Comparative Genomic Analysis of the pPT23A Plasmid Family of Pseudomonas syringae

Abstract

Members of the pPT23A plasmid family of Pseudomonas syringae play an important role in the interaction of this bacterial pathogen with host plants. Complete sequence analysis of several pPT23A family plasmids (PFPs) has provided a glimpse of the gene content and virulence function of these plasmids. We constructed a macroarray containing 161 genes to estimate and compare the gene contents of 23 newly analyzed and eight known PFPs from 12 pathovars of P. syringae, which belong to four genomospecies. Hybridization results revealed that PFPs could be distinguished by the type IV secretion system (T4SS) encoded and separated into four groups. Twelve PFPs along with pPSR1 from P. syringae pv. syringae, pPh1448B from P. syringae pv. phaseolicola, and pPMA4326A from P. syringae pv. maculicola encoded a type IVA T4SS (VirB-VirD4 conjugative system), whereas 10 PFPs along with pDC3000A and pDC3000B from P. syringae pv. tomato encoded a type IVB T4SS (tra system). Two plasmids encoded both T4SSs, whereas six other plasmids carried none or only a few genes of either the type IVA or type IVB secretion system. Most PFPs hybridized to more than one putative type III secretion system effector gene and to a variety of additional genes encoding known P. syringae virulence factors. The overall gene contents of individual PFPs were more similar among plasmids within each of the four groups based on T4SS genes; however, a number of genes, encoding plasmid-specific functions or hypothetical proteins, were shared among plasmids from different T4SS groups. The only gene shared by all PFPs in this study was the repA gene, which encoded sequences with 87 to 99% amino acid identityamong 25 sequences examined. We proposed a model to illustrate the evolution and gene acquisition of the pPT23A plasmid family. To our knowledge, this is the first such attempt to conduct a global genetic analysis of this important plasmid family.

The species Pseudomonas syringae comprises a group of plant-associated bacteria that act either as epiphytes or as plant pathogens causing important diseases with significant economic consequences (35). Although the P. syringae species as a whole causes plant diseases on a multitude of agriculturally important plant species, individual P. syringae strains typically have a more limited host range (3, 25, 35). In the past decade, significant progress in unveiling the mechanisms of pathogenesis of P. syringae and other plant pathogens has been made (3, 9, 36, 51). With either complete or draft sequences of three P. syringae pathovars currently available, this organism is an attractive model for molecular studies of plant-pathogen interactions (9, 10). A functional hypersensitive response and pathogenicity (hrp pathogenicity island [PAI]) type III secretion system (TTSS) that directs the delivery of effector proteins into host cells has been shown to be the key pathogenicity factor required for P. syringae to colonize and parasitize host plants (3, 9, 40, 46, 51). While the complete repertoire of effectors of any one P. syringae strain is still unknown, several recent studies have revealed that this number can be as large as 58 in P. syringae pv. tomato DC3000 (10, 16, 24, 50, 72). To date, more than 150 effector genes have been identified in P. syringae, and it has been suggested that variations in host specificity may be due to differences in the effector complements of individual P. syringae strains (3, 9, 29, 30).

TTSS effectors are encoded by genes that are either linked with the hrp PAI, dispersed throughout the chromosome, or harbored on indigenous plasmids (2, 4, 10, 37, 38, 52, 64). In some cases, alleles of individual effector genes, such as avrPphE, are found in either chromosomal or plasmid locations in different P. syringae strains such as P. syringae pv. tomato DC3000 and P. syringae pv. maculicola M6 (10, 52). Commonly, effector genes are associated with mobile genetic elements and presumably have been acquired by horizontal gene transfer (9, 32, 44).

Most strains of P. syringae contain one to several indigenous plasmids, with the majority of these plasmids belonging to the well-described pPT23A plasmid family (26, 58, 59, 64). Plasmid-borne PAIs and effector gene clusters, along with mobile elements, have been found on several of these pPT23A family plasmids (PFPs) and have been demonstrated to play a role in the virulence of P. syringae (1, 37, 52, 61). Besides effector proteins, PFPs have also been shown to encode other virulence and fitness determinants, including proteins conferring resistance to copper and antibiotics and tolerance to UV radiation, proteins involved in phytotoxin and hormone production, and chemotaxis transducer proteins (1, 7, 27, 28, 45, 47, 57, 62, 65, 66, 67). The role of PFPs in pathogen virulence and epiphytic and in planta bacterial growth has been demonstrated in a number of P. syringae pathosystems (37, 52, 67).

Many PFPs are known to be conjugative plasmids (26, 62, 67, 68). Of the recently sequenced PFPs from P. syringae, two completely different conjugation or type IV secretion systems (T4SS) are presented (10, 61, 64). In pPSR1 from P. syringae pv. syringae, pPh1448B from P. syringae pv. phaseolicola, and pPMA4326A from P. syringae pv. maculicola, the putative T4SS is a type IVA system, i.e., the VirB-VirD4 conjugation system, encoded by the virB1 through virB11 and virD4 genes (61, 64). In contrast, the T4SS encoded by both pDC3000A and pDC3000B from P. syrinage pv. tomato DC3000 is a type IVB system, i.e., the tra conjugation system (10). In addition to functioning in conjugation, some bacterial T4SSs are capable of delivering effector proteins or toxins into host cells during infection, thereby acting as virulence factors (6, 11, 15, 19, 22, 60). Furthermore, recent studies have provided evidence that T4SSs and TTSSs may have overlapping functions (48). However, whether T4SSs present on PFPs play a role in virulence and how these T4SSs are distributed among various PFPs are still unknown.

Despite the obvious importance of plasmids to the overall biology of P. syringae, research is still lacking in this area, especially in terms of defining genetic and evolutionary relationships among native plasmids of P. syringae (67). However, with the full sequences of eight PFPs from four pathovars currently available, this genomic information, combined with previous experimental data, makes it possible for comparative genetic analyses of this important group of plasmids. To our knowledge, a global genetic analysis of the pPT23A plasmid family has not been conducted. Here, we used a macroarray containing 161 known PFP-carried genes and other important genes from various genomes of P. syringae to rapidly estimate and compare the gene contents of 23 newly analyzed PFPs, along with eight sequenced PFPs, from 12 P. syringae pathovars belonging to four genomospecies. This analysis revealed that PFPs can be distinguished by the conjugative transfer system encoded and that this feature is also linked with the propensity to carry additional gene sets. Based on our analysis and previous reported data, we propose a model for the evolution of the pPT23A plasmid family.

MATERIALS AND METHODS

Bacterial strains and plasmids.

The bacterial strains and plasmids utilized in this study are listed in Table 1. P. syringae strains were grown overnight at 28°C with continuous shaking at 250 rpm in mannitol-glutamate broth (43) supplemented with 0.1% yeast extract. Antibiotics were added to the culture medium at the following concentrations: rifampin, 100 μg ml⁻¹; kanamycin, 30 μg ml⁻¹; ampicillin, 100 μg ml⁻¹.

TABLE 1.

P. syringae strains and plasmids used in this study

P. syringae pathovar	Genomospecies	Strain	Host	Plasmida	Estimated size (kb)	Origin or reference
syringae	I	A2	Ornamental pear	pPSR1	72.6	64
		B86-17	Bean	pB8617A	70	63
		HS191	Millet	pCG131	60	59
		4918	Butterfly pea	pPSS4918	67	63
aesculi	II	0893-23	Buckeye	pPA0893A	120	63
				pPA0893B	100
				pPA0893C	70
eriobotryae	II	301062	Loquat	pPSE	100	MAFFc
lachrymans	II	1188-1	Cucumber	pPSL1188	67	63
phaseolicolab	II	1449B	Bean	pPSPh1449B	154	37
		1448	Bean	pPh1448A	131.95
				pPh1448B	51.7
savastanoi	II	0485-9	Oleander	pPS0485A	100	63
				pPS0485B	85
				pPS0485C	50
		0693-10	Oleander	pPS0693A	75	63
tabacib	II	0893-29	Tobacco	pPSTA0893A	82	63
apii	III	1089-5	Celery	pPSA1089A	80	58
maculicola	III	M6	Crucifers	pFKN	39.5	52
		ES4326	Crucifers	pPMA4326A	46.7	61
				pPMA4326B	40.1
		88-10	Cauliflower	pPSM8810	89	70
		90-32	Cauliflower	pPSM9032A	95	70
				pPSM9032B	50
tomato	III	DC3000	Tomato	pDC3000A	73.6	10
				pDC3000B	67.5
		UPN2A	Tomato	pPT23A	100	59
		UPN2B	Tomato	pPT23B	85	58
		OK-1	Tomato	pOK1A	100	59
garcaeb	IV	2708	Coffee	pPG2708	40	NCPPBc
porrib	IV	3364	Leek	pPP3364	80	NCPPBc

^aGenBank accession numbers are as follows: pDC3000A, NC_004633; pDC3000B, NC_004632; pFKN, NC_002759; pPSR1, NC_005205; pPMA4326A, NC_005918; pPMA4326B, NC_005919. For pPh1448A and pPh1448B, partially completed genome sequences are available at the TIGR website (http://www.tigr.org).

^bStrains contain additional plasmids which were not analyzed in the present study.

^cNCPPB, National Collection of Plant Pathogenic Bacteria; MAFF, Ministry of Agriculture, Food, and Fisheries.

Plasmid DNA extraction and purification.

For large-scale plasmid preparation, plasmid DNAs were extracted from 100- or 250-ml overnight cultures of P. syringae by a modified alkaline lysis method (41). Briefly, bacterial cells were resuspended in E buffer (40 mM Tris-acetate, 2 mM EDTA [pH 7.9]) and lysed by the addition of 2 volumes of lysing buffer (50 mM Tris, 3% sodium dodecyl sulfate [pH 12.6]). Cell lysates were then incubated at 65°C for 40 min, followed by extractions with phenol-chloroform and chloroform. Plasmid DNA in the supernatants was then precipitated with ethanol and resuspended in TE buffer. For purification, plasmids were separated in 0.7% agarose gels and then isolated from the agarose gel with the QIAEX II agarose gel extraction kit (QIAGEN Inc., Valencia, Calif.) according to the manufacturer's instructions.

Macroarray printing.

All genes printed on the macroarray and their brief descriptions are listed in Table 2. Oligonucleotide primer sequences for the amplification of genes from four PFPs were selected from the published sequences of plasmid pFKN of P. syringae pv. maculicola M6, pPSR1 of P. syringae pv. syringae A2, and pDC3000A and pDC3000B of P. syringae pv. tomato DC3000 (10, 52, 64). Primers to amplify other genes on the chromosome or plasmids of P. syringae were generated with sequences in the National Center for Biotechnology Information databases. All of the oligonucleotide primer sequences used in this study and the expected sizes of the PCR products are available upon request. We used purified plasmid templates of pPSR1, pFKN, pDC3000A, and pDC3000B and genomic DNA preparations from P. syringae pv. maculicola ES4326, P. syringae pv. tomato DC3000 and PT23, and other P. syringae pathovars as sources for the sequences printed on the array. The expected size and purity of each individual sequence utilized were confirmed by gel electrophoresis, and PCR products were gel purified prior to use.

TABLE 2.

Groups of genes printed on the macroarray based on plasmid and bacterial origin

Gene or IS	Description of product	Functional groupa
pPSR1
rulB	Ultraviolet light resistance protein B	3
orf14	Surface anchor	4
orf18-19	Hypothetical protein	6
orf22	Hypothetical protein	6
orf23-24	ISPS1-a/b	3
orf29-30	Conserved protein	6
orf31	Peptidase	7
virB1	VirB1	1
virB5	VirB5	1
virB6	VirB6	1
virB8	VirB8	1
virB9	VirB9	1
virB11	VirB11	1
nic	Nickase	4
orf20	Unknown	6
orf21	Chemotaxis	3
orf25	Chemotaxis	3
dapC	Aminotransferase, succinyl-diaminopimelate	7
dapE	Desuccinylase	7
hopPmaD	HopPmaD	3
pyoS3	Pyocin S3 immunity protein	7
virB4	VirB4	1
orf	Putative transcriptional regulator	5
virB11	VirB11	1
virB2	VirB2	1
virB3	VirB3	1
virB7	VirB7	1
virB10	VirB10	1
virD4	VirD4	1
pDC3000A
rulA	Ultraviolet light resistance protein A	3
A0004	Site-specific recombinase	4
avrPpiB2	Avirulence protein AvrPpiB2	3
A0008	Partition protein, ParB family	4
A0009	ATPase, ParA family	4
A0011	Resolvase (putative)	4
A0013	ISPsy4, transposase	3
A0014	ATPase, ParA family	4
A0017	Hypothetical protein	6
A0021	Resolvase (putative)	4
A0022	Hypothetical protein	6
A0023	Hypothetical protein	6
A0024	Hypothetical protein	6
A0025	Conserved hypothetical protein	6
A0028	Mobilization protein MobB	4
A0029	Mobilization nuclease MobA	4
A0031	Conserved hypothetical protein	6
lsc-3	Levansucrase	3
A0034	GGDEF domain protein	7
A0035	GntR family transcriptional regulator	5
A0037	Major facilitator family transporter	4
A0040	Conserved hypothetical protein	6
pilT	PilT protein	4
A0043	TraH protein (putative)	2
traI	Relaxase	2
A0045	Type IV pilus biogenesis protein	2
A0047	Endonuclease	7
traT	TraT protein	2
A0065	Exclusion-determining protein	7
trbA	TrbA protein	2
trbB	TrbB protein	2
A0068	TraG/TraD conjugal transfer protein	2
A0070	Hypothetical protein	6
A0071	GntR family transcriptional regulator	5
pDC3000B
B0075	TrbB protein (putative)	2
trbC	TrbC protein	2
B0077	GntR family transcriptional regulator	5
B0003	Conserved hypothetical protein	6
B0005	Phosphoesterase family protein	7
B0007	PilM protein (putative)	4
B0009	Membrane protein (putative)	7
B0012	PFAM protein family HMM	7
B0013	Conserved hypothetical protein	6
B0016	Conserved domain protein	6
stbB	Plasmid stability protein StbB	4
stbC	Plasmid stability protein StbC	4
B0020	Hypothetical protein	6
B0021	Conserved hypothetical protein	6
B0025	Conserved hypothetical protein	6
B0027	Hypothetical protein	6
B0030	Conserved hypothetical protein	6
B0032	MobC protein	4
B0042	Conserved hypothetical protein	6
B0046	Conserved hypothetical protein	6
B0047	Hypothetical protein	6
B0049	PilT protein (putative)	4
B0050	TraH protein (putative)	2
B0051	TraI protein	2
B0052	Type IV pilus biogenesis protein	2
B0053	Hypothetical protein	6
traK-1	TraK protein	2
B0057	DNA primase TraC	2
traL-1	TraL protein	2
traM-1	TraM protein	2
traN-1	TraN protein	2
traO-1	TraO protein	2
traP-1	TraP protein	2
traQ-1	TraQ protein	2
traR-1	TraR protein	2
traU-1	TraU protein	2
traW-1	TraW protein	2
traX-1	TraX protein	2
traY-1	TraY protein	2
B0072	Conserved hypothetical protein	6
B0073	PIN domain protein	7
pFKN
orf2	AvrPphE	3
orf4	Unknown function	6
avrRpm1	AvrRpm1	3
orf7	Unknown	6
orf8	Unknown	6
parA	Putative partition protein	4
orf11	Unknown	6
orf12	Chemotaxis proteins	3
orf13	Unknown	6
orf14	Unknown	6
orf15	Unknown	6
orf16	Inorganic phosphatase	7
orf17	Unknown	6
orf18	Putative sugar transporter	7
orf20	Unknown	6
orf21	Putative lipoprotein	4
orf22	Putative integrase	4
orf23	Putative transcriptional regulator	5
orf24	Putative transcriptional regulator	5
orf26	Putative transcriptional regulator	5
orf27	Putative transcriptional regulator	5
orf28	Unknown	6
orf29	Putative transcriptional regulator	5
P. syringae pathovarsb
iaaH	Indoleacetamide hydrolase	3
iaaM	Tryptophan 2-monooxygenase	3
iaaL	IAA-lysine synthetase	3
cfl	Coronatine biosynthesis protein	3
virPphAPgy	Avirulence protein (virPphAPgy)	3
avrPphDPgy	AvrPphDPgy	3
avrA	Avirulence protein A	3
avrF	Avirulence protein F	3
efe(pETH2)	Ethylene-forming enzyme	3
avrPpiE	AvrRps4 protein	3
hrmA	HrmA	3
IS801	IS 801	3
P. syringae pv. tomato     DC3000 and PT23
hslV	Heat shock protein HslV	7
hslU	Heat shock protein HslU	7
hrpK	HrpK	3
hrpS	Transcriptional regulator HrpS.	3
HrpL	RNA polymerase sigma factor HrpL	3
gacS	Response regulator	3
hrpR	Transcriptional regulator HrpR	3
hrcC	Type III secretion protein HrcC	3
avrD	Avirulence protein D
HopPsyApto	HopPstApto	3
HopPtoB	HopPtoB	3
HopPtoA1	HopPtoA1	3
HrpApto	HrpApto	3
HrpWpto	HrpWpto	3
HrpZpto	HrpZpto	3
HolptoT	HolptoT	3
HolPtoX	HolPtoX	3
avrRpt2	AvrRpt2	3
avrEpto	Avirulence protein AvrE	3
avrPto	AvrPto	3
avrPphB1	AvrPphB1	3
repA	Replication protein A	4
P. syringae pv.     maculicola ES4326
HopPmaA	HopPmaA	3
HopPmaB	HopPmaB	3
HopPmaD	HopPmaD	3
HopPmaG	HopPmaG	3
HopPmaH	HopPmaH	3
HopPmaI	HopPmaI	3
HopPmaJ	HopPmaJ	3
HopPmaK	HopPmaK	3
HopPmaL	HopPmaL	3
HolPmaN	HolPmaN

^aFunctional groups are as follows: 1, type IVA secretion system (VirB-VirD4); 2, type IVB secretion system (tra); 3, hrp, hop, avr, vir, and fitness-related genes; 4, repA and plasmid-related genes; 5, transcriptional factor; 6, unknown, hypothetical genes; 7, others.

^bP. syringae pathovars include P. syringae pv. syringae, glycinea, pisi, and savastanoi.

PCR products of selected genes were diluted 1:1 in a denaturing solution (1 M NaOH, 5 M NaCl) and placed in a 96-well microtiter plate just before printing. The denatured PCR products were then deposited in duplicate on a positively charged 7.5- by 11.5-cm nylon membrane with the Beckman 96-pin high-density replicating tool (Biomek 2000 laboratory automation workstations; Beckman Coulter, Inc., Fullerton, Calif.). After spotting was completed, the membranes were denatured again in 1.5 M NaCl-0.5 M NaOH and neutralized in 1.5 M NaCl-0.5 M Tris-HCl (pH 7.5)-1 mM EDTA. DNA fragments were then cross-linked to the membrane with a UV transilluminator (120 mJ; Stratalinker; Stratagene, La Jolla, Calif.).

Plasmid labeling and macroarray hybridization.

Probes were generated from whole plasmids as the DNA template. Purified plasmids were first denatured by being mixed with 1/10 volume of 2 M NaOH-2 mM EDTA for 5 min at room temperature. The denatured plasmids were precipitated with ethanol and resuspended in water for direct labeling. Plasmid labeling with [³²P]dCTP was accomplished with the Random Primers DNA labeling system (Invitrogen, Carlsbad, Calif.) according to the manufacturer's instructions. Macroarray hybridizations were carried out by standard DNA hybridization techniques. Hybridizations at 65°C, followed by high-stringency washes, were performed as described previously (71). Seven genes (repA, virB2, virB3, virB7, virB10, virD4, and the gene encoding the IS801 transposase) that were not printed on the array were labeled individually with digoxigenin-11-dUTP (Genius kit; Boehringer Mannheim, Indianapolis, Ind.) according to the manufacturer's instructions for use as probes and were hybridized to all denatured, purified plasmids blotted on separate nylon membranes.

Analysis of repA sequences.

The repA genes of individual PFPs were amplified by using the primer pair 532 and 1588 (59), which flank a fragment of 1,399 bp containing 1,279 bp of the repA coding region plus 120 bp upstream of the putative start codon (26), and the remaining coding sequences of repA were amplified with degenerate primer pair repA-F1 (5′-AGCTTCAAGAYCAGGGMAA-3′) and repA-R2 (5′-ARRTCCATCARYCGGTCRAA-3′).PCR amplifications with the primer pair 532 and 1588 were performed in a 50-μl reaction volume containing 1× PCR buffer, 4 mM MgCl₂, 0.75 mM (each) deoxynucleoside triphosphate, 1 pmol of each primer, 1.0 U of Taq polymerase, and 1 μl of purified plasmid DNA. PCR was carried out as follows: 1 cycle at 94°C for 5 min; 35 cycles at 94°C for 1 min, 55°C for 1 min, and 72°C for 1.5 min; and then 72°C for 10 min. PCR amplifications with the primer pair repA-F1 and repA-R2 were the same as those for the primer set 532 and 1588 except that 2 mM MgCl₂ and an annealing temperature of 50°C were used. PCR products were purified with a gel extraction kit (QIAGEN Inc.), and the purified fragments were cloned into the pGEM-T Easy vector (Promega, Madison, Wis.) and sequenced at the Genomics Technology Support Facility at Michigan State University. The RepA protein sequences encoded by plasmids pDC3000A, pDC3000B, pPMA4326A, pPMA4326B, pPT23A, pPSR1, and pFKN were obtained from GenBank (accession numbers AE016855, AE016854, AY603979, AY603980, AJ224509, AY342395, and AF359557, respectively). The RepA sequences encoded by pPh1448A and pPh1448B were obtained from The Institute for Genomic Research (TIGR) website (www.tigr.org).

Phylogenetic analyses.

Amino acid alignments used for phylogenetic analyses were done with ClustalW, version 1.83 (European Bioinformatics Institute, Cambridge, United Kingdom). The neighbor-joining method was utilized to analyze sets of aligned amino acid sequences to generate phylogenetic trees (54). Trees were visualized with Treeview, version 1.6.6.

Nucleotide sequence accession numbers.

Sequences for the repA gene from newly analyzed PFPs that were generated in this study were deposited in GenBank with accession numbers AY768793 through AY768807.

RESULTS

Construction and hybridization analyses of a macroarray.

The genes printed on the macroarray could be divided into seven groups, including type III secretion and effector genes, known avirulence genes, virulence genes, and fitness-related genes (50 genes); T4SS genes (type IVA, 12 genes; type IVB, 21 genes); plasmid-specific function genes (20 genes); transcription factor genes (8 genes); genes encoding unknown or hypothetical protein known to be encoded on PFPs (33 genes); and other genes (17 genes). In total, fragments of 134 known genes carried on plasmids and 37 genes carried on chromosomes were amplified by PCR from various P. syringae strains (Table 2). Hybridizations of macroarrays, with ³²P-labeled whole plasmids as probes, yielded consistent, repeatable results (Fig. 1). The hybridization results were visually scored based on the size of the spot and the degree of the hybridization signal compared to the background signal (blank spot and chromosomal genes used as negative controls, such as heat shock protein genes hslV and hslU) (Fig. 1).

FIG. 1.

Representative plasmid macroarray hybridization images for PFPs containing the type IVB (A) and type IVA (B) secretion system genes. A total of 164 genes were printed in duplicate on a 7.5- by 11.5-cm membrane. The position of each gene on the array was labeled as a combination of P (plate), C (column), and R (row). (A) Array hybridized with pPSM8810 from P. syringae pv. maculicola strain 88-10; (B) array hybridized with pPSS4918 from P. syringae pv. syringae strain 4918. Arrowheads, spots containing the blank and chromosomal genes used as negative controls.

Twenty-four PFPs (including pPSR1) were isolated and purified from 12 pathovars of P. syringae (Table 1). One, two, or even three plasmids belonging to the pPT23A plasmid family were recovered from single strains of P. syringae as reported previously (10, 59, 61, 67). In silico analyses of eight fully sequenced PFPs (pDC3000A, pDC3000B, pFKN, pPSR1, pPh1448A, pPh1448B, pPMA4326A, and pPMA4326B) were also conducted, and data were presented along with the 23 newly analyzed PFPs. The in silico data for pPSR1 were verified by hybridization of the purified plasmid to the macroarray. The estimated sizes of the plasmids examined ranged from 40 to 150 kb (Table 1). Since agarose gel separation with 0.7% gels could effectively distinguish plasmids within only a few kilobases, such as pDC300A and pDC3000B (data not shown), we found that, with the QIAEX II agarose gel extraction kit, plasmids up to 150 kb could be easily isolated and purified from agarose gels. This method was easier to utilize and generated plasmids of higher purity than methods using traditional CsCl-ethidium bromide gradients for plasmid purification.

Identification of four distinct subgroups of the pPT23A plasmid family.

Recent P. syringae genome and plasmid sequencing efforts have shown that PFPs can encode either of two distinct T4SSs, a type IVA (VirB-VirD4) system encoded by virB1 through virB11 and virD4 or a type IVB Tra (IncI) conjugation system (10, 15, 19, 61, 64). In our study, we included 12 genes from the pPSR1 type IVA system and 21 genes from the type IVB systems from pDC3000A and pDC3000B on the macroarray (Table 2).

We found that 12 PFPs encoded a type IVA secretion system similar to that encoded by pPSR1, pPMA4326A, and pPh1448B and that five plasmids (pB8617A, pCG131,pPSS4918, pPP3364, and pPG2708) hybridized to all 12 genes of the VirB-VirD4 system (Table 3). Several plasmids hybridized to 11 of 12 genes of the type IVA secretion system; for example, plasmid pPSPh1449B hybridized to all of the type IVA genes except virD4 and plasmids pPSE and pPSL1188 hybridized to all of the type IVA genes except virB6 (Table 3). Other type IVA system variant plasmids found included the three plasmids from P. syringae pv. savastanoi 0485-9, with pPS0485A lacking hybridization to virB7, and pPS0485B and pPS0485C, exhibiting overall weak signals for hybridization to the virB genes and lacking hybridization to virB7 and virD4 (Table 3). With a hybridization pattern similar to that of pPS0485B and pPS0485C, plasmid pPS0693A, from a different P. syringae pv. savastanoi strain, hybridized to virB1 to virB4 and virB8 to virB11 but not to the other type IVA secretion system genes (Table 3). It is interesting that plasmids pPSR1, pB8617A, pCG131, pPSS4918, pPP3364, and pPG2708, harbored within P. syringae strains from genomospecies I and IV, encoded the complete type IVA secretion system whereas plasmids harbored within P. syringae strains from genomospecies II (pPSPh1449B; pPSE; pPSL1188; pPS0485A, -B, and -C; and pPS0693A) encoded apparently incomplete type IVA systems. Two exceptions to the association of the type IVA secretion system and host genomospecies were plasmids pPh1448B from P. syringae pv. phaseolicola (genomospecies II) and pPMA4326A from P. syringae pv. maculicola (genomospecies III), which both encoded the complete type IVA secretion system.

TABLE 3.

Distribution of virB-virD4 genes on 31 P. syringae plasmids

P. syringae pathovar and plasmid(s)c	Hybridizationa for:
	virB1	virB2	virB3	virB4	virB5	virB6	virB7	virB8	virB9	virB10	virB11	virD4
syringae (I)
pPSR1	+	+	+	+	+	+	+	+	+	+	+	+
pB8617A	+	+	+	+	+	+	+	+	+	+	+	+
pCG131	+	+	+	+	+	+	+	+	+	+	+	+
pPSS4918	+	+	+	+	+	+	+	+	+	+	+	+
maculicola (III)
pPMA4326Ad	+	+	+	+	+	+	+	+	+	+	+	+
porri (IV)
pPP3364	+	+	+	+	+	+	+	+	+	+	+	+
garcae (IV)
pPG2708	+	+	+	+	+	+	+	+	+	+	+	+
phaseolicola (II)
pPh1448Bd	+	+	+	+	+	+	+	+	+	+	+	+
pPSPh1449B	+	+	+	+	+	+	+	+	+	+	+	−
eriobotryae (II)
pPSE	+	+	+	+	+	−	+	+	+	+	+	+
lachrymans (II)
pPSL1188	+	+	+	+	+	−	+	+	+	+	+	+
savastanoi (II)
pPS0485A	+	+	+	+	+	+	−	+	+	+	+	+
pPS0485B	+	+	+	+	±	±	−	±	±	±	±	−
pPS0485C	+	±	±	±	±	±	−	±	±	±	±	−
pPS0693A	+	+	+	+	−	−	−	+	+	+	+	−
16 plasmidsb	−	−	−	−	−	−	−	−	−	−	−	−

^a+, strong hybridization; ±, weak hybridization; −, negative.

^b16 plasmids included pDC3000A (in silico analysis only), pDC3000B (in silico analysis only), pPT23A, pPT23B, pOK1A, pPSM8810, pPSM9032A, pPSM9032B, pPSA1089A, pFKN (in silico analysis only), pPMA4326B (in silico analysis only), pPh1448A (in silico analysis only), pPSTA0893, pPA0893A, pPA0893B, and pPA0893C.

^cRoman numerals indicate the genomospecies of P. syringae that the pathovars belong to.

^dIn silico analysis only.

Ten plasmids (pPT23A, pPT23B, pOK1A, pPSM8810, pPSM9032A, pPSM9032B, pPSA1089A, pPSTA0893, pPSE, and pPP3364) hybridized to all 21 genes belonging to the putative type IVB secretion system encoded by plasmids pDC3000A and pDC3000B (Table 4). Among them, seven plasmids were from P. syringae strains that belong to genomospecies III, two (pPSTA0893 and pPSE) were from genomospecies II strains, and one (pPP3364) was from a genomospecies IV strain. Note that plasmids pPSE and pPP3364 also contained genes encoding complete or partial type IVA secretion systems (see above; Table 3). We could not distinguish multiple plasmids on the gels from repeated preparations of pPSE and pPP3364, and also restriction enzyme analysis (single-enzyme digestions) with five enzymes suggested that our estimated sizes for these plasmids (Table 1) were correct. Thus, we assume that these two plasmids encoded both T4SSs (type IVA and type IVB). Two plasmids (pPG2708 and pPSPh1449B) that hybridized to 12 and 11 of the type IVA secretion system genes, respectively, hybridized to at least four of the type IVB secretion system genes (Table 4).

TABLE 4.

Detection of homologs of tra genes on 31 P. syringae plasmids

P. syringae pathovar and plasmid(s)c	Hybridizationa for:
	traH	traI	type IV pilus gene	traK	traC	traL	traM	traN	traO	traP	traQ	traR	traT	traU	traW	traX	traY	trbA	traB	trbC	traG
tomato (III)
pDC300Ad	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	−	+
pDC3000Bd	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	−	+	+	−
pPT23A	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
pPT23B	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
pOK1A	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
maculicola (III)
pPSM8810	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
pPSM9032A	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
pPSM9032B	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
apii (III)
pPSA1089A	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
tabaci (II)
pPSTA0893	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
eriobotryae (II)
pPSE	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
porri (IV)
pPP3364	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
phaseolicola (II)
pPSPh1449B	+	−	−	−	−	−	−	+	−	−	−	−	−	−	−	−	−	+	+	−	−
garcae (IV)
pPG2708	−	±	+	−	−	−	−	+	±	−	−	−	−	±	±	±	+	−	−	−	+
aesculi (II)
pPA0893A	+	+	−	−	−	−	−	+	−	−	−	−	±	−	−	±	−	−	−	+	+
pPA0893B	+	+	−	−	−	−	−	±	−	−	−	−	±	−	−	±	−	−	−	+	+
pPA0893C	+	+	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	+
14 plasmidsb	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−

^a+, strong hybridization; ±, weak hybridization; −, negative.

^b14 plasmids included pPSR1, pB8617A, pCG131, pPSS4918, pPh1448A (in silico analysis only), pPh1448B (in silico analysis only), pPMA4326A (in silico analysis only), pPMA4326B (in silico analysis only), pFKN (in silico analysis only), pPSL1188, pPS0485A, pPS0485B, pPS0485C, and pPS0693A.

^cRoman numerals indicate the genomospecies of P. syringae that the pathovars belong to.

^dIn silico analysis only.

Of 31 plasmids examined in this study, only six plasmids did not encode a complete or almost fully complete T4SS. These included three plasmids from P. syringae pv. aesculi (pPA0893A, pPA0893B, and pPA0893C), which did not hybridize to any type IVA genes and which hybridized only to three to seven type IVB genes (Tables 3 and 4). In addition, plasmid pPh1448A from P. syringae pv. phaseolicola carried only truncated virB4 and virB9 genes of the type IVA secretion system, pPMA4326B from P. syringae pv. maculicola ES4326 did not carry any conjugative-transfer-related genes (61), and pFKN from P. syringae pv. maculicola M6 carried only one unique gene, orf28, which appears to comprise remnants of three separate genes fused together (52). As reported by Stavrinides and Guttman (61), portions of orf28 have high nucleotide sequence homology with virB1 and virB4 and the putative relaxase gene (orf47 or orf54) from both pPMA4326A and pPSR1, respectively, and this gene may have arisen from two deletion events resulting in the deletion of most of a type IVA secretion system gene cluster.

In summary, our hybridization results clearly demonstrated that PFPs could be separated into four major groups depending on the presence or absence of genes encoding a T4SS and the type of T4SS encoded (type IVA, type IVB, type IVA and type IVB, lack of a T4SS, or only remnants of a T4SS).

Detection of known virulence and mobile genes on pPT23A family plasmids.

We included 15 hrp-dependent outer protein (hop) and hop-like (hol) genes, 13 avirulence genes (avr), and eight hrp and hrc genes on the macroarray. Many of these genes are known to inhabit only chromosomal locations in P. syringae and were included both as controls for chromosomal DNA contamination in our plasmid preparations and on the odd chance that one or more of these genes may inhabit both P. syringae chromosomes and plasmids depending on the strain studied. In this study, we observed only hybridization to effector genes with known plasmid locations, and thus data are only presented for these genes.

Except for four plasmids (pDC3000B, pPSTA0893, pPh1448B, and pPSS4918), all of the other 27 plasmids examined contained one to four avr or hop genes (Table 5). Homologs of hopPmaA were detected on seven plasmids (pPT23A, pPT23B, pOK1A, pPSM9032A, pPSM9032B, pPh1448A, and pPMA4326B), all of which also encoded a type IVB secretion system except pPh1448A and pPMA4326B. In contrast, homologs of hopPmaD were found on four plasmids (pPP3364, pPG2708, pPSR1, and pCG131), all of which also encoded a type IVA secretion system (Tables 3 to 5). The most widely distributed avr genes on the 31 PFPs were avrD and avrPpiB2, with homologs on 13 plasmids each. Homologs of the avrA, virPphA, avrPphD, avrPphE, avrPphB1, and avrRps4 genes were found on three, five, six, six, three, and four plasmids, respectively (Table 5). No homology to avrRpm1 on PFPs other than pFKN was found. Furthermore, none of the plasmids examined in this study hybridized to avrE, avrF, avrPto, and avrRpt2 (Table 5).

TABLE 5.

Detection of homologs of avr genes on 31 P. syringae plasmids

Plasmid	Hybridizationd for:
	avrA	virPphApgy	avrD	avrPphDpgy	avrPphE	avrPpiB2	avrPphB1	avrRps4	hopPmaD	hopPmaA	avrRpm1	4 avr genese
pDC3000Aa,b	−	−	−	−	+	+	−	−	−	−	−	−
pDC3000Ba,b	−	−	−	−	−	−	−	−	−	−	−	−
pPT23A	−	−	+	−	−	−	−	+	−	+	−	−
pPT23B	−	−	+	+	−	−	−	−	−	+	−	−
pOK1A	+	+	+	+	−	−	−	−	−	+	−	−
pFKNa	−	−	−	−	+	−	−	−	−	−	+	−
pPMA4326Aa	−	−	−	−	−	−	−	−	−	−	−	−
pPMA4326Ba	−	−	−	−	−	−	−	−	−	+	−	−
pPSM8810	−	−	−	−	+	+	+	−	−	−	−	−
pPSM9032A	−	−	−	−	−	+	+	−	−	+	−	−
pPSM9032B	−	−	−	−	−	+	+	−	−	+	−	−
pPSA1089A	−	−	+	+	−	−	−	−	−	−	−	−
pPSTA0893	−	−	−	−	−	−	−	−	−	−	−	−
pPSE	−	−	−	−	+	−	−	−	−	−	−	−
pPP3364	−	−	+	+	−	±	−	+	±	−	−	−
pPSPh1449B	−	+	+	+	−	−	−	+	−	−	−	−
pPh1448Aa,c	−	−	+	+	−	−	−	+	−	+	−	−
pPh1448Ba	−	−	−	−	−	−	−	−	−	−	−	−
pPG2708	−	−	−	−	−	±	−	−	+	−	−	−
pPSR1	−	−	−	−	−	+	−	−	+	−	−	−
pB8617A	−	−	−	−	−	+	−	−	−	−	−	−
pCG131	−	−	−	−	−	+	−	−	+	−	−	−
pPSS4918	−	−	−	−	−	−	−	−	−	−	−	−
pPSL1188	−	−	−	−	−	+	−	−	−	−	−	−
pPS0485A	−	−	+	−	−	+	−	−	−	−	−	−
pPS0485B	−	−	−	−	−	+	−	−	−	−	−	−
pPS0485C	−	+	+	−	−	−	−	−	−	−	−	−
pPS0693A	−	+	+	−	−	−	−	−	−	−	−	−
pPA0893A	+	+	+	−	+	−	−	−	−	−	−	−
pPA0893B	+	−	+	−	+	−	−	−	−	−	−	−
pPA0893C	−	−	+	−	−	+	−	−	−	−	−	−

^aIn silico analysis.

^bOther avr genes on pDC3000 are hopPtoS1 and hopPtoT1.

^cThe other avr gene on pPh1448 is holPtoQ.

^d+, strong hybridization; ±, weak hybridization; −, negative.

^eavrE, avrF, avrPto, and avrRpt2.

We also assessed the distribution of 14 known virulence-associated, fitness-related, and mobile genes, besides avr and hop genes, in this study (Table 6). These genes included UV radiation resistance genes (rulA from pDC3000A and rulB from pPSR1), the levansucrase gene (lsc from pDC3000A), methyl-accepting chemotaxis transducer genes (mcp genes from pPSR1 and pFKN), the phytotoxin coronatine gene, indole-acetic acid (IAA) and ethylene biosynthetic genes (cfl from pPG4180A; iaaH, iaaM, and iaaL from pIAA1 and pIAA2; and efe from pETH2), regulatory genes (hrmA and gacS from P. syringae), and insertion sequence (IS) transposase (IS801 from pPG4180A and ISPsy1-a/b and ISPsy4 from pPSR1 and pDC3000A) genes.

TABLE 6.

Detection of virulence and fitness-related genes on 31 P. syringae plasmids

Plasmid	Hybridizationb for:
	rulA	rulB	lsc	mcp	cfl	iaaL	iaaH	iaaM	efe, hrmA, gacS	IS801	ISPs1-a/b	ISPsy4
pDC300Aa	+	+	+	−	−	−	−	−	−	−	−	−
pDC3000Ba	−	+	−	−	−	−	−	−	−	−	−	+
pPT23A	+	+	−	−	+	−	−	−	−	+	−	+
pPT23B	−	−	−	±	−	−	−	−	−	+	−	+
pOK1A	+	+	+	−	+	−	−	−	−	+	−	+
pFKNa	−	+	−	+	−	−	−	−	−	+	−	−
pPMA4326Aa	−	+	−	−	−	−	−	−	−	−	−	−
pPMA4326Ba	−	+	−	−	−	−	−	−	−	+	+	−
pPSM8810	+	+	−	−	−	−	−	−	−	+	+	+
pPSM9032A	+	+	−	−	−	−	−	−	−	−	−	+
pPSM9032B	+	+	−	−	−	−	−	−	−	−	−	+
pPSA1089A	−	−	−	−	−	−	−	−	−	+	−	−
pPSTA0893	+	+	+	−	−	−	−	−	−	−	−	±
pPSE	−	−	±	−	−	−	−	−	−	+	−	−
pPP3364	+	+	+	−	+	−	−	−	−	+	−	+
pPSPh1449B	+	+	+	±	+	−	+	+	−	+	−	±
pPh1448Aa	−	−	+	+	−	−	−	−	−	+	−	+
pPh1448Ba	+	+	−	−	−	−	−	−	−	−	−	−
pPG2708	+	+	−	−	−	−	−	−	−	−	−	−
pPSR1	+	+	−	+	−	−	−	−	−	−	+	+
pB8617A	+	+	−	−	−	−	−	−	−	−	−	−
pCG131	+	+	−	−	−	−	−	−	−	−	−	−
pPSS4918	+	+	−	+	−	−	−	−	−	+	−	−
pPSL1188	+	+	−	−	−	−	−	−	−	−	−	−
pPS0485A	+	+	−	−	−	+	+	+	−	+	−	±
pPS0485B	+	+	−	−	−	+	+	+	−	+	−	±
pPS0485C	−	−	−	±	−	+	+	+	−	+	−	+
pPS0693A	+	+	−	−	−	+	+	+	−	+	+	+
pPA0893A	−	+	+	−	+	−	+	±	−	+	−	+
pPA0893B	−	+	+	−	−	−	+	±	−	+	−	+
pPA0893C	−	+	−	−	−	−	+	±	−	+	−	+

^aIn silico analysis.

^b+, strong hybridization; ±, weak hybridization; −, negative.

As shown in Table 6, no homology to hrmA, gacS, and efe genes on the 31 PFPs studied was observed. The most widely distributed of the genes studied were rulA, rulB, and the IS801 and ISPsy4 genes, with homology on 19, 26, 19, and 19 plasmids, respectively (Table 6). Nine, seven, and four plasmids carried homologs of lsc, mcp, and the ISPs1-a/b genes, respectively. The phytotoxin coronatine cfl gene was found on plasmids pPT23A and pOK1A as reported previously (59), whereas another three plasmids (pPP3364, pPSPh1449B, and pPA0893A) also hybridized to the cfl gene, although the presence of the entire coronatine biosynthetic cluster was not determined. All three IAA biosynthetic genes (iaaL, iaaH, iaaM) were found on four plasmids (pPS0485A, -B, and -C and pPS0693A) isolated from two strains of P. syringae pv. savastanoi, an organism in which a plasmid location for the iaa genes was previously reported (67). It is also interesting that iaaH and iaaM were found on four additional plasmids (pPSPh1449B and pPA0893A, -B, -C). Both the rulA and rulB genes were detected on 19 of 31 plasmids examined, whereas another seven plasmids (pDC3000B; pFKN; pPMA4326A; pPMA4326B; and pPA0893A, -B, and -C) harbored only the rulB gene (Table 6). We suspected that, as in pPMA4326A, pPMA4326B, pFKN, and pDC3000B, the rulB gene was truncated in pPA0893A, -B, and -C. Besides an intact rulAB operon, plasmid pDC3000A contains another truncated rulB gene (GenBank accession number NC_004632; PSPTOA0006). The same was true for plasmids pPMA4326A and pPMA4326B, in which part of the same rulB gene was found (61).

Replication gene (repA) of the pPT23A plasmid family.

The defining characteristic of PFPs is that all of these plasmids carry the major replication gene repA, indicating that PFPs descended from a common ancestor. Surprisingly, the repA gene was the only gene on the array shared by all the plasmids examined. To further analyze the phylogenetic relationship of PFPs, we cloned and sequenced repA from 16 PFPs from this study. A phylogenetic analysis was performed, and an unrooted phylogenetic tree was constructed based on deduced amino acid sequences of RepA from these 16 PFPs plus nine previously sequenced full-length repA genes (Fig. 2). The repA gene was highly conserved, with the deduced amino acid sequences of all 25 RepA proteins having at least 87% identity and 90% similarity. Plasmids coexisting within a bacterial strain encoded sequences with either high amino acid identity (pPSM9032A and pPSM9032B, 97%; pPMA4326A and pPMA4326B, 96%) or lower identity (pDC3000A and pDC3000B, 88%; pPT23A and pPT23B, 90%). Similar results were reported previously for the RepA proteins encoded by pPph1448A and pPph1448B (61). Interestingly, the percentages of similarity between the RepA proteins encoded by pDC3000B and pPT23A (99%) and between those encoded by pDC3000A and pPT23B (99%) were also among the highest observed. Furthermore, the RepA proteins encoded by the three plasmids from P. syringae pv. savastanoi strain 0485-9 were almost identical (99% identity) (Fig. 2). Relationships deduced from the RepA phylogenetic analysis show that three major plasmid lineages exist; however, these lineages do not agree with strain groupings based on analyses of chromosomal DNA (25, 55, 56) as each grouping contains plasmids from strains in at least two genomospecies (Fig. 2). In addition, the plasmid groupings based on RepA analyses also do not delimit the four plasmid groupings characterized earlier in this paper based on T4SS (Fig. 2).

FIG. 2.

Phylogenetic tree constructed by using the neighbor-joining method indicating relationships among RepA proteins encoded by the pPT23A plasmid family. The GenBank accession numbers for RepA protein sequences encoded by plasmids pDC3000A, pDC3000B, pPMA4326A, pPMA4326B, pPT23A, pPSR1, and pFKN areAE016855, AE016854, AY603979, AY603980, AJ224509, AY342395, and AF359557, respectively. The RepA sequences encoded by pPh1448A and pPh1448B were obtained from the TIGR website (www.tigr.org). The scale is shown in number of substitutions per site. The genomospecies that the P. syringae pathovar host of each plasmid belongs to is indicated as a superscript.

Plasmids from the same P. syringae pathovar or strain showed similar hybridization patterns.

The remaining 77 genes on the macroarray included genes for plasmid maintenance, stability, and transmission; transcriptional regulator genes; genes encoding hypothetical and conserved hypothetical proteins; and genes with no known homology. This group included 46 genes from pDC3000A and pDC3000B, 20 genes from pFKN, and 11 genes from pPSR1. Overall, the hybridization results showed no obvious trend to how these 77 genes were distributed on the different plasmids examined in this study. However, we noticed that the 10 plasmids encoding the complete type IVB secretion system hybridized to a majority of genes from pDC3000A, pDC3000B, and pFKN and did not hybridize to any of 11 miscellaneous genes from pPSR1 (Table 7). Interestingly, 7 of these 10 plasmids were recovered from P. syringae pv. tomato or maculicola, as were plasmids pDC3000A, pDC3000B, and pFKN. Plasmids which encoded the type IVA secretion system hybridized to some of the miscellaneous genes found on pPSR1, but only small numbers of the genes from pDC3000A and pDC3000B (Table 7). pPSE and pPP3364, which harbored both type IVA and IVB secretion system genes, shared some miscellaneous plasmid genes with pDC3000A, pDC3000B, pFKN, and pPSR1 (Table 7).

TABLE 7.

Numbers of miscellaneous genes from previously sequenced P. syringae PFPs hybridizing to selected PFPs

Plasmid	No. of genes hybridizing on the macroarraya for:
	pDC3000A/B	pPSR1	pFKN
Type IVB
pPT23A	20	0	4
pPT23B	25	0	6
pPSM8810	39	0	3
pPSM9032A	19	0	3
pPSA1089A	18	0	4
pPSTA0893	23	0	8
Type IVA
pPSR1	3	11	4
pB8617A	3	2	9
pCG131	6	4	6
pPSS4918	4	2	4
pPSL1188	5	3	10
pPSPh1449B	15	3	5
pPG2708	5	2	1
Both T4SSs
pPSE	12	4	3
pPP3364	26	3	5

^aThe total numbers of miscellaneous genes included on the macroarray were as follows: pDC3000A/B, 46 genes; pPSR1, 11 genes; pFKN, 20 genes.

DISCUSSION

Our DNA macroarray analysis enabled us to rapidly survey the distribution of 161 genes among diverse members of the pPT23A plasmid family. This study revealed that PFPs could be divided into distinct subgroups based on the T4SS encoded, with additional plasmid variants apparently encoding both type IVA and IVB secretion systems or T4SSs with partially deletions or not containing T4SS genes. A phylogenetic analysis utilizing RepA, however, indicated that carriage of a particular T4SS was an inconclusive character in defining the evolutionary history of individual plasmids, suggesting that T4SS-encoding genes can also move between plasmids via horizontal transfer and recombination.

The T4SS is one of the five major secretion systems in gram-negative bacteria and is defined as a translocation system ancestrally related to any conjugation system of gram-negative and -positive bacteria (12, 13, 48, 49, 69). T4SSs can be classified into three groups, i.e., the conjugation systems, DNA uptake and release systems, and effector translocation systems (13, 19). As one of the most well-studied T4SSs, the VirB-VirD4 system of Agrobacterium tumefaciens (transfer DNA transfer system) has served as a prototype for T4SSs and is indispensable for the infection process, functioning to deliver oncogenic transfer DNA and effector proteins to host plants, resulting in crown gall disease (13, 14, 39). The type IVA secretion system of A. tumefaciens consists of 11 proteins encoded by the virB operon and the VirD4 protein; the genetic organizations of these structural genes in plasmids pPSR1, pPh1448A, and pPMA4326A are similar, although the function of the genes on the P. syringae plasmids is still unknown (Fig. 3A). Since most of the PFPs containing genes of the type IVA system hybridized to virB1 to virB11 and virD4 (Table 3), we anticipate that the gene organization on these plasmids would be similar to that of pPSR1, pPH1448A, and pPMA4326A as well. Recent research has shown that other type IVA secretion systems encoded by genes carried either on plasmids or the chromosome of plant-pathogenic bacteria play a role in pathogenesis (6, 22). Examination of the arrangement of type IVA gene sets from plasmids or chromosomes of other plant-related bacteria suggests at least four different groupings (Fig. 3A). Group 1 consists of genes virB1 through virB11, with or without virD4, and includes Ti or Ri plasmids from Agrobacterium spp. and the P. syringae plasmids (Fig. 3A). Group 2 gene sets contain virB2, virB3, virB7, and/or virD4 with deletions while retaining the same gene order as group 1, an arrangement currently limited to Erwinia spp. (Fig. 3A) (6). Group 3 and 4 gene sets are currently limited to Xanthomonas campestris and Xanthomonas axonopodis strains, with the main differences including an altered placement of the virB1 gene in group 3 or a completely different gene organization combined with the maintenance of multiple virB6 genes in group 4 (Fig. 3A) (17).

FIG. 3.

(A) Schematic map for the type IVA secretion genes from plant-related bacteria. The type IVA secretion system is normally encoded by 11 virB genes and the virD4 gene, as represented by the A. tumefaciens pTi VirB-VirD4 reference system. Structural variations and gene rearrangements occur in different plasmids and/or genomes. Genes with similar functions were drawn with similar colors. Genes without color have no known functions. Solid lines indicate genes that are physically unlinked. A superscript “a” indicates that two or more virB6 genes with ISs were found either upstream or downstream of the operon. The GenBank accession numbers for plasmids or genomes are as follows: X. campestris pv. campestris ATCC 33913 genome (XCC) and plasmid pXcB, NC_003902 and NC_005240; X. axonopodis pv. citri 306 genome (XAC) and plasmid pXAC64, NC_003919 and NC_003922; Erwinia amylovora EA110 plasmid pEU30, NC_005247; Erwinia carotovora subsp. atroseptica SCRI1043 genome (ECA), BX950851; P. syringae pv. maculicola 4326 plasmid pPMA4326A, NC_005918; P. syringae pv. syringae A2 plasmid pPSR1, NC_005205; A. tumefaciens strain C58 plasmid pAT_washington, NC_003306; A. tumefaciens plasmid pTi_Washington, Sakura, NC_003308, NC_002147, and NC_002377; Agrobacterium rhizogenes plasmid pRi1724, NC_002575. (B) Structural conservation of type IVB secretion genes. Representatives of the type IVB subfamilies are the R64 and ColIb-P9 (IncI) tra systems. Solid lines indicate genes that are physically unlinked. *, essential genes for the type IVB secretion system (13). 1, pR64 and ColIb-P9; 2, pCTX-M3; 3, pEL60; 4, pDC3000A; 5, pDC3000B. The GenBank accession numbers for plasmids are as follows: P. syringae pv. tomato DC3000 plasmids pDC3000A and pDC3000B: NC_004633 and NC_004632; Citrobacter freundii plasmid pCTX-M3, NC_004464; E. amylovora plasmid pEL60, NC_005246; Salmonella enterica serovar Typhimurium plasmid pR64, NC_005014; ColIb-P9, NC_002122.

Ten PFPs from this study, along with pDC3000A and pDC3000B, were found to encode a type IVB secretion system, encoded by 21 genes with a structural organization highly similar to that found in other known conjugative plasmids (Fig. 3B). The GC content of most of the tra genes from pDC3000B ranged from 59% to 62%, values which are slightly higher than the overall GC content of pDC3000B (56.2%) but similar to that of repA from pDC3000B (58.9%) (10). In our previous study (64), we showed that the GC contents of the virB operon and virD4 from pPSR1 were similarly slightly higher than that of the pPSR1 plasmid overall. These observations suggest that both of these T4SS determinants may have been acquired by P. syringae PFPs from other organisms; however, because the sequence differences are relatively small, it is likely that both T4SS determinants have evolved and adapted to their P. syringae host.

In both animal and plant pathogens, genes specifically required for pathogenesis are located on well-characterized PAIs (2, 5, 18, 20, 23, 33, 34, 42). PAIs are generally regarded as large regions of chromosomal or plasmid DNA containing multiple virulence genes, which are flanked by repeated sequences and are characteristically distinct in GC content from the rest of the genome (31). In P. syringae, the chromosomal hrp PAI contains structural genes of the type III secretion system plus additional effector genes; several PAIs have also been identified on PFPs (1, 37, 38, 52). These PFP-carried PAIs contain either effector, toxin production, or other virulence genes as well as sequences for DNA mobility, such as those encoding integrases and transposases and IS elements (1, 37, 38, 52). Furthermore, bacterial plasmids can also contain so-called “fitness island” sequences (FIs), which consist of genes responsible for the epiphytic fitness of the bacteria (61). The fitness island usually includes genes such as integrase, resolvase, effector, and/or chemotaxis and UV resistance genes (61, 64). A central characteristic of island sequences is that they are acquired via horizontal transfer. In our study, 26 of 31 plasmids contained one to four predicted avr or hop genes and 19 of 31 plasmids contained rulAB and IS801 and other ISs were also widely distributed among those plasmids examined. It is possible that either PAIs or FIs could exist on these newly analyzed PFPs. These results further demonstrated that the pPT23A plasmids represent a mobile arsenal in P. syringae and contribute to the virulence and epiphytic fitness of the pathogen during interactions with host plants (67).

While the distribution of certain avr or hop genes among PFPs from multiple P. syringae pathovars is known, the functional significance of these observations is not well understood. In addition, the significance of the location of effector genes on plasmids or chromosomes has not been experimentally addressed. We found that certain avr genes, such as avrD and avrPpiB2, were widely distributed on PFPs, while other genes, such as avrRpm1 and avrPphB1, had a more limited distribution. Furthermore, some of the avr genes, such as avrRpt2, have not been found in a plasmid location to date (4, 67). Plasmids are autonomous genetic entities that tend to carry genes of ecological benefit to host bacteria under certain environmental conditions (21). The continued maintenance of plasmids in host bacterial lineages can be accomplished through selection of plasmid-carried genes or through infectious transfer of plasmids among host strains in a population (8). The maintenance of PFPs in almost all P. syringae strains studied to date is a testament to the ecological success of this plasmid. The conservation of effector genes among PFPs in distinct P. syringae pathovars and genomospecies may be functionally significant but likewise could be due to the horizontal transfer capabilities of these elements.

How have PFPs evolved in P. syringae? Macroarray hybridizations in this study suggest that these plasmids may carry a backbone of genes including repA, the gntR family transcriptional regulator gene, which is directly upstream of repA in pDC3000A, pDC3000B, pFKN, and pPSR1, and rulAB, the UV radiation tolerance determinant, which is located immediately downstream of repA in pPT23A (58). Other genes encoding plasmid-specific functions or conserved hypothetical proteins that are widely distributed among current PFPs also may comprise this backbone. Note that repA is the only gene conserved among all PFPs studied; thus other genes carried on a putative plasmid backbone could be lost from some PFP lineages. An ancestral plasmid containing this putative backbone of genes may have infected P. syringae and was maintained by selection for the rulAB locus, a fitness locus which increases the survival of epiphytic P. syringae strains on leaf surfaces (62, 65, 66). Such a plasmid may have inhabited P. syringae strains even prior to the divergence of modern pathovars. It is tempting to speculate that the acquisition of either the type IVA or type IVB secretion system occurred after the existence of the initial plasmid backbone because of the conservation of backbone genes among plasmids containing genes for either system.

Over time, as PFPs continued to evolve, the plasmids acquired additional genes, increasing virulence or ecological fitness in particular plant pathosystems. These genes tend to be located either between repA and rulA (for example, a PAI in pFKN and Tn5393 in pPSR1 [52, 64]), as insertions within rulB (4, 5), or in a location between genes encoding plasmid-specific functions and T4SSs (61, 64). Because of the predicted horizontal transfer capabilities of most PFPs, any novel gene acquired and fixed within one PFP genome could possibly be acquired by PFPs inhabiting other P. syringae strains or pathovars. Gene acquisition and recombination events would tend to disrupt gene order, further obscuring the evolutionary history of individual plasmids. Through mechanisms such as plasmid integration and excision, ecologically important genes may ultimately be deposited in host chromosomes, potentially stabilizing these determinants within a genome.

Retrospective evidence provided by sequence and hybridization analyses indicates that horizontal transfer, recombination, and gene duplication must play a role in the evolution of PFPs. For example, plasmids pDC3000A and pDC3000B inhabit the same P. syringae pv. tomato DC3000 strain and consist of extensively duplicated genomes although their individual repA genes are phylogenetically distinct (10). In this situation, it is likely that a distinct PFP was acquired by an ancestor to DC3000 containing a single resident PFP and, following recombination and deletion events with the resident DC3000 plasmid, generated the plasmid combination maintained currently. Highly similar PFPs can be maintained within a single cell as compatible plasmids; however, replicon incompatibility can be demonstrated in some cases (59). The introduction of an incompatible PFP into a new host could result in the acquisition of novel genes by the surviving plasmid, including situations such as that for pPSE and pPP3364, which encode both type IVA and IVB secretion systems. However, the mechanism of plasmid incompatibility of PFPs has not yet been elucidated.

In summary, we report a global genetic analysis of the predicted gene contents of 31 PFPs of P. syringae. Only the repA gene was found to be distributed among all 31 PFPs; thus, the possession of repA is the current defining parameter of plasmids in this family. Our results indicated that the plasmids could be grouped by the particular T4SS encoded. Phylogenetic analysis of the repA gene implied plasmid relationships that did not correlate with known phylogenetic relationships of their bacterial hosts (53, 55, 56), suggesting that horizontal transfer and recombination have contributed to the evolution of PFPs and to the current distribution of individual PFPs and PFP-carrried genes. Sequencing and comparative genomic analyses have contributed to our current understanding of the evolution of this plasmid family; however, it is also clear that additional comparative and functional genomic studies are necessary to add both defined figures and shadings to a future complete picture describing the biology of PFPs and their role in the ecology and pathogenesis of P. syringae.