Identification of Pseudomonas syringae pv. tomato genes induced during infection of Arabidopsis thaliana

Authors

  • Jens Boch,

    1. Department of Biology, Campus Box 1137, Washington University, 1 Brookings Drive, St Louis, MO 63130, USA.
    Search for more papers by this author
    • Present addresses: Martin-Luther-Universität, Institut für Genetik, 06099 Halle, Germany.

  • Vinita Joardar,

    1. Department of Biology, Campus Box 1137, Washington University, 1 Brookings Drive, St Louis, MO 63130, USA.
    Search for more papers by this author
  • Lisa Gao,

    1. Department of Biology, Campus Box 1137, Washington University, 1 Brookings Drive, St Louis, MO 63130, USA.
    Search for more papers by this author
    • Department of Biological Sciences, Purdue University, West Lafayette, IN 47907, USA.

  • Tara L. Robertson,

    1. Department of Biology, Campus Box 1137, Washington University, 1 Brookings Drive, St Louis, MO 63130, USA.
    Search for more papers by this author
    • § Department of Biology, St Louis University, St Louis, MO 63108, USA.

  • Melisa Lim,

    1. Department of Biology, Campus Box 1137, Washington University, 1 Brookings Drive, St Louis, MO 63130, USA.
    Search for more papers by this author
  • Barbara N. Kunkel

    Corresponding author
    1. Department of Biology, Campus Box 1137, Washington University, 1 Brookings Drive, St Louis, MO 63130, USA.
    Search for more papers by this author

Summary

Phytopathogenic bacteria possess a large number of genes that allow them to grow and cause disease on plants. Many of these genes should be induced when the bacteria come in contact with plant tissue. We used a modified in vivo expression technology (IVET) approach to identify genes from the plant pathogen Pseudomonas syringae pv. tomato that are induced upon infection of Arabidopsis thaliana and isolated over 500 in planta-expressed (ipx) promoter fusions. Sequence analysis of 79 fusions revealed several known and potential virulence genes, including hrp/hrc, avr and coronatine biosynthetic genes. In addition, we identified metabolic genes presumably important for adaptation to growth in plant tissue, as well as several genes with unknown function that may encode novel virulence factors. Many ipx fusions, including several corresponding to novel genes, are dependent on HrpL, an alternative RNA polymerase sigma factor that regulates the expression of virulence genes. Expression analysis indicated that several ipx fusions are strongly induced upon inoculation into plant tissue. Disruption of one ipx gene, conserved effector locus (CEL) orf1, encoding a putative lytic murein transglycosylase, resulted in decreased virulence of P. syringae. Our results demonstrate that this screen can be used successfully to isolate genes that are induced in planta, including many novel genes potentially involved in pathogenesis.

Introduction

Interactions between plant pathogenic bacteria and their hosts are complex and require that the pathogen both adapt to the in planta environment and modulate the physiology of the host (Alfano and Collmer, 1996). For example, to successfully infect and cause disease, bacterial plant pathogens must be able to grow epiphytically on plant surfaces, gain entry into host tissues, colonize the intercellular spaces between plant cells (the apoplast), obtain nutrients and multiply to high levels within the apoplast and cause damage to host tissue (Alfano and Collmer, 1996; Staskawicz et al., 2001). The identities of the pathogen factors involved in these processes are largely unknown, and how they function to promote parasitism and disease is not understood (Staskawicz et al., 2001).

Genetic analysis of Pseudomonas syringae, a well-known bacterial plant pathogen, has led to the identification of several loci involved in pathogenesis. These include the hrp/hrc genes, avr genes and genes that direct the production of toxins and other extracellular virulence factors (Alfano and Collmer, 1996; Bender et al., 1999; White et al., 2000). The hrp/hrc genes encode a specialized (type III) protein secretion apparatus homologous to those found in many Gram-negative animal pathogens, including Yersinia and Salmonella (Alfano and Collmer, 1997; Galan and Collmer, 1999). Type III secretion systems mediate the transfer of bacterial proteins (referred to as ‘effectors’) into the cytosol of host cells, where they modulate normal host cell processes (Staskawicz et al., 2001). In P. syringae, the hrp/hrc type III secretion system is required for both growth in susceptible host plants and activation of plant defence responses, often characterized by the hypersensitive response (HR), in resistant plants (Lindgren et al., 1986; Staskawicz et al., 2001). A number of type III effectors produced by phytopathogenic bacteria have been identified; however, their mode of action in plant cells is not understood.

The majority of known P. syringae effector genes were identified because their expression elicited host defence responses that prevented disease in some plant hosts (White et al., 2000). Hence, these genes are referred to as ‘avirulence’ (avr) genes. The abundance of avr genes in pathogen genomes and the implication that many of the corresponding proteins are translocated into plant host cells by the type III system suggest that these proteins may promote parasitism and disease. Consistent with this hypothesis is the finding that several avr genes contribute to virulence of P. syringae on susceptible plant genotypes (Chen et al., 2000; White et al., 2000; Guttman and Greenberg, 2001).

Pseudomonas syringae strains produce several additional factors that contribute to virulence. These include exopolysaccharides, cell wall-degrading enzymes, plant hormones and several low-molecular-weight phytotoxins, such as coronatine and syringomycin (Gross and Cody, 1985; Alfano and Collmer, 1996). These factors are believed to promote virulence by altering various aspects of host cell biology and physiology (Gross and Cody, 1985; Bender et al., 1999). However, mutations in genes directing the production of these various virulence determinants do not render P. syringae non-virulent, indicating that virulence is multifactorial. Given the complexity of interactions between P. syringae and its hosts, it is likely that there are many additional virulence factors yet to be discovered.

Many known P. syringae pathogenicity and virulence determinants, including hrp/hrc and avr genes, are induced upon inoculation into plant tissue (for examples, see Rahme et al., 1991; Innes et al., 1993; Salmeron and Staskawicz, 1993). Thus, we hypothesized that we could identify additional P. syringae genes important for colonization, growth in planta and disease production by screening for bacterial genes that are expressed after inoculation into plant tissue. This approach could also lead to the identification of novel virulence genes that might be missed in standard genetic screens, as mutations in such genes might give rise to subtle phenotypes.

To identify P. syringae pv. tomato genes that are induced in plant tissue, we developed a promoter-trap strategy to select directly for genes that are expressed during infection of Arabidopsis thaliana. Our approach is similar to the in vivo expression technology (IVET) strategies used by a number of laboratories studying pathogenic bacteria (for examples, see Osbourn et al., 1987; Mahan et al., 1993; Young and Miller, 1997; Gahan and Hill, 2000) and environmental microbes (Oke and Long, 1999; Rainey, 1999). The screen is based on complementation of a non-pathogenic P. syringae pv. tomato hrcC mutant that is defective for type III secretion, and therefore unable to grow in plant tissue. The strategy involves selecting for P. syringae pv. tomato promoters that can direct in planta expression of a promoterless hrcC gene, thus restoring growth in plant tissue. Here, we report that characterization of the first 79 P. syringae pv. tomato in planta-expressed (ipx) genes has resulted in the identification of many genes known or predicted to be involved in pathogenesis and several putative novel virulence factors.

Results

Development of an in planta expression system for P. syringae

To identify P. syringae pv. tomato (Pst) genes that are induced during growth in plant tissue, we carried out a modified IVET screen. As illustrated in Fig. 1, our strategy involves two main components: (i) a Pst strain carrying a deletion mutation in hrcC; and (ii) an in planta expression technology (pIPET) vector carrying promoterless copies of the wild-type Pst hrcC gene and the uidA (GUS) reporter gene (Jefferson et al., 1987). The hrcC gene encodes a conserved component of the type III secretion system that is essential for pathogenicity. Thus, hrcC mutants are unable to grow or cause disease in plant tissue (Alfano and Collmer, 1997; Deng et al., 1998). Our strategy was to clone wild-type P. syringae genomic DNA fragments into pIPET upstream of the promoterless hrcC gene and introduce the resulting library into the Pst hrcC deletion mutant. As the pIPET vector is based on pJP5603 (Penfold and Pemberton, 1992), a vector unable to replicate in Pst DC3000, the pIPET library constructs would integrate into the chromosome at sites homologous to the cloned genomic fragment. In most cases, recombination of the pIPET clones would be expected to result in duplication of the homologous region in the genome without causing an interruption in the gene of interest (Fig. 1).

Figure 1.

Overview of the IVET screen for P. syringae pv. tomato (Pst) genes induced in plant tissue. A library of chromosomal DNA fragments from Pst strain DC3000 was constructed in the pIPET vector and introduced into the Pst DC3000 ΔhrcC strain TLR1. Twenty independent pools of ≈1000 Pst TLR1::pIPET strains each were used to inoculate A. thaliana plants by vacuum infiltration. The plants were placed in a growth chamber, and the enriched bacterial populations were isolated from the leaf tissue 4 days after inoculation. The recovered bacteria were then used to infect A. thaliana plants in a second enrichment cycle. Bacteria recovered after the second cycle were screened for GUS activity after growth on rich bacteriological media (KB) plates using the indicator substrate Xgluc to allow for the identification of strains carrying constitutive promoters. Colonies with no or low GUS activity on KB were selected, and the plasmids were recovered for further characterization.

The resulting bacterial strains would then be pooled and inoculated into A. thaliana leaf tissue and the infected plants placed in a growth chamber. The bacteria would be isolated from the plants 4 days later, pooled and used to inoculate A. thaliana plants again for a second round of enrichment (Fig. 1). The majority of bacterial strains would not be expected to grow within the leaf tissue, as they carry a non-functional hrcC gene. However, those strains carrying pIPET plasmids containing P. syringae promoters that could direct expression of hrcC early during the infection process (e.g. upon inoculation into plant tissue) would be predicted to multiply, and thus be enriched. The resulting IPET strains could then be recovered from plant tissue, their pIPET plasmids isolated and the promoters of interest characterized.

The pIPET vector and the hrcC deletion mutant of Pst strain DC3000 used in this study were constructed as described in Experimental procedures. The hrcC deletion mutant (TLR1) was unable to proliferate within plant tissue (Fig. 2), did not cause disease symptoms on A. thaliana and tomato (data not shown) and was unable to elicit the hypersensitive response (HR) on the non-host plant Nicotiana tabacum (data not shown).

Figure 2.

Growth of P. syringae pv. tomato (Pst) TLR1 avrPtofor::pIPET (hrcC+) in plant tissue. The growth of the indicated P. syringae strains was monitored in A. thaliana leaves 0, 2 and 4 days after inoculation. Data points represent the means of three measurements ± SEM. Similar results were obtained in a second independent experiment. The strains used were Pst DC3000 (open triangles), TLR1 (ΔhrcC, circles), JB200 (TLR1 avrPtofor::pIPET, squares) and JB201 (TLR1 avrPtorev::pIPET, filled triangles).

To confirm that the pIPET vector and the selection scheme were functioning properly, we carried out control experiments using the promoter of avrPto, a gene previously shown to be induced upon inoculation into plant tissue (Salmeron and Staskawicz, 1993). A 950 bp fragment carrying the avrPto promoter was introduced into the pIPET vector in both forward and reverse orientations with respect to hrcC, and the resulting constructs were transferred into TLR1. The strain carrying the avrPto promoter in the forward direction (avrPtofor::pIPET; JB200) was complemented for the hrcC mutant phenotype, grew to wild-type levels in A. thaliana, caused disease on A. thaliana and tomato plants and elicited an HR on tobacco (Fig. 2; data not shown). In contrast, the pIPET derivative carrying the avrPto promoter in the reverse orientation (avrPtorev::pIPET; JB201) did not complement the hrcC mutant (Fig. 2). Thus, the selection scheme functions as expected. In addition, these results demonstrate that a Pst DC3000 promoter normally induced during growth in plant tissue directs the expression of hrcC early enough during the infection process to restore wild-type growth and virulence to the hrcC mutant.

To ensure that this selection scheme would result in enrichment of strains expressing a functional hrcC gene, we verified that the growth defect of hrcC mutant cells could not be complemented in trans by bacteria expressing a functional hrcC gene. This was assessed in a reconstruction experiment by inoculating A. thaliana plants with a mixture of JB200 (avrPtofor::pIPET; SpecR KanR) and TLR1 (hrcC::Δ; SpecR) cells, using a ratio of one JB200:1000 TLR1 cells. Analysis of bacterial growth revealed that strain JB200 was enriched at least 1000-fold after two in planta growth cycles of 4 days each (data not shown), indicating that strains expressing hrcC can be individually enriched within a population of IPET strains lacking a functional hrcC gene.

Isolation of P. syringae promoters that are expressed in planta

A library containing random genomic fragments (1–5 kb in length, average insert size 2.4 kb) from Pst DC3000 was constructed in pIPET (see Experimental procedures). Twenty separate pools of ≈1000 individual clones each were introduced into strain TLR1, and each resulting pool of Pst cells was used to inoculate A. thaliana to enrich for strains that could grow in plant tissue as described above (Fig. 1). To distinguish clones carrying promoters that are induced in plant tissue from those carrying constitutively expressed promoters, the bacteria were grown on rich media (KB), and GUS expression was analysed using Xgluc, a colorimetric substrate for the GUS reporter (see Experimental procedures). Before the enrichment experiment, the pIPET library contained ≈50% constitutive clones. After two rounds of enrichment in A. thaliana tissue, ≈90–95% of the clones recovered from plant tissue expressed GUS activity on KB plates, and thus appeared to contain constitutive promoters (data not shown). Strains that exhibited no or very low GUS activity when grown on rich media were selected for further analysis. These strains were likely to carry pIPET clones with promoters that are induced in planta and will be referred to as ‘in planta-expressed’ (ipx) promoters in this study.

From each of the 20 enrichment experiments, we chose 20–32 strains (resulting in a total of 532) for further analysis. As a first step towards analysing the ipx promoters, the GUS expression patterns of the ipx fusions were qualitatively assessed on rich, minimal and hrp-inducing medium, a minimal medium often used to induce the expression of P. syringae hrp/hrc and avr genes in culture (Huynh et al., 1989; Rahme et al., 1992; Salmeron and Staskawicz, 1993). The majority of the ipx fusions exhibited GUS activity on hrp-inducing medium and/or on minimal medium containing succinate (Table 1; data not shown). This suggests that these minimal media effectively mimic the conditions encountered by P. syringae in the extracellular spaces of A. thaliana leaves. Seventy-three fusions exhibited very low promoter activity when grown under these conditions (data not shown). To determine whether these 73 strains carried ipx fusions that are expressed in plant tissue, we monitored their ability to express hrcC in planta. This was done by testing these strains individually for the ability to elicit tissue collapse indicative of an HR on tobacco. Fourteen of the 73 strains tested elicited an HR on tobacco, and thus express hrcC in planta (data not shown). The remaining 59 clones (≈10% of the initial 532 enriched clones studied) did not exhibit detectable hrcC activity in the HR assay. These findings indicate that this IVET strategy works well to enrich for in planta-expressed promoters, but that fusions that are not strongly induced in planta can also be isolated in this screen. Thus, before further detailed analysis, fusions obtained in this screen should be verified by an independent assay for in planta expression. For the purposes of this study, we focused on fusions demonstrated to express hrcC, in hrp-inducing medium, minimal media or in planta, for further characterization.

Table 1. P. syringae ipx fusions expressed during growth in A. thaliana.
LocusSimilar protein, organism, % identityaPredicted functionb hrp boxcHrpLd
  • a. Bacterial proteins with most similarity to predicted ipx proteins. chp, conserved hypothetical protein; hp, hypothetical protein; GenBank or SWISSPROT database accession numbers are shown in parenthesis. DNA sequences corresponding to the ipx genes are available as Supplementary material and at: http://www.biology.wustl.edu/P.syringae.IVET.

  • b. Predicted function assigned based on predicted or known function of similar proteins identified in BLAST searches.

  • c. DNA sequence extending ≈1.5 kb upstream of the fusion point in the ipx clones was scanned for the presence of the conserved ‘hrp box’ sequence motif GGAAC-N (16–17)-CACNNA found in HrpL-dependent promoters (Xiao and Hutcheson, 1994).

  • d. Regulation by HrpL was determined in E. coli as described in Experimental procedures; +, increased GUS activity in E. coli carrying pYXL2B ( Xiao et al., 1994); –, no increased GUS activity; NT, not tested.

  • e.

    Two or more independent ipx fusions to the same gene were identified.

  • f.

    ipx fusion not induced in hrp-inducing medium but was induced in minimal media containing succinate.

  • g. Strain did not elicit tissue collapse (an HR) when inoculated into tobacco leaves at 5 × 10 7cfu ml−1.

  • h.

    ipx32 does not map to the CEL ORF region.

  • i.

    ipx fusion not induced in culture; strain elicited an HR when inoculated onto tobacco.

  • j.

    No hrp box found within 5 kb upstream of the ipx fusion point.

  • k.

    At the fusion point in ipx51, 66, 77 and 78, there is a putative ORF running in the opposite orientation with respect to hrcC. Thus, the nature of the promoters driving the expression of these fusions is unclear.

Known or predicted role in virulence
ipx1–6Hrp/Hrc, P. syringae pv. tomato, 100%
(HrpJ AAG33876; HrpG AAC34755; HrcQb
AAG33883; HrpA AAB00126;HrpKe AAF71489)
Type III secretion++
ipx7VirPphA, P. syringae pv. phaseolicola, 48% (AAD47203)Virulence++
ipx8AvrPpiB, P. syringae pv. pisi, 100% (CAA59280)Unknown++
ipx9AvrPphD, P. syringae pv. phaseolicola, 72% (CAC16699)Unknown++
ipx10,11eCEL ORF1, P. syringae pv. tomato, 100% (AAF71498)Lytic murein transglycosylase+
ipx12,13eCEL ORF3, P. syringae pv. tomato, 100% (AAF71501)Unknown++
ipx14CEL ORF7, P. syringae pv. tomato, 100% (AAF71506)Unknown++
ipx15CEL ORF8, P. syringae pv. tomato, 100% (AAF71507)Unknown++
ipx16–28Cfl/Cfa, P. syringae pv. glycinea, 89–95% (Cfl: AAC43314e,
Cfa1: AAB41298, Cfa5: PC4426, Cfa6: AAD03047e,
Cfa7: AAD03048e, Cfa8: AAC38656, Cfa9: AAC38657)
Coronatine biosynthesis
ipx29fg[link]SyrE, P. syringae pv. syringae, 50% (AAC80285)Syringomycin synthetase
ipx30fAlgA, Pseudomonas aeruginosa, 79% (P07874)Alginate biosynthesis
ipx31PlyD, Bacillus subtilis, 43% (BAA12119)Pectin lyase++
ipx32hCEL ORF5, P. syringae pv. tomato, 74% (AAF71504)UnknownNT
ipx33fg[link]PA 3920, P. aeruginosa, 71% (AAG07307)Probable metal transporting ATPase
ipx34CatF, P. syringae pv. syringae, 90% (AAC61659)Catalase
No established role in virulence
ipx35chp PA5547, P. aeruginosa, 55% (AAG08932)Putative phosphotransferase j +
ipx36fAmpG, P. aeruginosa, 47% (AAD21100)Muropeptide transporter
ipx37FadD, Pseudomonas putida, 86% (AAF02529)Long-chain fatty acid CoA ligase
ipx38AceE, P. aeruginosa, 90% (AAG08400)Pyruvate dehydrogenase
ipx39,40ehp from Streptomyces coelicolor, 35% (CAC13072)Putative hydrolase
ipx41PA2787, P. aeruginosa, 26% (AAG06175)Carboxypeptidase++
ipx42fDcdA, E. coli, 62% (AAA83861)Diaminopimelate decarboxylase
ipx43IlvI, P. aeruginosa, 70% (AAG08082)Acetolactate synthase
ipx44GcvH1, P. aeruginosa, 78% (AAG08599)Glycine cleavage complex protein H
ipx45chp PA2993, P. aeruginosa, 50% (AAG06381)Thiamine biosynthesis++
ipx46iPA4861, P. aeruginosa, 55% (AAG08246)ATP-binding component of ABC-transporter
ipx47SC1B2.04, S. coelicolor, 25% (CAB92560)Carboxylase
ipx48PA1396, P. aeruginosa, 47% (AAG04785)Two-component sensor
ipx49PA5512, P. aeruginosa, 58% (AAG08897)Two-component sensor
ipx50fg[link]TnpA, P. syringae pv. tomato, 98% (AAF71487)Transposase
Homology to proteins with unknown function or no homology
ipx52Rhs family protein, Salmonella enterica, 34% (CAD08754)Unknown++
ipx53fPM1365, Pasteurella multocida, 45% (AAK03449)Unknown j +
ipx54chp VC2671, Vibrio cholerae, 23% (AAF95812)Unknown++
ipx55fhp PA1837, P. aeruginosa 69% (AAG05226)Unknown
ipx56fg[link]PM1475, P. multocida, 38% (AAK03559)Unknown
ipx57fhp PA2437, P. aeruginosa, 70% (AAG05825)Unknown
ipx58fchp PA3664, P. aeruginosa, 81% (AAG07052)Unknown
ipx59ihp PA3258, P.◊aeruginosa, 72% (AAG06646)Unknown
ipx60fchp PA3922, P. aeruginosa, 75%(AAG07309)Unknown
ipx61fg[link]chp PA0550, P. aeruginosa, 68% (AAG03939)Unknown
ipx62hp PA0862, P. aeruginosa, 72% (AAG04251)Unknown
ipx63chp PA3413, P. aeruginosa, 58% (AAG06801)Unknown
ipx64fchp PA1730, P. aeruginosa, 82% (AAG05119)Unknown
ipx65hp from Mesorhizobium loti, 37% (BAB54235)Unknown
ipx51No homologyUnknownk
ipx6670No homologyUnknownk++
ipx71No homologyUnknown j +
ipx7279No homologyUnknownk

To eliminate sibling strains carrying identical ipx fusions or strains with more than one integrated pIPET vector, DNA hybridization experiments were performed with the 532 isolated strains using a pIPET-specific DNA probe. Only one representative strain from each of several groups apparently carrying identical fusions was chosen for further characterization. Twenty strains were found to carry more than one pIPET vector and were not analysed further. The resulting DNA blots were also hybridized with probes from the hrp/hrc region to identify strains carrying ipx fusions with promoters from the hrp/hrc cluster, which have previously been demonstrated to be induced in planta (Rahme et al., 1992). Thirty-six independent hrp/hrc fusions were identified in this manner and were not characterized further. A subset of the remaining 268 apparently unique ipx strains containing fusions demonstrated to express hrcC was chosen for further analysis (Table 1).

Sequence analysis of ipx fusions

To facilitate identification of the promoters driving expression of hrcC in the ipx strains, pIPET plasmids were isolated from 79 independent strains. The pIPET plasmids were transferred from Pst DC3000 into Escherichia coli by retrotransfer, a method based on triparental mating (Rainey et al., 1997). The resulting E. coli strains were used to prepare pIPET plasmid DNA, which was then subjected to DNA sequence analysis. Approximately 500 bp of Pst DC3000 genomic sequence upstream of hrcC was determined for each of the 79 fusions, and the resulting sequence data were used to search for similarities with published sequences in GenBank and unpublished Pst DC3000 genomic sequence available from TIGR (www.TIGR.org). We were thus able to identify most of the genes defined by our ipx fusions. As summarized in Table 1, 34 of the ipx fusions are to genes with known or predicted roles in virulence. This includes six strains (ipx1–6) carrying fusions in the hrp/hrc region that were not detected in the prescreen for hrp/hrc genes. Of the remaining fusions, 16 correspond to genes with no known virulence function and 14 to genes encoding hypothetical proteins with unknown function. Fifteen fusions appear to define unique genes, as they exhibit no similarity to entries in GenBank (Table 1). Of the 79 fusions sequenced, we found seven instances in which the same gene was identified by two or more independent fusions (Table 1).

ipx genes known or predicted to be involved in virulence

The nucleotide sequence analysis of the pIPET inserts revealed that several of the ipx promoters correspond to genes known to be involved in P. syringae pathogenesis. These include hrp/hrc genes (Galan and Collmer, 1999), genes encoding putative effectors of the hrp/hrc type III secretion system located both within the conserved effector locus (CEL) flanking the hrp/hrc cluster (Alfano et al., 2000) and elsewhere in the genome (virPphA, avrPpiB and avrPphD; Cournoyer et al., 1995; Jackson et al., 1999; Arnold et al., 2001a), biosynthetic genes for the P. syringae phytotoxin coronatine (cfl, cfa; Bender et al., 1999) and algA, which directs the synthesis of the exopolysaccharide alginate (Peñaloza-Vázquez et al., 1997).

Additionally, several ipx fusions are to novel genes that would be predicted to be involved in virulence, including a gene encoding a putative pectin lyase (ipx31), which may be involved in degrading plant cell walls. The ipx29 fusion identifies a non-ribosomal peptide-synthetase with homology to SyrE, a protein involved in the synthesis of the phytotoxin syringomycin in P. syringae pv.◊syringae (Guenzi et al., 1998). However, as Pst does not synthesize syringomycin (Völksch and Weingart, 1998; Bultreys and Gheysen, 1999), it is possible that the ipx29 gene is involved in the synthesis of other novel phytotoxins or antimicrobial toxins (Völksch and Weingart, 1998). The ipx33 fusion defines a putative metal-transporting ATPase that exhibits high homology to genes conferring copper resistance (Mellano and Cooksey, 1988). Copper resistance is a trait frequently acquired by plant pathogenic bacteria, as crops are commonly treated with copper bactericides (Agrios, 1997). The ipx34 fusion defines a gene encoding a periplasmic catalase (CatF). Catalase is predicted to contribute to virulence as it may protect bacteria from active oxygen species produced by plants in response to microbial attack (Klotz and Hutcheson, 1992; Felix et al., 1999). The importance of catalase in the virulence of a plant pathogen has been demonstrated for Agrobacterium tumefaciens, in which tumour formation was significantly attenuated in a catalase mutant (Xu and Pan, 2000).

Expression of a subset of ipx fusions is controlled by HrpL

One approach to classify ipx genes further is to determine whether they belong to the pathogenesis-related regulon governed by HrpL. In P. syringae, expression of many genes involved in pathogenesis, such as hrp/hrc and avr genes, is governed by the alternative RNA polymerase sigma factor HrpL (Rahme et al., 1992; Innes et al., 1993; Salmeron and Staskawicz, 1993; Xiao et al., 1994), and it is likely that HrpL also regulates the expression of additional genes involved in virulence. One way to identify ipx genes potentially regulated by HrpL is to examine the promoter regions of these genes for the presence of an ‘hrp box’, a regulatory element believed to be recognized by HrpL (Xiao and Hutcheson, 1994). For each of the 69 different ipx genes (defined by the 79 ipx fusions) shown in Table 1, we scanned genomic sequences (available at www.TIGR.org) corresponding to ≈1500 nucleotides upstream of the hrcC fusion point for the presence of the consensus ‘hrp box’ sequence motif (GGAAC-N(16–17)-CACNNA; Xiao and Hutcheson, 1994). Twenty-two of the 69 ipx genes identified, including 10 potentially novel virulence genes, contained a well-conserved ‘hrp box’ motif (Table 1), suggesting that expression of these genes is directly regulated by HrpL.

To assess directly whether expression of any of these ipx promoters are governed by HrpL, the pIPET plasmids were introduced into an E. coli strain carrying hrpL from P. syringae pv. syringae under the control of the lac promoter (pYXL2B; Xiao et al., 1994). For comparison, each pIPET plasmid was also introduced into an isogenic E. coli strain lacking hrpL. The GUS activity of the E. coli strains was then assayed qualitatively on agar plates containing Xgluc. In control experiments, E. coli carrying both the hrpL plasmid and a pIPET clone containing a hrp/hrc promoter (e.g. ipx4) gave rise to dark blue colonies, whereas E. coli carrying the hrp/hrc fusion, but no hrpL, remained white (data not shown).

The results from two independent experiments assaying HrpL-directed expression of the ipx fusions in this E. coli system are summarized in Table 1. All but one of the ipx genes containing an ‘hrp box’ (CEL orf1; ipx10, 11) exhibited HrpL-dependent expression in these experiments, indicating that this assay provides a reliable method for the identification of genes that are directly regulated by HrpL. The HrpL-regulated ipx fusions include the hrp/hrc fusions, the avrPpiB, avrPphD and virPphA fusions and the fusions to three of the genes in the CEL (ipx12–15; Table 1). Interestingly, several fusions that exhibited no similarity to entries in the databases (ipx66–71), similarities to hypothetical proteins (ipx35, 41, 45, 53, 54) and a putative Rhs family protein (ipx52) also appear to be regulated by HrpL (Table 1). These results suggest that the genes identified by these fusions belong to an HrpL-dependent regulon and may encode novel virulence factors, including as yet unidentified effector proteins for the hrp/hrc type III secretion system. Unexpectedly, two independent fusions to CEL orf1 (ipx10, 11) were not expressed in the E. coli HrpL assay, even though there is a well-conserved ‘hrp box’ in the promoter region for this gene (Table 1; Alfano et al., 2000). In addition, fusions to three genes (ipx35, 53, 71) that do not contain a recognizable ‘hrp box’ within ≈5 kb of the hrcC fusion point exhibited HrpL-dependent expression in this assay (Table 1).

The ipx fusions are induced after infection of plant tissue

To verify that the ipx promoters are induced in planta, the GUS reporter gene activity for eight ipx strains was monitored directly after inoculation into A. thaliana. The majority of the strains tested exhibited clear induction of promoter activity after infection of plant tissue (Table 2). The maximum relative GUS activity was observed at 12 h after infection for all strains tested, with the exception of the negative control JB201 (avrPtorev::pIPET) and ipx37 (fadD), which displayed low expression throughout the experiment. Interestingly, several of the fusions induced in planta exhibited only moderate (ipx31) or no (ipx10, 17 and 75) HrpL dependence in the E. coli expression assay (Table 1; data not shown). These findings suggest that other regulatory factors, in addition to HrpL, may be involved in directing the in planta expression of several genes identified in this screen.

Table 2. Expression of ipx fusions after inoculation of A. thalianaa.
     
Strain or
fusion
SimilarityGUS activity
t0b
GUS activity
t12b
Fold
induction
  • a.

    Similar results were obtained in a second independent experiment.

  • b. Pst TLR1::pIPET strains were grown at 28°C in KB medium, washed, vacuum infiltrated into A. thaliana leaves at a concentration of 5 × 10 8 cfu ml−1, and the inoculated plants were then placed in a growth chamber. GUS expression was measured at 0, 6, 12 and 24 h after infiltration. GUS activity is shown in μmol of 4-methylumbelliferyl produced min−1 10−9 cfu and was highest at 12 h after infiltration for all fusions. Data represent the mean of four measurements ± SEM.

JB200 avrPto for 0.68 ± 0.02548.6 ± 9.5871.3
JB201 avrPto rev 0.35 ± 0.0350.57 ± 0.121.6
ipx8 avrPpiB 0.54 ± 0.02534.0 ± 4.5762.5
ipx10CEL orf10.35 ± 0.077.12 ± 0.325.9
ipx17 cfl 0.61 ± 0.0559.04 ± 0.6714.8
ipx31 plyD 0.60 ± 0.06532.3 ± 1.7353.7
ipx37 fadD 2.25 ± 0.0952.52 ± 0.121.1
ipx53PM13650.30 ± 0.052.59 ± 0.078.7
ipx68No homology1.04 ± 0.21514.5 ± 1.614.0
ipx75No homology0.44 ± 0.052.57 ± 0.325.9

Construction and analysis of insertion mutations in ipx genes

To investigate the roles of Pst DC3000 genes induced upon infection, we chose to disrupt four ipx genes by insertional mutagenesis and assay the resulting mutants for alterations in virulence. These were ipx10 (CEL orf1), ipx8 (avrPpiB), ipx31 (plyD) and ipx37 (fadD). CEL orf1 was chosen because its location in the conserved ef-fector locus flanking the hrp/hrc cluster suggests that it may be important in pathogenesis (Alfano et al., 2000). Although other genes in the CEL have been shown to contribute to Pst DC3000 virulence (Alfano et al., 2000), the role of CEL orf1 in pathogenesis has not been described in detail. We chose to generate and characterize a mutation in the ipx gene encoding the avrPpiB homologue, as its contribution to virulence in Pst DC3000 has not been addressed. avr genes are predicted to contribute to virulence, but this has been demonstrated in only a few cases (White et al., 2000). The putative pectin lyase gene was chosen for mutational analysis, as pectolytic enzymes have been demonstrated to be involved in pathogenesis, especially for plant pathogenic bacteria that cause soft rot diseases (Alfano and Collmer, 1996). Although at least one pectate lyase gene has been identified in P. syringae pv. lachrymans (Bauer and Collmer, 1997), the role of pectolytic enzymes in the pathogenesis of P. syringae strains such as Pst DC3000 that cause necrotic lesions (e.g. leaf spotting diseases) has not been explored. The fourth gene chosen for mutational analysis was ipx37, a fadD homologue. Although this gene does not appear to be strongly induced upon inoculation into plant tissue (Table 2), a fadD homologue, rpfB, in Xanthomonas campestris pv.◊campestris has been implicated in the production of a diffusible signal molecule that regulates the expression of secreted virulence factors (Barber et al., 1997). The Pst DC3000 fadD gene defined by ipx37 exhibits 54% amino acid identity to the X. campestris rpfB gene.

To disrupt these genes, an integrational vector containing a genomic DNA fragment internal to the open reading frame (ORF) was constructed for each gene and then introduced into the Pst DC3000 genome (see Experimental procedures). The effect of these mutations on pathogenesis was assayed by analysing disease symptom formation and bacterial growth on A. thaliana and tomato, as well as elicitation of the HR on tobacco. Disruption of CEL orf1 resulted in a reduction in virulence on tomato. This was most obvious in the reduced severity of disease symptoms on tomato plants infiltrated with a low inoculum of the mutant strain (JB205; Fig. 3A and B). Tomato (Rio Grande 76S) plants infiltrated with wild-type Pst DC3000 exhibited many individual dark brown lesions (Fig. 3A), severe wilting of lower leaves and extensive necrosis of upper leaves (Fig. 3B). In contrast, the CEL orf1 mutant produced fewer disease lesions (Fig. 3A) and caused essentially no wilting and little necrosis of lower leaves. The younger leaves of plants inoculated with JB205 exhibited wilting and necrosis similar to that observed on plants inoculated with the wild-type strain (Fig. 3B). The CEL orf1 mutant also exhibited a reproducible decrease in disease symptom production on A. thaliana, but the reduced virulence was less dramatic than that observed on tomato (data not shown). The reduced virulence of the CEL orf1 mutant was also reflected in a small but reproducible reduction in growth on tomato by the sixth day after inoculation (Fig. 3C). We were unable to detect a significant difference in growth on A. thaliana in multiple experiments (data not shown). Mutations in avrPpiB (JB206), plyD (KP391) and fadD (KP415) did not result in a significant reduction in virulence on tomato (Fig. 3D and E) or A. thaliana (data not shown). None of the four mutants exhibited an alteration in the HR elicited on tobacco plants (data not shown). The observation that the avrPpiB, plyD and fadD mutants did not exhibit significant alterations in pathogenesis suggests that there may be a large amount of genetic and/or functional redundancy among Pst DC3000 virulence factors, and that candidate virulence genes such as these are unlikely to be identified by loss of function mutant screens.

Figure 3.

Figure 3.

Virulence of P. syringae pv. tomato (Pst) DC3000 ipx mutants on susceptible tomato plants.

A. Disease lesions on tomato (Rio Grande 76S) leaf 6 days after pipette infiltration with 5 × 104 cfu Pst DC3000 and JB205 (CEL orf1::Δ).

B. Tomato plants 4 days after vacuum infiltration with 1 × 104 cfu ml–1Pst DC3000 and JB205 (CEL orf1::Δ).

C. Growth of Pst DC3000, JB205 (CEL orf1::Δ) and JB206 (avrPpiB::Δ).

D. Growth of Pst DC3000 and KP391 (plyD::Δ).

E. Growth of Pst DC3000 and KP415 (fadD::Δ).

For measurement of bacterial growth in plant tissue, tomato plants were inoculated by vacuum infiltration with the indicated strains at an initial density of 1 × 104 cfu ml–1. The concentration of bacteria in the plant leaves were assayed 0, 3 and 6 days after inoculation. Data points represent the means of three (C and D) or five (E) independent measurements ± standard deviation of the mean (SDM). Similar results were seen in at least one additional independent experiment for each ipx mutant. The data shown in (C) and (D) were obtained in the same experiment, and thus the data shown for the Pst DC3000 control are identical.

Figure 3.

Figure 3.

Virulence of P. syringae pv. tomato (Pst) DC3000 ipx mutants on susceptible tomato plants.

A. Disease lesions on tomato (Rio Grande 76S) leaf 6 days after pipette infiltration with 5 × 104 cfu Pst DC3000 and JB205 (CEL orf1::Δ).

B. Tomato plants 4 days after vacuum infiltration with 1 × 104 cfu ml–1Pst DC3000 and JB205 (CEL orf1::Δ).

C. Growth of Pst DC3000, JB205 (CEL orf1::Δ) and JB206 (avrPpiB::Δ).

D. Growth of Pst DC3000 and KP391 (plyD::Δ).

E. Growth of Pst DC3000 and KP415 (fadD::Δ).

For measurement of bacterial growth in plant tissue, tomato plants were inoculated by vacuum infiltration with the indicated strains at an initial density of 1 × 104 cfu ml–1. The concentration of bacteria in the plant leaves were assayed 0, 3 and 6 days after inoculation. Data points represent the means of three (C and D) or five (E) independent measurements ± standard deviation of the mean (SDM). Similar results were seen in at least one additional independent experiment for each ipx mutant. The data shown in (C) and (D) were obtained in the same experiment, and thus the data shown for the Pst DC3000 control are identical.

Discussion

To gain a better understanding of the molecular mechanisms governing P. syringae pathogenesis, we performed a modified IVET screen to identify bacterial genes expressed during the susceptible interaction between Pst DC3000 and A. thaliana. Although only a fraction of the in planta-expressed genes potentially isolated in this screen have been characterized to date, this screen has already resulted in the identification of a large number of known virulence genes, novel potential virulence genes, metabolic and biosynthetic genes presumably important for survival and growth in leaf tissue and genes with unknown function (Table 1).

Genes known or predicted to be involved in virulence

Approximately 40% of the ipx fusions that we analysed in detail identify genes known or predicted to be involved in pathogenesis (Table 1). It is not surprising that we identified a large number of fusions to the hrp/hrc region, as these genes are known to be induced upon inoculation of bacteria into plant tissue (Rahme et al., 1991). Likewise, we anticipated identifying genes that encode known or predicted effectors of the type III secretion system, such as avr and vir genes and members of the conserved effector locus (CEL), as these genes are known or expected to be co-regulated with the hrp/hrc locus (Xiao et al., 1994; He, 1998; Alfano et al., 2000). The fact that our IVET strategy selected efficiently for several known genes of this class both validates the approach and suggests that many of the other genes identified in this study are also involved in pathogenesis.

Type III-dependent factors. The two avr genes identified in this screen, avrPpiB (ipx8) and avrPphD (ipx9) have been identified previously in Pst DC3000 (Cournoyer et al., 1995; Arnold et al., 2001a,b). Although avr genes confer the ability to elicit defence responses on plants carrying the corresponding resistance gene (White et al., 2000), it is hypothesized that they provide a selective advantage to the pathogen, perhaps by promoting virulence on susceptible hosts. However, only a few avr genes have been demonstrated to contribute significantly to pathogen virulence (White et al., 2000). We constructed an avrPpiB mutant Pst DC3000 strain (JB206), but were unable to detect a significant alteration in virulence towards A. thaliana or tomato in this strain (Fig. 3). The absence of a virulence phenotype for this mutant is not surprising, as many type III effector proteins are predicted to play redundant or overlapping roles in parasitism (Alfano and Collmer, 1996).

The identification of a gene with similarity to virPphA of P. syringae pv. phaseolicola (ipx7) in Pst DC3000 is exciting. virPphA was initially identified based on its virulence activity in P. syringae pv. phaseolicola and was then subsequently shown to function as an avr factor in soybeans (Jackson et al., 1999). Several groups have also recently reported the discovery of this virPphA-like gene in Pst DC3000 (G. Martin, personal communication, S.-Y. He, personal communication), and its contribution to virulence is currently under investigation.

Alfano et al. (2000) have proposed that many of the genes in the CEL encode type III effectors, as (i) most of the ORFs in the CEL are preceded by a hrp box; and (ii) the deletion of several of these ORFs (including avrE, hrpW and CEL orfs 2–5, but not CEL orf1) from the Pst DC3000 genome resulted in a strain with significantly reduced virulence on tomato. Our identification of ipx fusions to four different genes in the CEL, and demonstration that expression of these genes (with the exception of CEL orf1; see below) is regulated by HrpL (Table 1), further strengthens this hypothesis. Interestingly, we also identified an ipx fusion to a gene with similarity to CEL orf5 (ipx32) that does not map to the CEL and does not contain a conserved ‘hrp box’ in its regulatory region (Table 1). It is unclear whether this gene encodes a new virulence factor.

Our finding that CEL orf1 is not expressed in E. coli carrying the P. syringae hrpL gene (Table 1) is unexpected, given that the ORF is preceded by a well-conserved ‘hrp box’ (Table 1; Alfano et al., 2000), and that genetic studies indicate that expression of CEL orf1 in P. syringae is dependent on hrpL (Lorang and Keen, 1995). We have verified that the pIPET clones carrying the ipx10 and ipx11 fusions contain DNA extending more than 1 kb upstream of the ‘hrp box’, and thus should carry sufficient upstream regulatory sequences to direct normal expression of CEL orf1. This is the only example out of the 21 ipx genes with an ‘hrp box’ that was not induced in the E. coli HrpL expression assay (Table 1). These findings suggest that additional regulatory factors not present in E. coli may be required for the expression of CEL orf1.

CEL ORF1 exhibits similarity to murein lytic transglycosylases (38% amino acid identity to E. coli MltD; Alfano et al., 2000), proteins predicted to encode peptidoglycan hydrolases. It has been hypothesized that this protein may contribute to the virulence of P. syringae by facilitating insertion of the type III secretion apparatus through the peptidoglycan layer of the bacterial envelope (Alfano et al., 2000). Our finding that the CEL orf1 mutant strain (JB205) has reduced virulence (Fig. 3) is consistent with this hypothesis. However, we observed only a small decrease in virulence for this strain, suggesting that there may be one or more additional genes in Pst DC3000 encoding a similar function that compensates, in part, for the loss of CEL ORF1 activity. This subtle reduction in virulence may also account for the fact that, in a previous study, no significant change in virulence was observed for a Pst DC3000 CEL orf1 mutant (referred to as avrE transcription unit II by Lorang and Keen, 1995).

Pectin lyase. One of the intriguing genes identified in this screen encodes a putative pectin lyase (PlyD, ipx31). Pectin, a polysaccharide, is a major structural component of primary plant cell walls. Cell walls serve not only to provide shape and strength to plant cells, but also function as a barrier to pathogen entry, and presumably also to the establishment of direct cell–cell contact between pathogen and plant. The induction of pectolytic enzymes could facilitate the assembly of functional type III secretion complexes that allow transport of bacterial proteins through the plant cell wall and cytoplasmic membrane (Alfano and Collmer, 1997; Hu et al., 2001). Alternatively, these enzymes may contribute to degradation or softening of plant cell walls, increased osmotic fragility and eventual plant cell death (Gross and Cody, 1985). Although several P. syringae pathovars are reported to exhibit pectolytic activity (Gross and Cody, 1985), the role of pectolytic enzymes in the pathogenesis of P. syringae is not understood. Our finding that the plyD-like gene defined by ipx31 is induced during growth in plant tissue and is regulated by HrpL (Table 1) suggests that this gene is involved in pathogenesis.

The Pst DC3000 plyD mutant (KP391) did not exhibit reduced virulence when inoculated onto tomato or A. thaliana plants (Fig. 3; data not shown). It is possible that, like many other plant pathogenic bacteria, Pst DC3000 possesses multiple pectolytic enzymes and that a subtle reduction in virulence resulting from mutation of one of these was not detectable in these experiments (Alfano and Collmer, 1996). To date, only one other Pst DC3000 protein with similarity to a pectolytic enzyme has been reported. HrpW, one of the proteins encoded in the CEL, is a type III-secreted protein that contains a domain with similarity to pectate lyases (Charkowski et al., 1998). However, although HrpW specifically binds to pectate, the protein lacks pectolytic activity. Mutation of the Pst DC3000 hrpW gene did not result in a detectable reduction in virulence on tomato (Charkowski et al., 1998). We have carried out a preliminary search of the available PstDC3000 unfinished genome sequence (www.TIGR.org) for additional pectolytic genes and found one additional gene encoding a putative pectolytic enzyme with similarity to PlyA from Glomerella cingulata (Q00374, 38% amino acid identity/51% amino acid similarity; J. Boch and B. Kunkel, unpublished).

Coronatine biosynthetic genes. Thirteen of the ipx fusions are to genes in the cfl/cfa operon, which encodes enzymes that catalyse synthesis of coronafacic acid, a component of the P. syringae phytotoxin coronatine (COR). Given that the cfl/cfa locus in Pst DC3000 is very large (spanning ≈19 kb; F. Alarcón-Chaidez and C. Bender, personal communication), it is not surprising that we isolated multiple fusions to this region. COR is believed to function as a molecular mimic of the endogenous plant hormone methyl jasmonate (Bender et al., 1999) and may promote colonization of host tissue by altering plant defence signalling pathways (Mittal and Davis, 1995; Kloek et al., 2001). Our observation that the cfl/cfa operon is induced within 6 h after infiltration into plant tissue (Table 2; data not shown) is consistent with previous findings (Li et al., 1998), and supports the hypothesis that COR is important at an early stage of infection.

The regulation of COR gene expression in Pst DC3000 is not fully understood. Recent studies indicate that the expression of Pst DC3000 cfa genes is induced by plant signals (Li et al., 1998) and possibly also by temperature (Rohde et al., 1998), a factor that has a major effect on COR production in P. syringae pv. glycinea (Bender et al., 1999). Our data suggest that cfa gene expression is not directly regulated by HrpL (Table 1). Consistent with this is the finding that there are no sequences resembling an ‘hrp box’ in the regulatory region for the cfl and cfa genes (F. Alarcon-Chaidez and C. Bender, unpublished). However, the observation that an element resembling an ‘hrp box’ is present upstream of corR (V. Joardar and B. Kunkel, unpublished), a regulatory gene that governs COR gene expression in P. syringae pv. glycinea, suggests that COR biosynthesis in Pst DC3000 is coordinated with the expression of the hrp/hrc system.

Genes involved in adaptation for growth in plant tissue

Little is known about the environment faced by P. syringae upon colonization of plant tissue. The identification of ipx fusions to genes likely to be involved in amino acid biosynthesis (ipx42–43) suggests that the apoplastic concentration for some amino acids may be low. This finding is consistent with previous genetic studies demonstrating that many P. syringae mutants impaired for amino acid biosynthesis are unable to grow or cause disease on plant tissue (Cuppels, 1986; Kloek et al., 2000; D. Brooks, A. Kloek and B. Kunkel, submitted). These results may also reflect the fact that some amino acids, such as isoleucine (ipx43), are precursors for phytotoxins (e.g. coronatine; Bender et al., 1999).

Maintenance and/or modification of the bacterial cell envelope may be important for adaptation to changing environmental conditions encountered during pathogenesis (Heithoff et al., 1997; Young and Miller, 1997; Graham and Clark-Curtiss, 1999). In this study, we identified two fusions that define genes that may be involved in bacterial cell wall metabolism. The ipx36 gene encodes a protein with similarity to the muropeptide transporter AmpG, which is involved in the uptake of peptidoglycan turnover products (Holtje, 1998). The ipx35 fusion defines a cicA-like gene, which has been proposed to be involved in regulating cell wall synthesis and morphology in Caulobacter crescentus (Fuchs et al., 2001). The recent finding that a mutation in the N-acetylmuramyl-L-alanine amidase (ampD) gene of Ralstonia solanacearum results in decreased virulence indicates that functional peptidoglycan recycling is important for the virulence of plant pathogens (Tans-Kersten et al., 2000).

The possibility that our screen may predominantly have enriched for genes that are expressed relatively early during infection may account for the fact that we did not identify more genes obviously involved in nutrient acquisition or stress responses. Future studies aimed at identifying genes that are induced at later times during disease interactions may provide additional insight into the metabolic and biosynthetic pathways required for adaptation and growth in plant tissue.

Identification of potential new virulence genes

The largest class of ipx fusions identifies genes that have similarity to proteins with unknown functions (ipx52–65) or show no homology to entries in the databases (ipx51, 66–79). We are excited by the possibility that several of these genes may encode novel virulence factors. One class of fusions appears to be dependent on HrpL. Many genes whose expression is directed by HrpL are believed to be involved with type III secretion, by encoding either structural components of the apparatus or substrates that are secreted via the system (Xiao et al., 1994; Xiao and Hutcheson, 1994). Thus, some of the HrpL-dependent novel genes may encode new effector proteins of the hrp/hrc secretion system. As the hrp/hrc system plays a pivotal role in pathogenesis, it is of prime interest to identify the substrates of this system and to gain a better understanding of the function of these proteins in promoting parasitism and disease.

The regulatory pathways governing induction of P. syringae genes upon inoculation into host tissue are not well understood. Presumably, expression of the genes involved in pathogenesis is activated upon perception of cues (e.g. pH, nutritional cues, specific plant signals) present in the host environment. Based on our analysis of 79 ipx fusions, we have found little evidence in support of the hypothesis that P. syringae pathogenesis genes are induced in response to plant-specific signals. Only two of these fusions (ipx46 and ipx59) appeared to be expressed only in planta, as they did not exhibit GUS activity in culture and yet triggered an HR when inoculated onto tobacco (Table 1). Thus, as the majority of the ipx fusions characterized in this study are induced in hrp-inducing or minimal media, our data suggest that these genes are expressed in response to environmental or nutritional cues present in plant tissue. In the case of genes regulated by HrpL, in planta expression is governed by hrpR and hrpS, whose products are related to the NtrC family of regulatory proteins (Grimm and Panopoulos, 1989; Xiao et al., 1994). However, as HrpR and HrpS lack the regulatory domain found in other members of the NtrC family of two-component response regulators, the mechanism by which HrpR and HrpS sense and respond to environmental cues is unclear (Xiao et al., 1994). In this regard, it will be interesting to analyse the role of the two putative two-component sensors (ipx48, 49) isolated in this screen (Table 1). Future studies on the regulation of ipx genes may help to elucidate the mechanisms that govern how P. syringae senses and responds to the host environment.

Our results indicate that there is at least one additional class of in planta-expressed genes, those whose expression is independent of (or is not solely regulated by) HrpL (Table 1). These genes are especially exciting, as they may define novel virulence factors that constitute one or more new regulatory groups. Further analysis of these novel genes, their regulation and their potential role in pathogenesis should contribute to our understanding of the complex interactions between plant pathogens and their hosts.

Experimental procedures

Bacterial strains and growth conditions

Pseudomonas syringae pv. tomato (Pst) strain DC3000 (Cuppels, 1986) was grown on King’s B (KB; King et al., 1954) or NYG medium (Daniels et al., 1984) at 28°C. For β-glucuronidase expression studies, bacteria were grown on KB medium or a modified hrp/hrc-derepressing minimal medium (MM; Huynh et al., 1989) supplemented with 1.3 mM sodium citrate and 10 mM fructose (hrp-inducing), 10 mM glutamate (non-hrp-inducing) or 10 mM succinate (non-hrp-inducing). E. coli DH5α(λpir) was used as a host in cloning and for proliferation of plasmids with an R6K origin (Miller and Mekalanos, 1988). The following antibiotics were added to the growth media as appropriate: nalidixic acid (Nal, 50 μg ml−1), rifampicin (Rif, 100 μg ml−1), kanamycin (Km, 25 μg ml−1), ampicillin (Amp, 100 μg ml−1), spectinomycin (Sp, 100 μg ml−1), cycloheximide (Cyclo, 50 μg ml−1), tetracycline (Tet, 10 μg ml−1) and chloramphenicol (Cm, 30 μg ml−1).

Plant material and inoculation procedures

Arabidopsis thaliana ecotype Columbia (Col-0) was grown from seed in growth chambers under an 8 h photoperiod at 21–24°C and 75–80% relative humidity. Tobacco (Nicotiana tabacum cv. Xanthi NN) and tomato (Lycopersicon esculentum cv. Rio Grande 76S) were grown from seed in the greenhouse. Inoculation of A. thaliana and tomato plants with bacteria to analyse disease symptoms was carried out either by dipping 3- to 5-week-old plants into bacterial suspensions of 3–6 × 108 cfu ml−1 containing the surfactant Silwet L-77 (Kunkel et al., 1993) or by vacuum infiltration (Whalen et al., 1991). Pipette infiltrations to assay the HR on tobacco plants were carried out with bacterial strains suspended in 10 mM MgCl2 at a density of 5 × 107 cfu ml−1 (Whalen et al., 1991). Leaves were scored for tissue collapse ≈20 h after inoculation. Bacterial growth in A. thaliana and tomato was determined as described previously (Whalen et al., 1991).

The IVET screen was performed by growing the library of Pst TLR1::pIPET strains on KB agar, resuspending them to 1 × 105 cfu ml−1 in 10 mM MgCl2 and vacuum infiltrating them into A. thaliana. The plants were incubated for 4 days, the leaf tissue harvested, and the bacteria isolated by grinding the leaf tissue in 10 mM MgCl2. Dilutions were plated on KB-Cyclo-Rif-Sp-Kan to select for enriched IPET strains.

Construction of the P. syringae pv. tomato DC3000 hrcC deletion mutant and the pIPET vector

The Pst DC3000 strain containing a deletion at hrcChrcC), TLR1, was constructed using a suicide plasmid carrying a deletion construct with an Spr-Ω cassette (pTR20) to replace the wild-type copy of hrcC on the chromosome. pTR20 was generated in several cloning steps. Chromosomal DNA (2.8 kb) from Pst DC3000 flanking the 5′ end of the hrcC gene was obtained as a HindIII–StuI fragment from the cosmid pDC7-20 (obtained from C. Boucher, D. Dahlbeck and B. Staskawicz) and ligated into pBluescript-II KS+ (Stratagene), resulting in plasmid pTR14. The Spr-Ω cassette from pUC4::Ω (Prentki and Krisch, 1984) was inserted as a BamHI fragment into the BamHI site of pTR14, located at the 5′ end of the hrcC gene, creating plasmid pTR15. A 2 kb region of genomic DNA from Pst DC3000 flanking the 3′ end of the hrcC gene was amplified by polymerase chain reaction (PCR) using genomic DNA as a template and the following primers (5′-GGACTAGTCCGGCGGGATTACACTGTCGT GCAGTTAAAGG-3′; 5′-AAGGAAAAAAGCGGCCGCCCAC CTCGGCAAGCTGCTGGTTCTTCTGG-3′; SpeI and NotI restriction sites are underlined respectively), digested with SpeI and NotI and cloned into pTR15 cut with the same enzymes to generate pTR19. pTR19 was cleaved with SalI and SacI, and the resulting 9 kb fragment carrying the ΔhrcC marker replacement cassette was ligated into plasmid pJP5603 (Penfold and Pemberton, 1992) cut with the same enzymes. The resulting plasmid, pTR20, was introduced into Pst DC3000 by triparental mating using an E. coli helper strain carrying pRK2013 (Figurski and Helinski, 1979), and recombinants (KmR, SpR) in which pTR20 had integrated into the wild-type hrcC locus were isolated. These recombinants were screened for a second recombination event that eliminated the plasmid and left the ΔhrcC mutation in the chromosome (KmS, SpR) creating the ΔhrcC strain. The replacement of the wild-type hrcC gene by the deleted copy in strain TLR1 was verified by DNA blot analysis.

The IPET plasmid (pIPET) was generated using the mobilizable suicide plasmid pJP5603 (Penfold and Pemberton, 1992). A promoterless copy of the hrcC gene (′hrcC) was amplified by PCR from Pst DC3000 genomic DNA using the following primers: 5′-GCCGCTCGAGGAATTCTGAT TGATTGAAAACCGGGAGTGTGAAATGCGC-3′ and 5′-CCCCAAGCTTCGGCTTACGCGGTCAGTGCAT-3′ (XhoI, EcoRI and HindIII restriction sites are underlined, translational stop codons in all three reading frames are in italics, the putative Shine–Dalgarno sequence is in bold, and the start and stop codons of hrcC are indicated by the double underline). The resulting 2.3 kb PCR fragment was digested with XhoI and HindIII and ligated into pBluescript SK+, creating plasmid pTR4. A promoterless copy of the uidA gene (Jefferson et al., 1987) was amplified as a 1.8 kb fragment by PCR from plasmid pVJY-B26.5 using the following primers: 5′-GCCGAAGCTTTAGAGGATCTCGACCATGGT CCGTCCTGTAGAAACCCC-3′ and 5′-TCCCCCGGGTC TAGAAAATCATTGTTTGCCTCCCTGCTGCGGTTT-3′ (HindIII and XbaI restriction sites are underlined, an artificial Shine–Dalgarno sequence is in bold, and the start and stop codons of uidA are indicated by the double underline). The resulting fragment was cleaved with HindIII and XbaI and ligated downstream of the promoterless hrcC gene in pTR4, resulting in pTR11. pIPET was constructed by transferring the promoterless hrcC and uidA genes as a 4.1 kb KpnI–XbaI fragment from pTR11 into pJP5603.

Construction of the library of transcriptional fusions to hrcC and uidA

The IPET library of Pst DC3000 chromosomal DNA fragments was prepared by partially cleaving chromosomal DNA with Sau3AI and isolating DNA fragments 1–5 kb in size from an agarose gel. The ends were partially filled in with A and G, thus generating AG-5′ overhangs (Sambrook et al., 1989), a step included to prevent the ligation of multiple inserts. The fragments were ligated into pIPET, which was cut with XhoI and partially filled in with C and T to generate CT-5′ overhangs. The ligation products were transformed into E. coli DH5α(λpir), and 20 independent pools of ≈1000 transformants each were collected. Each pool of pIPET plasmids was then conjugated into Pst TLR1 to generate 20 independent pools of IPET strains. As the pIPET vector is unable to replicate in Pst DC3000 and as hrcC was deleted from the chromosome of TLR1, the ipx fusions were forced to recombine into the genome at the site of homology with the cloned genomic fragment (Fig. 1).

As a control, we also cloned a fragment containing the avrPto promoter into pIPET. The fragment was amplified by PCR from pPtDC5 (Ronald et al., 1992) using the following primers: 5′-CCCCTCGAGGTCGACGGATCT GAACCTGGTGGTCGGCCA-3′ and 5′-CCCCTCGAGACAT CATTGCCAGTTACGGTACGGGCTAGG-3′ (XhoI restriction sites are underlined). The resulting 960 bp product was cleaved with XhoI and ligated in both orientations into pIPET. The resulting plasmids were designated pJB110 (avrPtofor::pIPET, avrPto promoter directing expression of ′hrcC) and pJB111 (avrPtorev::pIPET, promoter in opposite orientation to ′hrcC), respectively, and each plasmid was introduced into strain TLR1 to generate strains JB200 and JB201 respectively.

Isolation of integrated IPET plasmids from the chromosome

To isolate the ipx–promoter fusions from the IPET strains, the integrated pIPET plasmids were transferred from the Pst DC3000 genome into E. coli DH5α(λpir) (NalR) by retrotransfer, a method based on triparental mating (Rainey et al., 1997). An E. coli strain (MM294A, Nals) carrying a CmR Kms derivative (Finan et al., 1986) of plasmid pRK2013 was used as a helper strain. Selection of DH5α(λpir) transconjugants was carried out at 37°C on LB plates containing kanamycin and nalidixic acid.

Methods used with nucleic acids

Routine manipulation of plasmid DNA, PCR and the detection of homologous sequences by DNA hybridization were carried out according to standard procedures (Sambrook et al., 1989). Chromosomal DNA was isolated with the Wizard genomic DNA purification kit (Promega). The protocol was scaled down for the simultaneous isolation of DNA from 96 strains in microtitre dishes. The probe hybridizing to pIPET was made by digoxigenin labelling a 1 kb EcoRI–HindIII fragment from pIPET containing vector sequence located adjacent to the XhoI cloning site (Fig. 1). The probe used to identify plasmids that contain DNA from the Pst DC3000 hrp/hrc cluster was made by isolating and labelling genomic HindIII fragments isolated from the genomic cosmid clone, pDC7-20 (C. Boucher, D. Dahlbeck and B. Staskawicz). Probes from cloned inserts in individual pIPET plasmids were made by excising the insert DNA from the plasmids with EcoRI and subsequently isolating and digoxigenin labelling the corresponding fragments (Genius kit; Boehringer Mannheim).

Sequencing reactions of double-stranded plasmid DNA were performed using the Big Dye Terminator cycle sequencing kit as recommended by the supplier (Perkin-Elmer). A DNA primer (5′-CCCGATCAACAATAAAGGCAAC-3′) corresponding to the 5′ end of the ′hrcC gene in pIPET was used to determine the nucleotide sequence of fragments cloned into the IPET plasmid. Database searches were conducted using the BLAST programs at NCBI. Homology searches with unpublished Pst DC3000 genomic sequences were conducted using preliminary sequence data obtained from The Institute for Genomic Research website at http://www.tigr.org.

Analysing HrpL-dependent expression of ipx fusions

pIPET plasmids were introduced into E. coli DH5α(λpir) strains carrying a plasmid-borne copy of hrpL from P. syringae pv. syringae under lacPO control (pYXL2B; Xiao et al., 1994) or the empty vector (pBluescript KS+). GUS activity of transformants was compared visually on LB agar plates containing Xgluc (20 μg ml−1). The origins of the pIPET- and hrpL-expressing plasmids are compatible.

Construction of gene disruptions in Pst DC3000

Cosmid clones containing wild-type genomic DNA corresponding to selected ipx fusions (ipx8, ipx10, ipx31 and ipx37) were isolated using chromosomal fragments present in the corresponding pIPET plasmids to probe a genomic cosmid library of Pst DC3000, which was constructed in the vector pRK7813 (Barta et al., 1992). The sequences of the putative ORFs defined by the ipx fusions were determined and used to design primers to amplify fragments predicted to be internal to the ORFs. Disruption of putative ORFs was done by subcloning these fragments into the suicide vectors pSUP202 (ipx8, ipx10 and ipx31; Simon et al., 1983) or pJP5603 (ipx37) and introducing the corresponding plasmids into Pst DC3000 by conjugational transfer. The resulting gene disruptions were confirmed by Southern or PCR analysis.

Enzyme assays

The β-glucuronidase (GUS) reporter gene (uidA) on pIPET was used to monitor promoter activity of ipx fusions. GUS activity of Pst TLR1::pIPET colonies grown on plates was monitored visually by lifting the colonies onto filter paper (Whatman) and incubating the filter in a solution containing 50 mM K2HPO4, 0.5% Triton X-100 and 1 mg ml−1 5-bromo-4-chloro-3-indolyl-β-D-glucuronide (Xgluc). The development of a blue colour indicated GUS activity. Quantitative GUS activity of bacteria grown in liquid culture was determined as described previously (Jefferson et al., 1987; Salmeron and Staskawicz, 1993) using 4-methylumbelliferyl-β-D-glucuronide (MUG) as a substrate and 0.2 M Na2CO3 as stop buffer. The activity was calculated in μmol of 4-methylumbelliferone (MU) produced min−1 10−9 cfu.

To monitor the expression of fusions in plant tissue, Pst strains containing ipx fusions were grown overnight on KB agar plates, resuspended in 10 mM MgCl2 at a concentration of 5 × 108 cells ml−1 and vacuum infiltrated into leaves of 5-week-old A. thaliana plants. At 0, 6, 12 and 24 h after infiltration, 0.5 cm2 disks were excised from the leaves with a cork borer and ground with a plastic pestle in 10 mM MgCl2. Serial dilutions were plated on NYG agar plates to determine the bacterial population sizes. Bacterial cells in the remainder of the sample were collected by centrifugation, and their GUS activity was determined using MUG as substrate and 0.2 M Na2CO3 as stop buffer (Salmeron and Staskawicz, 1993). Duplicate samples were taken for each strain and assayed for GUS activity twice each.

Supplementary material

The following material is available from http://www.blackwell-science.com/products/journals/suppmat/mole/mole2877/mmi2877sm.htm

Sequence of P. syringae ipx fusions.

Acknowledgements

We thank Amy B. Fearncombe for help with initiating the IVET screen, Robert F. E. Feissner for initial analysis of several ipx fusions, and Leeann E. Chandler for characterization of the plyD gene and construction of the plyD mutant. We are grateful to Steven W. Hutcheson for providing the hrpL-containing plasmid pYXL2B, Richard Jefferson and Rebecca Harcourt at CAMBIA for providing GUS vectors, and Valerie Oke for helpful discussion and comments on the manuscript. J. Boch was supported by a Deutsche Forschungsgemeinschaft postdoctoral fellowship. This work was supported by a grant from the Searle Scholars Program/The Chicago Community Trust and a David and Lucille Packard Fellowship for Science and Engineering awarded to B.N.K. Sequencing of the P. syringae pv. tomato DC3000 genome by The Institute for Genomic Research (TIGR) was accomplished with support from the National Science Foundation.

Ancillary