Identification of a pathogenicity island, which contains genes for virulence and avirulence, on a large native plasmid in the bean pathogen Pseudomonas syringae pathovar phaseolicola

Abstract

The 154-kb plasmid was cured from race 7 strain 1449B of the phytopathogen Pseudomonas syringae pv. phaseolicola (Pph). Cured strains lost virulence toward bean, causing the hypersensitive reaction in previously susceptible cultivars. Restoration of virulence was achieved by complementation with cosmid clones spanning a 30-kb region of the plasmid that contained previously identified avirulence (avr) genes avrD, avrPphC, and avrPphF. Single transposon insertions at multiple sites (including one located in avrPphF) abolished restoration of virulence by genomic clones. Sequencing 11 kb of the complementing region identified three potential virulence (vir) genes that were predicted to encode hydrophilic proteins and shared the hrp-box promoter motif indicating regulation by HrpL. One gene achieved partial restoration of virulence when cloned on its own and therefore was designated virPphA as the first (A) gene from Pph to be identified for virulence function. In soybean, virPphA acted as an avr gene controlling expression of a rapid cultivar-specific hypersensitive reaction. Sequencing also revealed the presence of homologs of the insertion sequence IS100 from Yersinia and transposase Tn501 from P. aeruginosa. The proximity of several avr and vir genes together with mobile elements, as well as G+C content significantly lower than that expected for P. syringae, indicates that we have located a plasmid-borne pathogenicity island equivalent to those found in mammalian pathogens.

Varietal resistance to halo-blight disease of bean (Phaseolus vulgaris L.) caused by Pseudomonas syringae pv. phaseolicola (Pph) is determined by gene-for-gene interactions involving five resistance (R) genes in the host and five matching avirulence (avr) genes in the pathogen. Depending on the presence or absence of functional avr genes, nine races of Pph have been distinguished (1, 2). The avr genes matching R1, R2, and R3 have been cloned and sequenced. Their full designations are avrPphF.R1, avrPphE.R2, and avrPphB.R3; the terminal R gene designation will not be used here (3–5). Both avrPphE and avrPphB are chromosomal, whereas avrPphF is located on a large plasmid in those races that cause the hypersensitive reaction (HR) in cultivars of bean with the matching R1 gene. The HR is a resistance response recognized by the rapid death of plant cells at inoculation sites and the restriction of microbial colonization (2). Additional avr genes located on the plasmid in Pph determine ability to elicit the HR in nonhost plants, avrPphC and a homolog of avrD (soybean interactors), and avrPphD that interacts with pea (6–8).

Certain avr genes, although recognized by their ability to activate plant defenses (the HR), also may have a role in pathogenicity in the absence of the interacting R gene in the host plant. In some cases there is a clear, qualitative effect on pathogenicity of mutations in avr genes, as with avrBs2 in certain races of Xanthomonas campestris pv. vesicatoria in pepper and avrRpm1 in P. syringae pv. maculicola in Arabidopsis (9, 10). However, the effect of avr gene mutations often is incomplete; for example, the avrE locus has a quantitative role in virulence in P. syringae pv. tomato strain PT23 but is dispensable in strain DC3000 (11). If the avr gene products do have a role in pathogenicity they may contribute to a redundancy of pathogenicity factors so that loss of a single avr gene product may not lead to total loss of ability to cause disease.

It has been proposed that gene-for-gene interactions are superimposed on an established basic, species-specific parasitism (12). Genes controlling such fundamental aspects of infection are located on well-characterized pathogenicity islands (PAIs) in animal pathogens (13, 14). There is increasing support for the hypothesis that bacteria pathogenic to animals have evolved from their nonpathogenic ancestors after acquisition of PAIs that typically contain large fragments of DNA that differ in G+C content from the rest of the genome (13). PAIs have been located on plasmids and also mapped to the chromosome. Examples of the former include the pCD1 plasmid in Yersinia pestis and of the latter, the islands SPI-1 and SPI-2 of Salmonella typhimurium (15, 16). In Salmonella, SPI-1 and SPI-2 both contain components of type III secretion systems (17, 18). This specialized protein delivery system is also the main component of the common hrp cluster found in plant pathogenic bacteria, which also may be considered a PAI. Mutations in hrp genes cause loss of the ability to elicit the HR in resistant plants (whether host or nonhost) and to cause disease symptoms (i.e., to be pathogenic) in susceptible cultivars. In Ralstonia solanacearum, hrp genes are located on a megaplasmid whereas they are chromosomal in Erwinia, Pseudomonas, and Xanthomonas (for recent reviews see refs. 19–21).

The proteins encoded by avr genes appear to be delivered via the type III secretion system into plant cells, where they act as elicitors of the HR. Evidence for this has been obtained by the transient expression of avr genes in plants, production of the Avr protein leading to activation of a rapid HR in an R gene-specific manner; examples include avrPphB and avrPphE (22). In animal pathogens, the type III secretion system is well recognized as a route for export of pathogenicity factors such as Yops in Yersinia, invasion plasmid antigens in Shigella, Sips and Sops in Salmonella, and Esps in enteropathogenic Escherichia coli (23–25). Circumstantial evidence from the existence of the type III system and activity of Avr proteins inside plant cells suggests that similar pathogenicity determinants or virulence factors, as encoded by PAIs in pathogens of animals, also should be present in phytopathogenic bacteria (19).

In addition to pathogenicity factors thought to be secreted through the type III hrp-dependent system and to establish basic parasitism, degradative enzymes, extracellular polysaccharides, and low molecular weight toxins are produced by many plant pathogens (26). Genes encoding the pathogenicity determinants that are hrp independent often are clustered, but they are not generally recognized as PAIs. An example is the cluster of genes involved in the production of phaseolotoxin by Pph. Secretion of phaseolotoxin causes the yellow haloes that develop around lesions in bean leaves (27, 28).

Races of Pph harbor plasmids ranging in size from 25 to 160 kb (29). The large native plasmid (designated pAV511 in race 7 strain 1449B) is known to carry several avr genes. To examine the function of pAV511 we attempted to cure the plasmid by expression of its origin of replication in trans. The successfully cured strains were found to be compromised in virulence toward bean. Here we report the identification of a region on pAV511 that controls virulence, which we describe as a PAI based on the presence of several virulence genes (vir) and association with the transposases IS100 and Tn501. We also characterize virPphA as the first (A) gene cloned for virulence functions from Pph.

MATERIALS AND METHODS

Bacterial Strains and Plasmids.

Sources of races of Pph are listed in ref. 1. The E. coli strain DH5α was used as host for clones in vectors pBluescript, pLAFR3, or pBBR1MCS-2 (2, 30).

DNA Extraction and Manipulation.

Total DNA was extracted by using a Puregene DNA isolation kit (Gentra systems, distributed by Flowgen, Lichfield, Staffs., U.K.). Plasmid DNA was extracted either by alkaline lysis miniprep (31) or with a midiprep extraction kit (Qiagen, Chatsworth, CA). Restriction endonucleases, ribonuclease A, and buffers (GIBCO/BRL) were used according to manufacturer’s instructions. Whole plasmids were extracted according to the method of Moulton et al. (32) and separated in 0.5% molecular biology agarose (Appligene, Strasbourg, France) midi gels for 4–5 h at 100 V, or for preparation of probe DNA in 0.5% ultra-pure low-melting point agarose (GIBCO/BRL). A gene library was constructed in the cosmid vector pLAFR3 (2).

Hybridization.

Separated DNA was transferred to Hybond N⁺ nylon membrane (ICN) by vacuum blotter (Appligene). Colony blotting was as described by Sambrook et al. (33), using BIOTRANS nylon colony blot. Probe DNA usually was purified by using a QIAEX II gel extraction kit (Qiagen), but pAV511 was excised from low-melting point gels. Restricted/amplified DNA was radiolabeled by using a High Prime radiolabeling kit (Boeringher Mannheim), and pAV511 was radiolabeled (1 h) in agarose by using Ready-to-Go beads (Amersham Pharmacia). Blots were hybridized (65°C, 16 h) with the labeled probes in hybridization solution (33), and then washed to high stringency (34).

Plasmid Transfer Procedures and Curing. Plasmid constructs were transferred from E. coli to rifampicin-resistant recipient strains of Pseudomonas by using a replica plate triparental mating procedure (35). Electroporations were carried out with a Gene Pulser (Bio-Rad) (36). Genomic clones were mutagenized by transposon insertion using Tn3 gus (37), and marker exchange mutants were obtained as described (in ref. 2).

Plasmid pPPY51 (which carries the replication gene from pAV505, a ca. 140-kb plasmid from race 4 strain 1302A; ref. 38) was electroporated into cells of 1449B to cure the homologous plasmid pAV511. Single-colony electroporants selected on King’s B agar plates (with 100 μg/ml ampicillin) were immediately subcultured to King’s B + ampicillin broth and grown overnight at 25°C. A loop full of the broth was further streaked on King’s B + ampicillin plates, and the plasmid profile of strains from single colonies was examined. Plasmid pPPY51 was cured from Pseudomonas cells by using a cold shock method, 7 days at 4°C (39).

PCR and DNA Sequencing.

Standard PCRs were performed with SuperTaq DNA polymerase and buffer (HT Biotechnologies, Cambridge, U.K.) by using either a DNA Thermal Cycler 480 or Gene Amp 2400 (Perkin–Elmer). Automated DNA sequencing was performed with an ABI 310 sequencer (Perkin–Elmer), and sequences were analyzed with a dnastar program. Database comparisons were made via the blast algorithm (40).

Pathogenicity Tests.

Bacteria were tested for pathogenicity and avirulence in pods and leaves of bean cultivars (41). For assessment of the growth of bacterial populations in leaves, inoculations with suspensions of 5 × 10⁸ colony-forming units/ml were made into unifoliate leaves, and tissue samples were taken with a 0.6-cm diameter borer (42). Tissues were homogenized in 10 mM MgCl₂ and dilutions were plated on selective medium. Pathogenicity tests in Phaseolus lunatus L., Pisum sativum L., and Nicotiana tabacum L. were as described (1, 2, 43), and on soybean (Glycine max L.) plants were inoculated by using a 1-ml syringe (without needle) to infiltrate bacterial suspensions of 5 × 10⁷ colony-forming units/ml into the underside of fully expanded primary leaves.

RESULTS

Strains of Race 7 Cured of the 154-kb Plasmid Have Reduced Virulence. Plasmid profiles from races of Pph are illustrated in Fig. 1A. The 154-kb plasmid pAV511 was cured from race 7 strain 1449B by the introduction of pPPY51, which carries a similar replication gene. Examination of plasmid profiles in several electroporants indicated loss of pAV511 (Fig. 1B). To confirm that the plasmid had been cured, and not forced into a recombination event with the chromosome, total and plasmid DNA were extracted and probed with the plasmid-borne avrPphF gene and the rep region on pPPY51. No hybridization was observed with either of the plasmid probes, and strains such as RW60 were used as plasmid-cured derivatives in further experiments.

Figure 1.

Curing the 154-kb plasmid from Pph race 7 strain 1449B. (A) Plasmid profiles from races of Pph; M gives profile from E. coli 39R861 that contains plasmids of known size (44). (B) The 154-kb plasmid pAV511 was cured from 1449B by using the rep gene cloned in pPPY51. Both RW50 and RW60 have been cured of pAV511, but RW50 retains pPPY51. The smaller plasmid in 1449B, RW50, and RW60 is designated pAV512. Chr, chromosomal DNA.

The cured strains were strikingly altered in virulence as summarized in Table 1. The appearance of inoculation sites in pods of differential cultivars is illustrated in Fig. 2A. In cvs. Canadian Wonder and Tendergreen, race 7 is normally fully virulent, causing water-soaked lesions, but the cured strains caused brown HR-like lesions. In cv. Red Mexican, loss of plasmid-borne avrPphF was expected to result in a shift to virulence but instead RW60 caused restricted orange/brown lesions phenotypically different from the wild-type HR (Fig. 2B). In contrast to the altered responses observed in other cultivars, in A43 loss of the plasmid had no effect on appearance of the HR governed by the chromosomal avrPphE and R2 gene interaction. The same pattern of loss of virulence in cured strains was apparent in pathogenicity tests on bean leaves and also on the alternative host P. lunatus (lima bean) cv. King of the Garden. However in nonhost tobacco and pea (as in bean cv. A43) an HR similar to that caused by the wild type was induced.

Table 1.

Reactions of strains on bean and soybean

Strain	Bean cultivar			Soybean cultivar
	Canadian Wonder	Tendergreen	Red Mexican	Osumi	Choska
P.s. pv. glycinea	HR	HR	HR	S	S
Pph race 7	S	S	HR*	HR	S−
RW60	HR	HR	HR†	N	N
RW60 (pAV518)	S−	S−	S−	HR	N
RW60 (pAV533)	S−	S−	S−	HR	N
Race 7∷virPphA	HRd	HRd	HR*	S−	S−
Tn3gus

S, fully susceptible water-soaked lesion; S−, slower development of lesions than S; HRd, delayed HR after initial water-soaking; N, null reaction, no cell browning. 

^*HR characteristic of the avrPphF/R1 interaction. 

^†HR in pods more orange in color than ∗. 

Figure 2.

Reaction phenotypes in bean pods and soybean leaves inoculated with Pph. (A) Pods of bean cultivars (from left to right): A43, Canadian Wonder, Red Mexican, and Tendergreen 3 days after stab-inoculation with (top to bottom site): Pph race 7 strain 1449B, cured strain RW60, complemented strains RW60 (pAV521) and RW60 (pAV518), and RW60 (pLAFR3) vector control. Note that pAV521 carries avrPphF, the avr gene matching resistance gene R1 in Red Mexican. (B) Time course of HR development in pods of cv. Red Mexican photographed 1, 2, 3, and 5 days after inoculation with 1449B (Upper) and RW60 (Lower); note the more orange lesion caused by the plasmid-cured strain. (C) Leaves of soybean cv. Osumi 10 days after inoculation with (left to right) 1449B, RW60, and RW60 (pAV518).

To determine whether the HR-like reaction caused after loss of pAV511 was hrp dependent, insertion mutants in hrpF (Tn3 gus; ref. 2) and hrpA (nonpolar mutation; G.T., unpublished work) were created in race 7, and the mutants then were cured of pAV511 as previously described. The hrp mutants of race 7 or the cured strain all failed to cause the HR or water soaking in any bean cultivar or nonhost tested. Delivery of signals leading to the formation of lesions by RW60 therefore depended on the presence of the hrp secretion system.

Cloning Virulence Factors.

Strains lacking pAV511 were able to produce proteases, extracellular polysaccharides, and phaseolotoxin to the same levels as the wild-type race 7 (data not shown). Production of these potential pathogenicity factors therefore was not compromised by loss of the 154-kb plasmid. To clone the putative vir genes from pAV511, a genomic library of race 7 was constructed by using the broad host range cosmid pLAFR3. Clones harboring plasmid DNA were identified by probing colony blots with pAV511. The 48 clones selected then were mobilized by triparental mating into strain RW60, and transconjugants were tested for their ability to restore virulence on bean pods. Nine cosmid clones modified the plant’s response to RW60, and they fell into two groups based on the phenotypes conferred. One group, typified by pAV521, was able to restore water-soaking ability to RW60 in cv. Tendergreen, and to a lesser extent cv. Canadian Wonder, and also restored the avrPphF/R1 phenotype of the HR in cv. Red Mexican, suggesting that avrPphF was present in these clones. The second group, typified by pAV518, conferred the ability to cause water-soaked lesions in cvs. Canadian Wonder, Tendergreen, and Red Mexican. In pods, the initial development of lesions caused by complemented RW60 in cvs. Canadian Wonder and Tendergreen was as rapid as observed with wild-type isolates of Pph, as shown in Fig. 2A. However, in contrast to fully virulent strains, after 5 days the water-soaked lesions caused by transconjugants of RW60 often developed some browning, leading to the S-classification used in Table 1. In leaves, RW60 caused a rapid HR in each cultivar and complementation (in the absence of avrPphF) was indicated by the slow development of susceptible symptoms, as quantified for cvs. Red Mexican and Tendergreen in Fig. 3. Restoration of virulence in transconjugants of RW60 harboring pAV518 and pAV521 also was demonstrated by increases in bacterial populations at inoculation sites (Table 2).

Figure 3.

Development of the HR and susceptible tissue collapse in leaves of cvs. Red Mexican and Tendergreen inoculated with suspensions of 2 × 10⁸ cells/ml of race 7 strain 1449B (black bars), cured strain RW60 (white bars), complemented strains RW60 (pAV521) (light gray bars) and RW60 (pAV518) (dark gray bars). Symptom scores assigned to infiltrated tissue were: 0, no reaction; 1, partial glazing; 2, 100% glazing over the inoculation site; 3, up to 50% tissue collapse; 4, between 50 and 100% collapse, 5, 100% collapse; 6, 100% collapse and browning, as observed during the HR. Data are the mean scores from sites in two leaves. Note that pAV521 carries avrPphF, the avr gene matching resistance gene R1 in Red Mexican.

Table 2.

Numbers of bacteria (×10⁻⁵) recovered from inoculation sites in leaves 4 days after inoculation

Cultivar	Bacterial strain*
	Race 7	RW60 pLAFR3	RW60 pAV518	RW60 pAV521
Canadian Wonder	92.0  ± 16.0	6  ± 0.7	43.1  ± 16.0	15.1  ± 5.0
Red Mexican	0.14	1.23  ± 0.3	12.5  ± 1.3	0.08
Tendergreen	139.5  ± 0.4	0.27  ± 0.1	33.6  ± 0.5	38.5  ± 2.0

Bacterial numbers are given per 0.6-cm diameter disc of leaf as means from three replicates ± SEM unless <0.001. 

^*Data from interactions giving the HR are italicized (see Table 1). 

Mapping, Transposon Mutagenesis, and Subcloning. Restriction mapping and hybridization experiments showed that all clones that restored virulence contained part of a region of about 30 kb, which was covered by pAV518 and pAV521. PCR and hybridization analysis revealed the presence of avr genes avrD and avrPphC as well as avrPphF within this region (Fig. 4). The two genomic clones were mutagenized with Tn3 gus to localize genes responsible for the partial restoration of virulence; more than 200 separate insertions were tested in each cosmid. Transposons located to several different sites within pAV518 abolished the restoration of virulence by the clone. In pAV521, only two insertions were similarly effective and sequencing located these in avrPphF and just downstream of the gene as indicated in Fig. 4.

Figure 4.

Characterized region of Pph plasmid pAV511 found to contain the PAI. The DNA common to genomic clones pAV518 and pAV521 is marked by the underlying line. The positions of single Tn3 gus insertions that abolished restoration of virulence by pAV521 (gray arrows) and pAV518 (black arrows) are marked. Note that insertions that did not affect virulence were not mapped.

Fragments carrying the regions implicated in the restoration of virulence in pAV518 and pAV521, which were bounded by convenient restriction sites, were subcloned into pLAFR3 or pBBR1MCS-2 for expression in Pph. A 1.8-kb BamHI fragment (designated pAV525), harboring a functional avrPphF gene, restored some water-soaking ability to RW60 on cv. Tendergreen. From pAV518, a 6.7-kb EcoRI fragment (designated pAV530) successfully restored pathogenicity to RW60 on cvs. Canadian Wonder, Tendergreen, and Red Mexican. The combination of insertion mutagenesis and subcloning therefore located determinants of virulence to specific regions in pAV518 and pAV521. Although some subclones significantly restored virulence, none was as fully effective as the genomic clones. For example, in pods of cv. Tendergreen RW60 (pAV530) caused water-soaked lesions but they developed slightly more slowly than those caused by RW60 (pAV518). The potential vir genes located by Tn3 gus mutagenesis appeared to have additive effects.

Sequencing vir Genes in pAV518. The contiguous 0.9-, 3.4-, and 6.7-kb EcoRI fragments from the right of pAV518 contained the site of several Tn3 gus insertions that abolished the restoration of pathogenicity by the genomic clone (Figs. 2 and 3). The 6.7-kb EcoRI fragment cloned as pAV530 possessed ability to restore virulence to RW60. Sequencing the 11-kb region of DNA revealed the presence of four possible ORFs, each of which was preceded by an upstream hrp box motif (indicating potential regulation by HrpL; ref. 45) as summarized in Table 3.

Table 3.

Features of the four ORFs other than transposases located in the PAI

	Putative promoter region*	Position relative to ORF†	Shine-Dalgarno sequence and start codon‡	G+C%§	Predicted protein size
ORF1¶	GGAACC-15N-CCAA	−63	GGAGAGTCTATATG	54.01	59.6kDa
ORF2	AGAAGC-15N-CCAC	−94	GCGCGAGCTGGTG	59.39	22.9kDa
ORF3	GGAACT-15N-CCAC	−31	GAGGATATGCGGTG	51.94	10.3kDa
ORF4	GGAACT-15N-CCAC	−59	GGAGAAAATCAGCATATG	53.09	35.4kDa

^*Nucleotides in bold indicate those sequences that are conserved according to the “hrp box” consensus (45). 

^†Position relative to translation start is given relative to the position of the 3′ C in the putative HrpL promoter sequence. 

^‡Nucleotides in bold indicate putative Shine-Dalgarno sequences, start codons are underlined. 

^§The genome of P.syringae pvs. is reported to be 59–61% G+C (46). 

^¶Designated virPphA. 

Transposon insertions that compromised virulence were located to sites within ORF1 and ORF4 and the putative promoter region of ORF3 (Fig. 5). The ORFs were subcloned to analyze their function. The smallest region achieving consistent restoration of water-soaking ability was a 1.8-kb NsiI–SspI fragment containing only ORF1, cloned as pAV536 (Fig. 5). The gene located as ORF1 therefore was designated virPphA as the first (A) gene for virulence cloned from Pph. Although clones containing ORF4 alone did not restore water-soaking ability, they did cause a clear delay in the onset of the HR normally observed in pods inoculated with RW60 (data not shown). No such effect was observed with ORFs 2 or 3 (Fig. 5). For details on subclones see Table 4, which is published as supplemental data on the PNAS web site, www.pnas.org.

Figure 5.

Location of ORFs for ORF1 (designated virPphA), ORF2, ORF3, and ORF4 within the 11 kb of DNA sequenced. The directions of transcription are indicated by horizontal arrowheads. Sites of Tn3 gus insertions abolishing restoration of virulence by pAV518 are indicated by vertical arrows. Sequences found to have similarity to IS100 and Tn501 are indicated. Clone pAV530 containing the 6.7-kb EcoRI fragment restored virulence almost as effectively (+++) as the genomic clone pAV518 (scored as ++++); pAV533 and pAV536 were slightly less effective. Clone pAV535, containing ORF4, consistently delayed the onset of the HR caused by RW60, a reaction scored as +.

The proteins predicted to be encoded by virPphA and the other potential vir genes represented by ORF3 and ORF4 are hydrophilic and lack similarity to other proteins in databases. As shown in Fig. 5, repeated and inverted sequences of 1,053 bp with predicted similarity to the IS100 transposase ORF from Y. pestis (64.8% amino acid identity; ref. 47) were located at either side of ORF4, flanking a region that also contained a homolog of the Tn501 transposase from Pseudomonas aeruginosa (62.5% amino acid identity; ref. 48). The G+C content of the sequenced region of pAV511 was 54%, significantly lower than the overall figures of 59–61% reported for pathovars of P. syringae (46). Individual ORFs, except ORF2, also had low G+C percentages (Table 3).

virPphA Acts as an avr Gene in Soybean.

Race 7 of Pph typically causes a rapid HR in soybean leaves but in certain cultivars such as Bragg or Choska it was found to be weakly virulent, causing symptoms similar to those produced by the recognized soybean pathogen P. syringae pv. glycinea. The plasmid-cured strain RW60 caused a null response on all cultivars of soybean tested. Induction of the HR was restored by the genomic clone pAV518 and subclones containing virPphA (Table 1). Only the transposon insertions in pAV518 that were located in virPphA prevented effects on soybean. Therefore, virPphA appeared to act as a typical avr gene, controlling expression of a rapid HR in soybean. Phenotypes observed are illustrated in Fig. 2C. The significance of virPphA in controlling host range was confirmed by the finding that virPphA mutants of strain 1449B, created by marker exchange of transposons L20 or J37 (Fig. 5), were able to colonize soybean. By contrast the 1449B∷virPphA Tn3 gus strains lost virulence to pods of bean cvs. Canadian Wonder and Tendergreen, producing a delayed HR phenotype after initial water soaking (Table 1).

DISCUSSION

Loss of the largest plasmid (pAV511) from Pph caused striking changes in virulence to bean. In particular, the cured strains of race 7, such as RW60, caused the HR in previously fully susceptible cvs. Canadian Wonder and Tendergreen (Fig. 2A). Elicitation of the HR by RW60 depended on the presence of a functional type III secretion system, which is believed to deliver Avr proteins that have been identified because of their interaction with matching R genes in the plant. Removal of the plasmid-encoded virulence factors therefore has revealed the potential presence of a second tier of “masked” avr genes. An intriguing question is how the vir genes may act to block phenotypic expression of avr genes, which function after loss of the pAV511 plasmid.

Possible routes to virulence are outlined in Fig. 6, in which avr genes functioning despite the presence of Vir factors are designated α and those revealed after plasmid loss β. Three modes of action of Vir factors are suggested: (i) β avr gene suppression, (ii) blocking β Avr protein transfer, and (iii) interference with signal transduction leading to the HR and other defense responses. In view of the ability of virPphA to act as an α avr gene in soybean and the presence of the HrpL promoter motif as found upstream of other avr genes, it seems probable that Vir proteins, like Avr proteins, function in the plant cell. Route 3 shown in Fig. 6 therefore is the most likely pathway to virulence. Characterization of the β avr genes will allow this possibility to be fully explored. Interference between avr gene products outside the bacterium already has been proposed to explain the epistasis of avrRpt2 over avrRpm1 in the P. syringae/Arabidopsis interaction (49).

Figure 6.

Possible mechanisms by which virulence (Vir) factors block the HR in bean caused by avr gene activity, which is masked in wild-type Pph. The masked avr genes are designated β. Routes to virulence are via 1) suppression of β avr gene expression; 2) blocking delivery of the encoded β avr protein, or 3) interference with generation of the HR after recognition. By contrast, α avr genes, such as avrPphE, are not affected.

Several features of the region controlling virulence characterized on pAV511 indicate that we have identified a PAI as defined by Hacker et al. (14). The region contains several vir genes; it is associated with transposases and has G+C content significantly lower than the genome overall (46). The presence of transposases with similarity to IS100 from the virulence plasmid of Yersinia pestis (48, 50) and Tn501 from P. aeruginosa (51) is particularly significant. The region containing IS100 homologs at its left and right borders, enclosing both the Tn501 transposase and the putative vir locus, ORF4, has the potential to act as a transposon in Pph. The association of avr genes with mobile elements of DNA either in the form of IS or Tn elements, or by location on plasmids, recently was highlighted by Kim et al. (52). They point out that three of four avr genes from P. syringae previously shown to have some Vir function, avrA, avrB, avrE, and avrRpm1, are positioned close to sequences related to transposable elements, in the latter case including Tn501. In the 154-kb plasmid of Pph (pAV511) the cluster of avr and vir genes would appear to have the potential to be highly mobile.

The experiments with soybean indicate a clear link between avr and vir gene function. We have identified virPphA as a gene that acts as a virulence determinant in one plant (bean) but an activator of resistance (HR) in another species (soybean). Clearly the nomenclature of genes with such dual functions is rather confusing! At present we propose to name genes on the basis of their first characterized function, therefore vir is applied to virPphA, which does not appear to act as an avr gene in bean. Several avr genes have activities that cross plant species barriers; for example, avrPphB and avrPpiA match R genes in bean, pea, soybean, and Arabidopsis, but in each plant they control the HR and not susceptibility (3, 8, 19). There are similarities between the properties of virPphA and the disease-specific (dsp) locus in Erwinia amylovora that comprises genes designated dspE and dspF (53) or dspA and dspB (54). The dsp genes are absolutely required for pathogenicity but not for elicitation of the HR. Mutation in the dsp locus causes E. amylovora to behave as a hrp mutant in pear and reduce the severity of tissue collapse during the HR in tobacco. The dspEF locus is homologous to the avirulence locus avrE in P. syringae pv. tomato, which determines host range, its presence leading to a strong HR in certain soybean cultivars (11).

An intriguing aspect of the complementation of plasmid-cured strains is the quantitative nature of virulence restoration achieved. Thus, genomic clones such as pAV518 apparently containing several vir genes were more effective than subclones or virPphA alone. It is envisaged that there is redundancy and multifactorial control of virulence and that Vir factors may interact with different targets, each of which contributes to the establishment of the resistance response. Such a quantitative interaction indicates the presence of targets within the plant that may have additive effects on the activation of the HR leading to resistance. Previous analysis of HR induction in response to Avr proteins has indicated a single recognition interaction, such as AvrPto binding to the resistance gene product Pto, and a subsequent signaling cascade leading to the HR. Variation in response has been attributed to affinity of binding between the Avr protein and its target (22), or regulation via modification of the signal transduction pathway and the so-called recognition rheostat (55, 56).

Our results reinforce the emerging similarities between plant and animal pathogens both in terms of the presence of virulence factors and their location on pathogenicity islands. In addition to the common type III secretion system there may be functional similarities in the use of virulence factors to subvert defense responses. For example, the Shigella invasion plasmid antigen B (IpaB) recently has been found to control apoptosis in macrophages (57). The HR is thought to be a form of programmed cell death (PCD) in plants with some similarities to apoptosis, including the involvement of caspases in its execution (58, 59). Shigella-induced apoptosis depends on the binding of IpaB to caspase-1 (57). Similar targets in PCD pathways in plants may allow subversion of the HR and lead to disease development as a result of the activity of Vir factors such as those encoded by virPphA.

ABBREVIATIONS

avr

avirulence gene

vir

virulence gene

HR

hypersensitive reaction

PAI

pathogenicity island

Pph

Pseudomonas syringae pv. phaseolicola

R

resistance gene

dsp

disease-specific gene