The RhaS activator controls the Erwinia chrysanthemi 3937 genes rhiN, rhiT and rhiE involved in rhamnogalacturonan catabolism

Authors

  • Nicole Hugouvieux-Cotte-Pattat

    Corresponding author
    1. Unité de Microbiologie et Génétique – Composante INSA, UMR CNRS-INSA-UCB 5122, bat Lwoff, 10 rue Dubois, Domaine Scientifique de la Doua, 69622 Villeurbanne Cedex, France.
    Search for more papers by this author

E-mail cotte@insa-lyon.fr; Tel. (+33) 4 72 43 15 53; Fax (+33) 4 72 43 15 84.

Summary

Erwinia chrysanthemi causes soft-rot diseases of various plants by enzymatic degradation of the pectin in plant cell walls. The linear regions of pectin are composed of an acidic sugar, d-galacturonic acid. The ramified regions of pectin also include neutral sugars, and are rich in l-rhamnose residues. E. chrysanthemi is able to degrade these polysaccharides, polygalacturonate and rhamnogalacturonate. In E. chrysanthemi, the production of pectinases acting on linear regions is induced in the presence of polygalacturonate by a mechanism involving the repressor KdgR. The induction of the two adjacent E. chrysanthemi genes, designated rhiT and rhiN, is maximal after the simultaneous addition of both polygalacturonate and l-rhamnose. The rhiT product is homologous to the oligogalacturonide transporter TogT of E. chrysanthemi. The rhiN product is homologous to various proteins of unknown function, including a protein encoded by the plant-inducible locus picA of Agrobacterium tumefaciens. Both rhiT and rhiN are highly induced during plant infection. Various data suggest that RhiT and RhiN are involved in rhamnogalacturonate catabolism. RhiN is able to degrade the oligomers liberated by the rhamnogalacturonate lyase RhiE. The induction of the rhiTN operon in the presence of polygalacturonate results from control by the repressor KdgR. The additional induction of these genes by rhamnose is directly mediated by RhaS, a protein homologous to the activator of rhamnose catabolism in Escherichia coli. The virulence of an E. chrysanthemi rhaS mutant towards different host plants was clearly reduced. In this phytopathogenic bacterial species, RhaS positively regulates the transcription of the rhaBAD operon, involved in rhamnose catabolism, of the rhiE gene and of the rhiTN operon. The regulator RhaS plays a larger role in E. chrysanthemi than in other enterobacteria. Indeed, the RhaS control is not restricted to the catabolism of rhamnose but is extended to the degradation of plant polysaccharides that contain this sugar.

Introduction

The skeleton of plant tissues is composed of a complex matrix of polysaccharides (Albersheim et al., 1996). Two complex networks, cellulose/hemicellulose and pectic polymers, endow the primary cell wall with most of its physical and biological properties. Pectins contain both linear and ramified regions. The linear regions consist of long stretches of α,4-linked d-galacturonate residues (polygalacturonate) (McNeil et al., 1984). Two types of ramified regions have been identified: rhamnogalacturonan I and rhamnogalacturonan II (Schols and Voragen, 1996). Rhamnogalacturonan I (or rhamnogalacturonate) is composed of a backbone of alternating l-rhamnose and d-galacturonate residues. Long chains of neutral polymers, arabinans and galactans, are attached as side-chains to approximately half the rhamnose residues of rhamnogalacturonan I (McNeil et al., 1980). Rhamnogalacturonan II is a very complex polysaccharide containing approximately 30 monosaccharidic units, including d-galacturonate, l-rhamnose and several uncommon sugars (O’Neill et al., 1996).

Many plant pathogenic fungi and bacteria produce extracellular enzymes involved in plant cell wall degradation. The enterobacterium Erwinia chrysanthemi, which causes soft-rot diseases of various plants, produces several types of pectinases that act on the linear regions of pectin: pectin methylesterases, pectin acetylesterases, exo- and endopectate lyases, exopolygalacturonases, etc. (Robert-Baudouy et al., 2000). Endopectate lyases are the main causative agent of the symptoms of maceration observed after plant infections. The specific Out system allows the bacteria to secrete at least 10 different pectinases (Bouley et al., 2001). The cleavage of polygalacturonate by this set of extracellular enzymes leads to the formation of oligogalacturonides of various lengths, which enter the periplasm by the specific porin KdgM (Blot et al., 2002). Periplasmic exodepolymerases further degrade these oligomers to give dimers and small oligomers (Shevchik et al., 1999a). These oligogalacturonides enter the cytoplasm by two transporters: TogMNAB, a multicomponent member of the ATP-binding cassette (ABC) transporter superfamily (Hugouvieux-Cotte-Pattat et al., 2001); and TogT, a member of the glycoside-pentoside-hexuronide (GPH) transporter family (Hugouvieux-Cotte-Pattat and Reverchon, 2001). Final degradation of the oligomers into monomers takes place in the cytoplasm (Shevchik et al., 1999b). Various data suggest that E. chrysanthemi is able to degrade not only the linear regions of pectin but also the ramified regions. Recently, an extracellular protein secreted by the Out system, RhiE, was shown to have rhamnogalacturonate lyase activity (Laatu and Condemine, 2003). This observation confirmed that E. chrysanthemi is able to cleave the rhamnogalacturonan I backbone. The potential steps involved in the uptake and catabolism of the rhamnogalacturonides have never been described in any organism.

The transcription of the E. chrysanthemi genes involved in pectin catabolism is tightly controlled (Hugouvieux-Cotte-Pattat et al., 1996). The main factor affecting their expression is induction in the presence of polygalacturonate. This induction is mainly mediated by the KdgR repressor (Nasser et al., 1994), which controls all the steps of pectin catabolism. In addition, the regulation of the pectinase genes involves other regulators, such as the repressors PecS and PecT, which respond to unknown signals, and the global regulator of sugar catabolism, CRP (catabolite regulator protein) (Reverchon et al., 1994; 1997; Surgey et al., 1996).

The strong regulation of E. chrysanthemi genes involved in pectin degradation allowed us to identify some pectinase genes on the basis of their induction by pectic derivatives (Shevchik and Hugouvieux-Cotte-Pattat, 1997; Shevchik et al., 1997). A screen using lacZ transcriptional fusions was performed to isolate polygalacturonate-inducible genes (pgi fusions) (Hugouvieux-Cotte-Pattat and Robert-Baudouy, 1989). As polygalacturonate is not a pure substrate, the inducing properties of four neutral sugars linked to pectic polymers, namely l-arabinose, d-galactose, l-rhamnose and d-xylose, were tested on the not-yet-identified pgi fusions. Three fusions were induced in the presence of rhamnose. Two of them affected a locus, designated as rhi, a characteristic of which is induction in the presence of both polygalacturonate and rhamnose. This locus contains two adjacent genes, rhiT and rhiN, probably involved in rhamnogalacturonan catabolism.

Results and discussion

Some pgi fusions are induced by neutral sugars

The pgi mutants isolated previously contain a lacZ transcriptional fusion induced in the presence of polygalacturonate (Hugouvieux-Cotte-Pattat and Robert-Baudouy, 1989). As polygalacturonate is not a pure substrate (content in galacturonic acid about 87%), it is perhaps not the real inducer of some weakly induced pgi fusions. Thus, the expression of 14 not-yet-identified pgi fusions was analysed in the presence of various neutral sugars linked to pectin: l-arabinose, d-galactose, l-rhamnose and d-xylose. None of the pgi fusions was induced in the presence of l-arabinose or d-xylose. In contrast, the fusions of the mutants PGI81 and PGI133 were induced in the presence of galactose (data not shown), and those of PGI1, PGI44 and PGI50 were induced in the presence of rhamnose (Table 1). PGI81 and PGI133 were unable to use galactose as a carbon source, and the corresponding insertions were shown to inactivate genes involved in galactose catabolism, galT and galE respectively. Polygalacturonate could contain traces of galactose either in a free form or, more probably, associated with polysaccharides.

Table 1. . Expression of some pgi fusions in E. chrysanthemi 3937.
Mutant (fusion)Growth conditionβ-Galactosidase activity
(nmol min−1 mg−1)
  1. The results reported are the average of at least three independent experiments with standard deviations corresponding to less than 20%, except for low activities (<20) for which standard deviations could reach 40%.

PGI50 (rha::lacZ )Glycerol   6
Glycerol + polygalacturonate  19
Glycerol + rhamnose2184
Glycerol + polygalacturonate + rhamnose1344
PGI1 (rhiN::lacZ )Glycerol  38
Glycerol + polygalacturonate 387
Glycerol + rhamnose 751
Glycerol + polygalacturonate + rhamnose3530
PGI44 (rhiN::lacZ )Glycerol  13
Glycerol + polygalacturonate 126
Glycerol + rhamnose 214
Glycerol + polygalacturonate + rhamnose2097

Analysis of the pgi fusions induced by rhamnose

The fusion of PGI50 was induced threefold in the presence of polygalacturonate but by more than 300-fold in the presence of rhamnose. When rhamnose and polygalacturonate were added simultaneously, the induction was lower than that obtained with rhamnose alone (Table 1). The mutant PGI50 was totally unable to use rhamnose as a carbon source for growth. The wild-type E. chrysanthemi strain 3937 is able to catabolize rhamnose but with a low growth rate. When rhamnose is used as the sole carbon source for growth, the doubling time is about 7 h, compared with 2 h with polygalacturonate. Analysis of the E. chrysanthemi genome (http://www.tigr.org/tdb/mdb/mdbinprogress.html) revealed the presence of homologues of the Escherichia coli genes rhaB, rhaA and rhaD involved in rhamnose catabolism (Fig. 1). As in E. coli, the rha genes of E. chrysanthemi probably constitute an operon, rhaBAD. In contrast, no rhaT homologue was found in the E. chrysanthemi genome. The absence of a rhamnose-specific transport system is sufficient to explain the low growth rate of E. chrysanthemi with rhamnose as the sole carbon source. As this sugar could enter only by non-specific transport systems, its uptake is probably the limiting step in rhamnose catabolism. The lack of rhaT in E. chrysanthemi suggests that this bacterium catabolizes rhamnose contained in oligosaccharides rather than free rhamnose.

Figure 1.

Physical map of the Erwinia chrysanthemi 3937 regions including the rhi and rha genes. The arrows indicate the position and the transcription direction of the genes.
A. Isolation of the E. chrysanthemi rhiN and rhiT genes. The restriction map of the main subclones is given. The grey region indicates the MudI1681 DNA, the black region indicates the uidA-Km cassette. and the thin line corresponds to E. chrysanthemi chromosomal DNA.
B. Organization of the rha genes in E. chrysanthemi and E. coli. The percentage identity between pairs of homologous proteins is indicated between the two regions.

The lack of growth of PGI50 on rhamnose, together with a high proximity (95% of co-transduction) between its Mu insertion and the rhaS gene suggested that the Mu insertion of PGI50 is situated in the rhaBAD operon (Fig. 1).

The two mutants PGI1 and PGI44 were not affected for growth with rhamnose as carbon source. The fusions of PGI1 and PGI44 were induced in the presence of either polygalacturonate or rhamnose, by about 10- and 20-fold respectively (Table 1). Moreover, a clear additive effect of induction was observed when both rhamnose and polygalacturonate were added simultaneously (Table 1), suggesting that both compounds are necessary for their full induction. The localization of the mutations PGI1 and PGI44 on the E. chrysanthemi chromosome revealed that these mutations are not situated in the region containing the rhaSR–rhaBAD cluster. Both are situated in the same locus, called rhi (for rhamnose inducible), which co-transferred with the cys-4, leu-1 and pro-374 markers. Three-factor cross indicated that rhi is located between the two first markers, very near to cys-4. About 40% of co-transduction between rhi and cys-4 was observed with the phi-EC2 transducing phage.

Identification of the rhiT and rhiN genes

To identify the gene(s) altered in the PGI1 and PGI44 mutants, the corresponding MudI1681 insertions were cloned by selection of the KmR phenotype of the Mu insertion. Restriction analysis of the DNA fragments isolated from PGI1 or PGI44 indicated that the two rhi mutations are situated in the same region; they are separated by less than 0.2 kb on the chromosomal DNA. The plasmid pI2224 contained a 9.5 kb PstI DNA fragment conferring the Lac+ phenotype resulting from the PGI1 fusion (Fig. 1). Determination of the nucleotide sequence of a 0.6 kb PstI–HindIII fragment (pI2276), containing the junction of the Mu insertion (Fig. 1), indicated that this insertion is situated at the beginning of a potential open reading frame (ORF) that has no homologues among the enterobacteria. To isolate adjacent DNA, a uidA-Km cassette was inserted in the EcoRV site situated 0.2 kb upstream from the Mu insertion (Fig. 1). After recombination of this cassette into the 3937 chromosome, a DNA fragment containing this insertion and adjacent chromosomal DNA was cloned by selection of the KmR phenotype of the uidA cassette (plasmid pN1832). Subsequently, the 2.4 kb EcoRV–PstI fragment adjacent to the cassette was isolated (pN1882) (Fig. 1).

Determination of the sequence of a 2.7 kb PstI fragment (from pI2276 and pN1882, Fig. 1) (GenBank accession number AJ292045) revealed the presence of a complete ORF named rhiN. The Mu insertions of the mutants PGI1 and PGI44 are situated inside this ORF, which begins with an ATG codon at position 302 (relative to the PstI site) and ends with TAA at position 1439. The rhiN gene encodes a 379-amino-acid protein with a calculated molecular mass of 44 174 Da. The rhiN gene is preceded by an ORF transcribed on the same DNA strand and ending with a TGA codon at position 220. The expression of this gene was tested after construction of a transcriptional fusion in the PstI site (Fig. 1). Similarly to rhiN, this fusion was induced in the presence of polygalacturonate or rhamnose, with an additive effect of both compounds (see below). The gene preceding rhiN was named rhiT. The whole sequence of rhiT was obtained from the E. chrysanthemi genome sequence (http://www.tigr.org/tdb/mdb/mdbinprogress.html). The rhiT gene encodes a 526-amino-acid protein with a calculated molecular mass of 58 453 Da. The rhiT gene is preceded by a gene involved in iron assimilation. The rhiN gene is followed by an ORF, transcribed on the other DNA strand, encoding a 324-amino-acid protein homologous to various hypothetical proteins. Construction of a gene fusion in the SalI site (Fig. 1) indicated that the expression of this gene is not induced in the presence of polygalacturonate or rhamnose (data not shown). Thus, the two adjacent genes rhiT and rhiN are the only genes of this cluster induced in the presence of both polygalacturonate and rhamnose.

The co-ordinated expression of rhiT and rhiN, as well as the proximity between the two genes, suggested a polycistronic organization. Northern blot analysis was performed using RNA extracted from a kdgR mutant grown in the presence of rhamnose to obtain a high level of transcription (see below). The 563 bp rhiT probe hybridized with an about 3 kb mRNA (data not shown), a size in agreement with a polycistronic transcription of the rhiT and rhiN genes.

Analysis of the RhiN homologues

The deduced RhiN amino acid sequence showed homologies (39% to 23% identity) with potential proteins of unknown functions from Agrobacterium tumefaciens (AGR_L_618p and AGR_L_3363p), Bacillus subtilis (products of yesR and yteR) and Yersinia pestis (y3225). In A. tumefaciens, AGR_L_3363p was formerly identified as a plant-inducible locus (orf1 of the picA locus; Rong et al., 1991). The picA inducer was shown to be derived from the pectic portion of plant cell walls and to consist of polysaccharides containing oligogalacturonides linked to neutral sugars (Rong et al., 1994). Analysis of the genomic regions surrounding the different rhiN homologues indicated that they are often linked to genes potentially involved in pectin catabolism. The genes encoding AGR_L_618p and AGR_L_3363p in A. tumefaciens are in the vicinity of genes encoding polygalacturonase homologues. Homologues of the gene kduD involved in polygalacturonate catabolism are found near genes encoding AGR_L_3363p in A. tumefaciens and y3225 in Y. pestis. AGR_L_3363p of A. tumefaciens and yesR of B. subtilis are situated near genes encoding elements of ABC transporters homologous to the oligogalacturonide transporter TogMNAB of E. chrysanthemi. Moreover, in B. subtilis, the gene yesR is situated in a region containing the genes yesT and yesY encoding homologues of rhamnogalacturonan acetylesterases, and yesW and yesX encoding homologues of rhamnogalacturonan lyases (McKie et al., 2001). These homologies suggest that the regions of A. tumefaciens, Y. pestis and B. subtilis encoding RhiN homologues are involved in the catabolism of linear or ramified regions of pectin.

Characterization of the RhiN protein

The rhiN gene was cloned into the pT7-7 vector, and the RhiN protein was overproduced in a recombinant E. coli strain. Analysis by SDS–PAGE of the plasmid-encoded proteins revealed the overproduction of a cytoplasmic 48 kDa protein (data not shown). RhiN constituted up to 80% of the cell protein content of the soluble fraction from strain BL21(DE3)/pI2673. Therefore, this extract was used directly for enzyme analysis.

When assayed for oxidase activity with various substrates, the RhiN extract showed no activity with pNP-Ara, pNP-Gal or oNP-Xyl and a very low hydrolase activity with pNP-Rha. No conditions were found that could improve this activity, which was reproducibly detected but only after several hours of incubation. Thus, pNP-Rha is not a suitable substrate for RhiN, and this residual activity probably indicates some specificity of RhiN towards rhamnosides.

The additive induction by polygalacturonate and rhamnose suggests that RhiT is physiologically involved in the catabolism of polysaccharides containing both rhamnose and galacturonate, such as rhamnogalacturonate. The sole E. chrysanthemi enzyme known to be involved in this catabolism is the extracellular rhamnogalacturonate lyase RhiE (Laatu and Condemine, 2003). RhiE cleaves rhamnogalacturonate and liberates oligomers with an unsaturation on the galacturonate residue situated at the non-reducing end. Such unsaturated oligomers, which absorb at 230 nm, are detected by spectrophotometry. RhiN was unable to liberate unsaturated products from rhamnogalacturonate (Fig. 2). However, as RhiN is a cytoplasmic enzyme, it is more probable that it acts on the oligomers liberated by RhiE after their entry into the cytoplasm. To test this hypothesis, a RhiE extract was incubated with rhamnogalacturonate until the OD230 increased by 0.010 units (about 10 min). Then, 1 µg of RhiN extract was added, and incubation was extended for 10 min. RhiN addition led to a decrease in OD230 (Fig. 2), indicating the disappearance of the unsaturated oligomers produced by RhiE. When RhiN was added simultaneously with RhiE, the increase in OD230 due to RhiE was also clearly reduced (Fig. 2). The scarcity of the substrate and the low sensitivity of this method prevented us from developing a reproducible enzyme assay for RhiN. However, the data obtained confirmed that RhiN acts on the unsaturated oligomers liberated by RhiE.

Figure 2.

Combination of RhiN and RhiE activities. Rhamnogalacturonate lyase activity was assayed by incubation of the protein(s) with rhamnogalacturonate (0.5 g l−1) in 0.1 M Tris-HCl buffer, pH 6.8. After incubation at 37°C for 10 min, the increase in absorbency at 230 nm was measured to detect the formation of unsaturated products. Extracts of RhiE (1 µg ml−1) and RhiN (1 µg ml−1) were tested individually but also in combination, either by the simultaneous addition of both proteins (a) or by pretreatment of the substrate with RhiE before the addition of RhiN (b).

Analysis of the RhiT sequence

The deduced RhiT amino acid sequence showed 42% identity with E. chrysanthemi TogT, an oligogalacturonide transporter of the GPH family that includes proteins involved in the uptake of complex sugars, mostly oligosaccharides (Saier, 2000). RhiT shares 23–24% identity with the glucuronide carrier protein GusB of E. coli, the melibiose carrier protein MelB of E. coli and the lactose permease LacP of Staphylococcus xylosus. Considering these scores, it is clear that RhiT does not belong to any of the previously described subgroups of the GPH family (‘GusB’, ‘MelB’ and ‘LacS’) (Poolman et al., 1996) but constitutes a novel subgroup with TogT. As the classification in subgroups is closely linked to the substrate specificity (Poolman et al., 1996), it is tempting to expect a relationship between the RhiT substrate and oligogalacturonides. The induction pattern of rhiT and its co-transcription with rhiN suggested that RhiT is physiologically involved in the uptake of rhamnogalacturonides resulting from the action of the extracellular rhamnogalacturonate lyase RhiE.

Occurrence of rhiN and rhiT in Erwinia spp.

To test for the presence of rhiN and rhiT homologues in other Erwinia species, polymerase chain reaction (PCR) experiments were performed using the appropriate oligonucleotides and chromosomal DNA from different strains of E. chrysanthemi or E. carotovora. DNA fragments of 1.9 kb and 1.1 kb, corresponding to rhiT and rhiN, respectively, were detected for all the E. chrysanthemi wild-type strains tested, i.e. 3937, B374, ENA49 and EC16 (data not shown). In contrast, no PCR product was obtained with the strains of E. carotovora ssp. carotovora or E. carotovora ssp. atroseptica. However, analysis of the genome of the E. carotovora ssp. atroseptica strain SCRI1043 (Sanger Institute, UK; http://www.sanger.ac.uk/Projects/E_carotovora) indicated the presence of rhiT and rhiN homologues (80% identity for both genes at the DNA level; 94% and 90% identity, respectively, at the protein level). Thus, the negative response obtained by PCR with E. carotovora resulted from a too high specificity of the oligonucleotides used. The two genes rhiT and rhiN appeared to be conserved in different pectinolytic Erwinia species.

Virulence of the rhiT, rhiN and rhaS mutants

The pathogenicity of the rhaS, rhiE, rhiT and rhiN mutants was tested on chicory leaves, potato tubers and plants of Saintpaulia ionantha (African violet). On chicory leaves, the rhiE, rhiT and rhiN mutants were obviously affected because the average length of macerated tissue decreased by 30–38% in comparison with the wild-type strain 3937 (Fig. 3A). Similarly, on potato tubers, the weight of rotted tissue decreased by 24–32% with these mutants (data not shown). Virulence of the rhaS mutant was even more affected, with maceration decreased by 77% on chicory leaves and 39% on potato tubers. These reductions in maceration were clearly correlated with decreased multiplication of bacteria during infection (Fig. 3B for chicory leaves). Thus, the RhaS-controlled genes are important for both maceration of the plant tissues and bacterial growth. Degradation of rhamnogalacturonan may favour the development of the soft-rot disease by a direct effect, contributing to the dissociation of the plant cell wall components, and by supplying bacteria with nutriments.

Figure 3.

Infection of chicory leaves with the rhi mutants, virulence, bacterial multiplication and expression of the gene fusions. Fifteen chicory leaves were infected for each strain, 3937 (wild type), A4278 (rhaS::Cm), A4279 (rhiE::uidA), A4280 (rhiN::uidA), A4281 (rhiT::uidA) and A1798 (pelD::uidA), and incubated at 30°C for 24 h. The pelD::uidA mutant, which has an attenuated virulence and a high in planta expression, was used as a control.
A. The length of rotted tissue was measured to estimate the disease severity.
B. The rotted tissue was recovered and used for the numeration of bacteria and (C) for the assay of β-glucuronidase in the case of strains containing a uidA fusion. The values reported are the average of the different leaves, and the standard deviations are indicated.

On Saintpaulia plants, 14 days after inoculation, the wild-type strain 3937 provoked 80% of systemic response, i.e. for eight plants out of 10, maceration progressed in other leaves because of the invasion of the plant vascular system by bacteria. In contrast, maceration remained localized to the inoculated leaf for all the plants infected with the rhiT and rhaS mutants. Thus, the rhiT and rhaS mutants are clearly non-invasive.

Transcription of the rhiN and rhiT genes

Transcriptional fusions in rhiN and rhiT were constructed by the insertion of a uidA-Km cassette (Bardonnet and Blanco, 1992) into rhiN and rhiT. The expression of these fusions was then followed in various conditions. During bacterial growth, the expression of both fusions increased about fourfold when cells entered the late exponential growth phase, and this increase was coincident with the production of pectate lyases (Fig. 4; data not shown). Thus, the expression of rhiN and rhiT is dependent on the cell density, as observed previously for the pectate lyase genes (Hugouvieux-Cotte-Pattat et al., 1992). In the absence of inducer, the expression of rhiT or rhiN showed a low basal level of expression (Table 2). Their transcription was stimulated six- to eightfold in the presence of polygalacturonate and 11- to 12-fold in the presence of rhamnose. In the presence of both compounds, induction ratios of 80–100 were reached, indicating an additive effect of each inducing compound. These data also demonstrated that the expressions of rhiT and rhiN are strictly co-ordinated, a feature typical of co-transcribed genes.

Figure 4.

Expression of the rhiN::uidA fusion during growth of E. chrysanthemi. The strain A3105 (rhiN::uidA) was grown in M63 supplemented with polygalacturonate. The OD600 was measured at intervals to estimate the cell density. The pectate lyase (PL) and the β-glucuronidase (GUS) specific activities were determined on each sample and are expressed as µmol and nmol of product, respectively, liberated min−1 mg−1 bacterial dry weight.

Table 2. . Expression of rhi::uidA transcriptional fusions in Erwinia chrysanthemi 3937.
Mutation(s)Growth conditionSpecific activity (nmol min−1 mg−1)
rhiN::uidArhiT::uidArhiE::uidArha::lacZ
  1. β-Glucuronidase and β-galactosidase activities are reported for uidA and lacZ fusions respectively. These results are the average of at least three independent experiments with standard deviations corresponding to less than 20%, except for very low activities (<20) for which standard deviations could reach 40%.

  2. PGA, polygalacturonate.

NoneGlycerol  14  21  5   6
Glycerol + rhamnose 156 2607982184
Glycerol + PGA  86 163  
Glycerol + PGA + rhamnose11392199  
Glucose   8  12  
Glucose + PGA + rhamnose 358 445  
crpGlucose   5   6  
Glucose + PGA + rhamnose   6   8  
kdgRGlycerol 124 184  
Glycerol + rhamnose12262081  
Glycerol + PGA 117 163  
Glycerol + PGA + rhamnose13902355  
rhaSGlycerol  12  18  4   6
Glycerol + rhamnose  13  18  4   5
Glycerol + PGA  93 149  
Glycerol + PGA + rhamnose  96 136  
kdgR, rhaSGlycerol  92 126  
Glycerol + rhamnose  98 145  
Glycerol + PGA  97 151  
Glycerol + PGA + rhamnose  90 150  

Concomitant work (C. H. Yang and N. T. Keen, personal communication) has shown that rhiT is induced during plant infection as the rhiT gene has been identified among a set of plant-inducible genes of E. chrysanthemi selected by the IVET method. Moreover, in A. tumefaciens, the rhiN homologue was identified in the plant-inducible locus picA (Rong et al., 1991). The expression of rhiT and rhiN transcriptional fusions was then analysed after infection of chicory leaves, and compared with that of the highly inducible pectate lyase gene pelD (Hugouvieux-Cotte-Pattat et al., 1992). Indeed, both rhiT and rhiN were highly transcribed in the macerated tissue after infection of either chicory leaves (Fig. 3C) or potato tubers (data not shown).

To identify the regulators controlling rhiT and rhiE transcription, the two fusions were transduced into strains containing mutations affecting pectate lyase production (Table 2). Their basal expression increased ninefold in the kdgR mutant, and they were no longer affected by the presence of polygalacturonate (Table 2). Thus, KdgR is responsible for the repression of rhiN and rhiT in the absence of polygalacturonate. In the crp mutant, the expression of rhiT or rhiN is very low even in the presence of the inducers (Table 2), indicating that CRP is involved in the activation of rhiT and rhiN transcription. This double regulation, a mechanism of repression by the specific regulator KdgR and of activation by the global regulator CRP, is typical of genes involved in pectin catabolism. In contrast, the regulators PecS and PecT, which control some pectinase genes, are not involved in rhiT and rhiN transcription as the expression of the fusions was not modified in pecS and pecT mutants (data not shown). In all these mutants, the two fusions remained inducible by rhamnose (Table 2; data not shown). This additional induction indicated the involvement of a regulatory mechanism that specifically responds to rhamnose.

Role of RhaS in the transcription of rhiT , rhiN and rhiE in E. chrysanthemi

In E. coli, the regulator RhaS is responsible for the induction of the rhaT gene and the rhaBAD operon in the presence of rhamnose. RhaS directly activates these two operons involved in the transport and catabolism of rhamnose respectively (Egan and Schleif, 1993; Via et al., 1996). In addition, a second regulator, RhaR, activates the expression of the rhaSR operon in the presence of rhamnose (Tobin and Schleif, 1990a). Analysis of the E. chrysanthemi genome for the presence of the rhamnose regulators revealed the presence of homologues of the E. coli genes rhaS and rhaR adjacent to the rhaBAD operon (Fig. 1). As in E. coli, these two genes probably constitute an operon, rhaSR.

In order to determine whether these activators could be involved in rhamnose induction of the rhi genes, the rhaS–rhaR region of E. chrysanthemi was cloned after PCR amplification, and an E. chrysanthemi rhaS::Cm mutant was constructed by reverse genetics. This polar insertion is also supposed to inactivate the downstream gene rhaR. The rhaS::Cm mutant has totally lost the capacity to grow with rhamnose as the sole carbon source. The rhaS::Cm mutation was transduced in strain PGI50, which contains a rhaBAD::lacZ fusion. As expected, the rhaS::Cm mutation abolished the rhamnose induction of this fusion (Table 2).

The rhaS::Cm mutation was then transduced into the different rhi mutants. In the presence of the rhaS mutation, the expression of rhiT or rhiN was not inducible by rhamnose; their induction by polygalacturonate was not affected (Table 2). Thus, in E. chrysanthemi, RhaS is responsible for the activation of rhiT and rhiN transcription in the presence of rhamnose. As the rhiE gene was previously shown to be rhamnose inducible (Laatu and Condemine, 2003), its expression was also tested in the rhaS mutant. The expression of a rhiE::uidA fusion was induced 150-fold by rhamnose in the parental strain, but it was expressed only at a low basal level in the rhaS mutant (Table 2). Thus, the RhaS activator of E. chrysanthemi controls the synthesis of proteins involved in rhamnogalacturonan degradation, such as the extracellular rhamnogalacturonate lyase RhiE.

Construction of a double kdgR rhaS mutant indicated that the effects of each mutation are additive in comparison with the two single mutants (Table 2). In the double regulatory mutant, the rhiT and rhiN fusions are partially derepressed, and they are no longer inducible in the presence of either rhamnose or polygalacturonate. Thus, the mechanisms of regulation of rhiT and rhiN by the two regulators KdgR and RhaS appear to be independent.

The existence of several RhaS-controlled genes in E. chrysanthemi led us to check their 5′ non-coding region for the presence of potential RhaS binding sites. Alignment of the 5′ non-coding end of rhaBAD, rhiT and rhiE revealed a conserved region of about 50 bp (Fig. 5A). The consensus deduced from this conserved region consists of two long imperfect inverted repeats of 18 bp separated by a less well-conserved GC-rich central region of 15 bp. This sequence presents obvious homology with the RhaS binding site identified in E. coli by analysis of the rhaBAD promoter, as an inverted repeat of two 17 bp half-sites separated by 16 bp of uncontacted DNA (Egan and Schleif, 1994). Most of the conserved nucleotides of the deduced consensus sequence (Fig. 5A) are situated in the external outer and inner major grooves shown to interact with RhaS (Bhende and Egan, 1999). The presence of this sequence in the 5′ non-coding end of rhaBAD, rhiT and rhiE is a strong indication that the E. chrysanthemi RhaS activator directly controls the transcription of these genes. To test this hypothesis, the RhaS protein was overproduced in E. coli containing the plasmid pI2861. However, almost all the overproduced protein formed insoluble aggregates (Fig. 6A). Limited solubility seems to be a general feature of regulators of the AraC family, including E. coli RhaS (Egan and Schleif, 1994). Crude soluble protein extracts contained a small quantity of RhaS (Fig. 6A). In DNA mobility shift assays, the mobility of a 325 bp DNA fragment containing the rhiT promoter region was retarded by theses crude extracts (Fig. 6B). No shift was observed with a similar extract from E. coli containing the vector pT7-6 (Fig. 6B). Moreover, this shift was dependent on the presence of rhamnose (Fig. 6B). These results indicate that RhaS is able to interact directly with the promoter region of rhiT and that rhamnose is the actual inducer in vivo.

Figure 5.

Sequences of the RhaS and RhaR binding sites in E. chrysanthemi and E. coli. Data for E. coli rhaB and rhaS regulatory regions are from Egan and Schleif (1994) and Tobin and Schleif (1990b) respectively.
A. The consensus of the RhaS binding site is deduced from three E. chrysanthemi sequences located upstream, namely rhaB, rhiE and rhiT. Capital letters represent highly conserved nucleotides, while lower case letters represent less-conserved nucleotides. Nucleotides of the E. coli rhaB regulatory region shown to be important for interaction with RhaS are underlined. The two half-sites of these binding sites are inverted repeats, as indicated by the arrows above the sequences. The sequences are divided to show the DNA grooves, with the outer and inner major grooves indicated.
B. The consensus of the RhaR binding site is deduced from comparison between the rhaS upstream sequences of E. coli and E. chrysanthemi. The nucleotides underlined in this consensus are conserved in both RhaS and RhaR binding sites.

Figure 6.

Overproduction and DNA-binding activity of the RhaS protein.
A. After overproduction of RhaS in E. coli, the protein samples were analysed by SDS–PAGE: soluble fraction from the strains BL21(DE3)/pI2861 (rhaS+) (lane 1) and BL21(DE3)/pT7-6 (lane 2); insoluble fraction from BL21(DE3)/pI2861 (lane 3). The position of the molecular weight markers is indicated. The arrow points to the RhaS protein.
B. Gel mobility shift assay of a 325 bp DNA fragment containing the rhiT promoter region. The 32P-labelled DNA fragment was incubated with 3 and 6 µg of soluble extracts from BL21(DE3)/pI2861 (lanes 1 and 2; 3 and 4) or from BL21(DE3)/pT7-6 (lanes 5 and 6) in the absence (lanes 1 and 2) or presence (lanes 3–6) of l-rhamnose. The arrow indicates the position of the free DNA fragment.

Analysis of the 5′ non-coding end of rhiT also revealed the presence of sequences homologous to the binding sites of KdgR and CRP (Fig. 7). The presence of three regulatory sites upstream of rhiT is in agreement with the in vivo regulation of the rhiTN operon by RhaS, KdgR and CRP (Table 2).

Figure 7.

Sequence of the rhiT regulatory region. The sequences underlined correspond to the stop codon of the preceding gene, the rhiT start codon, the Shine–Delgarno sequence (SD) and the sequences homologous to the KdgR and CRP binding sites. The consensus sequences of these sites are given below (R = G or A, Y = C or T). The −10 and −35 sites of two putative promoters are indicated. The arrows indicate two inverted repeats corresponding to a potential rho-independent transcription terminator and to the RhaS binding site respectively.

Examination of the 5′ non-coding end of rhaSR revealed the presence of a sequence showing 61% identity with the RhaR binding site identified in E. coli (Fig. 5B). The RhaR binding site shows a resemblance to the RhaS binding site (Tobin and Schleif, 1990b). The conservation of the two regulators RhaS and RhaR, and of the binding sites of these two regulators, suggests that rhamnose induction in E. chrysanthemi results from a cascade similar to that characterized previously in E. coli (Egan and Schleif, 1993). However, at the level of the rhiT promoter region, the mechanism of action of RhaS could be more complex than that described for the rhaB promoter as an additional regulatory protein, KdgR, is involved. Determination of the transcriptional start site(s) and of the hierarchy of binding of the different regulators is now necessary to understand better the mechanisms adjusting rhiTN transcription.

Conclusion

The absence of available substrates renders it difficult to analyse the function of the two proteins RhiN and RhiT. RhiN is an intracellular protein able to degrade the unsaturated rhamnogalacturonides generated by RhiE. Sequence homology suggests that RhiT is a transport system involved in the uptake of oligosaccharides related to oligogalacturonides. The fact that the expression of the rhiTN operon is induced in the presence of both polygalacturonate and l-rhamnose is a strong indication of a role for its products in the transport and cleavage of rhamnogalacturonides. The genes rhiT and rhiN are under the control of the two regulators, KdgR and RhaS. This double regulation allows for a coupling between the catabolism of the polymer and that of the corresponding monomers. Two types of monomers could be liberated intracellularly from rhamnogalacturonides: on one side, a galacturonate derivative, 4-deoxy-l-threo-5-hexosulose uronate, would be catabolized by the products of four genes controlled by KdgR, kduI, kduD, kdgK and kdgA; on the other side, rhamnose would be catabolized by the products of the rhaBAD operon controlled by RhaS.

In E. coli, the KdgR repressor controls the expression of only a few genes involved in the catabolism of a galacturonate derivative, 2-keto-3-deoxygluconate (kdgK, kdgA and kdgT). In contrast, in E. chrysanthemi, the KdgR repressor has a major role in controlling the genes involved in pectin degradation, i.e. genes encoding enzymes able to modify and cleave pectin or pectic oligomers (paeX, paeY, pemA, pemB, pelA, pelB, pelC, pelD, pelE, pelI, pelX, pelZ, pehV, pehW, pehX), the protein complex necessary for the secretion of pectinases (the out operon), transporters of the resulting oligomers or monomers (kdgM, togT, togMNAB, kdgT), the cytoplasmic enzymes involved in the cleavage of short oligomers (pelW, ogl) and the catabolism of the resulting monomer (kduI, kduD, kdgK, kdgA).

In E. coli, the RhaS activator controls the two operons rhaT and rhaBAD involved in rhamnose transport and catabolism. E. chrysanthemi does not need a rhaT gene because the substrate for the rhaBAD operon is not extracellular free rhamnose, but rather rhamnose released in the cytoplasm from the degradation of rhamnogalacturonides. In E. chrysanthemi, RhaS controls the rhaBAD operon but also a gene involved in the cleavage of rhamnogalacturonan, rhiE, and two genes, rhiT and rhiN, supposed to mediate the uptake of the resulting oligosaccharides and their intracellular cleavage respectively. Thus, in E. chrysanthemi, the regulator RhaS has an extended function, controlling not only rhamnose catabolism but also the steps involved in the degradation and catabolism of the rhamnose-rich polysaccharides of the plant cell wall.

Experimental procedures

Bacterial strains and genetic techniques

The bacterial strains of E. chrysanthemi or E. coli and the plasmids used in this study are listed in Table 3. The phi-EC2 generalized transducing phage was used for transduction (Resibois et al., 1984). For chromosomal mapping, various polyauxotrophic strains were used as recipients in conjugation with the donor strain containing the plasmid pULB110 (Van Gijsegem et al., 1985).

Table 3. . Bacterial strains, plasmids and oligonucleotides.
Strain/plasmidGenotype/phenotypeReference/origin
  • a

    . Restriction sites used for cloning are underlined.

E. chrysanthemi strains
 3937Wild typeLaboratory collection
 A350lmrT clacZ2Hugouvieux-Cotte-Pattat et al. (1989)
 PGI1lmrTclacZ2 rhiN::lacZ, KmHugouvieux-Cotte-Pattat et al. (1989)
 PGI44lmrTclacZ2 rhiN::lacZ, KmHugouvieux-Cotte-Pattat et al. (1989)
 PGI50lmrTclacZ2 rhaBAD::lacZ, KmHugouvieux-Cotte-Pattat et al. (1989)
 PGI81lmrTclacZ2 galT::lacZ, KmHugouvieux-Cotte-Pattat et al. (1989)
 PGI133lmrTclacZ2 galE::lacZ, KmHugouvieux-Cotte-Pattat et al. (1989)
 A1798lmrTclacZ2 pelD::uidA, KmHugouvieux-Cotte-Pattat et al. (1992)
 A2507lmrT clacZ2 crp::CmReverchon et al. (1994)
 A3105lmrTclacZ2 rhiN::uidA, KmThis work
 A3719lmrT clacZ2 kdgR::SmLaboratory collection
 A3946lmrTclacZ2 rhiE::uidA, KmLaatu and Condemine (2003)
 A4156lmrTclacZ2 rhiT::uidA, KmThis work
 A4157lmrT clacZ2 rhaS::CmThis work
 A4182lmrT clacZ2 rhaS::Cm kdgR::SmThis work
Other E. chrysanthemi strains
 B374Wild typeA. Toussaint
 EC16Wild typeN. T. Keen
 ENA49Wild typeV. Shevchik
E. carotovora ssp. carotovora strains
 SCRI193Wild typeM. Perombelon
 CC3-2Wild typeV. Shevchik
E. carotovora ssp. atroseptica strains
 SCRI31Wild typeM. Perombelon
 CA36AWild typeV. Shevchik
Plasmids
 pULB110RP4::Mu3A, ApR, TcRVan Gijsegem et al. (1985)
 pBR325ApR, TcR, CmRF. Bolivar
 pBS CmBluescript KS+, CmRStratagene
 pBS ApBluescript KS+, ApRStratagene
 pT7-6phi-10 promoter, ApRTabor and Richardson (1985)
 pT7-7phi-10 promoter, translation start for the T7 gene 10 protein, ApRTabor and Richardson (1985)
 pR1081pULB110 derivative, rhiN::lacZ+, KmRThis work
 pI2224pBR325 derivative with a 9.5 kb PstI fragment from pR1081, rhiN::lacZ +This work
 pI2276pBSCm derivative with a 0.6 kb PstI–HindIII fragment from p2224This work
 pN1832pBR325 derivative with a 6.2 kb PstI fragment, KmRThis work
 pN1882pBSAp derivative with a 2.4 kb PstI–HpaI fragment from pN1832, rhiN+This work
 pN2538pBSCm derivative with a 1.9 kb PCR fragment, rhaSR +This work
 pN2548pBSAp derivative with a 1.9 kb PCR fragment, rhiT+This work
 pI2549pBSCm derivative with a 1.2 kb PCR fragment, rhiN+This work
 pI2673pT7-7 derivative with a 1.2 kb fragment from pI2549, rhiN+This work
 pI2861pT7-6 derivative with a 1 kb fragment from pN2538, rhaS+This work
Oligonucleotidesa
 RhiNGGCTCTAGACATATGACCATTTTCCCTGTGAAAC 
 RhiNDCGGGATCCTTACAGATAAACGCGCAAATAC 
 RhiTGGCGAATTCTTCTGTCTTCAAACTGACC 
 RhiTDCATAACAGATATCCAGTTCCAGTTTCG 
 RhiTMCCTATATTGCCGACGTGGATGAAGTC 
 RhiTEGAACCGTAGGCCAGGTAATTCGCCAGC 
 RhaSGGCTCTAGACTACTATCCGGACATGCGG 
 RhaRDTCGGTACCCGTCGCCACCCTTGTTTAT 

Media and growth conditions

Cells were grown in LB or in M63 medium (Miller, 1972). When required, the media were solidified with agar (15 g l−1). E. chrysanthemi cells were usually incubated at 30°C and E. coli cells at 37°C. Carbon sources were added at 2 g l−1. Polygalacturonate (sodium salt from citrus) was from Sigma Chemicals. Rhamnogalacturonan was kindly supplied by Dr E. Bonnin (Bonnin et al., 2001). When required, antibiotics were added at the following concentrations: kanamycin (Km), 20 µg ml−1; ampicillin (Ap), 50 µg ml−1; chloramphenicol (Cm), 20 µg ml−1; tetracycline (Tc) 10 µg ml−1; streptomycin (Sm), 20 µg ml−1.

Isolation of R-prime plasmids

The pgi mutations were cloned by selection for the KmR marker of MudI1681. Plasmid pULB110, a kanamycin-sensitive RP4::mini Mu derivative (Van Gijsegem et al., 1985), generates R-prime derivatives containing an insert of chromosomal DNA. To isolate such R-prime plasmids, the E. chrysanthemi pgi mutants containing pULB110 were mated with the E. coli strain HB101, and transconjugants were selected on LB medium plates supplemented with Km and Sm.

Enzyme assays

Pectate lyase activity was determined by monitoring spectrophotometrically the formation of unsaturated products from polygalacturonate. The assay mixture consisted of 0.1 M Tris-HCl, pH 8.5, 0.1 mM CaCl2 and 0.5 g l−1 polygalacturonate in a total volume of 1 ml. The appearance of products was monitored at 37°C. The molar extinction coefficient of unsaturated oligogalacturonides at 230 nm was assumed to be 5200 (Moran et al., 1968). Specific activity is expressed as µmol of unsaturated products liberated min−1 mg−1 bacterial dry weight.

Rhamnogalacturonate lyase activity was measured at 37°C by monitoring the appearance of unsaturated products at 230 nm (Laatu and Condemine, 2003). The standard assay mixture consisted of 0.1 M Tris-HCl, pH 6.8, and 0.5 g l−1 rhamnogalacturonan in a total volume of 1 ml. The RhiE extract corresponded to the periplasmic fraction of an E. coli strain overproducing RhiE, BL21/T7-RhiE (Laatu and Condemine, 2003). For the analysis of the RhiN effect, the assay mixture was incubated in the presence of the RhiE extract for 10 min before the addition of RhiN.

β-Glucuronidase activity was measured by following the degradation of p-nitrophenyl-β-d-glucuronide into p-nitrophenol at 405 nm. β-Galactosidase activity was measured by monitoring the cleavage of o-nitrophenyl-β-d-galactoside at 420 nm. Specific activity of these enzymes is expressed as nmol of products liberated min−1 mg−1 bacterial dry weight.

Various substrates from Sigma Chemicals were used to test the potential glycosidase activity of RhiN: p-nitrophenyl-α-l-arabinopyranoside (pNP-Ara), p-nitrophenyl-α-d-galactopyranoside (pNP-Gal), p-nitrophenyl-α-l-rhamnopyranoside (pNP-Rha) o-nitrophenyl-β-d-xylopyranoside (oNP-Xyl). The substrates (1 mM in 0.1 M phosphate buffer, pH 6.5) were incubated at 37°C with enzyme extracts, and the release of p-nitrophenol or o-nitrophenol was followed at 405 or 420 nm respectively.

Recombinant DNA and RNA techniques

Preparation of plasmid or chromosomal DNA, restriction digestions, ligations, DNA electrophoresis and transformations were carried out as described previously (Sambrook et al., 1989). For nucleotide sequence analysis, deletions were generated with restriction endonucleases. The sequences were performed by Genome Express. Complementary data were obtained from the whole genomic sequence of the E. chrysanthemi strain 3937 (http://www.tigr.org/tdb/mdb/mdbinprogress.html).

Insertion of the uidA-Km cassette (Bardonnet and Blanco, 1992) in a gene in the correct orientation generates a transcriptional fusion with the reporter gene uidA encoding β-glucuronidase. The uidA-Km cassette was inserted into the EcoRV site of rhiN and into the PstI site of rhiT to generate rhiN::uidA and rhiT::uidA fusions respectively. The rhaS gene was inactivated by insertion of the CKC15 Cm cassette between the EcoRV and PstI sites situated in rhaS. Plasmids were introduced into E. chrysanthemi cells by electroporation. The insertions were integrated into the E. chrysanthemi chromosome by marker exchange recombination after successive cultures in low phosphate medium in the presence of the appropriate antibiotic (Roeder and Collmer, 1985).

PCR primers (Table 3) designed to clone rhiN, rhiT and rhaSR are complementary to the 5′ and 3′ ends of the corresponding ORFs; restriction sites were added when necessary. The 3937 chromosomal DNA was used as the template. The PCR products were purified (QIAquick PCR purification kit, Qiagen), digested and ligated to the digested vector.

Total cellular RNA was isolated using the ‘SV Total RNA Isolation System’ kit from Promega. RNA integrity was determined by electrophoresis on denaturing agarose gels in the presence of formaldehyde. For Northern blotting, about 30 µg of RNA was transferred to a nylon membrane. The 563 bp DNA probe was produced by PCR using the primers RhiTM and RhiTD (Table 3). Labelling was obtained by incorporation of [α-32P]-dCTP during PCR. Hybridization was performed at 42°C in the presence of 50% formamide, and filters were finally washed in 0.2× SSC−0.1% SDS at 68°C.

Overproduction of the RhiN and RhaS proteins

The rhiN and rhaS genes were overexpressed using the T7 promoter/T7 RNA polymerase system (Tabor and Richardson, 1985) after cloning in the pT7-7 and pT7-6 vectors respectively. Plasmid pI2673 (rhiN+) was obtained after digestion of pI2549 and pT7-7 by NdeI and BamHI. Plasmid pI2861 (rhaS+) was obtained after digestion of pN2538 and pT7-6 by XbaI and SacI. The resulting plasmids were introduced into E. coli BL21(DE3), which contains a chromosomal copy of the T7 RNA polymerase gene under the control of the lacUV5 promoter (Studier and Moffat, 1986). The BL21(DE3) transformed cells were grown at 30°C in LB supplemented with Ap. At an OD600 of 0.6, the synthesis of T7 RNA polymerase was induced by the addition of IPTG (1 mM), and the bacterial RNA polymerase was blocked by the addition of rifampicin (0.2 mg ml−1). After an additional 2 h of growth, cells were harvested by centrifugation for 2 min at 12 000 g and broken by sonication. Centrifugation for 5 min at 12 000 g was used to remove the insoluble elements. Supernatants corresponded to the soluble protein fraction. Protein extracts were analysed by SDS–PAGE (12% acrylamide) followed by staining with Coomassie blue G-250.

Gel mobility shift assay

The 315 bp DNA fragment containing the rhiT regulatory region was labelled using [α-32P]-dCTP during amplification by PCR with primers RhiTG and RhiTE. The binding assays were essentially performed as described by Ausubel et al. (1987). Binding reactions contained 40 000 c.p.m. labelled DNA fragment, 50 µg ml−1 non-specific competitor DNA poly-(dI–dC), 10 mM Tris-HCl (pH 7.4), 100 mM KCl, 50 µg ml−1 BSA, 5% glycerol, 1 mM dithiothreitol, 1 mM EDTA, 50 mM l-rhamnose and protein extracts in a final volume of 15 µl. After incubation at 30°C for 30 min, protein-bound and free DNA were resolved on non-denaturing polyacrylamide gels (4% acrylamide, 0.05% bisacrylamide), running in a low-ionic-strength buffer (10 mM Tris-acetate, 1 mM EDTA, pH 7.4). Results were visualized by autoradiography.

Pathogenicity tests

Chicory leaves were slightly wounded with a pipette tip before inoculation. At least 10 leaves were infected for each strain using 10 µl of bacterial suspension (106 bacteria) per inoculation site. After incubation at 30°C for 24 h, the length of rotted tissue was measured to estimate the disease severity. Bacterial cell numerations and β-glucuronidase assays were performed on the macerated tissues to estimate the bacterial multiplication and the expression of genes in plant infection conditions.

Potato tubers were infected with 5 µl of bacterial suspension (5 × 105 bacteria) by inoculation in short holes obtained by inserting sterile pipette tips into the tuber parenchyma to a depth of 5 mm. Ten tubers were inoculated with each strain and incubated at 30°C. After 48 h, tubers were sliced vertically through the inoculation point. The softened tissue was recovered and weighed to estimate the degree of maceration.

Two-month-old Saintpaulia ionantha plants were inoculated after wounding a leaf by depositing 50 µl of bacterial suspension (5 × 106 bacteria). Ten plants were inoculated with each strain and incubated at 25°C. Plants were observed after 14 days, to give sufficient time for the development of systemic infection. In our experimental conditions, most of the plants infected with the wild-type strain gave systemic infection, i.e. the whole plant was macerated. When the maceration remained limited to the infected leaf, it was considered to be a localized infection.

Nucleotide sequence accession number

The sequence data reported in this paper (rhiN of E. chrysanthemi) have been submitted to the EMBL, GenBank and DDBJ databases under accession number AJ292045.

Acknowledgements

Appreciation is expressed to Valerie James for reading the manuscript. I thank Estelle Bonnin for the gift of well-characterized rhamnogalacturonan, and Guy Condemine for the gift of the protein RhiE. I gratefully acknowledge members of the International Erwinia Consortium for the exchange of unpublished data concerning the unfinished E. chrysanthemi 3937 genome (http://www.ahabs.wisc.edu/~pernalab/erwinia/index.html). I thank Geraldine Effantin for assistance with some experiments, and William Nasser for advice on gel mobility shift assay. This work was supported by grants from the Centre National de la Recherche Scientifique and from the Ministère de l’Education Nationale et de la Recherche.

Ancillary