INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Most gram-negative plant-pathogenic bacteria contain clusters of genes termed hrp that are required for elicitation of a rapid localized defense response called the hypersensitive reaction (HR) in incompatible plants and that are required for pathogenicity in susceptible plants (1). Proteins encoded by hrp genes are involved in the regulation of the expression of other hrp genes and in a specialized secretion process called the Hrp or type III pathway (9). Harpins, a major class of proteins that travel the pathway (including HrpN of Erwinia species, HrpZ of Pseudomonas syringae, and PopA of Ralstonia solanacearum), elicit the HR when infiltrated into the apoplast of leaf tissue (reference 1 and references therein). They are heat stable, rich in Gly and/or Ser, lack Cys, and differ in their primary sequences. In Erwinia amylovora, mutation of hrpN results in substantially reduced Hrp phenotype (4, 6, 45).

E. amylovora causes the devastating fire-blight disease on many rosaceous plants, such as apple, pear, and cotoneaster. Cosmids pCPP430 and pCPP450, which harbor the hrp gene cluster of E. amylovora Ea321, enable Escherichia coli to elicit the HR in tobacco (7). The region of pCPP430 essential for the Hrp phenotype encodes two-component regulatory proteins, a ⁵⁴ enhancer-binding protein, a sigma factor, secretory proteins, and the HrpN harpin (11, 27, 42-45). In contrast, the locus next to hrp genes, designated dsp, contains pathogenicity genes, and P. syringae pv. glycinea containing the E. amylovora dsp locus causes the HR rather than disease in soybean plants (10). This locus encodes a Hrp-secreted protein and a probable chaperone of the secreted protein (8, 10, 17).

Additional HR elicitors in E. amylovora have been suspected based on the HR-variable phenotype of E. amylovora hrpN mutants (references 4 and 6; see also Table 1). We report here the identification and characterization of a novel harpin of E. amylovora, HrpW, the C-terminal domain of which surprisingly is homologous to fungal pectate lyases (PLs). We show that HrpW, the production of which is controlled by hrpL, is delivered by the E. amylovora Hrp pathway. HrpW elicits the HR in plants, and the HR necrosis is not due to the potential PL activity of HrpW. Finally, we provide evidence that HrpW is not required for the HR and pathogenicity, although when overexpressed it enhances the HR-eliciting activity of a hrpN mutant. Preliminary reports on E. amylovora HrpW have been made (28, 29), and, while this article was under revision, a paper describing HrpW from E. amylovora CFBP1430 (16) appeared.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Bacterial strains and plasmids. E. amylovora Ea321 and Ea273 are wild-type strains that infect pomaceous plants (7). Plasmids pCPP1152, pCPP1218, pCPP1219, pCPP1220, and pCPP1227 are subclones of pCPP1012, which was cloned from pCPP430 in pBluescript KS(+) (Stratagene, La Jolla, Calif.) (Fig. 1B). E. coli DH5 was used as the host of the plasmids. Strains of E. amylovora and E. coli were routinely grown in Luria broth medium at 28°C and 37°C, respectively.

View larger version (26K): [in this window] [in a new window]   FIG. 1.   Molecular structure of the region of E. amylovora genome containing hrpW. (A) Cosmids pCPP430 and pCPP450 that contain the regulatory and secretory region of the hrp cluster of E. amylovora. Arrow boxes above the cosmid clones indicate the transcriptional units; the names of the characterized operons are given above (references 6, 10, 11, 25, 27, 43-45, and this study). (B) Location of hrpW, which encodes a Gly-rich protein, locations of transposon insertions, and subclones of pCPP1012 used in the study. Boxes and arrow boxes indicate ORFs. Harpin genes are indicated by dark shading and avr homologs by light shading. Filled triangles are putative HrpL-dependent promoters. Restriction sites: B, BamHI; E, EcoRI; H, HindIII, Ea, EagI; Hp, HpaI.

Molecular biological techniques and sequencing analysis. General molecular biology procedures were performed by using standard techniques as previously described (34). Sequencing was done on an ABI 373A automated DNA sequencer at the Cornell University Biotechnology Program DNA Sequencing Facility. For DNA and protein sequence analyses, programs in the GCG software package, version 7.3 (Genetics Computer Group, Inc., Madison, Wis.) and DNASTAR (DNASTAR, Inc., Madison, Wis.) were used. PHD prediction was made by submitting the alignment shown in Fig. 2 to http://www.embl-heidelberg.de/.

Expression of hrpW in E. coli. The 1.4-kb HpaI fragment of pCPP1227 that contains hrpW was subcloned into pBC SK (Stratagene) such that hrpW is under the control of T710 promoter. The resulting plasmid, pCPP1232 (Fig. 1B), was introduced into E. coli DH5(pGP1-2) (40). Cells were incubated at 42°C to induce the expression of the T7 RNA polymerase gene, and newly synthesized proteins were radiolabelled with ³⁵S-Met as previously described (40). Resulting samples were resuspended in sample buffer and heated to 95°C for 3 min before sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) in a 10% polyacrylamide gel.

Partial purification of HrpW. HrpW, produced by heat-shock treatment of E. coli DH5(pGP1-2, pCPP1232) at 42°C, was purified by cutting out the area of the gel containing HrpW, eluting the protein with ELUTRAP (Schleicher & Schuell, Inc., Keene, N.H.), and desalting the HrpW-containing solution with Centriprep-30 (Amicon, Inc., Beverly, Mass.) and 5 mM potassium phosphate (KPO₄) buffer (pH 6.5), which would also reduce the amount of SDS in the solution. A similar preparation was made from E. coli DH5(pGP1-2, pBC SK) to be used for the negative control of further experiments.

Production of anti-HrpW antibodies. Polyclonal antibodies against HrpW were raised at the College of Veterinary Medicine, Cornell University, by injecting ca. 100 µg of HrpW into a rabbit three times at 2- to 3-week intervals. Blood containing the antiserum was collected 2 weeks after the final injection. The immunoglobulin fraction was cross-absorbed with heat-treated lysate of E. coli DH5(pGP1-2, pBC SK).

Immunodetection of HrpW. E. amylovora Ea321Rp (a rifampin-resistant derivative of Ea321), Ea321-K49 (hrpL::Tn10-miniKm) (44), Ea321-G84 (hrcC::Tn5gusA1) (27), Ea273Rp, Ea273-K49, and Ea273-G73 (hrcV::Tn5-gusA1) (43) were grown overnight in Terrific broth, transferred to an hrp minimal medium (23) at 10⁸ CFU/ml, and incubated at 20°C for about 24 h until the bacteria grew to ca. 10⁹ CFU/ml. Cultures were centrifuged at 17,500 × g for 10 min, and the cell pellet collected from 1/100 of the original culture was saved for gel loading. The supernatant was passed through a membrane filter (pore size, 0.2 µm; Whatman Inc., Fairfield, N.J.) after adding 1 mM PMSF and was concentrated 100-fold by using Centricon-10 and Microcon-10 (Amicon, Inc.) at 4°C. Both the cell and supernatant fractions were then subjected to SDS-PAGE in two 10% polyacrylamide gels.

Generation of an N-terminal fragment of HrpW. pCPP1232 was digested with BamHI and BstEII, and the ends of the 4.1-kb fragment were blunted by using the Klenow fragment and were self-ligated. The resulting plasmid, pCPP1254, which encodes the N-terminal 226 amino acids of HrpW and Ile-His residues derived from the vector sequence, was cloned in E. coli DH5 and then transferred to E. coli DH5(pGP1-2), generating E. coli DH5(pGP1-2, pCPP1254). The N-terminal fragment [HrpW(1-226)] was produced by heat-shock treatment and partially purified by holding the bacterial lysate in a boiling water bath for 10 min and removing the precipitate.

Transposon mutagenesis. hrpW mutants of pCPP1012 were created by using Tn5-gusA and TnphoA and then transferred into the genome of Ea321Rp and Ea321-T5 (hrpN::Tn10-miniKm) (45) as described previously (27, 43). Two Tn5-gusA-induced mutants, Ea321-G204 and Ea321-T5/G204, and three TnphoA-induced mutants, Ea321-P110, Ea321-P136, and Ea321-P114 (Fig. 1), were chosen for further analysis.

Complementation tests. The hrpW gene in pCPP1232 was recloned in pBluescript KS(+) behind the lac promoter, generating pCPP1233. pCPP1012 or pCPP1233 was introduced into harpin gene mutants of Ea321. Ea321-T5(pCPP1084) was used as a control for positive complementation (pCPP1084 contains wild-type hrpN [45]).

Plant assays. Elicitation of the HR was tested by infiltrating protein or bacterial preparations into the intercellular space of leaves of tobacco (Nicotiana tabacum L. `xanthi') and other plants (27). Cells were grown in either Luria broth (E. coli DH5 and MC4100) or an hrp minimal medium (E. amylovora Ea321 and Ea321-T5) (23) to 5 × 10⁸ CFU/ml and were resuspended in 5 mM KPO₄ buffer (pH 6.5) to 2 × 10⁸ CFU/ml (E. coli strains) or 5 × 10⁸ CFU/ml (E. amylovora strains). Metabolic inhibitors used were cycloheximide at 100 µM, LaCl₃ at 1 mM, and Na₃VO₄ at 50 µM. Pathogenicity tests in immature pear fruits and apple and pear seedlings were carried out essentially as described by Steinberger and Beer (39).

Southern blotting. Genomic DNA was digested with EcoRI, electrophoresed on a 0.7% agarose gel, transferred to an Immobilon-N membrane (Millipore Co.), and hybridized with the ³²P-labelled 1.4-kb HpaI fragment of pCPP1227 at 65°C for 24 h. The membrane was washed twice with a solution of 2× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate) and 1.0% SDS at 65°C, and then with 0.1× SSC until no radioactivity was detected in the wash solution. For low-stringency hybridizations, the membrane was incubated at 50°C and washed with 2× SSC at 45°C.

Pectic enzyme assay of HrpW. Heat-induced E. coli DH5(pGP1-2, pCPP1232) cells were pelleted, resuspended in one-tenth volume of 5 mM KPO₄ buffer (pH 6.5) or 10 mM Tris-HCl (pH 8.5), sonicated on ice, and centrifuged, and the PL activity of the supernatant was tested. Also, 50-fold-concentrated Ea321 culture supernatant was assayed. A dilute pectate lysase (PelE) of Erwinia chrysanthemi EC16 in 10 mM Tris-HCl (pH 7.8) was used as a positive control.

Nucleotide sequence accession number. The nucleotide sequence of hrpW and orfC was deposited in GenBank under the accession no. U94513.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Identification of hrpW, the gene encoding a glycine-rich protein. Infiltration of E. coli strains carrying pCPP430-T5 or pCPP450-T5, cosmids that have mutated hrpN genes resulting from insertions of Tn10-miniKm, still induced a weak HR in tobacco (26), suggesting the existence of another gene encoding an HR elicitor in the clones. The DNA downstream of hrpN where pCPP430 and pCPP450 overlap was therefore subcloned, and its sequence revealed four open reading frames (ORFs), designated orfA, orfB, orfC, and hrpW (Fig. 1B). A putative HrpL-dependent promoter (11, 27), CGGAACC-N₁₅-CCACTCAAT, was found 58 bp upstream of the hrpW start codon (44). Also, the stop codon of hrpW overlaps the start codon of orfC, which may encode a chaperone, and a potential transcription terminator, TAAAAAAACGCA-N₁₄-TGCGTTTTTTTA, is present between orfB and orfC. Therefore, hrpW and orfC appear to constitute an operon.

View larger version (62K): [in this window] [in a new window]   FIG. 2.   Alignment of HrpW with pectate lyases of N. haematococca mating type VI (F. solani f. sp. pisi) and of E. carotovora subsp. carotovora. The sequences were aligned by the PILEUP program (GCG software package, version 7.3) with default parameters, and the alignment was manually edited by using the LINEUP program in the same package. Conserved residues are boxed, highly conserved regions are underlined, and conserved Cys residues are indicated by asterisks. Shaded regions in HrpW may form -helices.

View larger version (41K): [in this window] [in a new window]   FIG. 3.   Expression of hrpW by a T7 RNA polymerase-directed gene expression system. Lanes: 1, E. coli DH5(pGP1-2, pBC SK); 2, E. coli DH5(pGP1-2, pCPP1232).

View larger version (22K): [in this window] [in a new window]   FIG. 4.   Immunoblot showing the hrp-dependent production and secretion of HrpW in E. amylovora Ea273. Lanes: 1, E. coli DH5(pGP1-2, pCPP1232); 2, HrpN; 3 and 4, wild type; 5 and 6, Ea273-K49 (hrp regulation mutant); 7 and 8, Ea273-G73 (hrp secretion mutant). Lanes 1, 3, 5, and 7 contain whole-cell preparations, and lanes 4, 6, and 8 contain supernatant preparations.

HrpW is an inactive homolog of PLs. Database searches with the BLAST algorithm (3) indicated that the C-terminal region of HrpW is homologous to PLA-D of Nectria haematococca mating type VI (Fusarium solani f. sp. pisi) (reference 19 and references therein). BLASTP P values from runs with default parameters were 4.0 × e¹⁴ to 3.0 × e¹⁰, and BESTFIT alignments indicated that HrpW is 27 to 33% identical to the fungal PLs (Z scores, 8.14 to 13.3). Also, database searches using the PLs of N. haematococca showed that they are homologs of Pel-3 and PelB of Erwinia carotovora subsp. carotovora (21, 31). These fungal PLs and E. carotovora Pel-3 and PelB, together with HrpW, form a group distinct from other PL families. From an alignment of the proteins, several highly conserved blocks were recognizable (Fig. 2). They share 20 identical residues, of which five are Gly. The PHD algorithm predicted -sheets and loops at the region corresponding to the PL-like domain of HrpW, except for the sequence at residues 329 to 336, which has a propensity to form an -helix (Fig. 2). Intriguingly, HrpW does not contain any Cys, which is abundant and often conserved among PL homologs. Furthermore, we were not able to detect PL activity of HrpW or E. amylovora by using the several assays described in Materials and Methods.

Production and secretion of HrpW are hrp dependent. Western blot analysis of E. amylovora Ea273 and Ea321 was performed to demonstrate that HrpW is an extracellular protein and its production and secretion are controlled by other hrp genes. As expected, anti-HrpW antibodies did not bind to HrpN (Fig. 4, lane 2). From the immunoblot, HrpW was detected mostly from the supernatant preparation of Ea273, indicating that HrpW is secreted (Fig. 4, lane 4). In agreement with the presence of a HrpL-dependent promoter sequence in front of the hrpW ORF, HrpW was not found in preparations of Ea273-K49 (hrpL mutant), suggesting expression of hrpW is HrpL-dependent (Fig. 4, lanes 5 to 6). In addition, HrpW was found only in the whole-cell preparation of the hrp secretion mutant, Ea273-G73 (Fig. 4, lane 7). This shows that secretion of HrpW is Hrp pathway dependent. Similar results were obtained from Ea321 and its hrp mutants (data not shown). Interestingly, HrpW was not detected in either the cell or the supernatant of an Ea321 secretion mutant Ea321-G84, suggesting that the mutant is affected in HrpW production.

HrpW induces rapid tissue necrosis in plants. From the predicted properties of HrpW, we inferred that it might be an HR elicitor. To test this possibility, the protein isolated from an SDS-PAGE gel of E. coli DH5(pGP1-2, pCPP1232) was infiltrated into tobacco leaves. The area infiltrated with the HrpW preparation began to collapse after 8 to 12 h, and typical tissue necrosis, indistinguishable from that elicited by HrpN, developed 24 to 36 h after inoculation (data not shown). In contrast, no tissue collapse was observed from the preparation of E. coli DH5(pGP1-2, pBC SK). HrpW induced tissue necrosis in tobacco at concentrations of 1.1 µM (50 µg/ml). HrpW also caused necrosis in African violet, geranium, tomato, and pepper plants, and in Kalanchoë diagremontiana and Arabidopsis thaliana (data not shown). On the other hand, neither HrpN nor HrpW induced obvious HR necrosis in leaves of soybean, apple, and pear. A heat-treated (100°C, 10 min; see Materials and Methods) preparation of HrpW retained the necrosis-inducing activity in tobacco leaves, indicating its heat-stable nature (Fig. 5A, panel 2). However, treatment of HrpW with 1 mg of pronase (protease type XIV; Sigma) per ml at 37°C for 1 h destroyed the activity.

Plant metabolism is required for the HrpW-induced HR; the PL-like domain of HrpW is not needed. A major question then was whether the tissue necrosis caused by HrpW is due to its potential PL activity, which was not detected in in vitro assays, or to its HR-eliciting ability, which is comparable to that of known harpins. Coinfiltration of HrpW with the metabolic inhibitors cycloheximide, lanthanum chloride, or sodium vanadate prevented development of the HR (Fig. 5A, panels 3 to 5); coinfiltration of HrpN with the inhibitors resulted in the same effects (Fig. 5A, panels 7 and 8). This indicates that active plant metabolism is needed for HR induction by HrpW. Tobacco leaves infiltrated with PelE of E. chrysanthemi EC16 or PLA-D of N. haematococca mating type VI also exhibited rapid tissue necrosis (Fig. 5A, panel 9, and data not shown), which occurred faster than that elicited by harpins. Also, the collapsed area was translucent, darker, softer, and easily crushed. Furthermore, PelE and PLA-D induced tissue necrosis irrespective of the presence of inhibitors (Fig. 5A, panels 10 and 11, and data not shown), indicating that the necrosis is the result of the enzyme action and not plant metabolism.

View larger version (97K): [in this window] [in a new window]   FIG. 5.   Tobacco leaves showing (A) suppression of HrpW-induced necrosis by inhibitors of plant metabolism, (B) elicitation of the HR following infiltration with heat-treated cell lysates of E. coli DH5 strains containing hrpW constructs, and (C) elicitation of the HR following infiltration with suspensions of a hrpN mutant of E. amylovora and the mutant strains containing hrpN or hrpW in plasmids. Panels in part A: 1, 5 mM KPO4 buffer (pH 6.5); 2, HrpW; 3, HrpW plus cycloheximide; 4, HrpW plus LaCl3; 5, HrpW plus Na3VO4; 6, HrpN; 7, HrpN plus cycloheximide; 8, HrpN plus Na3VO4; 9, PelE in 10 mM Tris-HCl (pH 7.8); 10, PelE plus cycloheximide; 11, PelE plus Na3VO4. Concentrations of HrpW and HrpN were 0.1 mg/ml. Panels in part B: 1, E. coli DH5(pGP1-2, pCPP1232) (pCPP1232 encodes full-length HrpW); 2, E. coli DH5(pGP1-2, pCPP1254) [pCPP1254 encodes HrpW(1-226)]; 3, E. coli DH5(pGP1-2, pBC SK). Panels in part C: 1, Ea321-T5 (hrpN mutant); 2, Ea321-T5(pCPP1084) (pCPP1084 contains hrpN); 3, Ea321-T5(pCPP1233) (pCPP1233 contains hrpW). Photo A was taken 36 h after infiltration; photos B and C were taken 48 h after infiltration.

hrpW is conserved among other Erwinia strains. The presence of hrpW in other bacteria was examined by Southern hybridization with Ea321 hrpW as a probe. Under high-stringency conditions, single bands of three sizes were observed from 11 strains of E. amylovora tested, including 4 pathogenic strains on Rubus plants and 3 isolated in Japan (Fig. 6, lanes 1 through 11). When low-stringency conditions were used, hybridizing bands were detected in addition from genomic DNA preparations of strains of other Erwinia species, including E. carotovora subsp. carotovora ATCC 15713, E. chrysanthemi EC16, and Erwinia salicis ATCC 15712 (Fig. 6, lanes 12 to 15), but not from those of P. syringae pv. tomato DC3000, R. solanacearum K60, Xanthomonas campestris pv. vesicatoria 21, and Xanthomonas oryzae pv. oryzae JH89031 (data not shown).

View larger version (39K): [in this window] [in a new window]   FIG. 6.   Southern blot showing that hrpW of E. amylovora Ea321 is present in other bacteria. Genomic DNA of strains was probed with a 1.4-kb HpaI fragment that contains Ea321 hrpW. High-stringency conditions were used for lanes 1 to 11; low-stringency conditions were used for lanes 12 to 15. Lanes: 1, Ea321; 2, Ea266; 3, Ea273; 4, Ea246; 5, Ea510; 6, Ea528; 7, Ea574; 8, Ea546; 9, Ea557; 10, Ea562; 11, Ea587; 12, E. carotovora subsp. carotovora ATCC 15713; 13, Erwinia mallotivora 1818; 14, E. salicis 1822; 15, pCPP2157 of E. chrysanthemi EC16.

hrpW is not essential for the Hrp phenotype of E. amylovora Ea321. To study the roles of hrpW in the HR and pathogenicity of E. amylovora, several transposon-induced mutants of hrpW were generated in Ea321. As previously reported (4, 45), Ea321-T5 (hrpN mutant) had a greatly reduced ability to elicit the HR in tobacco and to infect host plants. No difference was seen between the parental strain and hrpW mutants in terms of the HR induction in tobacco and virulence in immature pear fruits (Table 1) and shoots of pear and apple (results not shown). For example, both the parental strain and hrpW mutants induced complete necrosis of immature pear halves and produced large amounts of ooze even when only 10² cells were inoculated. Nor could we detect statistically significant differences between Ea321-T5 and Ea321-T5/G204 (hrpN and hrpW double mutant) in their HR-inducing ability in tobacco or symptom production in pears. Nevertheless, the double mutant seemed further reduced in the Hrp phenotype (Table 1). In complementation assays, when hrpN was provided to Ea321-T5 in trans, partial recovery of the Hrp phenotype was obvious, especially at higher inoculum doses (Table 1). However, neither pCPP1233 (which contains hrpW) nor pCPP1012 (which contains both hrpN and hrpW) complemented Ea321-T5/G204.

Overexpressed hrpW enhances the HR-eliciting ability of a hrpN mutant. Since no clear role of hrpW in the Hrp phenotype was observed following mutagenesis, we conducted a gain-of-function experiment. pCPP1233, a high-copy-number plasmid that carries hrpW, was introduced into the hrpN mutant, Ea321-T5, and the transformant was tested for the Hrp phenotype. Although the intensity of the HR was less than of that caused by Ea321-T5(pCPP1084), which contains the intact hrpN gene in trans, Ea321-T5(pCPP1233) induced more extensive HR than Ea321-T5 did (Fig. 5C), confirming the capacity of HrpW for HR elicitation. However, similar to Ea321-T5 and different from Ea321-T5(pCPP1084), Ea321-T5(pCPP1233) remained greatly reduced in virulence (data not shown).

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Our work demonstrates that E. amylovora produces two harpins, HrpN and HrpW. Analyses of HrpW indicate that it is a unique protein composed of an N-terminal harpin-like domain and a C-terminal PL-like domain. The presence of a PL-homologous harpin in E. amylovora is surprising because E. amylovora is believed to be nonpectolytic (35), and no pectic enzyme function has been suggested for harpins (1). HrpW-like harpins appear to be widely distributed among plant-pathogenic bacteria. In addition to Southern analysis, which suggested the presence of hrpW homologs in other Erwinia species, sequence comparison showed that the C-terminal domain of the HrpW harpin of P. syringae pv. tomato (12) is related to that of E. amylovora HrpW.

HR elicitation by HrpW depends on plant metabolism, and it does not require the PL-like domain (this work and reference 12). Although HrpW has a PL-homologous domain, it differs from PLs in several respects. HrpW lacks the N-terminal signal peptide recognized by the Sec machinery; rather, it is secreted via the type III pathway. Compared to its PL homologs, HrpW has an acidic pI and contains no Cys residues. Under our assay conditions, we could not detect any PL activity from HrpW. Similarly, no pectic enzyme activity has been found for PL-homologous pollen proteins (15). The failure to detect PL activity may be ascribed partly to the lack of Cys in HrpW, which might be needed for structural integrity and enzymatic function. Other possibilities include a different substrate specificity from its homologs or a different enzymatic function (for example, lysozymes and -lactalbumins [32]). Alternatively, HrpW may have only a pectate-binding function instead of the lyase function (30), as suggested by Charkowski and colleagues (12).

According to a forthcoming proposal (13), several evolutionarily related classes of pectic enzymes are apparent. One group of PLs, named class III, includes the four PLs of N. haematococca, Pel-3/PelB of E. carotovora subsp. carotovora, and the recently reported PelI of E. chrysanthemi (37). They show optimum activity at basic pH and at high Ca²⁺ concentrations. HrpW proteins of E. amylovora and P. syringae pv. tomato belong to this homology class. Also, a recent database search identified a PL homolog in Bacillus subtilis (GenBank accession no. Z99121 and AF017113) that contains only one Cys and has higher sequence similarity to HrpW proteins than to class III PLs.

Like HrpN, HrpW of E. amylovora can induce the HR following infiltration of purified HrpW into tobacco and by overexpressing hrpW in a hrpN mutant. However, our results reveal that the contributions of the two harpins to the Hrp function of E. amylovora are quite different. In contrast to the hrpN mutation, which significantly reduces the ability of E. amylovora to elicit the HR in nonhosts and to infect host tissues, hrpW mutants of E. amylovora were not affected in the Hrp phenotype. Perhaps compared to hrpN, the biological role of hrpW in the HR and pathogenicity is subtle and was not easily detectable under the assay conditions we used, or it may have a role in other hosts. Alternatively, hrpW could be a "vanishing gene" whose role(s) in pathogenesis has been replaced by hrpN during evolution. That HrpN is a major Hrp-secreted protein and HrpW is not (25) supports this hypothesis.

Harpins appear to target the plant cell wall. They elicit the HR when exogenously applied to plant tissue by infiltration. When harpins are added to tobacco cell suspension culture, K⁺ efflux and alkalinization of the medium, followed by cell death, occur (33, 45); these responses do not occur in protoplast culture (6, 22). In addition, antibodies against the HrpZ harpin localized HrpZ outside of plant cells and not in protoplasts, and the alkalinization and the localization were blocked by a chelating agent that extracts Ca²⁺ and pectin (22). The homology of HrpW to PLs and the ability to bind pectin (12) is consistent with a model in which the site of harpin action is the plant cell wall. Recent evidence suggests that several Avr proteins are transferred directly into the plant cell via the Hrp secretion machinery (41). However, to locate inside the plant cell, Avr proteins (or the secretion apparatus) must pass through the cell wall. Harpins may function in this step, perhaps by loosening the cell wall.

Functional redundancy appears to be typical for disease factors of phytopathogenic bacteria as shown for Avr proteins, PLs, and autoinducers. A similar conclusion could be drawn for harpins. It is interesting that hrpN and hrpW of E. amylovora are flanked by homologs of avrRxv of X. campestris pv. vesicatoria and avrE of P. syringae pv. tomato (Fig. 1B) (10, 25). Like several other Avr proteins, it is likely that these Avr homologs are introduced to the plant cell cytoplasm by the E. amylovora Hrp pathway. Thus, in E. amylovora, the region of the genome where harpin genes and avr homologs reside may constitute an arsenal of Hrp-delivered proteins used to bombard the host cell. Elucidating their specific locations in plants and functions in the HR and pathogenicity will help our understanding of the mechanisms of plant-bacterial interaction.