In vivo expression technology (IVET) analysis of rhizosphere-induced genes in the plant growth-promoting rhizobacterium (PGPR) Pseudomonas fluorescens SBW25 identified a homologue of the type III secretion system (TTSS) gene hrcC. The hrcC homologue resides within a 20-kb gene cluster that resembles the type III (Hrp) gene cluster of Pseudomonas syringae. The type III (Rsp) gene cluster in P. fluorescens SBW25 is flanked by a homologue of the P. syringae TTSS-secreted protein AvrE. P. fluorescens SBW25 is non-pathogenic and does not elicit the hypersensitive response (HR) in any host plant tested. However, strains constitutively expressing the rsp-specific sigma factor RspL elicit an AvrB-dependent HR in Arabidopsis thaliana ecotype Col-0, and a host-specific HR in Nicotiana clevelandii. The inability of wild-type P. fluorescens SBW25 to elicit a visible HR is therefore partly attributable to low expression of rsp genes in the leaf apoplast. DNA hybridization analysis indicates that rsp genes are present in many plant-colonizing Pseudomonas and PGPR, suggesting that TTSSs may have a significant role in the biology of PGPR. However, rsp and rsc mutants retain the ability to reach high population levels in the rhizosphere. While functionality of the TTSS has been demonstrated, the ecological significance of the rhizosphere-expressed TTSS of P. fluorescens SBW25 remains unclear.
Pseudomonas are efficient colonizers of plants and form intimate associations with host cells, leading to disease or enhanced growth and disease resistance. Two interactions between Pseudomonas and plants have attracted significant attention: plant parasitism by plant pathogenic Pseudomonas syringae (Alfano and Collmer, 1997); and stimulation of plant growth and induced systemic resistance (ISR) by plant growth-promoting Pseudomonas fluorescens and Pseudomonas putida (van Loon et al., 1998; Pieterse and van Loon, 1999). In both interactions, molecular exchanges at a cellular level result in local and systemic changes in gene expression and physiology of the host, and both pathogens and plant growth-promoting rhizobacteria (PGPR) possess common abilities to recognize, colonize and parasitize plants.
A major breakthrough in understanding pathogenicity and parasitism in P. syringae was the discovery and characterization of the hypersensitive response and pathogenicity (hrp) gene cluster (Alfano and Collmer, 1997). Hrp– mutants are unable to multiply in plant tissues and are non-pathogenic. They are also unable to elicit the rapid defence response known as the hypersensitive response (HR) when introduced into resistant or non-host plants. The HR is an active plant defence response that typically involves both localized programmed cell death and production of antimicrobial compounds 12–24 h after contact between plant cells and organisms, or elicitors that are recognized by surveillance mechanisms (Dangl et al., 1996). Non-pathogenic bacteria such as P. fluorescens and Escherichia coli generally fail to elicit the HR, even when inoculated at very high cell densities. The Hrp phenotype depends on a cluster of genes that encode a type III protein secretion system (TTSS) and associated regulatory, accessory and secreted proteins. Similar gene clusters have been identified in a wide range of plant and animal pathogens and there is increasing evidence that TTSS genes have been horizontally transferred (Hueck, 1998).
The TTSS of P. syringae secretes and targets proteins to the apoplast and the cytoplasm of host cells. However, direct evidence of TTSS-dependent secretion is only available for a limited number of proteins. In the absence of direct evidence for secretion, putative TTSS secreted proteins are identified on the basis of homology, TTSS-dependent phenotypes and the effect of transient expression of these proteins in planta (Collmer et al., 2000; Kjemtrup et al., 2000; Preston, 2000). TTSS secreted proteins fall into two main functional categories: translocation of TTSS secreted proteins into host cells (such as the hrp pilus HrpA; Wei et al., 2000); and effectors that act upon host cells, such as AvrRpm1 (Ritter and Dangl, 1995) and VirPphA (Jackson et al., 1999). The primary role of TTSSs in plant pathogens appears to be promotion of intercellular plant parasitism and endophytic growth. TTSS mutants of P. syringae and Erwinia chrysanthemi can still induce symptom development at high inoculum densities, but are impaired in endophytic growth and pathogenicity, particularly when inoculated at a low inoculum density (Bauer et al., 1994; Hirano et al., 1999; Penaloza-Vazquez et al., 2000).
Modulation of host metabolism by TTSS secreted proteins may play a crucial role in releasing nutrients for bacterial uptake, suppressing host defences or in rendering the apoplast of plant tissues hospitable to bacterial growth. Although most of the bacteria from which TTSSs have been identified and characterized are pathogenic, TTSSs have been identified in strains of the endophytic plant symbiont, Rhizobium (Meinhardt et al., 1993; Freiberg et al., 1997;Gottfert et al., 2001). TTSS mutants of Rhizobium are altered in host range and nodulation efficiency (Bellato et al., 1997; Viprey et al., 1998). In symbiotic Rhizobium, as in plant pathogens, the TTSS secretes factors that modulate host metabolism and enhance, or limit, interactions between bacteria and host cells.
PGPR strains of P. fluorescens and P. putida also modulate host metabolism, resulting in increased plant growth and disease resistance. Several mechanisms for plant growth promotion have been proposed, including production of phytohormones and antifungal compounds (Cook et al., 1995; Weller, 1988; Lugtenberg and Dekkers, 1999). However, PGPR, like pathogenic bacteria, are ultimately dependent on their ability to colonize plant surfaces and tissues, particularly roots. Although colonization is important for the systemic effects of PGPR on plant hosts, it is not known whether colonization and plant growth promotion by PGPR involves cell contact-dependent interactions with plant cells.
Recently, we have developed in vivo expression technology (IVET) strategies to study the mechanistic basis of rhizosphere colonization by a PGPR strain of P. fluorescens (isolate SBW25) (Rainey, 1999; Rainey and Preston 2000). Using these strategies we have isolated an extensive range of genes that show elevated levels of expression in the rhizosphere. A surprising discovery was a gene showing 51% identity to hrcC from P. syringae, which encodes the putative pore-forming outer membrane component of the type III secretion pathway. Subsequent analysis of the IVET-captured gene fusion revealed that the organization and sequence was similar to a region of the P. syringae hrp cluster extending from hrcJ through to hrcC (Rainey, 1999; this study). Here, we report further sequencing and characterization of the P. fluorescens SBW25 TTSS and provide proof of functionality. We also present evidence that TTSSs are widely distributed among plant-associated fluorescent Pseudomonas.
Isolation and sequence analysis of the P. fluorescens SBW25 type III (Rsp) gene cluster
Two cosmid clones, pGMP1 and pGMP2, were isolated from a P. fluorescens SBW25 library using primers designed from the previously isolated hrcC-like IVET fusion. Subsequent mapping and sequencing of these cosmids delimited a 20-kb cluster of TTSS-related genes, as shown in Fig. 1.
A total of 22 predicted open reading frames (ORFs) were identified and provisionally assigned the gene names rsp and rsc (for rhizosphere-expressed secretion gene). ORFs corresponding to widely conserved type III secretion genes were assigned a rsc gene designation and final letter matching the final letter of the corresponding conserved gene from Yersinia spp. according to convention (Bogdanove et al., 1996a). ORFs that displayed a significant degree of similarity only to P. syringae and Erwinia spp. hrp genes in terms of sequence and organization were designated rsp, and assigned the final letter of the corresponding hrp gene. The rsp designation was applied to ORFs that resemble ‘group I’ type III secretion genes of plant-associated bacteria and does not signify pathogenicity-related function. With the exception of rspA, rspF and rspO, all ORFs displayed a significant degree of similarity (as detected by blast searches and confirmed through randomized GAP analysis) to homologues from P. syringae and Erwinia amylovora. ORFs encoding peptides with similar features to hrpA, hrpF and hrpO from P. syringae are located between rspR–rspB, rspE–rspG and rspP–ORF2 respectively. The size, similarity to P. syringae and predicted function of the putative proteins encoded by the rsp cluster are listed in Table 1. The transcriptional organization of the rsp cluster, as predicted by sequence analysis and homology with P. syringae, is shown in Fig. 1A and E).
Table 1. Comparison of rsp and hrp cluster proteins in P. fluorescens SBW25 and P. s. pv. syringae 61.
. Gene designations for type III secretion genes in P. fluorescens SBW25 (bold, rsc/p) and P. s. pv. syringae 61 (hrc/p). Only the third and fourth letters of the gene name are given.
. Length of predicted peptides (a.a., amino acids). Products of alternate start sites are shown in parentheses. Predicted peptides with start codons other than AUG are annotated in superscript as V (GUG) and L (UUG). The hrcN-like P. fluorescens sequence overlaps the C-terminal end of the truncated hrpJ-like ORF2 and has no start codon.
. Percentage similarity and identity were calculated for full-length peptides using GAP (GCG).
d. Predicted function and properties of each P. fluorescens protein (cytoplasmic, inner membrane associated (IM); outer membrane associated (OM); or secreted) was independently obtained using blast (NCBI), GCG, Pfam (Sanger) and PSORT (Nakai and Kanehisa, 1991), and conformed to predictions and observations made for P. syringae ORFs.
. Partial sequence for a potential rscV gene was obtained as described in the results.
. The P. fluorescens SBW25 rsp cluster contains a truncated homologue of hrpJ, next to a short stretch of DNA with homology to hrcN. The region of homology between the translated P. fluorescens DNA and the corresponding P. syringae ORFs is listed in parentheses.
P. s. s. 61/Pss, P. syringae pv. syringae 61; P. f. SBW25, P. fluorescens SBW25; Pst, P. syringae pv. tomato; Psg, P. syringae pv. glycinea; Ea, E. amylovora.
σ54 -dependent, ATP-dependent
Transcriptional activator; cytoplasmic
Structural component of pilus; secreted
Accessory protein; secreted
IM/OM, lipoprotein (FliF)
133 (175Psg 193Ea)
Accessory protein; cytoplasmic
Accessory protein; OM
Negative regulator of hrp/rsp expression; cytoplasmic
Immediately adjacent to rspL is a large ORF encoding a protein that resembles the type III secreted proteins AvrE and DspE (DspA) from P. syringae, E. amylovora and Erwinia herbicola pv. gypsophilae (Lorang and Keen, 1995; Gaudriault et al., 1997; Bogdanove et al., 1998a; Mor et al., 2001) respectively. Single-pass sequencing indicates that this protein exhibits approximately 27% identity/35% similarity to AvrE. We propose that this ORF should be called ropE (rhizosphere-expressed outer protein), in line with convention, subject to further characterization and confirmation of TTSS-dependent secretion. The AvrE-like ORF is preceded by a small ORF (ORF1) that exhibits no significant homology to known proteins (Fig. 1E). ORFs flanking the cluster (copR–ggtB and gtp–ufaA1) are similar to proteins unlikely to be linked to type III secretion, and those on the right-hand side exhibit synteny with a 10.5-kb region of the Pseudomonas aeruginosa PAO1 genome (5231000–5241500) (Fig. 1A).
Given that horizontal gene transfer may play a role in the evolution of TTSS gene clusters, we examined the P. fluorescens SBW25 rsp cluster and flanking sequences (from ggtB to ctc, Fig. 1A) for the presence of insertion elements, inverted and direct repeats, tRNA genes and G + C content. Neither insertion sequences, nor duplicated regions, were detected, however, a scan for tRNA sequences using tRNAScan-SE (Lowe and Eddy, 1997) revealed a potential methionine-tRNAMet (Cove score, 90.17) between rspL and ropE. In addition, analysis of the G + C content of the cluster showed that the average G + C content of the eight genes spanning rspR to rspG is 52%. This is significantly less (P < 0.001) than both the (average) 61% G + C content of the 12 genes spanning rscC to rspL and the (average) 59.9% G + C content of P. fluorescens SBW25 (Spiers et al., 2001; see also http://www.plants.ox.ac.uk/sbw25/).
The P. fluorescens SBW25 rsp cluster is similar to the P. syringae hrp cluster, but has several unique features
The arrangement and orientation of the ORFs in P. fluorescens SBW25 bears strong similarity to P. syringae(Table 1, Fig. 1E), but there are four principal differences:
(i) The P. fluorescens SBW25 cluster lacks a significant portion of the hrpV operon (hrpJ, hrcV, hrpQ). Analysis of this region reveals a single ORF in P. fluorescens SBW25 that encodes a protein with homology to the N-terminal domain of HrpJ, which overlaps a short (160 bp) DNA sequence with homology to DNA encoding the C-terminal domain of HrcN. As the HrpJ-like ORF is significantly truncated, but may be transcribed, we refer to this protein as ORF2.
(ii) The P. fluorescens SBW25 cluster lacks a homologue of the hrpZ gene, which resides between hrpA and hrpB in the hrpJ operon of P. syringae (Preston et al., 1995).
(iii) The P. fluorescens SBW25 cluster harbours only a single response regulator homologue, rspR, whereas P. syringae has two (hrpR and hrpS) that are 64% identical to each other (Grimm et al., 1995; Deng et al., 1998).
(iv) In P. fluorescens SBW25 ropE lies adjacent to rspL, whereas in P. syringae avrE is located on the other side of the cluster, upstream of hrpRS (Alfano et al., 2000).
E. amylovora also carries a single homologue of hrpR/hrpS and lacks hrpZ, however, the sequence similarities and organization of the P. fluorescens SBW25 cluster are significantly closer to P. syringae than to E. amylovora.
The presence of truncated homologues of hrpJ and hrcN, and the absence of a homologue of hrcV, led us to consider the possibility that pGMP2 may have suffered a deletion or rearrangement during cloning. Primers were designed to amplify this region from genomic DNA. Polymerase chain reaction (PCR) products of the expected size based on pGMP2 sequence were obtained and sequence analysis revealed no evidence of a rearrangement within this region. The restriction map of this region was checked by DNA hybridization of P. fluorescens genomic DNA with probes generated from the ORF2–rspO region and hrcVPs. ORF2–rspO probes hybridized to 1.1 and 1.5 kb PstI fragments as predicted by DNA sequencing (Fig. 1D), while hrcVPssand hrcVPst probes hybridized weakly to a single 1.8 kb PstI fragment. Finally, conserved sequences in hrcVPsand hrcVEa were used to design primers to amplify hrcV homologous DNA from the P. fluorescens SBW25 genome. A 450-bp DNA fragment was obtained that had 56% identity to hrcVPss (a.a. 145–270) (Table 1). No product was obtained when pGMP1 or pGMP2 were used as template.
Taxonomic analyses and 16 s rDNA sequencing confirm that P. fluorescens SBW25 is a member of the P. fluorescens/putida group
Given the similarity of the P. fluorescens rsp cluster to the P. syringae hrp cluster, we validated the taxonomic identity of P. fluorescens SBW25 using routine phenotypic tests and sequenced the 16S rDNA gene. Positive test results for oxidase, arginine dihydrolase and gelatinase, combined with an inability to grow at 37°C, confirmed the identity of P. fluorescens SBW25 as a ‘saprophytic’Pseudomonas isolate belonging to the P. fluorescens/P. putida group (Stanier et al., 1966). In addition, a phylogenetic analysis based on the variable region of the 16S rDNA gene indicated that P. fluorescens SBW25 is more closely related to other P. putida and P. fluorescens isolates than it is to P. syringae(Fig. 2).
Rsp genes are positively regulated by RspL
In P. syringae, the hrp cluster is comprised of four operons regulated by the hrp-specific sigma factor, HrpL, and each HrpL-regulated operon is preceded by a ‘hrp-box’ motif (GGAACC−16 bp – CCAC) (Xiao and Hutcheson, 1994). ‘hrp-boxes’ are located approximately 50 bp upstream of each of the putative rspU, rspC and rspJ operons and ropE(Fig. 1E). There is no hrp box upstream of rspP as proposed in P. syringae, and no terminator between ORF2 and rspP, which suggests that ORF2, rspO and the rspU operon form a single operon in P. fluorescens SBW25, as has been proposed for E. amylovora (Bogdanove et al., 1996b). The hrp box preceding ropE is located within the coding sequence for ORF1, which suggests that ropE is transcribed as a single gene.
To test whether rsp genes were regulated by the HrpL homologue RspL we cloned rspL into the broad host range plasmid pML122, downstream of the constitutive nptII promoter (Xiao and Hutcheson, 1994) and introduced the resulting plasmid, pML122–RspL, into P. fluorescens SBW25 rscC–lacZY (Rainey, 1999). P. fluorescens SBW25 (pML122–RspL) colonies had a small, flat, wrinkled morphology and the cells clumped in broth culture. P. fluorescens SBW25 rscC–lacZY (pML122–RspL) showed a 20-fold increase in β-galactosidase activity relative to P. fluorescens SBW25 rscC–lacZY.
P. fluorescens SBW25 does not elicit a HR when infiltrated into plant leaves
P. fluorescens SBW25 did not elicit a HR or any visible symptom of pathogenic activity when infiltrated into a range of plant leaves at inoculum levels up to 109 colony-forming units (cfu) ml−1, as shown in Fig. 3. Plants tested included tobacco, tomato, sugar beet and Arabidopsis thaliana. Avirulent strains of P. syringae elicited a visible HR in all these plants when infiltrated into leaves at 107 cfu ml−1. No significant increase in bacterial population levels was observed in the first 24–48 h after infiltration of P. fluorescens SBW25 into A. thaliana or tobacco at 105 or 106 cfu ml−1, and less than 15% of the initial inoculum level could be recovered from infiltrated leaves after 7 ds.
P. fluorescens SBW25 strains carrying the hrp cluster of P. syringae pv. syringae 61 on the cosmid pHIR11 elicited an HR in tobacco at 107−8 cfu ml−1, as previously observed for P. fluorescens 55 (pHIR11) (Huang et al., 1988). However, strains of P. fluorescens SBW25 carrying pCPP2078 or pCPP2089, derivatives of pHIR11 with insertion mutations in the hrcU and hrcC genes, respectively (Huang et al., 1993), were unable to elicit a HR. Conversely, pGMP1 and pGMP2 did not complement the HR– phenotype of a polar hrcC mutant of P. syringae pv. tomato DC3000.
P. fluorescens SBW25 rsp genes are not induced in the leaf apoplast
The above results suggested that P. fluorescens SBW25 either does not produce an elicitor of the HR that is recognized by the plants tested, or that the TTSS and secreted proteins are not expressed when bacteria are infiltrated into leaves. Rainey (1999) observed that the rscC–lacZY fusion was expressed at a basal level in Luria–Bertani (LB) broth and induced in the sugar beet rhizosphere. We used the rhizosphere-expressed rscC–lacZY (hrcC–lacZY) fusion to examine rscC expression in P. fluorescens SBW25 in rich medium (LB), minimal medium (M9), P. syringae hrp-inducing minimal medium, which simulates the environmental conditions found in the leaf apoplast (Huynh et al., 1989), and in bacteria infiltrated into plant leaves. rscC was expressed at a similar basal level in all three media and was not induced when bacteria were infiltrated into plant leaves.
In order to test whether the inability of P. fluorescens SBW25 to elicit the HR and complement hrp mutations in pHIR11 was as a result of the low level of rsp gene expression in plant leaves, we inoculated plants with bacteria carrying the rsp-inducing plasmid pML122–RspL. P. fluorescens SBW25 (pML122–RspL), P. fluorescens (pCPP2078, pML122–RspL) and P. fluorescens (pCPP2089, pML122–RspL) failed to elicit a HR or cause disease symptoms when infiltrated into tobacco (Nicotiana tabacum), tomato, sugar beet and A. thaliana(Fig. 3).
P. fluorescens SBW25 (pAvrB, pML122–RspL) elicits an AvrB-dependent HR in A. thaliana ecotype Col-0
The TTSSs of P. syringae, Erwinia spp. and Xanthomonas campestris are able to secrete heterologous proteins (Bogdanove et al., 1998a; Ham et al., 1998; Rossier et al., 1999). In the absence of characterized TTSS secreted proteins from P. fluorescens, we tested whether P. fluorescens SBW25 could secrete the AvrB protein from P. syringae pv. glycinea and elicit an AvrB-dependent HR (Tamaki et al., 1988; Gopalan et al., 1996). Plasmids expressing avrB (pAvrB and pAvrB1) were transformed into P. fluorescens SBW25 and P. syringae pv. tomato DC3000 (a pathogen of tomato and A. thaliana). P. syringae pv. tomato DC3000 (pAvrB) elicited a HR in A. thaliana ecotype Col-0 at 107 cfu ml−1, but P. fluorescens SBW25 (pAvrB) was unable to elicit a HR, even at inoculum levels of 109 cfu ml−1. However, strains of P. fluorescens SBW25 carrying both pAvrB and pML122–RspL elicited the HR in A. thaliana when inoculated at 108 cfu ml−1(Fig. 3A). HR elicitation by P. fluorescens SBW25 (pML122–RspL, pAvrB) was not as consistently observed as the HR elicited by P. syringae, even at high inoculum levels, so we scored the responses of 10–20 individual leaves to confirm the HR for each strain tested. P. fluorescens SBW25 rspD::Tn(14), P. fluorescens SBW25 rscR::Tn(19) and P. fluorescens SBW25 rscT::Tn(28) mutants carrying both pAvrB and pML122–RspL did not elicit the HR (Fig. 3A). However, P. fluorescens SBW25 ropE::Tn(2) (pAvrB, pML122–RspL) was able to elicit the HR (Fig. 3A). P. syringae pv. tomato (pAvrB) and P. fluorescens SBW25 (pAvrB, pML122–RspL) consistently failed to elicit the HR when inoculated into the RPM1 mutant A. thaliana FN35B (rps3–1) which lacks the cognate R gene, RPM1, required for recognition of AvrB (Bisgrove et al., 1994; Grant et al., 1995), confirming that the HR elicited by P. fluorescens SBW25 (pAvrB, pML122–RspL) was AvrB-dependent and RopE-independent (Fig. 3A). rsp and rsc mutants were complemented and the HR phenotype restored by introducing pGMP1 or pGMP2 in trans.
Co-infiltration of inhibitors of plant metabolism such as 50 µm of sodium vanadate (ATPase) and 1.0 mM lanthanum chloride (Ca2+ channel blocker) inhibited the necrosis elicited by P. fluorescens SBW25 (pAvrB, pML122-RspL) in A. thaliana ecotype Col-0. In conjunction with AvrB and rsp-dependence and the 12–24 h lag before the visible manifestation of cell death, this strongly indicates that the necrosis observed resulted from induction of active plant metabolism and programmed cell death rather than non-specific phytotoxic activities or pectic enzyme production.
P. fluorescens SBW25 does not secrete AvrB to the extracellular milieu
In order to determine whether P. fluorescens SBW25 secreted AvrB to the extracellular milieu, we examined secretion of epitope-tagged AvrB-HA to the supernatant from P. fluorescens containing the plasmids pDSK220 and pME6010–AvrB-HA, using Coomassie-stained gels in conjunction with immunoblotting (Ham et al., 1998; Nimchuk et al., 2000). No evidence of AvrB secretion was obtained irrespective of whether rsp genes were induced using pML122–RspL. We also examined Coomassie-stained gels of supernatant preparations for evidence of secreted proteins that were uniquely associated with wild-type bacteria or bacteria carrying pML122–RspL, but absent in rsp/rsc mutants, however, none were detected. These results are in accordance with similar work on the TTSS in P. syringae which, unlike the TTSS of E. chrysanthemi, does not secret AvrB to the extracellular milieu (Ham et al., 1998; van Dijk et al., 1999).
P. fluorescens SBW25 elicits a host-specific HR in Nicotiana clevelandii when RspL is constitutively expressed
In order to confirm the HR phenotype observed in A. thaliana, we inoculated P. fluorescens SBW25 and strains carrying pML122–RspL, pAvrB or both into the AvrB-resistant plant Nicotiana clevelandii. Surprisingly, both P. fluorescens SBW25 (pML122–RspL) and P. fluorescens SBW25 (pAvrB, pML122–RspL) elicited a HR-like necrosis at 108 cfu ml−1, while P. fluorescens SBW25 and P. fluorescens SBW25 (pAvrB) gave no reaction (Fig. 3C and D). This result challenged our earlier conclusion regarding the absence of endogenous TTSS-dependent HR elicitors in P. fluorescens SBW25. The RspL-dependent necrosis induced by P. fluorescens SBW25 in N. clevelandii was blocked by inhibitors of plant metabolism (Fig. 3D) and was abolished in rsc/rsp mutants, indicating that the necrosis observed represented a rsp-dependent HR, rather than sensitivity to a toxic rsp-dependent factor. P. fluorescens SBW25 ropE::Tn(2) (pAvrB, pML122–RspL) also induced necrosis in N. clevelandii. In both A. thaliana and N. clevelandii, the P. fluorescens HR manifested at a slower rate than the P. syringae HR, with an extended watersoaking phase before leaf panels fully collapsed and dried out. This delay may be a consequence of the relative inefficiency of the TTSS, reduced competence of bacteria over-expressing RspL or the high levels of inoculum needed to obtain a response.
Rsp mutants of P. fluorescens SBW25 are not impaired in their ability to colonize the rhizosphere of sugar beet seedlings
Three independent rsc/rsp mutants and a ropE mutant were tested for their ability to colonize the rhizosphere of sugar beet seedlings using the model rhizosphere system and rhizosphere colonization assay described by Rainey (1999). All the mutants tested were able to reach high population levels (> 5 × 107 cfu) in the rhizosphere within 7 days after being inoculated onto seeds, and population levels observed at 7 days after inoculation were similar to those attained by wild-type P. fluorescens SBW25.
The rsc/rsp mutants were co-inoculated with wild-type P. fluorescens SBW25 at a 1:1 ratio in order to test their relative fitness in the rhizosphere. Population levels were measured at 0, 2 and 7 days after inoculation. No significant differences were observed (Fig. 4). The ability of the rsc/rsp mutants to colonize the phyllosphere of sugar beet seedlings over the same time period was also not significantly impaired (data not shown).
Type III secretion genes are widespread among plant-colonizing bacteria belonging to the P. fluorescens/putida group
In order to investigate whether TTSS genes are widely distributed among plant-colonizing bacteria belonging to the P. fluorescens/putida group, genomic DNA from a range of plant-associated Pseudomonas was hybridized with probes derived from various P. fluorescens SBW25 rsc genes, including rscU, rscJ and rscQ. A typical hybridization result from a rscU-probed Southern blot is shown in Fig. 5. From this selection of 12 previously studied strains (Haubold and Rainey, 1996), nine hybridized to the P. fluorescens-derived probes (100, 103, 119, 122, 130, 132, 151, 179, 185) (Fig. 5). In addition, probes were hybridized to DNA from a range of commonly used Pseudomonas strains. rsc probes hybridized weakly to DNA from P. syringae pv. syringae 61 and P. syringae pv. tomato DC3000. However, rsc probes hybridized strongly to DNA from the well-studied P. fluorescens and P. putida strains 55, WCS358r, WCS374r, WCS417r, WCS365, F113, AC10R and 277, and DNA from P. cichorii strains NCPPB3283, NCPPB3109, NCPPB907 and NCPPB943. Interestingly, rsc probes failed to hybridize to DNA from strain OE28.3. 16S rDNA sequences were obtained for all strains tested and the presence and strength of rsc hybridization signals mirrored their phylogenetic relationship. Strongest hybridization was observed for P. fluorescens 55, P. putida WCS417r and P. fluorescens/putida strains from cluster B3 (see Haubold and Rainey, 1996; strains 179, 132, 122, 103, 185, 151 in Fig. 5). 16S rDNA sequence data indicates that P. fluorescens SBW25, P. fluorescens 55 and P. putida WCS417r are closely related to strains within the B3 cluster and that P. fluorescens SBW25 has an identical 16S rDNA sequence to strain 132 (Haubold, 1997). ‘P. cichorii’ 119 also exhibited relatively strong hybridization to rsc probes, although 16S rDNA sequence indicates that ‘P. cichorii’ 119 is not closely related to the B3 strains or the four P. cichorii strains obtained from the NCPPB culture collection.
P. fluorescens has long been recognized as a widespread and successful colonizer of the plant rhizosphere. In recent years it has received increasing attention as a potential biocontrol agent and PGPR because of its ability to produce antibiotics and enter into beneficial interactions with host plants (Weller, 1988; Cook et al., 1995; Lugtenberg and Dekkers, 1999). However, while many different traits may contribute to plant growth-promoting activity, success as a PGPR is essentially dependent on an ability to competitively colonize plant roots. As part of our ongoing research into the genetic basis of rhizosphere competence in P. fluorescens SBW25, we have identified a gene cluster encoding a rhizosphere-expressed TTSS. This TTSS closely resembles TTSSs that promote plant parasitism and disease in plant pathogens such as P. syringae and E. amylovora. We have shown that the P. fluorescens SBW25 TTSS is able to deliver Avr proteins to plant cells and have provided evidence for the presence of TTSS secreted proteins in P. fluorescens SBW25. In addition, we have used probes derived from the P. fluorescens SBW25 rsc genes to obtain evidence that similar pathways are widespread in ‘saprophytic’Pseudomonas.
The P. fluorescens SBW25 rsp cluster bears greatest similarity to the hrp‘pathogenicity island’ of P. syringae, both at the level of amino acid sequence and in terms of genomic organization (Alfano et al., 2000). The cluster contains 8 of the 10 conserved hrc genes found in P. syringae (plus the truncated hrcN gene), and 10 of the 13 hrp genes. The cluster contains the conserved components of the group I hrp regulatory cascade (hrpR and hrpL) and a homologue of the secreted protein avrE. In contrast with other group I TTSS gene clusters, we predict that the non-regulatory rsp and rsc genes in the P. fluorescens SBW25 rsp cluster correspond to conserved components of the secretion machinery. The fact that all rsp and rsc mutants abolished HR elicitation lends support to this proposition.
In P. syringae, the hrp gene cluster is present as a single functional unit that can be cloned and expressed in E. coli and P. fluorescens and is sufficient to deliver proteins into plant cells (Alfano and Collmer, 1997). The core group of hrp genes is flanked by secreted proteins and it seems probable that recombination events have played a role in evolution of the locus (Alfano et al. 2000). With the exception of a tRNAMet located between rspL and ropE, the P. fluorescens SBW25 rsp cluster contains no sequences indicative of recent horizontal gene transfer. However, the atypical G + C content for the eight genes spanning rspR to rspG and absence of the central portion of the ‘hrpJ–hrcN′ operon, together with the tRNAMet, suggest an unusual evolutionary history.
The search for TTSS genes in other Pseudomonas isolates by Southern hybridization confirmed that P. fluorescens SBW25 rsp genes are distinct from genes typically found in P. syringae. Hybridization data indicates that TTSS genes are present in the genomes of all isolates examined from the B3 cluster described by Haubold and Rainey (1996), but hybridization was not observed for all P. fluorescens/putida strains. In addition, interrogation of uncompleted genome sequences at TIGR indicated its absence from P. putida strains KT2440 and PRS1. A recent analysis of hrp gene phylogeny in P. syringae suggests that the hrp genes in P. syringae pathovars are evolutionarily stable (Sawada et al., 1999). Our results indicate that the P. fluorescens rsp cluster is also evolutionarily stable, but that it may be confined to strains occupying a similar ecological niche (Preston et al., 1998).
Proteins secreted by TTSSs play a critical role in microbe–host interactions. Four characterized classes are known from P. syringae: Avr proteins, harpins, HrpA and AvrE. The DNA sequence of the P. fluorescens SBW25 rsp cluster revealed the presence of two putative secreted proteins, RspA and RopE, while a third unidentified Rsp-dependent elicitor protein was indicated by the HR elicited by both wild-type and ropE mutants of P. fluorescens SBW25 on N. clevelandii. That no candidate for this unidentified Avr is present in the P. fluorescens SBW25 rsp cluster (or located close by) is consistent with the genomic distribution of Avr proteins in P. syringae, in which elicitors are found both associated with the cluster and in other genomic locations (Kim et al., 1998). Studies of protein secretion and HR elicitation in P. syringae and E. amylovora have shown that neither AvrE nor DspE are required (Lorang and Keen, 1995; Bogdanove et al., 1998a; b) – a finding consistent with our observation that an AvrB-dependent HR is observed in a ropE mutant strain. However, the presence of homologues of this protein in several plant-associated bacteria with different life histories and different hosts suggests a broadly conserved role for this protein in plant–bacteria interactions.
To test the functionality of the P. fluorescens SBW25 TTSS, we exploited the known ability of TTSSs to export certain heterologous Avr proteins (Rosqvist et al., 1995; Bogdanove et al., 1998a; Ham et al., 1998; Anderson et al., 1999; Rossier et al., 1999). We introduced AvrB from P. syringae pv. glycinea into P. fluorescens SBW25 and observed a rsp/AvrB-dependent HR in leaves of AvrB-responsive A. thaliana ecotype Col-0. Significantly, HR elicitation by P. fluorescens SBW25 was dependent on induction of rsp gene expression, and a macroscopic HR was only observed with strains that over-expressed the rsp-specific ECF sigma factor RspL. Our use of this heterologous expression strategy in conjunction with the HR phenotype to demonstrate functionality of the TTSS may prove useful for the discovery of these clusters in other bacteria.
The lack of induction of the P. fluorescens SBW25 rscC gene in the leaf environment is a notable difference between P. fluorescens SBW25 and P. syringae. Our previous work showed that rscC is expressed in a model sugar beet rhizosphere (Rainey, 1999). It is therefore possible that the pathway delivers bacterial proteins into root cells with greater efficiency than into leaf cells. rsp expression may be induced in response to specific root or rhizosphere signals, as observed for the TTSS genes of Rhizobium and Ralstonia solanacearum (Kovacs et al., 1995; Viprey et al., 1998; Aldon et al., 2000), or expressed at a consistently low level that is sufficient for biological function, as proposed for the hrp genes of E. chrysanthemi (Ham et al., 1998). Secretion of hrp-dependent effector proteins by P. syringae is also controlled at a post-transcriptional level such that AvrB is contained within host cells and only delivered to host cells after cell contact. As we were unable to observe secretion of AvrB to the extracellular milieu, even when RspL was overexpressed, it seems probable that a similar regulatory mechanism is operating in P. fluorescens.
P. fluorescens and P. putida have been used as genetic backgrounds in which to study the TTSS of P. syringae cloned on pHIR11 (Huang et al., 1988; Alfano and Collmer, 1997). When expressed in P. fluorescens 55, pHIR11 behaves as if it was the only TTSS gene cluster in the bacterium; pHIR11 hrp mutants fail to elicit the HR, and there is no evidence for translocation of heterologous elicitor proteins by an endogenous TTSS. We obtained similar results when pHIR11 was expressed in wild-type P. fluorescens SBW25. Two main hypotheses can be proposed to explain this result. Firstly, as the cis-acting sequences that direct hrp and rsp expression are not fully characterized, it is possible that they have diverged to the extent that they are poorly recognized by a heterologous sigma factor. Secondly, it is possible that there is a certain level of expression of the rsp cluster when pHIR11 is present, but that rsp genes cannot complement mutations in pHIR11 or do not enable the secretion of heterologous proteins at sufficient levels to be detected in standard assays. This is consistent with our observation that at least 10-fold more bacteria were needed to elicit a RspL-dependent HR than were needed for elicitation of a pHIR11-dependent HR by P. fluorescens SBW25. It is also possible that the P. fluorescens SBW25 TTSS is unable to translocate the elicitor protein encoded by pHIR11, HopPsyA (HrmA). There is some speculation that HopPsyA, unlike AvrB, requires a chaperone to be secreted (Alfano et al., 1997; van Dijk et al., 1999) and it is conceivable that specificity for the P. syringae TTSS is encoded by this chaperone.
The ecological significance of type III secretion in non-pathogenic Pseudomonas is unknown. One possibility, based on knowledge of the role of type III secretion in P. syringae, rhizosphere expression of the P. fluorescens SBW25 cluster and presence of putative TTSS-secreted elicitors, is that type III secretion in P. fluorescens SBW25 facilitates parasitism of plant root tissues. If correct, then bacteria such as P. fluorescens SBW25 probably have a selective advantage in the rhizosphere because of access to a protected nutrient pool and/or ecological niche. To test this hypothesis, mutants were generated in both the rsp cluster and in a putative secreted protein (RopE) and the competitive ability of the mutants was determined in the rhizosphere of sugar beet seedlings. No significant effects on colonization were observed. However, the colonization assays were conducted on 1–2 week-old seedlings that had been maintained under ideal conditions of nutrients, moisture and light. Such conditions may not provide the selective regime necessary to reveal loss of fitness in rsp mutants. Moreover, we looked for a phenotype at the population level, whereas differences may be more evident at the level of the interaction between individual bacteria with plant root cells or as plants mature. In a recent study, Hirano et al. (1999) observed that hrcC and hrpJ mutants of P. s. pv. syringae B728a were not impaired in their ability to colonize seedlings immediately following germination, but were impaired in their ability to subsequently colonize and cause disease on leaves. The presence of TTSSs in PGPR also raises the issue of whether TTSS-mediated host-specificity affects the interactions of PGPR with host plants. Such specificity would make the selection of host plant species and genotype a critical factor in establishing the function of type III secretion in PGPR. We are currently developing more sensitive assays to rigorously address this issue for root-colonizing Pseudomonas.
Advances in the understanding of pathogenesis and symbiosis continue to erode the view that these are clearly defined and mutually exclusive lifestyles. The presence of a TTSS was, until recently, the hallmark of many pathogenic bacteria, but this view was challenged upon the discovery of TTSS genes in strains of Rhizobium (Meinhardt et al., 1993; Freiberg et al., 1997; Gottfert et al., 2001) and, more recently, of hrp-like genes in PGPR Pseudomonas (Karden et al., 1996; Rainey, 1999). The presence of a functional TTSS in P. fluorescens SBW25 and evidence of widespread distribution in related bacteria challenges the definition of TTSS function still further. The single attribute unifying bacteria that possess TTSSs is the capacity to live in intimate association with eukaryote hosts. P. fluorescens SBW25 colonizes root surfaces and intercellular spaces and has systemic effects on plant growth and disease resistance. The presence of a hrp-like TTSS in P. fluorescens suggests that this bacterium enters into a more intimate association with plant cells than previously thought.
Bacterial strains, plasmids and media
E. coli strains were grown in Luria–Bertani (LB) broth at 37°C. P. syringae and P. fluorescens were grown in LB or King’s medium B (KB) broth (King et al., 1954) at 28°C. For gene induction studies, bacteria were cultured in LB, M9 and hrp-inducing minimal medium (Huynh et al., 1989; Sambrook et al., 1989). E. coli DH5α (Clontech) was used for plasmid construction and maintenance. Construction of the cosmid library in E. coli XL1 MR from which pGMP1 and pGMP2 were obtained is described in Rainey (1999). For standard DNA manipulations pBSSK(–) from Stratagene was used. A summary of bacterial strains and plasmids is provided in Table 2. Antibiotics were used at the following concentrations unless otherwise stated (µg ml−1): ampicillin, 100; gentamicin, 10; kanamycin, 50; rifampicin, 25, spectinomycin, 50; and tetracycline, 12.5. CFC supplement (Difco) was used at half strength in LB agar to select for P. fluorescens recovered from the rhizosphere.
Cosmid pIJ3200 containing P. f. 25 rsp genes (rscN-rspR) and flanking DNA (GTP-bindingprotein-ufaA1), Tcr
Cosmid pIJ3200 containing P. f. 25 rsp locus (ropE-rspR) and flanking DNA (copR-ggtB (left), GTP-binding protein-ctc (right)), Tcr
Coding region of rspL amplified by PCR, with the introduction of an artificial ribosome binding site.Cloned in expressed orientation relative to nptII promoter in pML122 using XhoI and BamHI, Gmr
pGMP2 with transposon insertion rscR::Tn(19), Tcr, Kanr
pGMP2 with transposon insertion rscT::Tn(28), Tcr, Kanr
pGMP2 with transposon insertion rspD::Tn(14), Tcr, Kanr
pGMP2 with transposon insertion ropE::Tn(2), Tcr, Kanr
P. f. 25 rscR::Tn(19)
Transposon insertion 19 marker-exchanged into P. f. 25, Kanr
P. f. 25 rscT::Tn(28)
Transposon insertion 28 marker-exchanged into P. f. 25, Kanr
P. f. 25 rspD::Tn(14)
Transposon insertion 14 marker-exchanged into P. f. 25, Kanr
P. f. 25 ropE::Tn(2)
Transposon insertion 2 marker-exchanged into P. f. 25, Kanr
avrB-HA cloned into broad host range vector pME6010 as XhoI/EcoRI insert, Tcr
Recombinant DNA techniques were performed according to standard protocols (Sambrook et al., 1989). Restriction and modifying enzymes were obtained from BRL and New England Biolabs (NEB). Electroporation was performed using a Bio-Rad GenePulser according to the manufacturer's protocol (Bio-Rad). Triparental mating was carried out using helper plasmid E. coli HB101 (pRK2013) according to standard protocol. PCR reactions using Taq polymerase (Qiagen) were performed according to the manufacturer's protocol using purified plasmid or genomic DNA as a template. Oligonucleotide primers were obtained from Genosys and from MWG Biotech. Details of oligonucleotide sequences used for DNA amplification are available on request.
Southern hybridization analysis was performed using DNA probes derived from rspL–rspJ’, ‘rscN–rsp0, rscU, rscC, rscJ and rscQ by polymerase chain reaction (PCR). Probes were labelled using the ECL random prime labelling system (Amersham). Genomic DNA was transferred to Hybond-N membrane using alkali solution. Blots were hybridized for 16 h at 60°C, followed by washes in 1× SSC, 0.1% SDS and 0.5× SSC, 0.1% SDS at 60°C. Fluorescein-labelled probe was detected using the Gene Images CDP-Star detection module (Amersham).
DNA sequencing pGMP1 and pGMP2 were subcloned into pBSSK(–) using EcoRI, PstI, HindIII and SalI. Subclones were mapped, and selected subclones were sequenced using T3 and T7 primers. Additional sequence was obtained from pGMP1, pGMP2 and subclones using specific oligonucleotide primers, and by sequencing out from transposon insertions into pGMP2 as described below. All DNA sequencing was done using Big-Dye Terminators (Applied Biosystems) on an Automated DNA Sequencer, model 310 (Perkin Elmer). Details of oligonucleotides used in DNA sequencing and analysis are available on request. DNA sequence was assembled and analysed with the Genetics Computer Group sequence analysis software package. DNA sequence has been deposited in GenBank under the accession number AF292566. Four hundred and eighty-one nucleotides were sequenced from the variable region of the 16S rDNA gene from a range of Pseudomonas strains (Table 2) using conserved primers (Haubold, 1997). DNA sequence from P. fluorescensSBW25was deposited in EMBL under the accession number AJ310393.
Transposon mutagenesis of pGMP2 and P. fluorescens SBW25
Transposon mutagenesis of pGMP2 was carried out using the pGPS-1 plasmid and the GPS-1 genome priming system (NEB) according to the manufacturer’s protocol. Mutagenized cosmid DNA was transformed into ultracompetent E. coli DH5α (Clontech) and transformants plated onto selective media containing 25 µg ml−1 kanamycin. More than 250 independent cosmid mutants were streaked and stored for further characterization. Transposon insertions were mapped using restriction enzymes, and selected mutants were further characterized by sequencing the DNA flanking the insertion using N and S primers supplied by the manufacturer (Fig. 1C). Transposon mutations in pGMP2 were introduced into the genome of P. fluorescens SBW25 by marker-exchange mutagenesis. Cosmid DNA was electroporated into P. fluorescens SBW25 and kanamycin-resistant transformants were screened for tetracycline sensitivity. The site of each transposon insertion in pGMP2 and genomic DNA was confirmed by DNA amplification from purified template DNA, using specific primers to sequences flanking the transposon.
Tobacco (Nicotiana tabacum L ‘Samsun’), tomato (Lycopersicon esculentum Mill. ‘Moneymaker’), sugar beet (Beta vulgaris var. ‘Amethyst’) and Nicotiana clevelandii were cultivated under greenhouse conditions prior to HR assays. Arabidopsis thaliana ecotype Columbia (Col-0) and A. thaliana FN35B (rps3–1) plants and sugar beet seedlings were grown and maintained in a growth chamber at constant temperature (21°C) for HR and rhizosphere colonization assays. Bacteria were inoculated for HR assays as described by Gopalan et al. (1996). HR inhibitors were co-infiltrated at the following concentrations: cycloheximide, 100 µM, lanthanum chloride, 1 mM, sodium vanadate 50 µM. Low levels of necrosis were observed in A. thaliana in response to cycloheximide at the standard concentration, so serial dilutions of cycloheximide were used to confirm the inhibition of the HR. All HR results were confirmed in at least three independent experiments. Inoculation of bacteria onto sugar beet seeds and recovery of bacteria for rhizosphere colonization assays was performed as described by Rainey (1999). Sugar beet seeds were removed from the rhizosphere following germination and before recovery of bacteria from the rhizosphere. For competitive rhizosphere colonization assays, bacteria were inoculated at a ratio of 1:1 (∼104 bacteria per seed) and plated onto selective (kanamycin 25 µg ml−1, CFC) and non-selective (CFC only) media following recovery from the rhizosphere to determine relative population densities. Bacterial populations in the rhizosphere were measured for three seedlings of similar size, as described in the results.
Assay for β-galactosidase activity
β-galactosidase was assayed using the substrate 4-methylumbelliferyl-β-d-galactoside. Reaction products were detected using a fluorimeter at 460 nm with 365 nm excitation wavelength according to the manufacturer's instructions (DyNAQuant). Protein concentrations were determined using the Bio-Rad protein assay according to the manufacturer's instructions (Bio-Rad).
Preparation of protein samples and immunoblot analysis of secreted proteins
Bacteria were grown overnight on LB plates at 30°C with appropriate antibiotics and resuspended in 40 ml LB broth at OD600 0.15. Bacteria were cultured at 30°C in a rotary shaking incubator at 200 r.p.m. until the OD600 reached 0.8. Protein samples were prepared from supernatant and cell fractions according to the protocol described by Ham et al., 1998 and resuspended in 1x NuPAGE LDS sample buffer (Invitrogen). Protein samples were separated by electrophoresis through a 12% NuPAGE Novex Bis-Tris gel with MOPS running buffer and then either stained using Bio-Safe Coomassie stain (Bio-Rad) or electrotransferred to Immobilon-P membrane (Millipore) with the X Cell II Blot Module (Invitrogen). AvrB-HA was detected using Anti-HA-Peroxidase and rat monoclonal high affinity antibody (clone 3F10) (Roche) was detected using the ECL Detection system (Amersham).
We thank E. Holmes for assistance with phylogenetic analysis and tree construction, S. Gurr for comments on a draft of the paper, J. Baker for photography, I. Moore for assistance with photography of A. thaliana, and A. Collmer, A. Vivian, M. Grant, B. Lugtenburg, J. Vanderleyden, F. O′gara, C. Pieterse and J. Dangl for providing seeds and bacterial strains. This work was supported by the BBSRC (UK).