hrp gene expression in the phytopathogenic bacterium Ralstonia solanacearum GMI1000 is induced through the HrpB regulator in minimal medium and upon co-culture with plant cell suspensions. The putative outer membrane protein PrhA is specifically involved in hrp gene activation in the presence of plant cells and has been proposed to be a receptor of a plant-dependent signal transduction pathway. Here, we report on the identification of two regulatory genes, hrpG and prhJlocated at the right-hand end of the hrp gene cluster, that are required for full pathogenicity. HrpG belongs to the OmpR subclass of two-component response regulators and is homologous to HrpG, the activator of hrp genes in Xanthomonas campestris pv. vesicatoria. PrhJ is a novel hrp regulatory protein, sharing homology with the LuxR/UhpA family of transcriptional activators. As for HrpG of X. c. pv. vesicatoria, HrpG is required for hrp gene expression in minimal medium, but, in addition, we show that it also controls hrpB gene activation upon co-culture with Arabidopsis thaliana and tomato cell suspensions. In contrast, PrhJ is specifically involved in hrp gene expression in the presence of plant cells. hrpG and prhJ gene transcription is plant cell inducible through the PrhA-dependent pathway. From these results, we propose a regulatory cascade in which plant cell signal(s) sensed by PrhA are transduced to the prhJ gene, whose predicted product controls hrpG gene expression. HrpG then activates the hrpB regulatory gene, and, subsequently, the remaining hrp transcriptional units in all known inducing conditions.
To overcome plant defence responses and colonize their hosts, phytopathogenic bacteria produce an enormous artillery of virulence factors in response to multiple environmental and plant-derived signals (for a review, see Van Gijsegem, 1997). In the last decade, the significant advances made in the genetic and molecular characterization of these bacterial pathogenicity factors have led to the identification of hrp (hypersensitive response and pathogenicity) genes, key determinants controlling the ability to cause disease on compatible hosts and to induce a defence reaction called the hypersensitive response (HR) on non-host or resistant plants (Lindgren et al., 1986). hrp genes have been identified in most Gram-negative phytopathogenic bacteria, that is Pseudomonas syringae, Xanthomonas, Erwinia and Ralstonia solanacearum, and have been found to be clustered in large regions of ≈20–35 kb (reviewed by Lindgren, 1997). A hint concerning the function of hrp genes was provided by the discovery that several conserved hrp genes, recently renamed hrc genes (for hrp-conserved, Bogdanove et al., 1996), are homologous to essential pathogenicity determinants of diverse animal pathogens (such as Yersinia, Shigella and Salmonella). hrc gene counterparts in animal pathogens encode components of a protein secretion machinery termed the type III secretion system (Van Gijsegem et al., 1993; Alfano and Collmer, 1997; Hueck, 1998). In Yersinia species, several Yop proteins, which are secreted through the type III secretion system, are translocated into eukaryotic cells (for a review, see Cornelis and Wolf-Watz, 1997). In plant pathogens, a set of proteins, including harpins of E. amylovora and P. syringae, PopA of R. solanacearum and DspA/DspE of E. amylovora, is secreted in the extracellular medium in an Hrp-dependent manner (Wei et al., 1992a; He et al., 1993; Arlat et al., 1994; Gaudriault et al., 1997; Bogdanove et al., 1998). Moreover, recent evidence suggests that avirulence proteins are transported into the plant cell and that this transport is dependent on a functional Hrp type III secretion system (Gopalan et al., 1996; Scofield et al., 1996; Tang et al., 1996; Van den Akervecken et al., 1996; and for a review Mudgett and Staskawicz, 1998). However, the role of these secreted and/or injected proteins during the infection process still remains largely unknown.
Expression of most hrp genes is suppressed during bacterial growth in complex media, but is induced in planta (Arlat et al., 1992; Rahme et al., 1992; Schulte and Bonas, 1992). Induction of hrp gene expression is also observed during growth in synthetic minimal media, where the level of expression depends on the nature of the carbon source provided (Arlat et al., 1992; Rahme et al., 1992; Wei et al., 1992b). In certain species, other factors, such as temperature, pH and osmolarity, also induce hrp gene expression (Rahme et al., 1992; Wei et al., 1992b; Xiao et al., 1992). All these environmental factors are thought to mimic physiological conditions encountered by bacteria during plant infection. Despite these common aspects, two different regulatory networks of hrp gene activation can be found. In P. syringae, three regulatory genes hrpR, hrpS and hrpL are involved. hrpS and hrpL have also been described in E. amylovora. The HrpR and HrpS proteins are highly similar to each other and share homology to σ54 enhancer binding proteins (Xiao et al., 1994). HrpR and HrpS positively regulate expression of hrpL, which encodes an alternative sigma factor of the extracytoplasmic subfamily. HrpL in turn activates all other hrp genes (Xiao et al., 1994; Wei and Beer, 1995). In R. solanacearum and X. campestris pv. vesicatoria two related regulatory proteins HrpB and HrpX, respectively, belong to the AraC/XylS family of regulators (Genin et al., 1992; Wengelnik and Bonas, 1996) and have been proposed to bind to the so-called PIP boxes that are found in several hrp promoters (Wengelnik and Bonas, 1996). In X. campestris pv. vesicatoria, an additional regulatory gene, hrpG, related to the OmpR subfamily of response regulators of two-component signal transduction systems, was also found to control the expression of the hrpX gene (Wengelnik et al., 1996).
Research in our laboratory is focused on hrp genes of Ralstonia solanacearum (formerly Pseudomonas solancearum) (Yabuuchi et al., 1995), the causal agent of bacterial wilt that affects several economically important solanaceous plants (Hayward, 1991). In R. solanacearum GMI1000, the hrp gene cluster is organized in seven hrp transcription units that comprise at least 20 genes (Arlat et al., 1992; Van Gijsegem et al., 1995). Transcription units 1, 2, 3, 4 and 7 are induced in minimal media through the hrpB regulatory gene. In contrast, expression of units 5 and 6, located at the right-hand end of the hrp gene cluster, is hrpB-independent and is not affected by media conditions (Genin et al., 1992). Recently, we reported that expression of hrp transcription units 1, 2, 3 and 4 of R. solanacearum is also induced when bacteria are co-cultured with plant cell suspensions (Marenda et al., 1998). This induction, which is also mediated by HrpB, was shown to be dependent on PrhA, a protein showing similarities with TonB-dependent siderophore receptors. Despite this homology, PrhA does not seem to be involved in iron starvation sensing and does not control hrp gene expression in minimal medium. In view of these results, we proposed that PrhA acts as a receptor for plant-specific signal(s) at the top of a novel regulatory pathway (Marenda et al., 1998).
In this paper, we report on the identification of two regulatory genes hrpG and prhJ located at the right-hand end of the hrp gene cluster. We demonstrate that both regulatory genes are required for full pathogenicity and control hrp gene expression upon co-culture with plant cell suspensions, whereas only hrpG is required for full induction in minimal medium. hrpG and prhJ gene transcription is plant cell inducible and PrhA-dependent. Our results allow us to define these novel regulatory genes as components of the PrhA-dependent signal transduction pathway.
Nucleotide sequence of the hrpG and prhJ genes
Previous studies had established that the hrp gene cluster of R. solanacearum GMI1000 was organized in at least seven transcriptional units (Arlat et al., 1992; Van Gijsegem et al., 1995). Units 5 and 6 located towards the right-hand end of the hrp region were defined by the Tn5-B20 insertions 1425 and 1423 respectively (Fig. 1). Mutants generated by these insertions induced a partial and delayed HR after infiltration into tobacco leaves and were hypoaggressive on compatible tomato plants (Arlat et al., 1992). To further characterize this region we have determined the nucleotide sequence of transcription unit 5 and of its downstream region. Data concerning unit 6 will be reported elsewhere.
The nucleotide sequence was determined for the 1675 bp EcoRI/SmaI DNA fragment shown in Fig. 1 (EMBL databank accession number AJ006694). This sequence is located 1 kb downstream of the hrpY gene corresponding to transcription unit 7 (Van Gijsegem et al., 1995) (Fig. 1). Sequence analysis led to the identification of an open reading frame from nucleotide 95–859, termed hrpG. Two methionine codons, nine amino acids apart, were detected as potential translational initiation sites. Although no consensus Shine–Dalgarno sequences were found upstream of these ATG codons, we propose the first methionine codon as the translational start site, based on sequence homology data that will be shown below. Therefore, hrpG is predicted to encode a 255-amino-acid polypeptide with a calculated molecular mass of 28.3 kDa. The Tn5-B20 insertion 1425 was mapped with the previously reported orientation after nucleotide 309 within the hrpG coding sequence. Thus, hrpG belongs to transcription unit 5 originally described by Arlat et al. (1992) (Fig. 1).
Analysis of the nucleotide sequence located downstream of hrpG revealed an additional open reading frame (ORF) from positions 991–1512, which was predicted to be oppositely transcribed (Fig. 1). This ORF, named prhJ, potentially encodes a polypeptide of 174 amino acid residues with a molecular mass of 19.1 kDa. The start codon is preceded by a putative ribosome binding site centred 8 bp upstream. A putative rho-independent transcription terminator was found in the intergenic region (131 nucleotides long) of hrpG and prhJ. Rich A+T stretches are present in the DNA sequence 5′ of hrpG and of prhJ, suggesting the existence of promoter regions for each gene.
HrpG and PrhJ show sequence similarity to regulatory proteins
The amino acid sequences of hrpG and prhJ were examined for homology to known proteins in the databases using the BLASTP algorithm (Altschul et al., 1990). This search revealed that HrpG shares significant similarities with response regulators of the OmpR subfamily of two-component signal transduction systems (Stock et al., 1989; Parkinson and Kofoid, 1992) (Fig. 2). The highest sequence similarity (36% identity, 55% similarity) was observed with HrpG of X. campestris pv. vesicatoria a protein described as a transcriptional activator of the hrp gene cluster in this bacterium (Wengelnik et al., 1996). The significant N-terminal conservation observed between HrpG of R. solanacearum and its homologue in X. campestris pv. vesicatoria suggested the choice of the translational initiation site. Among the members that originally defined the OmpR subclass, HrpG was also found to be related to PhoB regulators from Bradyrhizobium japonicum (Minder et al., 1998) and Escherichia coli (Makino et al., 1986), to the OmpR proteins from E. coli (Wurtzel et al., 1982) and Salmonella typhimurium (Liljestrom et al., 1988) and to ChvI from Rhizobium spp. (Osteras et al., 1995). The highest sequence conservation of HrpG with its homologues was observed in the C-terminal part corresponding to the helix–turn–helix domain (up to 54% identity and 59% similarity over a 95-amino-acid region) (Fig. 2A). The N-terminus of HrpG shares similarities to receiver domains of response regulators. In addition, we could identify an aspartic residue (Asp-60) in the HrpG amino acid sequence aligned to the phosphorylatable Asp-55 of OmpR (Brissette et al., 1991; Delgado et al., 1993) (Fig. 2A).
PrhJ shares homology to regulatory proteins of the LuxR/UhpA family of transcriptional activators. The most significant similarities were observed with the C-terminal helix–turn–helix domain of NarL and NarP from E. coli (Stewart et al., 1989; Rabin and Stewart, 1993), DegU from Bacillus subtilis (Kunst et al., 1988) and UhpA from S. typhimurium (Island et al., 1992) (Fig. 2B), all of which are members of the FixJ subfamily of response regulators. Values of up to 43% identity and 58% similarity were obtained in the C-terminal alignments of PrhJ to these proteins. However, no similarity to receiver domains of two-component response regulators or to other domains of regulatory proteins was observed in the N-terminal region of PrhJ. Significant similarities were also observed with helix–turn–helix domains of activators that are not response regulators, such as MalT from E. coli (Cole and Raibaud, 1986) and MoaR from Klebsiella aerogenes (Azakami et al., 1993) (Fig. 2B). Interestingly, no homology was found to other hrp regulatory genes described to date.
hrpG and prhJ genes are required for pathogenicity
As mentioned above, strain GMI1425 carrying a hrpG ::Tn5-B20 insertion was originally reported as having a leaky phenotype in both its ability to elicit the HR on tobacco leaves and to induce disease on tomato plants (Arlat et al., 1992). To study the role of the hrpG and prhJ genes in pathogenicity, mutations in each gene were generated by introducing an antibiotic resistance cassette (Ω cassette) (Prentki and Krisch, 1984). These insertions carried by plasmids pBBL103Ω and pBBL102Ω were marker exchanged into strain GMI1000 of R. solanacearum to obtain strains GMI1578 (hrpG ::Ω) and GMI1579 (prhJ ::Ω).
Strains GMI1578, GMI1579 and GMI1425 were tested for HR elicitation on tobacco leaves. Unlike the previous results obtained with GMI1425, no visible HR symptoms could be seen, even at inoculation densities of up to 108 cfu ml−1 and at 72 h post infiltration. No HR symptoms were seen with GMI1578, whereas GMI1579 showed a leaky phenotype, i.e. induction of a partial and delayed HR after 48–72 h at 108 cfu ml−1 (data not shown). Mutations in the prhA locus of R. solanacearum are also characterized by a delayed HR-inducing ability on tobacco (Marenda et al., 1998). To compare the degree of leakiness between the prhJ and the prhA mutants, we analysed, in the same test, the HR symptoms produced by GMI1579 and the prhA::Tn5-B20 mutant GMI1567. At all concentrations tested, the HR induced by GMI1579 (prhJ ::Ω) was more restricted and delayed than the one triggered by strain GMI1567 (data not shown).
We then evaluated the aggressiveness of strains GMI1578, GMI1579 and GMI1425 towards tomato and Arabidopsis thaliana plants. At inoculation densities of 108 and 107 cfu ml−1, GMI1578 and GMI1425 were totally non-pathogenic on tomato plants, whereas GMI1579 was hypoaggressive (Fig. 3A). Phenotypes were also checked on A. thaliana accession Col-V plants, described as a susceptible host for R. solanacearum GMI1000 (Deslandes et al., 1998). Again, GMI1578 and GMI1425 were completely unable to develop disease symptoms at concentrations of 108 cfu ml−1, whereas mutant GMI1579 was strongly affected for pathogenicity (Fig. 3B).
Complementation assays were carried out to confirm that the observed phenotypes were due to the Ω mutations. These assays also allowed us to determine the transcriptional organization of this region. pLAFR6 derivative plasmids carrying the hrpG or prhJ coding sequences, each one under the control of its corresponding potential promoter region, were constructed (Fig. 1). The presence of synthetic trp terminators flanking the polylinker sequence of this vector ensures expression of each cloned gene from its 5′ control region. These plasmids were introduced into GMI1578, GMI1425 and GMI1579 and the resulting strains checked for complementation of the HR phenotype on tobacco. Introduction of plasmid pBBL12 carrying the coding sequence of hrpG in a 1.5 kb Pst I DNA fragment (Fig. 1) restored HR elicitation in mutants GMI1578 and GMI1425, but not in GMI1579 (data not shown). Mutant GMI1579 was complemented by plasmid pBBL13 harbouring the prhJ gene in a 1.4 kb Sal I/BamHI DNA fragment (Fig. 1). The transfer of pBBL13 into GMI1578 and GMI1425 did not restore the HR ability of these strains (data not shown). These complementation assays thus show that pBBL12 and pBBL13 carry functional hrpG and prhJ genes respectively. Only partial complementation was observed in pathogenicity tests towards tomato and Arabidopsis plants. This was probably due to plasmid instability in the absence of selection pressure in planta.
From these assays we conclude that both the hrpG and prhJ genes are required for full HR elicitation and pathogenicity. In addition, mutant complementation analysis indicates that the hrpG and prhJ genes define two monocistronic operons.
hrpG, but not prhJ, is required for hrp gene expression in minimal medium
Minimal medium has been described as an inducing condition for R. solanacearum hrp gene expression (Arlat et al., 1992). In addition, results obtained by Genin et al. (1992) indicated that in minimal medium the expression of the hrpB regulatory gene is controlled by upstream activating factors. The homologies reported in this paper with transcriptional activators, and specifically with the hrpG regulatory gene of X. campestris pv. vesicatoria, suggest that HrpG and probably PrhJ may be implicated in the regulation of R. solanacearum hrp genes.
To test whether the transcriptional activation of hrpB (unit 1) and hrp units 2, 3 and 4 is controlled by the hrpG and prhJ genes in minimal medium, we measured in an hrpG or a prhJ mutant background, the β-galactosidase activity associated with Tn5-B20 genomic insertions in units 1, 2, 3 and 4. For this purpose, we inserted by marker exchange the Ω cassette in the hrpG gene into strains GMI1475 (hrpB ::Tn5-B20), GMI1494 (hrpK ::Tn5-B20), GMI1492 (hrcU ::Tn5-B20) and GMI1389 (hrpV ::Tn5-B20). Derivative strains affected in the prhJ gene were obtained in the same way. As shown in 4Fig. 4A, the low hrpB gene transcription exhibited in minimal medium was slightly reduced in both an hrpG and a prhJ mutant background, but was most affected in the absence of hrpG. This effect was more clearly observed in the highly activated units 2, 3 and 4, whose levels of expression were reduced in the hrpG mutant background. This reduction fluctuated between 3.5- and 20-fold depending on the hrp transcriptional unit monitored. However, when the prhJ gene was disrupted, no significant difference in the β-galactosidase activity associated with these units was observed (Fig. 4A). This latter result is reminiscent of that obtained when the expression of these same Tn5-B20 insertions was tested in minimal medium in a prhA mutant background (Marenda et al., 1998).
Our results indicate that in minimal medium HrpG probably acts upstream of HrpB as a transcriptional activator of hrp gene expression. In contrast, hrp gene induction seems to be independent of PrhJ in these conditions.
hrpG and prhJ genes control hrp gene expression in bacteria–plant cell suspension co-cultures
Recently, new in vitro activating conditions for hrp gene expression in R. solanacearum have been described (Marenda et al., 1998). Upon co-culture of bacteria with tomato or A. thaliana cell suspensions, hrp gene transcription is specifically induced via the putative outer membrane receptor protein PrhA. This discovery suggested the existence of specific plant signal(s) controlling hrp gene expression in this bacterium (Marenda et al., 1998). Therefore, we tested the effect of an hrpG or of a prhJ mutation on hrp gene expression in co-culture with plant cell suspensions. For this, we compared the β-galactosidase activities produced by the hrp::Tn5-B20 fusions in units 1, 2, 3 and 4 (described above) with those induced by the corresponding hrp::Tn5-B20/hrpG ::Ω and hrp::Tn5-B20/prhJ ::Ω double mutants previously generated.
Insertion 1475 (unit 1) tested upon co-culture with tomato and Arabidopsis cell suspensions exhibited an strong induction (more than 10-fold) versus the activity observed in minimal medium (Fig. 4B and C). In the presence of plant cells, fusions 1494, 1492 and 1389 corresponding to units 2, 3 and 4, respectively, showed slightly higher or similar values to those obtained in minimal medium. An expression analysis of several Tn5-B20 insertions in transcriptional units 2, 3 and 4 confirmed that on average insertions 1494, 1492 and 1389 are representative of the ratio of activation of hrp units. Upon co-culture with tomato cell suspensions the β-galactosidase activity associated with the hrp::Tn5-B20 insertions was abolished in an hrpG mutant background and strongly reduced in the prhJ mutant strains (Fig. 4B). When the same double mutants were tested for induction in co-culture with Arabidopsis cell suspensions, similar results were obtained (Fig. 4C). These results were confirmed by monitoring activities from additional Tn5-B20 insertions in each hrp transcriptional unit (data not shown). β-Galactosidase activities were restored when plasmids pBBL12 and pBBL13 carrying functional copies of the hrpG and prhJ genes, respectively, were introduced appropriately into the hrp::Tn5-B20/hrpG ::Ω and hrp::Tn5-B20/prhJ ::Ω double mutants (data not shown). We also analysed whether the expression of the prhA gene was altered when the hrpG or prhJ genes were disrupted. No change in the activity of the prhA::Tn5-B20 insertion 1567 was detected in any case (data not shown), indicating that prhA expression is independent of HrpG and PrhJ.
Taken together these experiments establish that both HrpG and PrhJ control hrp gene transcription in the presence of plant cells but have different functions. HrpG is a key transcriptional activator being required in all known activating conditions: minimal medium and plant cell suspension co-culture. PrhJ specifically controls the plant cell dependent hrp gene expression, and is not involved in hrp gene activation in minimal medium. This latter behaviour resembles that observed for PrhA, suggesting that PrhJ and PrhA are involved in a common regulatory pathway.
hrpG and prhJ gene expression is plant cell inducible
As the hrpG and prhJ genes were shown to control hrp gene expression when bacteria are grown in the presence of plant cell suspensions, we then investigated whether hrpG and prhJ gene expression was induced in these conditions. Previous results had demonstrated that hrpG gene transcription (monitored by the Tn5-B20 insertion 1425) was not activated under minimal medium conditions (Genin et al., 1992). When we examined the β-galactosidase activity associated with insertion 1425 upon co-culture with Arabidopsis and tomato cell suspensions, a strong induction of hrpG gene transcription compared to that exhibited in minimal medium, Gamborg or MS media was observed (Fig. 5A). This activation obtained in an hrpG mutant background indicates that, in the presence of plant cells, upstream activating factors/genes are involved in hrpG gene induction.
We then investigated the pattern of prhJ gene expression in the same conditions. As no Tn5-B20 insertion in prhJ was available, a prhJ ::lacZ transcriptional fusion was generated. The prhJ gene was mutagenized in E. coli to obtain plasmid pBBL1011 carrying a Tn5-B20 insertion in the correct orientation (for details see Experimental procedures). This Tn5-B20 insertion, termed 1577 (Fig. 1), was localized by sequence analysis after nucleotide 1250 and then marker exchanged into the genome of the GMI1000 wild-type strain to obtain GMI1577.
As for hrpG, prhJ gene transcription in strain GMI1577 was detected at a low level both in rich and minimal medium conditions. When GMI1577 was co-cultured with Arabidopsis cell suspensions a sevenfold induction in the level of prhJ gene expression was observed compared with the value obtained in Gamborg medium (Fig. 5B). The same insertion tested upon co-culture with tomato cells was significantly less activated (twofold higher than the activity observed in rich medium, and fourfold the one displayed in minimal and MS medium), indicating that cells of distinct plants promote different levels of activation.
PrhA is required for hrpG and prhJ gene expression
To investigate which were the upstream activating factors/genes controlling hrpG and prhJ gene expression and to tentatively define their relative positions in the regulatory cascade, we generated mutant strains by insertion into strain GMI1425 of an Ω cassette in the prhJ or prhA genes. GMI1535, a GMI1425 derivative strain carrying an hrpB ::Ω insertion (Genin et al., 1992) was also tested in this assay. Disruption of the hrpB gene did not affect hrpG gene expression upon co-culture with either Arabidopsis or tomato cells (Fig. 6). In contrast, the absence of PrhJ dramatically reduced the level of induction of hrpG in both Arabidopsis and tomato cell culture conditions. hrpG gene transcription was also affected in a prhA::Ω mutant, but this reduction was less important in tomato than in Arabidopsis cell co-culture (Fig. 6). This latter result resembles the effect that a prhA mutation exerts on the expression of other hrp transcriptional units upon co-culture with tomato cells, where hrp gene induction was higher than in the co-culture with Arabidopsis (Marenda et al., 1998). Complementation experiments were performed to restore hrpG gene expression by the introduction of plasmids pBBL13 and pMM44 (carrying functional copies of prhJ and prhA respectively) in the prhJ ::Ω or the prhA::Ω mutant (data not shown). In addition, we tested these double mutants in minimal medium, and no change was observed on the basal transcription of hrpG (data not shown). From these results, we conclude that the plant cell-induced transcription of hrpG is controlled by both the prhJ and prhA gene products.
We then studied the relation of prhJ with other known hrp regulatory genes. We generated double mutants by marker exchange of insertion 1577 into the genome of mutants GMI1525 (hrpB ::Ω), GMI1578 (hrpG ::Ω), and GMI1575 (prhA::Ω). These double mutants were then co-cultured with Arabidopsis cell suspensions. The results clearly showed that disruption of the hrpB or hrpG genes had no effect on prhJ gene expression in these conditions, whereas the mutation in prhA reduced the level of prhJ gene induction (Fig. 7). The expression was restored when plasmid pMM44 was introduced in the prhJ ::Tn5-B20/prhA::Ω double mutant (data not shown). When we checked the mutants upon co-culture with tomato cell suspensions, strains carrying the Ω insertion in the hrpB and hrpG genes showed similar β-galactosidase activity to that of strain GMI1577, whereas in the prhA mutant prhJ gene expression was again reduced to the level obtained in rich and minimal medium (Fig. 7).
Altogether these data indicate that the regulatory cascade leading to hrp gene expression requires the activity of two regulatory proteins HrpG and PrhJ. Genetic evidence suggests that the plant cell signal sensed by PrhA is transduced to PrhJ, which directly or indirectly activates hrpG gene transcription. HrpG, in turn, controls the expression of the remaining hrp units through the hrpB regulatory gene in both plant cell co-culture and minimal medium conditions.
HrpG and PrhJ, two novel regulators of the R. solanacearum hrp gene cluster
In this paper we report the finding of two new regulatory genes hrpG and prhJ located towards the right-hand end of the hrp gene cluster of R. solanacearum GMI1000. HrpG belongs to the large family of transcriptional activators of the two-component signal transduction systems. These regulatory systems comprises a sensor kinase protein that autophosphorylates at a conserved histidine residue of its transmitter domain in the presence of an environmental signal. The phosphoryl group is then transferred to a conserved aspartic residue in the receiver domain of the cognate response regulator, whose phosphorylation state modulates the activity of its output domain to trigger the adaptive response (Albright et al., 1989). HrpG is related to OmpR-like response regulators, which consist of a conserved N-terminal receiver domain and a C-terminal output domain containing a helix–turn–helix motif involved in transcriptional activation. The helix–turn–helix domain of HrpG appears to be highly conserved with the homologous proteins. However, despite the homologies observed with receiver domains and the presence of a putative phosphorylatable aspartic acid residue Asp-60, several key amino acids such as those corresponding to Asp-11, Asp-12 and K-105 of OmpR are missing in the HrpG receiver domain. The absence of these amino acids, which are essential for signalling and phosphorylation (Parkinson and Kofoid, 1992), prompts us to question whether phosphorylation at Asp-60 would be a possible event in HrpG. Several examples of response regulators lacking these residues in their receiver domains have been described, such as the FlbD regulator of Caulobacter crescentus (Ramakrishnan and Newton, 1990) and HrpG of X. campestris pv. vesicatoria (Wengelnik et al., 1996), to which HrpG from R. solanacearum shows the highest sequence similarity. At least for FlbD, the importance of the equivalent aspartic residue and indirect evidence that phosphorylation controls its activity have been demonstrated (Wingrove et al., 1993). In the case of HrpG of R. solanacearum, experiments of in vitro phosphorylation as well as the study of point mutations in the Asp-60 residue would clarify this aspect. The identification of its sensor kinase partner would also help to elucidate this point, although unlike most two-component regulatory pairs, no sensor kinase protein has been identified in the vicinity of the hrpG gene (our unpublished results).
PrhJ is a novel hrp regulatory protein with a notable C-terminal helix–turn–helix motif similar to the DNA binding domain of transcriptional activators of the LuxR/UhpA family. Among the proteins sharing similarities with PrhJ are members of the FixJ subfamily of response regulators. However, PrhJ should not be considered as a response regulator, because no homology to receiver domains was observed in its N-terminal part.
Different regulatory roles for HrpG and PrhJ in hrp gene expression
The role of the hrpG and prhJ genes in pathogenicity and hrp gene expression was then investigated. An hrpG mutant strain was unable to induce the HR on tobacco leaves and to provoke disease symptoms on Arabidopsis and tomato plants. In minimal medium, hrp gene expression was strongly reduced in an hrpG mutant strain, although a residual activation was still observed. This result is in agreement with that reported for HrpG of X. campestris pv. vesicatoria (Wengelnik et al., 1996). In addition, we found that hrp gene induction in the presence of Arabidopsis and tomato cells was completely abolished in an hrpG mutant background. From these results, we conclude that HrpG plays a central regulatory role, being required for hrp gene activation in the presence of both plant cell-derived signals and stimuli of the metabolic state.
Unlike hrpG, disruption of the prhJ gene did not affect expression of hrp genes in minimal medium, whereas hrp gene activation was significantly reduced upon co-culture with plant cell suspensions. This behaviour, together with the hypoaggressive phenotype of the prhJ mutant towards host plants, resembles the effect exerted by a prhA mutation on R. solanacearum hrp gene activation and pathogenicity (Marenda et al., 1998). For these reasons, we consider prhJ as a prh gene (for plant regulatory hrp) specifically involved in transduction of plant cell-derived signal(s).
Concerning the pathogenicity phenotypes, it is also interesting to note the differential behaviour displayed by the prhJ mutant strain on Arabidopsis versus tomato plants. Whereas on tomato this strain is only hypoaggressive, it is practically unable to induce any disease symptoms on Arabidopsis plants. As PrhJ is not involved in activation in minimal medium and only appears to respond to plant signals, plant cell-derived signal(s) must be essential for bacterial virulence in Arabidopsis, and the additional environmental stimuli detected by R. solanacearum inside this host presumably have a reduced contribution to pathogenicity. In contrast, bacteria might encounter multiple activating signals in tomato plants, where nutrient starvation and other metabolic signals probably still play an important role in hrp gene activation and pathogenicity in addition to plant derived signal(s).
PrhJ, HrpG and HrpB: a regulatory cascade transducing plant signal(s) sensed by PrhA
The requirement of HrpG for hrp gene expression in minimal medium and upon co-culture with plant cells prompted us to investigate whether its own expression was induced in these conditions. Interestingly, unlike the basal transcription observed in minimal medium, we found that hrpG gene expression was highly induced upon co-culture with Arabidopsis and tomato cells. This observation could explain the strong activation of hrpB detected in the presence of plant cells. But, unexpectedly, this high hrpB induction does not trigger, in turn, the same ratio of activation in the remaining hrp units. The explanation of this paradox requires further investigation, but preliminary results suggest that the hrpB gene product could modulate its own expression and the final level of activation of hrp genes (our unpublished data).
The expression of the prhJ gene was also significantly induced in the presence of Arabidopsis cell suspensions, whereas this gene was only weakly activated with tomato cells. This result could suggest either a different level of induction by cell suspensions of distinct plants or the existence of other mechanism of activation of PrhJ.
Taking advantage of the plant cell dependent induction of hrpG and prhJ gene expression, we demonstrate that PrhA is required for hrpG and prhJ gene expression in the presence of plant cells. From these results, we postulate that plant cell signal(s) sensed by the PrhA receptor are transduced to PrhJ. This information is then transferred to the hrpG gene, whose predicted product is required for activation of the hrpB regulatory gene and hence the remaining hrp transcriptional units in the presence of plant signals, but also in response to nutrient conditions (Fig. 8).
This model of regulation raises several comments. Interestingly, it appears clearly that the lower in the regulatory cascade the mutation occurs, the stronger the impact is on the observed phenotype (hrp gene expression and pathogenicity). This is consistent with the integration of multiple input signals at different levels of the regulatory hierarchy. In this model, the functional role of PrhA and PrhJ (top of the cascade), although important in certain conditions (pathogenicity on Arabidopsis), could be bypassed in others because additional signals (i.e. minimal medium) and/or regulatory components can activate the hrpG/hrpB gene expression. In contrast, both HrpG and HrpB proteins (bottom of the cascade) are required in all conditions, in agreement with the complete Hrp− phenotype of the corresponding mutants.
Regarding the mechanism of signal transduction in this regulatory cascade, Marenda et al. (1998) proposed that receptors other than PrhA could be involved in the activation of the plant cell-dependent pathway. In this current work, we report additional evidence that supports this hypothesis: hrp gene expression in plant cell co-culture and pathogenic phenotypes are more attenuated with the prhJ mutant than with the prhA one. These observations suggest that PrhJ can respond to some inducing signal(s) that is not recognized by the PrhA receptor, and therefore implies that the integration of signals by PrhJ may involve multiple receptors. However, how is the plant signal/s transduced from the putative outer-membrane receptor PrhA in order to intracellularly activate the expression of the prhJ gene?. This mechanism, which is still unknown, will probably require the action of additional components.
In addition, we need to consider where and how the information of the nutrient and metabolic state transduced by the ‘minimal medium pathway’ is integrated in the cascade. As HrpG is required for hrp gene expression in minimal medium, one possibility is that the metabolic signals are integrated at the HrpG level. In this case, a mechanism of phosphorylation of HrpG by a sensor kinase protein would conciliate the fact that hrpG gene expression is not activated by minimal medium signal(s) while its gene product is required in these conditions. Integration of minimal medium signals could also occur in an intermediate step between HrpG and HrpB, because direct activation of hrpB gene expression by HrpG has not been demonstrated. Another possibility could be that HrpB is directly activated by stimuli from the nutrient state and this would compensate for the basal hrpB gene expression level directed by HrpG. This latter hypothesis, together with the fact that HrpB can slightly activate its own expression in minimal medium (Genin et al., 1992), could also explain the residual hrp gene expression observed in an hrpG mutant strain in these conditions.
Our data show that hrp gene expression in R. solanacearum is controlled by a regulatory cascade that comprises multiple components and integrates different signals. In addition, these results support the classification of hrp gene clusters in two subgroups proposed by Alfano and Collmer (1997) and based on differences on sequence homologies, gene organization and regulatory mechanisms. The regulatory proteins here described do not exhibit homology to the HrpR, HrpS and HrpL regulators identified in hrp gene clusters from subgroup I. In contrast, X. campestris pv. vesicatoria and R. solanacearum, both belonging to subgroup II, share common hrp activators, HrpB/HrpX and now HrpG. It will be interesting to determine whether the remaining components of the R. solanacearum hrp regulatory cascade are also present in X. campestris pv. vesicatoria.
The proposed model for hrp gene regulation in R. solanacearum presents several similarities to the complex regulatory cascade that controls the expression of the type III secretion machinery in S. typhimurium. In this pathogen, metabolic and host cell signals are transduced by rather complicated cascades that interact in order to activate their pathogenic determinants (Cotter and Miller, 1998). Environmental factors (such as the extracellular Mg2+ concentration) regulate the expression of virulence genes through the PhoP/PhoQ two-component regulatory system (Miller et al., 1989). On the other hand, the multistep cascade that transduces the host cell-derived signals in this bacterium, comprises a member of the FixJ subfamily of transcriptional activators (SirA) controlling the expression of an OmpR-like regulator (HilA), which regulates an AraC-like activator (InvF) (Johnston et al., 1996; Bajaj et al., 1996). However, the mechanism of transduction of host signals to SirA still remains to be elucidated.
Besides the known targets of HrpB, HrpG and PrhJ, one attractive hypothesis is that these regulatory proteins activate additional genes required for pathogenicity. Production of virulence factors such as plant cell-degrading exoenzymes and extracellular polysaccharides in R. solanacearum exemplifies another complex regulatory network that comprises several activators and inducing signals (Schell, 1996; Flavier et al., 1997a,b). An unidentified plant signal has also been proposed to influence the expression of some of these regulatory proteins (Allen et al., 1997). It would be interesting to investigate whether the hrp and additional virulence factor regulatory networks interact; and, if so, we would need to determine the level of interaction and the signals that are involved. The implications of a putative connection between all these pathogenic determinants in R. solanacearum would be of major interest in the global study of bacterial virulence.
Bacterial strains, plasmids, phages and growth conditions
The relevant characteristics of the bacterial strains used in this work are listed in Table 1. E. coli strains were grown at 37°C in Luria Bertani medium (Sambrook et al., 1989). R. solanacearum strains were grown at 28°C in B medium or minimal medium (Boucher et al., 1985) supplemented with glutamate 20 mM at final concentration. When needed, antibiotics were added to the media at the following final concentrations (mg l−1): kanamycin (Km) 25 (E. coli ), 50 (R. solanacearum); spectinomycin (Sp) 25 (E. coli ), 40 (R. solanacearum); tetracycline (Tc) 10; ampicilin (Ap) 100. Plasmids were introduced in R. solanacearum by triparental mating using pRK2013 or pRK2073 as conjugative helper plasmids by the protocol described in Arlat et al. (1992). Bacteriophage T4 was used to counterselect E. coli in interspecific matings. R. solanacearum was transformed according to the method of Boucher et al. (1985).
Plasmid extraction, restriction enzyme digestions, agarose gel electrophoresis and DNA cloning were performed by standard methods (Sambrook et al., 1989). Both strands of the DNA region coding for hrpG and prhJ were sequenced by the dideoxy chain terminator method. Serial deletions of plasmid pBBL121 were generated by exonuclease III-nuclease S1 treatment. DNA and protein sequence manipulations were carried out with the GCG (Wisconsin Sequence Analysis Package, Version 9.0) software package. Databases searches were carried out with the BLASTX and BLASTP algorithms (Altschul et al., 1990). Search of conserved protein domains was performed with the PRODOM service (Corpet et al., 1998). The GAP program was used for the protein sequence comparisons. The terminator sequence was found by the GCG program TERMINATOR. The alignments of multiple sequences shown in Fig. 2 were carried out with the CLUSTALW version 1.6 program (Thompson et al., 1994).
Tn5-B20 mutagenesis was performed as described in Arlat and Boucher (1991). Tn5-B20 insertions in plasmid pBBL101 were positioned by restriction fragment analysis. Plasmid pBBL1101, which carries an insertion in the prhJ gene, was linearized and marker exchanged into R. solanacearum strains. Correct genomic insertion was confirmed by Southern blotting. The precise location of insertions 1425 and 1577 was determined by sequencing of the adjacent region to the Tn5-B20 transposon in plasmids pBBL832 and pBBL1101, respectively, as described in Genin et al. (1992).
Construction of the hrpG::Ωand prhJ::Ωmutants and hrp::Tn5-B20/Ω double mutants
The hrpG ::Ω and prhJ ::Ω mutants were obtained by insertion of the Ω cassette (Spr/Smr) (Prentki and Krisch, 1984) from plasmid pHP45Ω into the Sal I and XhoI restriction sites located in the hrpG and prhJ genes respectively. For that purpose, plasmid pBBL103 carrying the entire sequence of the hrpG gene was digested with Sal I, blunt-ended in its extremities with the DNA Blunting kit from Amersham and ligated to the Ω cassette previously restricted with SmaI. To generate the prhJ ::Ω mutant, the SmaI-digested Ω cassette was cloned into the XhoI site of plasmid pBBL102 after blunt-end treatment. Plasmids pBBL102 and pBBL103 with the corresponding Ω insertions were linearized by restriction with XbaI and used to transform R. solanacearum GMI1000 and diverse hrp::Tn5-B20 or prh ::Tn5-B20 derivative strains. Alternatively, strains GMI1425 and GMI1577 were transformed with genomic DNA from GMI1575 (prhA::Ω). The genomic insertion of all the resulting strains were confirmed by Southern blot using total genomic DNA.
Ralstonia solanacearum strains were tested for HR ability by infiltration of bacterial cultures adjusted to 107 and 108 cells ml−1 into tobacco (cultivar Bottom special) leaf parenchyma as described by Boucher et al. (1985). Pathogenicity tests on tomato (Lycopersicum esculentum cv. Marmande) were performed by inoculating bacterial suspensions of 5 × 109 and 5 × 108 cfu per plant according to Arlat and Boucher (1991). Disease tests on A. thaliana accession Col-V (provided by Lehle Seeds Company) were performed with bacterial inocula of 2 × 109 cfu per plant according to the procedure of Deslandes et al. (1998). Wilting symptoms were daily scored using an arbitrary disease index from 1 to 4 according to plant state (Arlat and Boucher, 1991).
Plant cell cultures and bacteria–plant cell co-cultures
The A. thaliana At-202 (accession Col-0) (Callard and Mazzolini, 1997) and the tomato Msk8 (Felix et al., 1991) cell suspensions were grown in Gamborg B5 (Flow Laboratories) and T-MSMO (Sigma) medium respectively (Felix et al., 1991; Marenda et al., 1998). For the bacteria–plant cell co-cultures, samples of 10 ml of Arabidopsis and tomato cells suspensions in exponential phase were inoculated with R. solanacearum cultures as described in Marenda et al. (1998). After 16 h of incubation at 28°C, the mixture was filtered and the bacteria recovered for the β-galactosidase tests. The β-galactosidase activity was measured according to Miller (1972) and values are expressed in Miller units.
We thank Dominique Douilhac for technical assistance and Sylvie Camut, Claudette Icher and Jean-Luc Pariente for plant preparation. We also thank Daniel Kahn and Frédérique Van Gijsegem for helpful discussions. We are in debt to Matthieu Arlat, Jacques Batut and Clare Gough for critical reading of the manuscript. This work was supported by projects BIO4-CT-97-2244 from the European Commission and AIP-188 Microbiologie from the Institut National de la Recherche Agronomique. B.B. is the recipient of a Marie Curie TMR postdoctoral grant from the European Commission. M.M. was the recipient of a grant from the Ministère de l'Enseignement Supérieur et de la Recherche.