Homology and functional similarity of an hrp-linked pathogenicity locus, dspEF, of Erwinia amylovora and the avirulence locus avrE of Pseudomonas syringae pathovar tomato

Abstract

The “disease-specific” (dsp) region next to the hrp gene cluster of Erwinia amylovora is required for pathogenicity but not for elicitation of the hypersensitive reaction. A 6.6-kb apparent operon, dspEF, was found responsible for this phenotype. The operon contains genes dspE and dspF and is positively regulated by hrpL. A blast search revealed similarity in the dspE gene to a partial sequence of the avrE locus of Pseudomonas syringae pathovar tomato. The entire avrE locus was sequenced. Homologs of dspE and dspF were found in juxtaposed operons and were designated avrE and avrF. Introduced on a plasmid, the dspEF locus rendered P. syringae pv. glycinea race 4 avirulent on soybean. An E. amylovora dspE mutant, however, elicited a hypersensitive reaction in soybean. The avrE locus in trans restored pathogenicity to dspE strains of E. amylovora, although restored strains were low in virulence. DspE and AvrE are large (198 kDa and 195 kDa) and hydrophilic. DspF and AvrF are small (16 kDa and 14 kDa) and acidic with predicted amphipathic α helices in their C termini; they resemble chaperones for virulence factors secreted by type III secretion systems of animal pathogens.

Erwinia amylovora causes fire blight of apple, pear, and other rosaceous plants and elicits plant defense responses in nonhost plants. Required for these interactions are the clustered bacterial hrp genes, encoding regulatory proteins (ref. 1; Z.M.W., B. J. Sneath, and S.V.B., unpublished data), a large set of proteins broadly conserved among plant and animal pathogens and constituting a type III secretion pathway (known as the “Hrp pathway” in phytopathogenic bacteria; refs. 2 and 3), and at least two proteins secreted via the Hrp pathway (4, 5). hrp genes, present in all Gram-negative necrogenic plant pathogens, were discovered by transposon mutagenesis of Pseudomonas syringae pathovars and were named for the “hypersensitive reaction” (HR) and “pathogenicity” (reviewed in ref. 6). The HR is a manifestation of plant defense characterized by rapid necrosis at the site of pathogen ingress.

Pathogen avirulence (avr) genes (for a review see ref. 7) generate signals that trigger defense responses leading to disease resistance in plants with corresponding resistance (R) genes. Typically, avr genes are isolated by expressing a cosmid library from one pathogen in another pathogen and screening for narrowed host range. avr genes traditionally have been considered as negative determinants of host specificity at the race-cultivar level, but some, including the avrE locus from the bacterial speck pathogen Pseudomonas syringae pathovar (pv.) tomato (8), may restrict host range at the pathovar–species or species–species level (9, 10). Many avr genes, including avrE, are hrp regulated. avrE and avrPphE (11) are physically linked to hrp genes. Only a few avr genes (such as avrE), however, play detectable roles in pathogen fitness or in virulence in hosts tested (12–16), and the selective force driving the maintenance in pathogen genomes of many of these host-range-limiting factors has remained a mystery.

When expressed in trans, the avrE locus renders P. syringae pv. glycinea, which causes bacterial blight of soybean, avirulent in each of 10 tested cultivars (17). The locus comprises two convergent transcription units, one preceded by a putative σ⁵⁴ promoter and the other by a hrp box (17, 18), a sequence found upstream of many hrp and avr genes that are positively regulated by the alternate sigma factor HrpL (1, 18). Expression of both transcripts requires hrpL. The avrE locus contributes quantitatively to the virulence in tomato leaves of P. syringae pv. tomato strain PT23, but not of strain DC3000 (15, 17).

Transposon mutagenesis of E. amylovora revealed, linked to the hrp gene cluster, a “disease specific” (dsp; see ref. 19) region required for pathogenicity but dispensable for HR elicitation. Through sequencing and further mutagenesis, we have defined a two-gene apparent operon, the dspEF locus, responsible for this phenotype. Here, we present an analysis of the genes, including the finding that they are homologous with genes in the avrE locus. In addition, we show that the dspEF locus converts P. syringae pv. glycinea to avirulence in soybean, and that avrE restores pathogenicity to dsp mutant strains of E. amylovora. We discuss the implications of these findings with respect to the nature, evolution, and potential usefulness of bacterial genes encoding proteins involved in infection of plants.

MATERIALS AND METHODS

Recombinant DNA Techniques.

DNA was isolated, cut by using restriction enzymes, and ligated, and transformed into Escherichia coli according to procedures described by Sambrook et al. (20). A P. syringae pv. tomato DC3000 genomic library was constructed and screened by using colony hybridization also as described (20). The library was constructed by using pCPP47, a low-copy-number, broad-host-range cosmid (21). Except where noted, E. coli DH5 and E. coli DH5α were used as hosts for DNA clones, and pBluescript or pBC plasmids (Stratagene) were used as vectors. E. amylovora was transformed by electroporation as described (22). Plasmids were mobilized into E. amylovora and P. syringae by using pRK2013 (23).

Nucleotide Sequencing and Analysis.

The nucleotide sequence of the dsp region of E. amylovora strain Ea321 was determined by using subclones of pCPP430 (24). The nucleotide sequence of the avrE locus was determined by using subclones of pCPP2357, a clone selected from a P. syringae pv. tomato DC3000 genomic cosmid library based on hybridization with the hrpRS operon of P. syringae pv. syringae, and the finding, based on partial sequencing, that it contained the avrE locus. Nucleotide sequencing was performed by the Cornell Biotechnology Sequencing Facility on a Model 377 Sequencer (Perkin–Elmer/Applied Biosystems Division, Foster City, CA). Sequence analyses were performed by using the programs of the gcg 7.1 software package (Genetics Computer Groups, Madison, WI) and dnastar (DNAstar, Madison, WI). Database searches were performed by using blast (25).

Expression of DspE and DspE′ in E.

coli. The dspEF locus was cloned in two pieces into pCPP50, a derivative of pINIII¹¹³-A2 (26) with an expanded polylinker (D. W. Bauer and A.J.B., unpublished data), yielding pCPP1259. Expression in pCPP1259 is driven by the lpp promoter of E. coli, under the control of the lac operator. An intermediate clone, pCPP1244, extending from the start of the locus to the BamHI site in the middle of dspE, also was isolated. E. coli DH5α strains containing pCPP1259 and pCPP1244 were grown in Luria–Bertani (LB) medium at 37°C to an OD₆₂₀ of 0.3. Isopropylthio-β-d-galactoside then was added to 1 mM, and the cells further incubated until reaching an OD₆₂₀ of 0.5. Cells were concentrated 2-fold, lysed, and subjected to SDS/PAGE as described (20). Cells containing pCPP50 were included for comparison. Proteins were visualized by Coomassie blue staining.

Deletion Mutagenesis of dspE.

We deleted 1,554 bp from the 5′ HindIII-BamHI fragment of dspE in pCPP1237 by using unique StuI and SmaI sites. The mutagenized clone then was inserted into the suicide vector pKNG101 (27) by using E. coli SM10λpir as a host, yielding pCPP1241. The mutation, designated Δ1554, then was transferred into E. amylovora strains by using marker eviction as described previously (2). By using two BstEII sites blunted with Klenow fragment, 1,521 bp were deleted from the 3′ HindIII fragment of dspE in pCPP1246. This mutation, Δ1521, was transferred into E. amylovora strains as above.

Pathogenicity Assays.

For E. amylovora strains, cell suspensions of 5 × 10⁸ colony-forming units (cfu) per ml were pipetted into wells cut in immature Bartlett pear fruit, or stabbed into Jonamac apple and cotoneaster shoot apices, and assays were carried out as described previously (28, 29). For P. syringae pv. glycinea strains, panels of primary leaves of 2-week-old soybean seedlings (Glycine max, cultivar Norchief) were infiltrated with bacterial suspensions of 8 × 10⁵ cfu/ml as for the HR assay, below. Plants were then covered with clear plastic bags for 1 day and incubated under fluorescent lights (16 hr/day) at 22°C for 5–7 days. Leaves were scored for necrosis and chlorosis.

Bacterial Population Assays.

Cotoneaster shoot tips, 10 cm long, that had been inoculated with E. amylovora strains were homogenized in 5 mM KPO₄ buffer, pH 6.8, at 5 days postinoculation. Inoculated pear fruits were homogenized at 7 days postinoculation. Homogenates were plated in a dilution series on LB agar with antibiotics (rifampicin, 25 μg/ml; tetracyclin, 10 μg/ml; kanamycin, 50 μg/ml) as appropriate to determine bacterial populations. Triplicate shoots or fruits were assayed individually for each strain tested.

HR Assays.

Tobacco leaf panels (Nicotiana tabacum L. ‘xanthi’) were infiltrated with bacterial cell suspensions as described previously (4, 30). Primary leaves of 2-week-old soybean seedlings (secondary leaves emerging) were infiltrated with bacterial cell suspensions as for tobacco. Plants were scored for HR (tissue collapse) after 24–48 hr on the laboratory bench. E. amylovora strains were suspended in 5 mM KPO₄ buffer, pH 6.8, and P. syringae strains in 10 mM MgCl₂.

GUS Assays.

Cells were (i) grown in LB to an OD₆₂₀ of 0.9–1.0; (ii) grown in LB to an OD₆₂₀ of 0.5, then washed and resuspended in an hrp-gene-inducing minimal medium (Hrp MM; ref. 31) to an OD₆₂₀ of 0.2 and incubated at 21°C for 36 hr to a final OD₆₂₀ of 0.9–1.0; or (iii) grown in LB to an OD₆₂₀ of 0.5, washed and concentrated 2-fold in 5 mM KPO₄ buffer, pH 6.8, and then transferred to freshly cut wells in pear halves and incubated as for the pathogenicity assay for 36 hr. Cells were assayed for β-glucuronidase (GUS) activity essentially according to Jefferson (32). For the cells in LB or Hrp MM, 50 μl were mixed with 200 μl GUS extraction buffer (50 mM NaHPO₄, pH 7.0/10 mM 2-mercaptoethanol/10 mM Na₂EDTA/0.1% sodium lauryl sarcosine/0.1% Triton X-100) containing 2 mM 4-methylumbelliferyl β-d-glucuronide as substrate and incubated at 37°C for 100 min. For cells in pear fruit, the tissue surrounding the well was excised by using a #4 cork borer and homogenized in 5 mM KPO₄ buffer, pH 6.8. Two hundred microliters of homogenate was mixed with 800 μl of GUS extraction buffer with substrate and incubated as above. Reactions were stopped by adding Na₂CO₃ to a final concentration of 0.2 M in a total volume of 2 ml. Fluorescence was measured by using a TKO 100 Mini-Fluorometer (Hoefer). For all samples, cell concentration was estimated by dilution plating, and fluorometric readings were converted to pmol of substrate hydrolyzed per 10⁸ cfu/min, after Miller (33).

RESULTS

The “Disease-Specific” (dsp) Region of E. amylovora Consists of a 6.6-kb, Two-Gene Apparent Operon.

Mapping of previous transposon insertions (ref. 34; C. H. Zumoff, D. W. Bauer, B. J. Sneath, Z.M.W., and S.V.B., unpublished data) that abolish pathogenicity but not HR-eliciting ability confirmed the presence of the “disease-specific” (dsp) region downstream of the hrpN gene in strain Ea321 as reported in strain CFBP1430 (19). The sequence of approximately 15 kb of DNA downstream of hrpN from Ea321 was determined, revealing several ORFs (Fig. 1). One large ORF was found that encompassed the region to which all our dsp insertions mapped. This ORF was present in an apparent 6.6-kb operon containing another, smaller ORF downstream. The two ORFs were designated dspE and dspF, and the operon, the dspEF locus. dspE is preceded (beginning 70 bp upstream of the initiation codon) by the sequence GGAACCN₁₅CAACATAA, which matches the HrpL-dependent promoter consensus sequence, or “hrp box” of E. amylovora (1, 3) and strongly resembles the hrp box of P. syringae hrp and avr genes (18). Immediately downstream of dspF is A/T-rich DNA, followed by an ORF highly similar to the Salmonella typhimurium gene spvR, a member of the lysR family of regulatory genes (35). Immediately upstream of the dspEF locus is an Hrp-regulated gene, hrpW, encoding a second harpin (5).

Figure 1.

The dspEF locus of E. amylovora: mutagenesis, complementation and heterologous expression constructs, and homology with and restoration of mutants by the avrE locus of P. syringae. Dashed boxes are uncharacterized ORFs; a solid triangle indicates an hrp box, and an open triangle indicates another promoter. Thick lines delineate the DNA for which sequence was accessioned. (A) The dsp/hrp gene cluster of E. amylovora in pCPP430. Operon names and types of proteins encoded are indicated at the top. B, BamHI; E, EcoRI; H, HindIII. Half-arrows indicate internal promoters without similarity to the hrp box consensus. (B) The region downstream of hrpN containing the dspEF locus. Circles mark deletion mutations and representative transposon insertions: black, nonpathogenic and HR⁺ or HR-reduced (dsp); gray, reduced virulence and HR; white, wild type. T104 lies within the area marked by the dashed double arrow. K, Tn5miniKm; P, Tn5phoA; T, Tn10tet^r; Δ, deletion mutation. The gray box is A/T-rich DNA. (C) Clones and subclones of the dspEF locus. Plasmid designations are indicated at the left, and vector-borne promoters are indicated at the right. Restriction sites used for subcloning not shown above are shown in parentheses. A “+” aligned with a circle representing a mutation in B indicates that the subclone complements that mutation. (D) The avrE locus (transcription units III and IV) of P. syringae pv. tomato DC3000 in pCPP2357. Percent amino acid identity of the predicted proteins AvrE and AvrF to DspE and DspF, respectively, are indicated. Solid rectangles are transcriptional terminators (inverted repeats). Ability to restore mutations depicted in B are indicated, aligned as for complementation data in C.

The deduced product of dspE contains 1,838-aa residues and is hydrophilic. The predicted molecular mass, 198 kDa, was confirmed by expression in E. coli (Fig. 2). Expression of an intermediate clone containing only the 5′ half of dspE yielded a protein of corresponding predicted mobility, suggesting that the N-terminal half of the protein might form an independently stable domain. DspF, predicted to be 16 kDa, acidic (pI, 4.45), and predominantly α-helical, with amphipathic α-helices in its C terminus, is physically similar to virulence factor chaperones of animal-pathogenic bacteria (36).

Figure 2.

Expression of the full-length and the N-terminal half of DspE in recombinant E. coli DH5α. Lysates of cells carrying either pCPP1259, containing the entire dspEF locus (lane A); pCPP50, the cloning vector (lane B); or pCPP1244, containing only the 5′ half of the dspE gene (lane C), were subjected to SDS/PAGE (7.5% acrylamide) followed by Coomassie staining. Bands corresponding to DspE (lane A) and the N-terminal half of DspE (lane C) are marked by arrows. Migration of molecular mass markers is indicated on the left.

dspE Is Required for Fire Blight.

Two in-frame deletions within dspE (Fig. 1) were made in Ea321 and Ea273 (low- and high-virulence strains, respectively). The first (Δ1554) corresponds to amino acid residues G₂₀₃ to G₇₂₀ and the second (Δ1521) to amino acid residues T₁₀₆₄ to V₁₅₇₀. Each deletion abolished the ability of both strains to generate fire blight symptoms (necrosis) and bacterial ooze when inoculated to immature pear fruit (Fig. 3). The mutants also failed to cause fire blight when inoculated to apple and cotoneaster shoots (not shown). Populations of dsp mutant strains isolated from cotoneaster shoots after 5 days were equivalent to that of a hrpL regulatory mutant strain (“K49”; ref. 1). The Δ1554 deletion mutants of Ea321 and Ea273 were restored to full virulence by pCPP1237, a clone carrying only the overlapping 5′ half of dspE, further suggesting that the N terminus of the protein forms a stable domain (Figs. 1 and 3).

Figure 3.

The role of the dspE gene in pathogenicity and HR elicitation. (A) Immature pear fruit 4 days after inoculation with (left to right) strains Ea321, Ea321dspEΔ1554, or Ea321dspEΔ1554 harboring the 5′ half of dspE on pCPP1237. (B) Norchief soybean leaf 24 hr after infiltration with 5 × 10⁸ cfu/ml suspensions of (1) Ea321, (2) Ea321dspEΔ1554, (3) Ea321hrpN∷Tn5 (ref. 4), and (4) Ea321hrpL∷Tn5 (ref. 1). (C) Tobacco leaf 48 hr after infiltration with parallel dilution series of suspensions of strains Ea321 (Left) and Ea321dspEΔ1554 (Right). The concentrations infiltrated (top to bottom) are 1 × 10¹⁰, 1 × 10⁹, 5 × 10⁸, 1 × 10⁸, and 5 × 10⁷ cfu/ml. (D) As for C, except the more virulent strain, Ea273, and corresponding mutant Ea273dspEΔ1554 were used, and concentrations ranged from 5 × 10⁹ to 5 × 10⁵ cfu/ml in log increments.

The dspEF Locus Contributes Quantitatively and in a Strain-Dependent Fashion to HR Elicitation by E. amylovora in Tobacco and Is Not Required for HR Elicitation by E. amylovora in Soybean.

Transposon insertions in the dsp region of E. amylovora strain Ea321 reduce the ability of this strain to elicit the HR in tobacco (data not shown). Dilution series of suspensions of dspEΔ1554 mutant strains of Ea321 and Ea273 were infiltrated into tobacco leaves alongside their wild-type parents to define precisely the role of dspE in HR elicitation (Fig. 3). All strains were capable of eliciting the HR, but Ea321dspEΔ1554, on a per-cell basis, was roughly one-tenth as effective as the wild type. pCPP1237 restored full HR-eliciting ability to this strain (not shown). There was no noticeable difference in HR-eliciting ability in tobacco between Ea273 and Ea273dspEΔ1554. Ea321dspEΔ1554, infiltrated at a standard concentration, elicited wild-type HR in Acme, Centennial, Harasoy, and Norchief soybean leaves (Fig. 3).

The dspEF Locus Is Hrp-Regulated.

A promoterless uidA gene construct (D. W. Bauer) was cloned downstream of the dspE fragment in pCPP1241 that was used to introduce the Δ1554 mutation (Fig. 1) into wild-type strains of E. amylovora (this construct consists of a 3′-truncated dspE gene with the internal deletion). The resulting plasmid, pCPP1263, was mobilized into Ea321 and Ea273. Pathogenic strains, in which plasmid integration had preserved an intact copy of dspE, and nonpathogenic strains, in which the native copy of dspE had been mutated, were isolated. All strains were assayed for GUS activity in LB and in Hrp MM, and pathogenic strains were assayed for activity in pear fruit. High levels of activity were obtained from strains incubated in Hrp MM and pear, but not LB. The level of expression in Hrp MM was equivalent to that of a hrcV-uidA fusion (“G73”; ref. 1) used as a positive control. There were no significant differences in levels of expression of the dspE-uidA fusion in the wild-type and dspE mutant backgrounds (data not shown), indicating that dspE likely is not autoregulated. Expression of the dspE-uidA fusion in hrpL mutants of Ea321 and Ea273 in Hrp MM was two orders of magnitude lower than that in HrpL⁺ strains. Data for Ea273 and derivatives are shown in Fig. 4.

Figure 4.

Expression of a promoterless GUS construct fused to dspE in E. amylovora Ea273. Ea273 and Ea273dspE∷uidA (a merodiploid containing both a wild-type dspE and a truncated dspE fused to the uidA gene; solid bars) were grown in LB or Hrp MM, or inoculated to immature pear fruit. Ea273dspE∷uidAhrpL∷Tn5 (darkly shaded bar) and Ea273hrcV∷Tn5uidA (lightly shaded bar) were also grown in hrp MM. Values shown represent means of triplicate samples normalized for bacterial cell concentration. Standard deviations are represented by lines extending from each bar. The mean values for three samples of Ea273 in each assay were subtracted from, and standard deviations added to, the corresponding values obtained for the other strains.

dspE and dspF Are Homologous with Genes in the avrE Locus of Pseudomonas syringae pv. Tomato.

A blast (37) search of the genetic databases revealed similarity in the dspE gene to a partial sequence of the avrE locus of P. syringae pv. tomato (17). A cosmid library of P. syringae pv. tomato DC3000 genomic DNA was constructed, and a clone overlapping the hrp gene cluster and containing the avrE locus was isolated (pCPP2357). The complete nucleotide sequence of the avrE locus was determined, revealing homologs of dspE and dspF (Fig. 1). The dspE homolog, alone in an operon previously designated transcription unit III, encodes a 195-kDa, 1,795-aa protein 30% identical to DspE. The dspF homolog, at the end of the opposing operon previously designated transcription unit IV, encodes a 14-kDa, 129-aa protein 43% identical to DspF. We designate these genes avrE and avrF, respectively. The aligned C-terminal halves of DspE and AvrE (starting from V₈₄₅ of DspE) show greater conservation (33% identity) than the N-terminal halves (26% identical). AvrE contains an ATP-/GTP-binding-site motif (“P-loop”; ref. 38) at residues A₄₅₀ to T₄₅₇ and a putative leucine zipper at residues L₁₇₇₂ to L₁₇₉₃. These features are not present in DspE, however, and their functional significance in AvrE, if any, is unclear. Amino acid identities are distributed equally throughout the DspF and AvrF alignment, and AvrF shares the predicted physical characteristics of DspF. Upstream of avrF, completing the operon, is a 2.5-kb gene with no similarity to sequences in the genetic databases.

The dspEF Locus Functions as an Avirulence Locus.

The dspEF locus was cloned into pML122 (39) downstream of the nptII promoter, and this construct, pCPP1250, was mobilized into P. syringae pv. glycinea race 4 (gift of N. T. Keen, Univ. of California, Riverside). The resulting strain, but not a control strain containing pML122, elicited the HR in soybean cultivars Acme, Centennial, Harasoy, and Norchief; in Norchief plants incubated under conducive conditions, race 4 harboring pCPP1250 did not cause symptoms of disease, whereas the control strain caused necrosis and chlorosis that spread from the point of inoculation (Fig. 5).

Figure 5.

Transgeneric avirulence function of the dspEF locus and restoration of dspE mutants with the avrE locus. Norchief soybean leaves were either (A) infiltrated with 1 × 10⁸ cfu/ml suspensions of P. syringae pv. glycinea race 4 carrying pCPP1250 (containing the dspEF locus) (Left) or pML122 (the cloning vector) and photographed after 24 hr at room temperature (Right), or (B) infiltrated with 8 × 10⁵ cfu/ml suspensions of the same strains and photographed after 7 days at 22°C and high relative humidity. Tissue collapse is apparent on both leaves where the strain carrying pCPP1250 was infiltrated. On the leaf incubated for 7 days, chlorosis extending beyond the infiltrated area, typical of disease, is apparent on the half infiltrated with the strain carrying the vector only. The dark section on the side of the leaf infiltrated with the strain carrying pCPP1250 is a shadow caused by a buckle in the leaf. Pear halves are shown (C) 10 days after inoculation with (left to right) Ea273, Ea273dspEΔ1521(pCPP2357, containing the avrE locus), or Ea273dspEΔ1521(pCPP2357avrE∷Tn5uidA), and (B) cross-sectioned through the well 10 days after inoculation with Ea321dspEΔ1521(pCPP2357) (Left) and Ea321dspEΔ1521(pCPP2357avrE∷Tn5uidA) (Right). Although greatly reduced relative to wild type, water soaking and necrosis are apparent around and ooze can be seen within the wells of fruit inoculated with dspE strains carrying the intact avrE locus. Fruit inoculated with dspE strains carrying a disrupted clone of avrE is symptomless and shows no ooze.

The avrE Locus Restores Pathogenicity to dspE Mutants.

Cosmids pCPP2357 (carrying the avrE locus) and pCPP2357avrE∷Tn5uidA (Fig. 1) were mobilized into dspEΔ1521 mutants of Ea273 and Ea321, and the resulting transconjugants, and wild-type strains, were inoculated to immature pear fruit (Fig. 5). Ea273dspEΔ1521(pCPP2357) cells increased in number 10-fold over 7 days, generating ooze, water soaking, and slight necrosis in and immediately surrounding the sites of inoculation. Virulence was much lower than that of wild-type cells, which increased 5 × 10³-fold, resulting in copious ooze and necrosis throughout the fruit. Ea273dspEΔ1521(pCPP2357avrE∷Tn5uidA) cells did not increase in number and generated no symptoms, indicating that the observed restoration of pathogenicity was avrE-specific. Similar results were observed for transconjugants of Ea321dspEΔ1521 (Fig. 5) and Ea321dspEΔ1554 (not shown).

DISCUSSION

We characterized the dspEF locus of E. amylovora and discovered that it is homologous with the avrE locus of P. syringae pv. tomato and is essential for E. amylovora pathogenicity. In contrast, avrE plays only a quantitative role in virulence in P. syringae pv. tomato strain PT23 and is completely dispensable in strain DC3000 (the source of the clone used here; refs. 15 and 17). We found that the dspEF locus and the avrE locus function similarly and function transgenerically: like avrE, the dspEF locus confers avirulence when expressed in P. syringae pv. glycinea, and the avrE gene can partially substitute for the dspE gene in mutant strains of E. amylovora. Our findings provide a striking example of dual functionality for Avr-like effector proteins of plant pathogenic bacteria. Further, the data indicate that the relative contribution of homologous virulence/avirulence genes to disease depends on the genetic background in which they are expressed. Our results therefore suggest that many avr genes for which no virulence phenotype yet has been detected have functions that can promote infection.

How can differences in genetic background evolve that lead to such dramatic differences in the virulence phenotype of avr gene homologs in different bacteria? Alfano and Collmer (6) proposed a model in which coevolution of pathogen and host plant(s) favors proliferation and redundancy of virulence factors through modification of preexisting factors and acquisition of others from heterologous pathogens, while conserving the virulence-factor-delivery system (the Hrp secretion system). According to this model, the more coevolved a pathogen with its host(s), the less likely is any single virulence factor to be critical for pathogenicity. The phenotypic difference between a dspE mutation and an avrE mutation may result from and reflect a difference in the extent or nature of the coevolution with plant hosts experienced by E. amylovora and by P. syringae. Evolution of corresponding R genes and modification of targets of pathogen virulence factors (that would lead to modification, substitution, and redundancy of the factors) are likely to have occurred more over time in the numerous herbaceous hosts typically infected by P. syringae pathovars than in the relatively fewer and more slowly reproducing woody hosts with which E. amylovora presumably evolved. Alternatively or additionally, E. amylovora may have acquired the dspEF locus and the linked hrp gene cluster more recently than P. syringae acquired the hrp-linked avrE locus, allowing less time for coevolution leading to modification or the development of redundant function. In support of this idea, the harpin-encoding genes of these two bacteria show a phenotypic difference similar to that of dspE and avrE. hrpN mutants of E. amylovora are drastically reduced in virulence or are nonpathogenic (4, 40), whereas hrpZ mutants of P. syringae show little or no difference in disease-causing ability from the wild type (41). These results suggest that E. amylovora generally has evolved fewer redundant virulence functions than P. syringae.

Localization of the dspE and dspF gene products during the plant–bacterial interaction will be important, in light of the absolute requirement for the dspEF locus in pathogenicity. Several reports (reviewed in ref. 42) provide indirect yet compelling evidence that a number of Avr proteins are localized to the plant cell interior via the Hrp pathway in much the same way as virulence proteins of animal pathogenic bacteria are delivered into host cells (43). It remains to be determined whether the avirulence function of the dspEF locus depends on secretion through the Hrp pathway. This seems likely considering the physical similarity of DspF (and AvrF) to chaperones required for type III secretion of virulence factors from animal pathogenic bacteria (36).

The dspEF locus is the first-described avirulence locus in E. amylovora. We have also found a homolog of avrRxv from Xanthomonas campestris (44) near the dspEF locus (5). Monogenic (R-gene-mediated) resistance to fire blight has not been reported, but differential virulence of E. amylovora strains on apple cultivars has been observed (45). Also, some strains of E. amylovora infect Rubus spp. and not pomaceous plants, and vice versa (46). Whether DspE, the AvrRxv homolog, or other similar proteins play a role in these specificities awaits determination.

Although the dspEF locus triggers defense responses in soybean when expressed in P. syringae pv. glycinea, it is not required for the HR of soybean elicited by E. amylovora. Nor is hrpN required (Fig. 3). It is possible that E. amylovora must have either dspE or hrpN to elicit the HR in soybean. We have observed, however, that, in contrast to its effect on many other plant species (47), infiltrated harpin (HrpN) does not elicit the HR in soybean, suggesting the alternative explanation that E. amylovora harbors another avr gene recognized by this plant. A cell-free DspE and DspF preparation also failed to elicit the HR when infiltrated into soybean leaves, raising the possibility that one or both of these proteins trigger defense responses from within the plant cell (D. W. Bauer and S.V.B., unpublished data).

Recognition of E. amylovora avirulence signals in soybean suggests the presence of one or more specific R genes. A dspEF-specific R gene might be useful for engineering apple and pear for resistance to fire blight. R-gene-mediated resistance to the apple scab pathogen Venturia inaequalis (48) and successful transformation of apple with attacin E for control of fire blight (49) attest the feasibility of such an approach. R-gene-mediated resistance to apple scab has been overcome in the field (50), but the requirement for the dspEF locus in disease favors relative durability of a corresponding R gene (12). Avirulence screening of dspEF and other E. amylovora genes in pathogens of genetically tractable plants such as Arabidopsis could broaden the pool of candidate R genes and hasten their isolation. A similar approach could be used to isolate R genes effective against other pathogens of woody plants. Furthermore, if the dspEF locus is as widely conserved as is suggested by its homology with the avrE locus, a corresponding R gene could be effective against a variety of pathogens both of woody and herbaceous plants.

ABBREVIATIONS

dsp

disease specific

hrp

hypersensitive reaction and pathogenicity

HR

hypersensitive reaction

R gene

resistance gene

pv.

pathovar

cfu

colony-forming units

GUS

β-glucuronidase

LB

Luria–Bertani

Hrp MM

hrp-gene-inducing minimal medium

Note Added in Proof

Recently, Gaudriault et al. (51) characterized the dsp locus of Erwinia amylovora strain 1430. They designated the genes corresponding to dspE and dspF as dspA and dspB, respectively.