Characterization of the Xanthomonas axonopodis pv. glycines Hrp Pathogenicity Island

Abstract

We sequenced an approximately 29-kb region from Xanthomonas axonopodis pv. glycines that contained the Hrp type III secretion system, and we characterized the genes in this region by Tn3-gus mutagenesis and gene expression analyses. From the region, hrp (hypersensitive response and pathogenicity) and hrc (hrp and conserved) genes, which encode type III secretion systems, and hpa (hrp-associated) genes were identified. The characteristics of the region, such as the presence of many virulence genes, low G+C content, and bordering tRNA genes, satisfied the criteria for a pathogenicity island (PAI) in a bacterium. The PAI was composed of nine hrp, nine hrc, and eight hpa genes with seven plant-inducible promoter boxes. The hrp and hrc mutants failed to elicit hypersensitive responses in pepper plants but induced hypersensitive responses in all tomato plants tested. The Hrp PAI of X. axonopodis pv. glycines resembled the Hrp PAIs of other Xanthomonas species, and the Hrp PAI core region was highly conserved. However, in contrast to the PAI of Pseudomonas syringae, the regions upstream and downstream from the Hrp PAI core region showed variability in the xanthomonads. In addition, we demonstrate that HpaG, which is located in the Hrp PAI region of X. axonopodis pv. glycines, is a response elicitor. Purified HpaG elicited hypersensitive responses at a concentration of 1.0 μM in pepper, tobacco, and Arabidopsis thaliana ecotype Cvi-0 by acting as a type III secreted effector protein. However, HpaG failed to elicit hypersensitive responses in tomato, Chinese cabbage, and A. thaliana ecotypes Col-0 and Ler. This is the first report to show that the harpin-like effector protein of Xanthomonas species exhibits elicitor activity.

Many gram-negative plant-pathogenic bacteria possess two sets of genes that modulate their interactions with plants. The avirulence gene determines host specificity based on gene-for-gene interactions, and the hrp (hypersensitive reaction and pathogenicity) genes are involved in pathogenicity and the induction of hypersensitive responses (HRs) in nonhost plants (6). The nine hrp genes, which are highly conserved in plants and bacterial pathogens of animals, are known as the hrc (hrp conserved) genes (5). The hpa (hrp-associated) genes contribute to pathogenicity and to the induction of HR in nonhost plants but are not essential for the pathogenic interactions of bacteria with plants (19). These genes are generally clustered in a chromosomal region that spans 20 to 30 kb, and most of the Hrp and Hrc proteins function as type III protein secretion systems (16).

Type III secretion systems mediate the translocation of effector proteins across the bacterial membrane and into the host and are often important for virulence and in the modulation of host defense responses (16). The hrp-hrc regions are now designated pathogenicity islands (PAIs) in various plant-pathogenic bacteria (2). PAIs contain many virulence genes, are present only in pathogenic bacteria, have different G+C contents compared with the host bacterial DNA, are often flanked by direct repeats, are bordered by tRNA genes and/or cryptic mobile genetic elements, and are unstable (21).

Compared to animal pathogens, relatively few effector proteins have been reported in plant-pathogenic bacteria. The Hrp type III secreted proteins include effectors that are essential for pathogenicity, avirulence proteins, and nonspecific elicitors. Included in this group are many avirulence proteins, including PopA, PopB, and PopC of Ralstonia solanacearum (3, 20), HopPsyA of Pseudomonas syringae pv. syringae (12), HrpZ from P. syringae pv. syringae (22), HrpW, DspA, and HrpN of Erwinia amylovora (17, 30, 46), HrpN_Pnss of Pantoea stewartii subsp. stewartii (1), HrpN_Ech of Erwinia chrysanthemi (4), and HrpN_Ecc of Erwinia carotovora (32).

The so-called harpin proteins HrpN and HrpW of E. amylovora and HrpZ and HrpW of P. syringae pathovars are well-known HR elicitors. In E. amylovora, HrpN is a major HR elicitor for tobacco, and an hrpN mutant was nonpathogenic for pear plants and did not elicit HR (46). The HrpW of E. amylovora has also been reported to be an HR elicitor for tobacco, but the hrpW mutant retained the wild-type ability to elicit HR in nonhosts and to cause disease in hosts (30). hrpZ and hrpW mutants of P. syringae pathovars were only slightly reduced in HR elicitation activity in tobacco, whereas HR activity was significantly reduced in an hrpZ hrpW double mutant. However, the double mutant retained the ability to cause disease symptoms on host plants (11). Two xanthomonad effector proteins, HrpB2 and HrpF, which are essential for pathogenicity were identified in Xanthomonas campestris pv. vesicatoria (38). Furthermore, it has been reported that XopA (Xanthomonas outer protein A) is important for growth in planta and for full avirulence (33). However, no harpin-like proteins with elicitor activity have been reported in the xanthomonads.

The hrp-hrc gene cluster is conserved in P. syringae pathovars and has a tripartite mosaic structure that is composed of a cluster of type III secretion genes, which are bounded by exchangeable effector and conserved effector loci that contribute to parasitic fitness and pathogenicity (2). Since the Hrp PAIs of P. syringae were isolated from different pathovars, it is possible to compare their gene organization and function (2). However, among the xanthomonads, the Hrp PAI of X. campestris pv. vesicatoria has mainly been characterized (6, 33). To expand our knowledge regarding the Hrp PAIs of xanthomonads, we studied genes that were involved in pathogenicity and encoded effector proteins in X. axonopodis pv. glycines, which is the causal agent of bacterial pustule on soybean.

Previously, we isolated cosmid clones that complemented two hrp mutants of X. axonopodis pv. glycines and demonstrated that the Hrp PAI spanned approximately 25 kb of the bacterial genome (35). In the present study, we compared the Hrp PAI sequences of X. axonopodis pv. glycines with those of other xanthomonads. We discovered that, in contrast to the P. syringae Hrp PAI, the concept of a tripartite mosaic architecture was not applicable to the xanthomonad Hrp PAI. We also found that the HpaG protein, which is encoded in the Hrp PAI region of X. axonopodis pv. glycines, is an elicitor. This is the first report to show that the harpin-like effector protein of Xanthomonas exhibits elicitor activity.

MATERIALS AND METHODS

Bacterial strains, plasmids, and growth conditions.

The bacterial strains and plasmids used in this study are listed in Table 1. All of the X. axonopodis pv. glycines strains used in these experiments were derivatives of the parent strain 8ra. P. syringae pv. syringae 61 and its hrcQ_A mutant Pss61-2084 were obtained from Steven W. Hutcheson at the University of Maryland. Escherichia coli cells were cultivated at 37°C in Luria broth (LB; USB) or on LB agar plates. The X. axonopodis pv. glycines strains were grown at 28°C in LB or on YDC (1% yeast extract, 2% calcium carbonate, and 2% d-glucose) agar plates. Antibiotics were used at the following concentrations: ampicillin, 100 μg/ml; chloramphenicol, 15 μg/ml; kanamycin, 25 μg/ml; nalidixic acid, 25 μg/ml; and spectinomycin, 25 μg/ml for E. coli; kanamycin, 50 μg/ml; spectinomycin, 50 μg/ml; and rifampin, 50 μg/ml for X. axonopodis pv. glycines. Tetracycline was used at 10 μg/ml for E. coli and at 2 μg/ml for X. axonopodis pv. glycines.

TABLE 1.

Bacterial strains and plasmids

Strain or plasmid	Characteristicsa	Source or reference
Escherichia coli
DH5α	F− φ80dlacZΔM15 Δ(lacZYA-argF)U169 endA1 deoR recA1 hsdR17(rK− mK+) phoA supE44 λ−thi-l gyrA96 relA1	Gibco-BRL
S17-1	Tra+recA Spr	42
C2110	polA Nalr	43
BL21(DE3)	F−ompT hsdSB(rB− mB−) gal dcm(DE3)	Novagen
Xanthomonas axonopodis pv. glycines
8ra	Wild type, Rifr	E. J. Braun
8raG	hrpG::Ω, hrpG mutant	This study
8-13	8ra hrpF::Omegon-Km	35
26-13	8ra hrcC::Omegon-Km	35
1-34/G1-34	8ra hrpF::Tn3-gus/8raG hrpF::Tn3-gus	This study
1-43/G1-43	8ra hpaB::Tn3-gus/8raG hpaB::Tn3-gus	This study
1-44/G1-44	8ra hrcU::Tn3-gus/8raG hrcU::Tn3-gus	This study
1-46/G1-46	8ra hpaC::Tn3-gus/8raG hpaC::Tn3-gus	This study
1-49/G1-49	8ra hrpE::Tn3-gus/8raG hrpE::Tn3-gus	This study
2-47/G2-47	8ra hpaF::Tn3-gus/8raG hpaF::Tn3-gus	This study
3-28/G3-28	8ra hpaD::Tn3-gus/8raG hpaD::Tn3-gus	This study
3-33/G3-33	8ra hrcC::Tn3-gus/8raG hrcC::Tn3-gus	This study
3-38/G3-38	8ra hrcJ::Tn3-gus/8raG hrcJ::Tn3-gus	This study
4-2/G4-2	8ra hrpD5::Tn3-gus/8raG hrpD5::Tn3-gus	This study
4-8/G4-8	8ra hrcV::Tn3-gus/8raG hrcV::Tn3-gus	This study
70-1/G70-1	8ra hpaG::Tn3-gus/8raG hpaG::Tn3-gus	This study
9-21/G9-21	8ra hpaH::Tn3-gus/8raG hpaH::Tn3-gus	This study
Plasmids
pBluescript II SK(+)	Phagemid, pUC derivative Ampr	Stratagene
pLAFR3	Tra− Mob+ RK2 replicon Tetr	44
pLAFR3ΔE	As pLAFR3 but without EcoRI site in multilinker	This study
pLAFR6	As pLAFR3 but without lacZα, contains multilinker of pUC18 flanked by synthetic trp terminators, Tetr	25
pRK2013	Tra+, ColE1 replicon, Kmr	15
pHoKmGus	Promoterless β-glucuronidase gene, Kmr AmprtnpA	7
pSShe	Cmr	43
pHP45Ω	Ω cassette, Spr Smr	36
pT7-7	T7 promoter-based expression vector, Ampr	45
pGA16	27.8-kb DNA fragment from strain 8ra cloned into pLAFR3	This study
pGA161	5.2-kb partial EcoRI-BamHI fragment, including hpaH to hpaC genes, from pGA16 cloned into pLAFR3	This study
pGA16037L	3.5-kb insert including the entire hrcC cloned into pLAFR6	This study
pGA166	9.9-kb partial EcoRI-SacI fragment, including hpaC to hrpB1 genes, from pGA16 cloned into pLAFR6	This study
pGA1604L	6.1-kb EcoRI fragment, including hrcU to hrcS genes, from pGA16 cloned into pLAFR3	This study
pGA410	6.3-kb partial EcoRI-XbaI fragment, including hrcQ to ORF1 genes, from pGA4 cloned into pLAFR3	This study
pGA411	4.1-kb XbaI-BamHI fragment, including hrpF, from pGA4 cloned into pLAFR6	This study
pGA406L	2.5-kb BamHI fragment, including hpaF, from pGA4 cloned into pLAFR3	This study
pGA33	22.4-kb DNA fragment from strain 8ra cloned into pLAFR3	This study
pGA4	23.5-kb DNA fragment from strain 8ra cloned into pLAFR3	This study
pGB1	22.5-kb DNA fragment, including hrpG and hrpX, from strain 8ra cloned into pLAFR3	This study
pGB106L	1.9-kb BamHI fragment from pGB1 cloned into pLAFR3	This study
pGB106Ω	2.0-kb Ω cassette from pHP45Ω cloned into pGA106L, hrpG::Ω	This study
pLGX3	3.9-kb partial BamHI-SmaI fragment from pGB1 including only hrpG and hrpX cloned into pLAFR3	This study
pLGXhpaG	0.75-kb PCR product including the entire hpaG cloned into pLGX3	This study
pCH85	23.0-kb DNA fragment, including hrpG and hrpX, from X. oryzae pv. oryzae KXO85 cloned into pLAFR3	This study

^aAmp^r, ampicillin resistance; Cm^r, chloramphenicol resistance; Km^r, kanamycin resistance; Nal^r, nalidixic acid resistance; Rif^r, rifampicin resistance; Sp^r, spectinomycin resistance; Tet^r, tetracycline resistance.

DNA manipulations.

Standard methods were used for DNA cloning, restriction mapping, and gel electrophoresis (41). The vector DNA was treated with the appropriate restriction enzymes as recommended, and DNA fragment extraction from gels was carried out as described by the manufacturer (Qiagen). DNA hybridizations were performed with the Gene Images labeling and detection system, according to the manufacturer's instructions (Amersham Biosciences). All other standard molecular biological methods were carried out as described by Sambrook et al. (41).

To complement hpaG mutant strain 70-1, pLGXhpaG carrying hrpG, hrpX, and hpaG was constructed as follows. We amplified 0.75 kb of hpaG containing only the coding region and the putative promoter region by PCR with primers HPAGH1 (5′-CCAAGCTTGAATACAGGTCTC-3′) and HPA1-2 (5′-CAACCTGCAGTTACTGCATCGA-3′). The amplified fragment was blunt ended with the Klenow fragment (Takara) and then ligated into the SmaI site of pBluescript II SK(+). The DNA sequences of the amplified fragments were confirmed by sequencing. The fragment was then generated as a HindIII fragment and cloned into the same site of pLGX3 carrying hrpG and hrpX.

For complementation analysis of other hrp, hrc, and hpa mutants, seven recombinant plasmids were constructed, as listed in Table 1: pGA161, pGA16037L, pGA166, pGA1604L, pGA410, pGA411, and pGA406L.

Mating and mutagenesis.

A genomic library of strain 8ra was constructed previously in pLAFR3, and all pLAFR3 derivatives were mobilized into X. axonopodis pv. glycines strains by triparental mating, as described previously (26). The cosmid clone pGA16, which complemented the hrp mutations, was mutagenized with the Tn3-gus transposon, as described by Bonas et al. (7). The insertion site and orientation of Tn3-gus in each mutant were mapped by restriction enzyme digestion analysis and by direct sequencing of the plasmid with the primer Tn3gus (5′-CCGGTCATCTGAGACCATTAAAAGA-3′), which allows sequencing out of the Tn3-gus transposon.

Marker exchange.

Mutagenized plasmids that carried Tn3-gus insertions were introduced individually into the parent strain 8ra by conjugation. The hrp::Tn3-gus, hrc::Tn3-gus, and hpa::Tn3-gus fusions were marker exchanged into strain 8ra, as described previously (39). All marker exchanges were confirmed by Southern hybridization analysis.

Plant assays.

Pathogenicity was determined by inoculating cotyledons and leaves of the susceptible soybean cultivar ′Pella' as described previously (26). The growth patterns in planta of X. axonopodis pv. glycines derivatives were determined as described previously (26). For the HR test, plants of pepper (Capsicum annuum L. cv. Dabokkun or Chokwang), tomato (Lycopersicon esculentum Mill. cv. Seokwang or Kwangsoo and six wild-type species that were obtained from the Tomato Stock Center at the University of California at Davis), tobacco (Nicotiana tabacum cv. Samsun NN and cv. Xanthi and N. glutinosa), Chinese cabbage (Brassica campestris L. cv. Karaksin-1), and Arabidopsis thaliana ecotypes Col-0, Cvi-0, and Ler were inoculated with approximately 2 × 10⁸ CFU/ml in a 0.85% NaCl solution, and the plant responses were observed 12 to 24 h after injection.

Construction of hrpG mutant strain 8raG.

We generated an hrpG mutant by insertion of an Ω cassette into the wild-type strain 8ra by marker exchange. The 1.4-kb fragment of the hrpX gene was PCR amplified with X. oryzae pv. oryzae genomic DNA as the template. Two PCR primers, hrpX1 (5′-GCAGCGATCTCTGCGTTGTC-3′) and hrpX2 (5′-GAAGTGCTGGCGATAGCCCT-3′), were used for the amplification. Twelve cosmid clones that carried the hrpG and hrpX regulatory genes from X. axonopodis pv. glycines 8ra were isolated from the genomic library by colony hybridization with the 1.4-kb fragment of hrpX from X. oryzae pv. oryzae as the probe. Among those, a 1.9-kb BamHI fragment, which included the hrpG gene from pGB1, was removed and subcloned into pLAFR3ΔE, generating pGB106L. The Ω cassette (which specifies spectinomycin resistance) from pHP45Ω (36) was inserted into the unique EcoRI site within the hrpG sequence in pGB106L to give pGB106Ω. This plasmid was introduced into X. axonopodis pv. glycines 8ra to generate the hrpG::Ω mutant strain (8raG) by marker exchange mutagenesis. We then introduced Tn3-gus fusions that had been constructed in the hrcC, hrcJ, hrcU, hrcV, hrpD5, hrpE, hrpF, hpaB, hpaC, hpaD, hapF, hpaG, and hpaH genes into wild-type strain 8ra and the hrpG mutant strain 8raG by marker exchange and measured the β-glucuronidase activities.

β-Glucuronidase assays.

The β-glucuronidase enzyme assay was performed as described previously, with some modifications (27). X. axonopodis pv. glycines strains were grown in hrp induction medium, XVM2 medium (50), for 24 h, centrifuged, resuspended in GUS extraction buffer, and lysed by sonication with a VCX-400 sonicator (Sonics & Materials Inc.). The extract was used in the β-glucuronidase enzyme assay with 4-methylumbelliferyl glucuronide as the substrate. Fluorescence was measured at 365-nm excitation and 460-nm emission in a TKO100 fluorometer (Hoefer Scientific Instruments). One unit of β-glucuronidase was defined as 1 nmol of 4-methylumbelliferon released per bacterium per minute.

Overexpression and purification of HpaG, Hpa1, XopA, and HrpN.

hpaG of X. axonopodis pv. glycines 8ra, hpa1 of X. oryzae pv. oryzae PXO86, xopA of X. campestris pv. vesicatoria 82-8, and hrpN of E. amylovora Ea321 were PCR amplified and cloned into the NdeI and BamHI sites of pT7-7. The DNA sequences of the amplified fragments were confirmed. E. coli BL21(DE3) strains harboring individual cloned genes in pT7-7 were grown in LB, and each protein was overexpressed following isopropylthiogalactopyranoside (IPTG) induction. The cells were harvested by centrifugation and resuspended in 50 mM Tris-HCl (pH 8.0), sonicated, and boiled for 15 min. After centrifugation, the partially purified proteins were loaded onto the anion-exchange column Mono Q (Amersham Biosciences), equilibrated with 50 mM Tris-HCl (pH 8.0), and eluted with a linear gradient of NaCl with a fast protein liquid chromatography system (Amersham Biosciences). Each protein was further purified by gel filtration with a Superdex 200 FPLC column (Amersham Biosciences). Purified protein concentrations were measured by the method of Bradford with bovine serum albumin as the standard (8).

Secretion analysis of HpaG.

For the HpaG secretion assay, X. axonopodis pv. glycines 8ra(pLGX3), hrcU mutant strain 1-44(pLGX3), hpaG mutant strain 70-1(pLGX3), and strain 70-1(pLGXhpaG) were grown in XVM2 broth medium with tetracycline for 24 h, subcultured in 500 ml of XVM2 broth medium containing tetracycline and bovine serum albumin (50 μg/ml) for 24 h, and then centrifuged twice. The cell pellet was washed twice in sterile distilled water, resuspended in 5 ml of 50 mM Tris-HCl (pH 8.0) buffer, and lysed by sonication. The culture supernatant was precipitated with trichloroacetic acid at a final concentration of 10%, and the protein pellet was washed three times with 100% ethanol. The protein pellet was resuspended in Laemmli buffer at a 1,000-fold concentration. For immunoblot analysis, the cell lysates and the supernatant proteins were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (15% acrylamide gel) and transferred to a polyvinylidene difluoride membrane (Amersham Biosciences). The mouse polyclonal anti-HpaG antibody was used as the primary antibody, and alkaline phosphatase-conjugated rabbit anti-mouse immunoglobulin G (Pierce) was used as the secondary antibody in the Western blots. Positive signals were detected with One-Step nitroblue tetrazolium-5-bromo-4-chloro-3-indolylphosphate solutions (Pierce).

DNA sequencing and data analysis.

The inserted DNA in pGA16 was digested with appropriate restriction enzymes and subcloned into pBluescript II SK(+) prior to sequencing. Universal and reverse primers were used for the primary reactions, and synthesized primers were then used to sequence both strands completely. All of the DNA sequencing reactions to identify Omegon-Km (Ω-kanamycin resistance cassette) insertions were carried out with the primer HR-1 (5′-TGCTCAATCAATCACCGG-3′). The DNA sequence data were analyzed with the BLAST program at the National Center for Biotechnology Institute (18), MEGALIGN software (DNASTAR), and GENETYX-WIN software (Software Development Inc.). The tRNA gene was identified with the tRNAscan-SE algorithm (31).

Nucleotide sequence accession number.

The complete DNA sequence of the Hrp PAI of X. axonopodis pv. glycines 8ra was deposited in the GenBank database under accession number AF499777.

RESULTS

Isolation of Hrp PAI of X. axonopodis pv. glycines.

To isolate the Hrp PAI of X. axonopodis pv. glycines, we mobilized the strain 8ra genomic library into the nonpathogenic mutants 8-13 and 26-13 and isolated a cosmid clone (pGA16) that complemented the mutant phenotype. Restriction enzyme digestion analysis showed that pGA16 had an insert of approximately 27.8 kb. We also isolated cosmid clones pGA4 and pGA33, which overlapped the sequences at the ends of pGA16. Southern hybridization analysis confirmed that the inserted fragments were colinear with the X. axonopodis pv. glycines 8ra genome (data not shown). Restriction enzyme and Southern hybridization analyses showed that pGA16 and pGA4 overlapped in the 11.8-kb region between hpaP and hpaF and that pGA16 and pGA33 overlapped in the 5.8-kb region between hrcC and hpaH (Fig. 1). The complete DNA sequence of the 27.8-kb insert in pGA16 contained 26 putative open reading frames (ORFs) and part of hpaF (Fig. 1). The region had an overall G+C content of 62%, which is lower than the average G+C content of 65% for xanthomonads (13).

FIG. 1.

Genetic organization and restriction map of the Hrp PAI of X. axonopodis pv. glycines 8ra, which was cloned in pGA16, pGA4, and pGA33. Open arrows indicate the positions and orientations of the hrp, hrc, and hpa genes. Black rectangles above open arrows indicate the PIP boxes. Vertical bars in the pGA16 map indicate the positions and orientations of the Tn3-gus insertions, and the major phenotypes of the mutants are represented below the restriction map. B, BamHI; E, EcoRI; H, HindIII; X, XbaI. Enzyme sites from the vector are shown in parentheses.

The complete DNA sequence of the Hrp PAI region revealed nine hrp genes, nine hrc genes, eight hpa genes, and one tRNA gene. In addition, the sequence contained one ORF that had no apparent role in the bacterium-plant interaction. All of the predicted Hrp, Hrc, and Hpa proteins of X. axonopodis pv. glycines 8ra are compared with related proteins in Table 2. In total, there were seven predicted plant-inducible promoter (PIP) boxes (6) in the Hrp PAI region: Four genes (hpaG, hrpB1, hrcU, and hrcQ) had perfect PIP boxes (TTCGC-N₁₅-TTCGC), while the hpaH, hrpF, and hpaF genes had imperfect PIP boxes (hpaH, TTCGC-N₁₅-TTCGT; hrpF, TTCGC-N₈-TTCGT; hpaF, TTCGC-N₁₆-TTCGC) in their putative promoter regions (Fig. 1).

TABLE 2.

Comparison of Hrp, Hrc, and Hpa proteins of X. axonopodis pv. glycines 8ra with related proteins

X. axonopodis pv. glycines 8ra proteina	% Identity with X. axonopodis pv. glycines protein
	X. axonopodis pv. citrib	X. campestris pv. campestrisc	X. campestris pv. vesicatoriad	X. oryzae pv. oryzaee	Ralstonia solanacearumf	Yersinia enterocoliticag	Burkholderia pseudomalleih	Flagellar biosynthesisi	Others
HpaH (187)	Hpa2 (98.6)	Hpa2 (90.6)	HpaH (94.3)	Hpa2 (79.0)		YsaHj (34.8)	Unknownk (52.2)		IpgFl (33.6)
HpaG (133)	Hpa1 (93.2)	Hpa1 (31.4)	XopA (50.4)	Hpa1 (66.2)
HpaC (141)
HrcC (607)	HrcC (99.7)	HrcC (85.1)	HrcC (97.5)	HrcC (95.2)	HrcC (47.4)	YscC (26.7)	SctC (43.9)		PulDm (16.6)
HrcT (276)	HrcT (100)	HrcT (86.6)	HrcT (97.1)	HrcT (97.5)	HrcT (47.1)	YscT (24.1)	SctT (45.0)	FliR (19.5)
HrpB7 (169)	HrpB7 (98.2)	HrpB7 (66.3)	HrpB7 (91.1)	HrpB7 (91.1)
HrcN (442)	HrcN (99.5)	HrcN (91.0)	HrcN (97.7)	HrcN (97.7)	HrcN (69.0)	YscN (56.5)	SctN (69.7)	FliL (11.0)
HrpB5 (233)	HrpB5 (100)	HrpB5 (78.1)	HrpB5 (96.1)	HrpB5 (93.1)	HrpF (33.5)		SctL (31.3)
HrpB4 (209)	HrpB4 (100)	HrpB4 (69.9)	HrpB4 (94.7)	HrpB4 (93.3)	HrpH (19.6)
HrcJ (253)	HrcJ (99.6)	HrcJ (87.0)	HrcJ (98.0)	HrcJ (92.5)	HrcJ (54.5)	YscJ (29.5)	SctJ (48.2)	FliF (18.6)
HrpB2 (130)	HrpB2 (100)	HrpB2 (72.3)	HrpB2 (96.2)	HrpB2 (96.9)	HrpJ (27.3)		Unknownn (19.5)
HrpB1 (151)	HrpB1 (100)	HrpB1 (83.4)	HrpB1 (96.0)	HrpB1 (96.0)	HrpK (33.1)
HrcU (357)	HrcU (99.2)	HrcU (82.9)	HrcU (NA)	HrcU (92.2)	HrcU (48.7)	YscU (32.2)	SctU (44.7)	FlhB (31.7)
HrcV (640)	HrcV (99.7)	HrcV (94.5)	HrcV (NA)	HrcV (97.2)	HrcV (65.6)	LcrD (44.5)	SctV (63.4)	FlhA (32.8)
HpaP (204)	HpaP (98.5)	HpaP (58.1)	HpaP (NA)	HpaP (88.2)	HpaP (26.9)		Unknowno (22.5)
HrcQ (304)	HrcQ (98.4)	HrcQ (74.0)	HrcQ (90.7)	HrcQ (87.8)	HrcQ (25.3)	YscQ (16.4)	SctQ (26)	FliN (14.6)/FliY (6)	FDR3p (31.3)
HrcR (200)	HrcR (100)	HrcR (96.5)	HrcR (100)	HrcR (98.0)	HrcR (78.0)	YscR (50.5)	SctR (69.5)	FliP (36.5)
HrcS (86)	HrcS (100)	HrcS (97.5)	HrcS (98.8)	HrcS (97.7)	HrcS (62.8)	YscS (34.9)	SctS (51.2)	FliQ (24.4)
HpaA (271)	HpaA (98.9)	HpaA (60.9)	HpaA (77.1)	HpaA (74.5)
HrpD5 (312)	HrpD5 (99.0)	HrpD5 (75.0)	HrpD5 (91.7)	HrpD5 (89.4)	HrpW (29.8)		SctD (28.8)
HrpD6 (80)	HrpD6 (97.5)	HrpD6 (77.5)	HrpD6 (81.3)	HrpD6 (81.3)
HrpE (93)	HrpE (88.2)	HrpE (47.8)	HrpE (81.7)	HrpE (87.1)
HpaB (156)	HpaB (100)	HpaB (89.7)	HpaB (93.6)	HpaB (92.9)			Unknownq (51.7)
HpaD (19)			E3 (31.6)
ORF1 (124)
HrpF (805)	HrpF (98.1)	HrpF (68.9)	HrpF (84.3)	HrpF (82.0)	PopF1 (40.2)/ PopF2 (35.5)				NolXr (47.8)
HpaF (646)	HpaF (99.5)		HpaF (72.6)/HpaG (86.1)	HpaF (83.1)	PopC (22.4)

^aNumber of amino acids shown in parentheses. NA, not available in the database.

^bGenBank accession no. AE008923.

^cGenBank accession no. AE008922.

^dFrom ORF1 to HrpB1, GenBank accession no. U33548; from HrcQ to HpaG, no. AF056246.

^eFrom Hpa2 to HrpB1, GenBank accession no. AF026197; from HrcU to HpaB, no. AB045311; from HrpF to HpaF, no. AB045312.

^fGenBank accession no. AL646053.

^gpYVe8081-encoded proteins; from YscC to YscS, GenBank accession no. AF336309.

^hGenBank accession no. AF074878.

ⁱE. coli; GenBank accession no. NC_000913.

^jChromosomally encoded protein; GenBank accession no. AAF82325.

^kGenBank accession no. AAD05172.

^lShigella flexneri; GenBank accession no. AAL72337.

^mKlebsiella pneumoniae; GenBank accession no. AAA25126.

ⁿGenBank accession no. AAK73237.

^oGenBank accession no. AAD11409.

^pFe-deficiency-related protein from Zea mays (induced by iron deficiency); GenBank accession no. AAK53546.

^qGenBank accession no. AAK73244.

^rSinorhizobium fredii; GenBank accession no. NP_444155.

Core region of xanthomonad Hrp PAI is highly conserved but flanking regions are variable.

The Hrp PAI core region encoded mainly genes that were necessary for type III secretion systems, and the cluster was delimited by hrcC and hpaB in X. axonopodis pv. glycines (Fig. 1). The 20 genes, from hrcC to hpaB, of the core region were present in all five Xanthomonas species tested (Fig. 2) and highly conserved among xanthomonads, with more than 80% identity at the amino acid level except for HpaA, for which there was relatively low identity among the different species (Table 2).

FIG. 2.

Comparisons of Hrp PAIs of five Xanthomonas species. The Hrp PAI core regions and variable flanking regions of X. axonopodis pv. glycines, X. axonopodis pv. citri, X. campestris pv. campestris, X. campestris pv. vesicatoria, and X. oryzae pv. oryzae are represented. The positions and orientations of the hrp, hrc, and hpa genes are shown by open and color-filled arrows. Each gene is named above or below the arrows. The colors represent homologues of the encoded proteins. The organization and size of the genes are depicted based on nucleotide sequence data from the GenBank database. The following sequences were used: X. axonopodis pv. citri genes (GenBank accession no. AE008923); X. campestris pv. campestris genes (AE008922); ORF1 to HrpB1 of X. campestris pv. vesicatoria (U33548); HrcQ to HpaG of X. campestris pv. vesicatoria (AF056246); Hpa2 to HrpB1 of X. oryzae pv. oryzae (AF026197); HrcU to HpaB of X. oryzae pv. oryzae (AB045311); and HrpF to HpaF of X. oryzae pv. oryzae (AB045312).

The HrpB7, HpaA, HrpD6, HrpE, and HpaB proteins were found only in xanthomonads and not in the closely related Hrp PAI of Ralstonia solanacearum (Table 2) (13, 14, 24, 40). Homologs of X. axonopodis pv. glycines hpaH and hpaG were present upstream from the Hrp PAI core region in all five Xanthomonas species (Fig. 2). However, there were no mobile elements in this region in X. axonopodis pv. glycines and pv. citri or in Xanthomonas campestris pv. campestris (Fig. 2). Homologs of the hrpF and hpaF genes of X. axonopodis pv. glycines were identified downstream from the Hrp PAI core region and were highly conserved in Xanthomonas species. However, the hpaF homolog was not present in X. campestris pv. campestris. Instead, the hrpW homolog from P. syringae pathovars and Erwinia amylovora was present in this region. In two X. axonopodis pathovars and X. campestris pv. vesicatoria, the tRNA^Arg gene was located downstream of the Hrp PAI core region. In contrast, tRNA^Arg was absent from the Hrp PAI regions of X. campestris pv. campestris and Xanthomonas oryzae pv. oryzae (Fig. 2).

Phenotypes of hrp, hrc, and hpa mutants.

The pGA16 plasmid was mutagenized with Tn3-gus to investigate the functions of the proteins that are encoded by genes in the Hrp PAI of X. axonopodis pv. glycines. Restriction enzyme analysis and direct sequencing of the mutagenized plasmids mapped a total of 15 insertions in the Hrp PAI region (Fig. 1). Each marker-exchanged mutant was assayed for pathogenicity and the ability to induce HR in pepper plants. All of the mutant strains were complemented in pathogenicity and the ability to induce HR with the appropriate clone. pGA161 complemented the hpaH and hpaC mutants; pGA16037L complemented the hrcC mutant; pGA166 complemented the hrcJ mutant; pGA1604L complemented the hrcU and hrcV mutants; pGA410 complemented the hrpD5, hrpE, hpaB, and hpaD mutants; pGA411 complemented the hrpF mutant; and pGA406L complemented the hpaF mutant (data not shown).

The hrp and hrc genes tested were essential for full pathogenicity and for the induction of HR (Fig. 3A and B). The hpa mutants, which included hpaG, hpaC, and hpaF mutants, had significantly reduced virulence but induced HR in pepper plants (Fig. 3A and B). By contrast, the hpaH mutant showed significantly reduced virulence and did not induce HR in pepper plants (Fig. 3A and B). The wild-type strain 8ra and its hrp, hrc, and hpa mutants did not induce HR in tobacco plants (Fig. 3B). To enhance the credibility of the HR bioassays, cells of P. syringae pv. syringae 61 and its hrcQ_A mutant Pss61-2084 were infiltrated into four different nonhost plants. The wild-type strain 61 induced HR in all four nonhost plants, and its hrcQ_A mutant did not, as expected (Fig. 3B). In order to correlate pathogenicity with bacterial growth in planta, we inoculated approximately 5 × 10⁵ CFU of the wild-type strain 8ra, the nonpathogenic mutant strain 8-13, and the hpaF mutant strain 2-47 into Pella cotyledons. The hrpF mutant strain 8-13 increased about 10-fold in number during the first day of infection and then remained constant for 7 days, whereas strain 8ra increased 1,000-fold (Fig. 4). The hpaF mutant strain 2-47 increased in planta by about 100-fold during the first 7 days, but the cells grew more slowly and to a lower density than wild-type strain 8ra (Fig. 4).

FIG. 3.

(A) Effects of mutations in hrc, hrp, and hpa genes of X. axonopodis pv. glycines 8ra on pathogenicity. a, water control; b, 8ra; c, hrcC mutant 3-33; d, hpaG mutant 70-1; e, hpaH mutant 9-21; f, hpaF mutant 2-47. The soybean leaves and cotyledons were photographed 12 days and 8 days after inoculation, respectively. The soybean leaves that were inoculated with strain 8ra had typical symptoms, i.e., pustule formation surrounded by a chlorotic halo (b). No symptoms developed following infection with either strain 3-33 (c). The hpaG, hpaH, and hpaF mutant strains produced a few pustules underneath the leaves, and the symptoms developed slowly (d, e, and f). In the cotyledon assays, yellow chlorotic areas developed on the wounded detached cotyledons that were inoculated with strain 8ra (b). However, no yellow chlorotic patches were observed in infections with strains 3-33 (c). The cotyledons that were inoculated with the hpaG, hpaH, and hpaF strains (d, e, and f) had small yellowish regions, but symptoms developed very slowly compared to infections involving the wild-type strain. (B) Responses of nonhost pepper, tomato, tobacco, and Chinese cabbage plants to infection with hrp, hrc, and hpa mutants of X. axonopodis pv. glycines 8ra. Responses of pepper (Capsicum annuum L. cv. Chokwang) (a), tomato (Lycopersicon esculentum Mill. cv. Seokwang) (b), tobacco (Nicotiana tabacum cv. Samsun NN) (c), and Chinese cabbage (Brassica campestris L. cv. Karaksin-1) (d). Sites: 1, water; 2, 8ra; 3, hrpG mutant 8raG; 4, hpaH mutant 9-21; 5, hpaG mutant 70-1; 6, hpaC mutant 1-46; 7, hrcC mutant 3-33; 8, hrcU mutant 1-44; 9, hrpF mutant 1-34; 10, hpaF mutant 2-47; 11, Pss61; 12, hrcQ_A mutant Pss61-2084). The pepper, tomato, and Chinese cabbage leaves were photographed 24 h after injection.

FIG. 4.

Growth patterns of X. axonopodis pv. glycines parent strain 8ra, hrpF mutant 8-13, and hpaF mutant 2-47 in soybean (cv. Pella) cotyledons. Bacterial numbers were determined daily after inoculation. The data are shown as the average values for five samples, and the vertical bars indicate the error ranges.

Since all of the hrp mutants induced HR in the two cultivars of Lycopersicon esculentum Mill. cv. Seokwang and cv. Kwangsoo (Fig. 3B), we used other Lycopersicon species to determine whether various tomato species responded differently to X. axonopodis pv. glycines. All of the nonpathogenic mutants were tested for HR by injection into wild-type tomato plants, such as L. chmielewskii, L. hirsutum, L. panviflorum, L. pennelli, L. peravianum, and L. pinpinellifolium. All of the mutants induced HR in all of the wild-type tomato plants tested. When Chinese cabbage was used as the nonhost plant, the wild-type strain 8ra and all hpa mutant strains except the hpaC mutant strain 1-46 induced HR, and all hrp and hrc mutants failed to induce HR (Fig. 3B). The hpaC mutant strain 1-46 exhibited no detectable HR after 24 h (Fig. 3B). This indicates that hpaC has an important role in incompatible interactions between X. axonopodis pv. glycines and Chinese cabbage.

Expression of hrp, hrc, and hpa genes.

To determine whether the hrp, hrc, and hpa genes of X. axonopodis pv. glycines were regulated by HrpG and HrpX, as previously observed in X. campestris pv. vesicatoria and X. oryzae pv. oryzae, we measured the expression levels of the hrp, hrc, and hpa genes in both the wild-type strain 8ra and the hrpG mutant strain 8raG. All of the genes except hpaH were expressed at high levels in the wild-type background when the cells were grown in hrp induction medium, whereas the expression levels were very low in the 8raG derivatives (Fig. 5). The expression of all of the 8raG derivatives was complemented by pLGX3 carrying X. axonopodis pv. glycines hrpG and hrpX (Fig. 5). The hpaF gene was the most strongly expressed gene, while hpaH expression was very low under the conditions used in our experiments. Taken together, our results indicate that all of the hrp, hrc, and hpa genes of X. axonopodis pv. glycines are regulated by HrpG and HrpX.

FIG. 5.

Expression levels of hrp, hrc, and hpa genes in X. axonopodis pv. glycines 8ra, 8raG, 8ra(pLGX3), and 8raG(pLGX3) after growth in XVM2. β-Glucuronidase activities were measured as described in Materials and Methods and are averages of three experiments. One unit of β-glucuronidase was defined as 1 nmol of 4-methylumbelliferon released per bacterium per minute.

HpaG is an Hrp type III-secreted elicitor.

Since HpaG has the features of harpins, which are found in P. syringae pathovars, Erwinia species, and Ralstonia solanacearum, we investigated whether HpaG had HR elicitor activity in various nonhost plants. The wild-type strain 8ra elicited HRs in pepper (Capsicum annuum L. cv. Chokwang), tomato (Lycopersicon esculentum Mill. cv. Kwangsoo), Chinese cabbage, and A. thaliana ecotype Cvi-0. Interestingly, purified HpaG (1 μM) elicited HRs in pepper and A. thaliana ecotype Cvi-0, whereas it failed to elicit HRs in tomato, Chinese cabbage, and A. thaliana ecotype Col-0 and Ler (data not shown). In tobacco plants, HpaG elicited HRs, but the wild-type strain 8ra did not (Fig. 6). To compare the elicitor activities of HpaG with those of other xanthomonad HpaG homologs and a harpin protein, we purified Hpa1 of X. oryzae pv. oryzae, XopA of X. campestris pv. vesicatoria, and HrpN of Erwinia amylovora and injected them separately into tobacco leaves. HRs were observed clearly at 1 μM and partial HRs were observed at 0.5 μM HpaG and HrpN (Fig. 6A). However, XopA did not exhibit any elicitor activities, and Hpa1 showed clear HR at concentrations above 5 μM (Fig. 6A).

FIG. 6.

Comparison of HpaG activity with other known harpin HrpN and harpin-like XopA and Hpa1 (A). Effect of HpaG on the ability to induce HR in tobacco leaves. Sites: 1, 50 mM Tris-HCl (pH 8.0); 2, 8ra(pLAFR3) (2 × 10⁸ CFU/ml); 3, 8ra(pLGX3) (2 × 10⁸ CFU/ml); 4, hrcU mutant 1-44(pLGX3) (2 × 10⁸ CFU/ml); 5, HpaG (1 μM) in 50 mM Tris-HCl (pH 8.0); 6, hpaG mutant 70-1(pLGX3) (2 × 10⁸ CFU/ml); 7, 70-1(pLGXhpaG) (2 × 10⁸ CFU/ml) (B). The tobacco leaves were photographed 24 h after injection.

Western blot analysis was performed to examine HpaG secretion via Hrp type III secretion systems. HpaG was detected in the cell lysates and culture supernatants of strain 8ra harboring pLGX3 (Fig. 7). Although HpaG was not detected in the cell lysates and culture supernatants of the hpaG mutant strain 70-1(pLGX3), it was detected in both the cell lysates and culture supernatants of strain 70-1(pLGXhpaG). HpaG was not detected in the culture supernatant of hrcU mutant strain 1-44 carrying pLGX3, whereas it was detected in cell lysates of this strain (Fig. 7). Therefore, we conclude that HpaG is an Hrp type III-secreted effector protein.

FIG. 7.

Secretion analysis of HpaG. All strains used in this analysis harbored additional copies of the hrpG and hrpX gene in pLAFR3. The approximate sizes of the proteins are shown at the right. Lanes: 1, total cell protein of 8ra(pLGX3); 2, total cell protein of hrcU mutant 1-44(pLGX3); 3, total cell protein of hpaG mutant 70-1(pLGX3); 4, total cell protein of 70-1(pLGXhpaG); 5, supernatant protein of 8ra(pLGX3); 6, supernatant protein of 1-44(pLGX3); 7, supernatant protein of 70-1(pLGX3); 8, supernatant protein of 70-1(pLGXhpaG).

In comparison with other HpaG homologs, HpaG showed high-level homology with Hpa1 of X. axonopodis pv. citri (93.2% identity) and Hpa1 of X. oryzae pv. oryzae (66.2% identity). XopA (50.4% identity) of X. campestris pv. vesicatoria and Hpa1 (31.4% identity) of X. campestris pv. campestris lacked 16 and 9 amino acid residues, respectively, in the internal region of HpaG (Fig. 8). HpaG (21% glycine), Hpa1 (22% glycine) of X. axonopodis pv. citri, and Hpa1 (26% glycine) of X. oryzae pv. oryzae had relatively high ratios of glycine residues, whereas XopA (8% glycine) and Hpa1 (13% glycine) of X. campestris pv. campestris had lower ratios (Fig. 8).

FIG. 8.

Alignment of HpaG of X. axonopodis pv. glycines with Hpa1 of X. axonopodis pv. citri, Hpa1 of X. oryzae pv. oryzae, XopA of X. campestris pv. vesicatoria, and Hpa1 of X. campestris pv. campestris. The alignment was produced with the Clustal X program. Asterisks (*), colons (:), and periods (.) indicate identical amino acid residues, conserved residues, and similar residues, respectively.

Wild-type strain of X. axonopodis pv. glycines requires additional copies of hrpG and hrpX to induce HR in tobacco plants.

Among the various nonhost plants, wild-type strain 8ra was able to induce HRs in pepper, tomato, Chinese cabbage, and A. thaliana ecotype Cvi-0. However, wild-type strain 8ra did not induce HR in tobacco plants (Fig. 3B and 6). In order to identify the factor from X. oryzae pv. oryzae that elicited HR in tobacco plants, we introduced a cosmid library of X. oryzae pv. oryzae KXO85 into X. axonopodis pv. glycines 8ra. Among the transconjugants that harbored genomic clones, we isolated a cosmid clone, pCH85, that conferred on strain 8ra the ability to induce HRs in tobacco plants. We subcloned the 8.5-kb EcoRI fragment from pCH85 into pLAFR3 and sequenced the fragment.

Interestingly, we found that this DNA fragment contained the hrpG and hrpX genes of X. oryzae pv. oryzae. This finding led us to introduce pLGX3 into the wild-type strain 8ra. The resulting strain, 8ra(pLGX3), induced HRs in tobacco plants (Fig. 6). The expression levels of hrp, hrc, and hpa genes were increased 8- to 35-fold in the presence of the additional copies of the hrpG and hrpX genes (Fig. 5). The expression levels of all of the 8raG derivatives carrying pLGX3 were lower than those of 8ra derivatives carrying the same plasmid but higher than those of 8ra derivatives (Fig. 5). This indicates that the wild-type strain of X. axonopodis pv. glycines 8ra requires additional copies of the hrpG and hrpX genes on a multicopy plasmid to induce HRs in tobacco plants. To determine whether HpaG had a major role in inducing HR in tobacco plants, strain 70-1(pLGX3) was injected into tobacco leaves. Strain 70-1(pLGX3) induced no HR, strain 1-44(pLGX3) failed to induce HR, and strain 70-1(pLGXhpaG) showed clear HR in tobacco plants (Fig. 6).

DISCUSSION

In this study, we isolated the Hrp PAI of X. axonopodis pv. glycines and compared it with the PAIs of four related Xanthomonas species. The characteristics of the region, such as the presence of many virulence genes, the low G+C content, and bordering tRNA genes, satisfied the criteria for a PAI of a bacterium (21). The Hrp PAI core region was composed of 20 genes, from hrcC to hpaB, and encoded a type III secretion system that was highly conserved (>90% similarity) among xanthomonads. Some of the proteins encoded by the Hrp PAI core region have been characterized previously in X. campestris pv. vesicatoria (10, 24, 33, 38, 48). A model has been proposed for X. campestris pv. vesicatoria Hrp proteins in type III secretion systems; however, the exact functions of most of these proteins remain to be determined (38). Five transcriptional units have been reported in the Hrp PAI core region of X. campestris pv. vesicatoria, and genes within these units are regulated by HrpG and HrpX (47, 50). Our findings are similar, in that the genes of the Hrp PAI core region of X. axonopodis pv. glycines are regulated by HrpG and HrpX.

The phenotypes of the hrc, hrp, and hpa mutants of X. axonopodis pv. glycines were similar to those of their respective X. campestris pv. vesicatoria mutants. However, there were some discrepancies in earlier comparisons of X. axonopodis pv. glycines and X. campestris pv. vesicatoria. First, Oh and colleagues reported that a mutation in hrcU did not abolish the ability to induce HR in pepper plants (34). However, we found that a hrcU mutation did abolish the ability to induce HR in pepper plants. We believe that the phenotype of our hrcU mutant (strain 1-44) is consistent with that of X. campestris pv. vesicatoria. Second, all of the hrp and hrc mutants of X. axonopodis pv. glycines retained the ability to induce HRs in tomato plants. This suggests that the bacterium uses hrp-independent systems or elicitors to induce HRs in tomato plants. Since it was conceivable that toxic substances produced by X. axonopodis pv. glycines are responsible for inducing HR in tomato leaves, we infiltrated culture supernatants into tomato and tobacco leaves. We found no HR in either tomato or tobacco leaves (J.-G. Kim and I. Hwang, unpublished data). Therefore, it is very unlikely that toxic substances produced by X. axonopodis pv. glycines induced HR in tomato leaves.

Mutations in the hpaB and hpaD genes of X. axonopodis pv. glycines reduced bacterial pathogenicity, although an ORF1 mutant retained full pathogenicity. Therefore, this region may not contribute qualitatively to pathogenicity, and may contain some redundancies. The corresponding regions in other Xanthomonas species are poorly characterized compared to the Hrp PAI core region genes. The exact contributions to parasitic fitness of hpaD and ORF1 of X. axonopodis pv. glycines and of the E3 gene of X. campestris pv. vesicatoria are unclear.

HpaF is one of the bacterial leucine-rich repeat (LRR) proteins, and homologs are present in the Hrp PAI region of other Xanthomonas species with the exception of X. campestris pv. campestris (Fig. 2). Although the LRR motif is of considerable biological interest, the mechanism underlying LRR activity is unknown (28, 29). The LRR proteins appear to have essential roles in the pathogenicity of animal-pathogenic bacteria. The outer membrane protein YopM of Yersinia pestis, which consists of 13 tandem LRR repeats, is an effector protein that is secreted by the type III secretion pathway and plays an important role in the initial stages of infection (37). The InlB surface protein of Listeria monocytogenes carries eight tandem LRR repeats that are sufficient for entry into mammalian cells (9). Ours is the first report that a mutation in the hpaF gene of a plant-pathogenic bacterium affects pathogenicity and symptom development.

PopC of R. solanacearum and HpaG of X. campestris pv. vesicatoria, which have similar LRR domains, have undetermined roles, since knockout mutations in these genes had no effect on either pathogenicity or HR induction (20, 33). Southern analysis showed that other bacterial strains (Ralstonia solanacearum, Agrobacterium tumefaciens, Pseudomonas syringae pathovars, Pseudomonas fluorescens, Bacillus subtilis, and E. coli) did not hybridize with the hpaF-containing probe, whereas Erwinia amylovora Ea321 showed two weak signals (data not shown). This indicates that HpaF belongs to the bacterial LRR-type protein group and is unique to xanthomonads. Based on the structures and possible functions of known LRR proteins (29), the LRR motif of HpaF may be involved in the binding of a specific ligand of the host cell cytoplasm during disease development. Thus, we speculate that HpaF disturbs the host defense mechanisms that are directed against invading pathogenic bacteria, thereby delaying symptom development. Interestingly, although hpaF was expressed at high levels in XVM2 medium, we did not detect secreted HpaF by Western blot analysis (J.-G. Kim and I. Hwang, unpublished data). Assuming that HpaF is indeed an effector protein, this indicates that HpaF may not be stable in the culture medium that we used. Noël et al. also failed to detect the HpaF homolog in X. campestris pv. vesicatoria, and they concluded that hpaG RNA was not translated under the conditions used (33).

The region downstream from the Hrp PAI core region of X. axonopodis pv. glycines contains three genes, hrpF, hpaF, and tRNA^Arg. It has been reported that the hrpF mutant has a typical hrp phenotype and that HrpF is an effector protein in X. campestris pv. vesicatoria that is not crucial for the secretion of other effector proteins (38). Since we obtained similar results with X. axonopodis pv. glycines (J.-G. Kim and I. Hwang, unpublished data), we believe that HrpF does not affect the secretion of other effector proteins into the supernatant in the secretion assay and that it may have other functions that are crucial to full pathogenicity and the ability to induce HRs in nonhost plants.

In this regard, it is worth considering the model proposed by Rossier et al., which proposes that HrpF functions as a translocator of effector proteins into the host cell (38). The findings that the hrpW homologs of P. syringae and Erwinia amylovora are located between the hrpE and hrpF genes in X. campestris pv. campestris and that the tRNA^Arg gene is absent from both X. campestris pv. campestris and X. oryzae pv. oryzae indicate that this region varies depending on the species and pathovar.

The region upstream from the hrcC gene showed variability in five Xanthomonas species and encoded effector proteins, and in inserion sequence elements. All of the Hpa proteins in the region, i.e., HpaH, HpaG, and HpaC, influenced the pathogenicity of X. axonopodis pv. glycines. Since similar phenotypes have been observed for mutants of X. campestris pv. vesicatoria and X. oryzae pv. oryzae (33, 51), we believe that these genes contribute to pathogenicity on different levels. The findings that hpaH was expressed at a very low level under XVM2 conditions and that the hpaH mutant lost the ability to induce HR in pepper plants indicate that small amounts of the HpaH protein can mediate effector functions. Nonetheless, the exact role of HpaH in the interactions between X. axonopodis pv. glycines and nonhost plants remains to be elucidated. Interestingly, the hpaC mutant elicited HRs in pepper and tomato plants but not in Chinese cabbage. This result implies that HpaC functions in a specific interaction between the bacterium and Chinese cabbage.

Given the fact that both ends of the Hrp PAI core region vary depending on the species and pathovar, we believe that the concept of tripartite mosaic structures composed of a cluster of type III secretion genes and bounded by exchangeable effector and conserved effector loci (as shown for P. syringae pathovars) is not applicable to xanthomonads. It appears that the core region of the xanthomonad Hrp PAI is highly conserved and, unlike the Hrp PAI of P. syringae, both flanking regions are relatively diverse.

In addition, we sequenced approximately 10 kb extending the end of the tRNA^Arg gene contiguously from pGA4. We found that the gene arrangement in this region is the same as that in X. axonopodis pv. citri but differs from that of X. campestris pv. campestris. No repeat sequences, insertion sequences, mobile elements, or plasmid genes were found downstream from the tRNA^Arg gene in X. axonopodis pv. glycines or X. axonopodis pv. citri. In X. campestris pv. campestris, the IS1478 transposase was located 8.5 kb downstream from hrpF. Interestingly, we found a truncated insertion sequence element showing 76% similarity with ISxcC1 transposase in the 47-codon interval between the hpaF and tRNA^Arg genes in the two X. axonopodis pathovars. Since we only observed part of the insertion sequence element and a frameshift mutation in the element, we believe that this could be evidence that PAI is unstable, as suggested by Hacker et al. (21). Therefore, we believe that this tRNA^Arg gene of X. axonopodis pv. glycines is the right end of the Hrp PAI, as suggested by the fact that about 75% of the virulence genes associated with bacterial pathogenicity are flanked by tRNA genes (23).

When we sequenced approximately 10 kb extending the end of the hpaH gene contiguously from pGA33, the gene arrangement in the two X. axonopodis pathovars was identical, while it was dissimilar in X. campestris pv. campestris. Several conserved hypothetical proteins were found upstream from hpaH in the two X. axonopodis pathovars. Since our Tn3-gus mutants, such as 4-12 and 9-14, retained wild-type phenotypes, we believe that this region is not part of Hrp PAI. Taking the data together, we believe the region between the hpaH and tRNA^Arg genes delimits the Hrp PAI of X. axonopodis pv. glycines.

Zhu et al. reported that the Hpa1 of X. oryzae pv. oryzae resembled the harpins of P. syringae pathovars and Erwinia species and the harpin-like PopA of R. solanacearum and showed that the hpa1 mutant of X. oryzae pv. oryzae had reduced pathogenicity (51). Recently, Hpa1 homologs were found in X. campestris pv. vesicatoria, X. axonopodis pv. citri, and X. campestris pv. campestris (13, 33). XopA of X. campestris pv. vesicatoria is a type III-secreted effector protein that is necessary for both growth in planta and full avirulence (33). Nevertheless, it is not clear if either Hpa1 or XopA is an HR elicitor on nonhost plants. Noël et al. reported that an XopA-glutathione S-transferase fusion protein did not induce an HR-like response in tobacco leaves (33). Here, we report that HpaG is a true HR elicitor, with activity equivalent to that of the Erwinia harpin HrpN, and that HpaG induces HRs in various nonhost plants.

This is the first HR-eliciting protein identified in xanthomonads. Similar to the phenotype of the hrpN mutant of Erwinia amylovora, the hpaG mutant that carried HrpG and HrpX induced negligible HRs in tobacco plants. However, the hpaG mutant induced HRs in pepper and tomato plants, and HpaG failed to elicit HRs in tomato and Chinese cabbage. This suggests that there are specific interactions between X. axonopodis pv. glycines and nonhost plants and that various elicitor proteins are responsible for inducing HR on nonhost plants. This hypothesis is supported by the recent finding that a homolog of HrpW, which is a harpin protein in Erwinia amylovora and P. syringae pathovars, was found during whole genome sequence analysis of two Xanthomonas species (13).

Among the HpaG homologs of xanthomonads, only HpaG from X. axonopodis pv. glycines exhibited a true harpin-like elicitor activity. Although all of the homologs exhibited high-level identity at the amino acid level, interesting differences were seen in the amino acid alignment of HpaG and XopA. XopA lacked 16 amino acid residues that corresponded to positions 59 to 74 in HpaG of X. axonopodis pv. glycines. We believe that this discrepancy is critical for HpaG homologs to act as elicitors on nonhost plants. Currently, we are attempting to determine the critical amino acid residues in HpaG that confer HR activity.

The wild-type strain 8ra of X. axonopodis pv. glycines and the other strains tested did not induce HRs in tobacco plants. The interactions of plants and plant-pathogenic bacteria have long been considered peculiar (26, 35). Unexpectedly, we found that cells of X. axonopodis pv. glycines required additional copies of hrpG and hrpX on multicopy plasmids to induce HR in tobacco plants. A similar phenomenon was observed during the mutational analysis of hrpG of X. campestris pv. vesicatoria (49). Three mutations in the hrpG gene of X. campestris pv. vesicatoria rendered constitutive the expression of hrp genes in medium that would normally suppress this expression and induced HR in tobacco plants (N. tabacum cv. Xanthi), while the wild-type strain induced only a weak chlorotic and necrotic reaction 2 to 3 days after infection (49). Therefore, we believe that the hrpG and hrpX genes are expressed at very low levels when X. axonopodis pv. glycines interacts with tobacco cells and that this results in very low expression levels of all of the genes under the control of HrpG and HrpX. We conclude that the expression levels of hrpG and hrpX in tobacco plants are important for inducing HR in plants. It will be interesting to discover the limiting factors in tobacco plants that cause low-level expression of hrpG and hrpX.