteracts part of the plant defense response. ROS (Reactive Oxygen Species) are important components of the host immune response [29], and many pathogens need to prevent and overcome oxidative stress in order to establish and maintain infections. The use of ROS as antimicrobial agents by the immune system is based on the high reactivity of this species with various cellular components, such as the MetE enzyme which is known to be very sensitive to ROS [30]. It is established that R. solanacearum encounters an oxidative environment during tomato infection [31] so it can be hypothesized that the specific hrpGdependent activation of metE can be a mean for the pathogen to overcome the toxic affect of the plant oxidative burst. The finding that HrpG also activates the transcription of the katE catalase involved in the direct detoxification of ROS [5] supports this hypothesis.

How Methionine Biosynthesis Contributes to Pathogenicity?
In addition to R. solanacearum, met auxotroph mutants have been described in other plant pathogens such as Pseudomonas syringae [32,33] and Agrobacterium tumefaciens [34]. Whereas A. tumefaciens and P. syringae pv phaseolicola met mutants were avirulent, P. syringae pv syringae met genes were not tested for virulence but were shown to be required for optimum epiphytic fitness [35]. Compared to these former studies, a major difference is that the metE or metH single mutants are not methionine auxotrophs and our observations incite to think that the methionine biosynthetic pathway contributes to R. solanacearum aggressiveness not only by promoting bacterial growth. The strongest argument supporting this view is that at 5 dpi, the metE and metR mutants reach population levels similar to the wild-type parent, whereas significant differences in the rate of symptom development are already visible at this time point. This indicates that bacterial multiplication in planta and wilt symptom production can be dissociated in metER mutants in the experimental conditions used. Future investigations should determine if this behavior is variable among various host plants and whether bacterial multiplication of metER mutants is directly linked to the methionine concentration inside these hosts. Finally, methionine biosynthesis appears to be an essential requirement for pathogenicity of phylogenetically distant R. solanacearum strains differing in their host range spectrum: a recent study led on strain IPO1609, which carries a large genomic deletion comprising the metER genes, was shown to be hypovirulent on potato and this reduced virulence was partly dependent on metER genes [36].
The reason for a specific requirement of methionine biosynthesis in disease symptom production is currently not understood. We investigated whether T3SS-dependent secretion was altered in a metE mutant but no quantitative difference in the secretion of T3E Gala7 and PopP1 was observed after Western blot analysis (data not shown). Similarly, no significant difference was recorded after monitoring on test plates the activity of plant cell wall-degrading enzymes such as polygalacturonase or cellulase secreted through the general secretion pathway. Methionine is the precursor of SAM (S-adenosyl methionine) [13], the principal methyl group donor in cells, which is implicated in numerous cellular processes. It is therefore possible that reduced methionine biosynthesis impacts multiple cellular activities. One possible hypothesis to explain the role of MetE in the control of virulence is that SAM is a substrate of the PhcB methyltransferase which synthesizes 3-hydroxy palmitic acid methyl ester (3-OH PAME), the R. solanacearum virulence quorum sensing signal [37]. A reduced methionine synthase activity at a critical step of infection would limit the biosynthesis of 3-OH PAME and consequently delay the production of virulence factors such as exopolysaccharide. Another possibility relies on the fact that methionine is the direct precursor of polyamines such as spermidine and spermine, which are small polycationic compounds associated with a variety of biological processes. During the recent years, evidence has accumulated on the role of polyamines in microbial pathogenesis, survival in the host and oxidative stress tolerance [38,39]. Finally, a third scenario is that during infection the MetE enzyme principally uses homocysteine and 5-methyl tetrahydropteroyltri-L-glutamate (methyltetrahydrofolate) of plant origin as substrates. This MetE-induced depletion of plant metabolites could alter the plant cell homeostasis and contribute to disease development. Future studies should address whether activation of the methionine pathway during plant infection by R. solanacearum is devoted to the synthesis of a specific factor or more generally contributes to bacterial fitness.

Bacterial Strains, Plasmids and Growth Conditions
Relevant characteristics of the plasmids and bacterial strains used in this work are listed in Table 1. E. coli strains were grown at 37uC in Luria Bertani medium [40]. R. solanacearum strains were grown in complete BG medium or in MP minimal medium supplemented with glutamate at 20 mM final concentration [41]. When needed, antibiotics were added to the media at the following final concentrations (mg/l): kanamycin (Km) 50 for R. solanacearum, spectinomycin (Sp) 40 for R. solanacearum, gentamycin (Gm) 10 for R. solanacearum, tetracyclin (Tc) 10 for R. solanacearum, ampicillin (Ap) 100 for E. coli.

DNA Manipulations and Genetic Constructs
Standard recombinant DNA techniques were performed as described previously [40]. Restriction enzymes, DNA ligase and other DNA enzymes were used according to the manufacturers' recommendations.
Creation of metE, metR, metH and cobN Disruption Mutants metE and metR genes were disrupted by insertion of an V cassette [42] conferring resistance to spectinomycin. A 960 bp fragment upstream metE was cloned after PCR amplification using primers metE-Xba 59-TCTAGAGGATCTCGAAGCGGAACAG-39 and metE-HindUP 59-AAGCTTGGTCATTTCCATGGCC-39. A 980 bp fragment downstream of metE was cloned downstream of the first fragment using primers metE-HindDW 59-AAGCTTCCGCGAGGGCCTGCCG -39 and metE-Not 59-GCGGCCGCCGGCGACATCCACG-39. The V cassette was inserted between the two fragments using HindIII restriction site, resulting in plasmid pLP45. A 1.6 kb fragment of metR gene was cloned after PCR amplification using primers metR-F 59-TTCGCGCACCTCGTCGAACA-39 and metR-R 59-GCGCCCCGCGTTCTGGAC-39. The V cassette was inserted at position 225 of the gene using EcoRV restriction site resulting in plasmid pLP71. Both constructions were linearized by XbaI and introduced into R. solanacearum wild type and mutant strains by natural transformation. Double recombination events were selected using spectinomycin resistance and checked by PCR. The metH and cobN (RSp0626) mutants were isolated from the R. solanacearum GMI1000 MutantsDb database. MutantsDb is a database of approx. 5000 mutants generated through random insertion of the transposon EZ-Tn5,KAN-2. TM from Epicentre Biotechnologies (Madison, WI). Localisation of each individual insertion on the genome was determined by sequencing of its flanking region using an adaptation of the TAIL-PCR protocol from Qian et al. [43] and automated BlastN comparison with the genome sequence. Correct insertion of the transposon integration in metH and cobN was confirmed by PCR and sequencing.

Creation of metE, metH and metR Reporter Strains
The metE::lacZ and metH::lacZ fusion were created by using integration plasmid pCZ367 [44]. An internal fragment of the metE gene was PCR amplified by using primers MetE-5 (59-AAGCTTATGAGCTCAAGGGCTGGCTG-39) and MetE-3 (59-TCTAGACTCCGAATAGCACATGTGCG-39) and cloned as a 938-bp HindIII-XbaI fragment into pCZ367. An internal fragment of the metHa gene was PCR amplified by using primers MetH-5 (59-AAGCTTCCTGGTGAACCTGGTCGG-39) and MetH-3 (59-ACCGGATCCAAGGCGTTC-39). These plasmids were used to transform R. solanacearum GMI1000 and a single integrative event was selected by using pCZ367 gentamicin resistance. Recombinant clones were then checked by PCR.
lacZ reporter fusions with the metE and metH genes were introduced into the different genetic backgrounds (DhrpG, hrpB:: V, and metR:: V) by transformation with the genomic DNA of the donor strains. Recombinant strains were selected on media with adequate antibiotics, and the correct genomic insertion of the lacZ fusion was checked by PCR.

Determination of Growth Rates
The growth rates of strains GMI1000 and the derivative mutants were measured in cultures growing in MP minimal medium supplemented with glutamate at a 20 mM final concentration. Overnight cultures grown at 28uC with 200 rpm shaking were used to inoculate 200 mL of fresh medium with an initial OD 600 at 0.1. Bacterial growth was performed in 96-well plates and monitored using the FLUOstar Omega microplate reader (BMG Labtech), measures of OD 600 was performed every 30 minutes. Six biological replicates were used for the non auxotrophic strains: wild-type strain, metE, metR and metH single mutants, and for metE, metR double mutant; three biological replicates were used for auxotrophic mutants: metE, metH and metR, metH double mutants. For each time point, the mean of the replicates was calculated.

Plant Cell Cultures and Bacteria-plant Cell Cocultures
The A. thaliana At-202 (accession Col-0) cell suspension lines were grown in Gamborg B5 (Flow Laboratories). For the bacteriaplant cell co-cultivation, 10 ml samples of Arabidopsis cell suspensions were inoculated with R. solanacearum [19]. After 16 h of incubation at 28uC, the co-culture was filtered and the bacterial cells were recovered for b-galactosidase assays.
b-Galactosidase Assays b-Galactosidase assays were performed as described previously [19]. b-Galactosidase activity was determined and expressed in Miller units [46].

Plant Tests
Pathogenicity tests by soil inoculation were performed by watering 4-week-old tomato plants (Lycopersicum esculentum cv. Marmande) with 50 ml of a suspension containing 10 7 CFU ml 21 . Pathogenicity tests by stem injection were performed by injecting 20 ml of a 10 6 CFU ml 21 bacterial suspension into the stems of 4week-old plants. Disease development was scored daily by using a disease index scale ranging from 0 for no symptoms to 4 for completely wilted plants.
To evaluate bacterial multiplication in plants, tomato stems were removed, weighed, sterilized, cut and placed for 30 min at 25uC in sterile water. Bacteria were enumerated by plating serial dilutions onto BG medium.

Statistical Analysis of the Data
For pathogenicity assays, within-group Kaplan-Meier survival estimates (the total number of plants surviving [i.e., with disease index scores below 2] out of the total number inoculated for each group) were computed for each time interval in order to build Kaplan-Meier survival curves for each group. The log rank test was used to perform between-group comparisons, testing the equivalence of the Kaplan-Meier survival curves for a pair of groups [47]. When p,0,05 the survival curves were considered as significantly different.
For bacterial multiplication in plants, comparison between the wild -type and each mutant strain was performed using a Student's t test; when p,0,05 the bacterial multiplication was considered as significantly different between the two conditions studied.