Complete DNA Sequence and Comparative Analysis of the 50-Kilobase Virulence Plasmid of Salmonella enterica Serovar Choleraesuis

Abstract

The complete nucleotide sequence of pKDSC50, a large virulence plasmid from Salmonella enterica serovar Choleraesuis strain RF-1, has been determined. We identified 48 of the open reading frames (ORFs) encoded by the 49,503-bp molecule. pKDSC50 encodes a known virulence-associated operon, the spv operon, which is composed of genes essential for systemic infection by nontyphoidal Salmonella. Analysis of the genetic organization of pKDSC50 suggests that the plasmid is composed of several virulence-associated genes, which include the spvRABCD genes, plasmid replication and maintenance genes, and one insertion sequence element. A second virulence-associated region including the pef (plasmid-encoded fimbria) operon and rck (resistance to complement killing) gene, which has been identified on the virulence plasmid of S. enterica serovar Typhimurium, was absent. Two different replicon regions, similar to the RepFIIA and RepFIB replicons, were found. Both showed high similarity to those of the pO157 plasmid of enterohemorrhagic Escherichia coli O157:H7 and the enteropathogenic E. coli (EPEC) adherence factor plasmid harbored by EPEC strain B171 (O111:NM), as well as the virulence plasmids of Salmonella serovars Typhimurium and Enteritidis. Comparative analysis of the nucleotide sequences of the 50-kb virulence plasmid of serovar Choleraesuis and the 94-kb virulence plasmid of serovar Typhimurium revealed that 47 out of 48 ORFs of the virulence plasmid of serovar Choleraesuis are highly homologous to the corresponding ORFs of the virulence plasmid of serovar Typhimurium, suggesting a common ancestry.

Plasmid-encoded gene products are required for full expression of virulence in many enteropathogenic bacteria, including those of the genera Shigella (53, 54) and Yersina (17, 20), as well as Salmonella (12, 33, 35, 47, 56). Nontyphoidal Salmonella serovars are important agents of gastroenteritis and can cause systemic infection, such as bacteremia (septicemia), in animals and humans. Many of these serotypes typically carry large plasmids which are essential to the production of systemic infection in animal models (21, 23). Although the virulence plasmids of these Salmonella strains are variable in size, ranging from 50 to 94 kb, their distribution is dependent on the serotype. For example, S. enterica serovar Choleraesuis, S. enterica serovar Enteritidis, S. enterica serovar Dublin, S. enterica serovars Gallinarum and Pullorum, and S. enterica serovar Typhimurium harbor the 50-, 60-, 80-, 90-, and 94-kb virulence plasmids, respectively.

Strains of serovar Typhimurium cured of the virulence plasmid are strongly attenuated in their subsequent spreading infection to the mesenteric lymph nodes, spleen, and liver (23), while the presence of the virulence plasmid of Salmonella does not appear to be required for bacterial adherence to and invasion of cultured eukaryotic cells or for colonization of the cecum or invasion of Peyer's patches in the mouse (24, 42). All of these virulence plasmids contain a highly conserved 8-kb region, which contains the spv (Salmonella plasmid virulence) locus that can confer complete virulence on a strain of serovar Typhimurium cured of the plasmid (25).

The spv region consists of spvR, a gene that encodes a transcriptional factor of the LysR family, and the spvABCD operon of structural genes (1, 2, 22, 25, 37). The spv operon is required for the systemic phase of disease in specific hosts, i.e., serovar Choleraesuis in pigs (15), serovar Dublin in cattle (39, 61), serovars Gallinarum and Pullorum in fowl (5, 6), and serovars Typhimurium and Enteritidis in mice (24, 33, 47). The importance of these genes for the establishment of a systemic infection by serovar Typhimurium has also been shown by in vivo expression technology, which has demonstrated that the genes are induced during infection of the animal (28), and by signature-tagged mutagenesis, which has identified them as essential virulence genes (29). Recently, it has been reported that SpvB is an ADP-ribosylating enzyme of an unknown host protein (49). However, the molecular functions of other Spv proteins have not yet been determined.

Other virulence-associated loci on the virulence plasmid of serovar Typhimurium include the pef (plasmid-encoded fimbria) operon, which has been implicated in bacterial adherence to intestinal epithelial cells and is required for fluid accumulation in infant mice (8, 19), and the rck (resistance to complement killing) gene, which encodes an outer membrane protein whose expression renders the bacteria host serum resistant (26, 27). In addition, a recent in vivo expression study of serovar Typhimurium using the gfp reporter gene has identified mig-5, a macrophage-inducible gene that codes for carbonic anhydrase. Insertional mutation in mig-5 resulted in a decrease in bacterial colonization in the mouse spleen, demonstrating that the gene product of mig-5 is a virulence factor (60). However, the presence of these genes in every serotype has not been proved and their role in pathogenesis remains unclear.

To establish virulence determinants of the large virulence plasmids of nontyphoidal Salmonella, we tried to determine the genetic organization of the 50-kb virulence plasmid, designated pKDSC50, of serovar Choleraesuis strain RF-1 (35). A detailed restriction map of this plasmid has already been made available (36). In addition, the spv region on pKDSC50 has been subjected to a detailed genetic analysis and sequenced (44–46). In this report, we present the entire DNA sequence of the 50-kb virulence plasmid, pKDSC50, from serovar Choleraesuis strain RF-1. The complete DNA sequence of the plasmid could provide important insight into the evolution and origin of the virulence plasmids of Salmonella serovars.

MATERIALS AND METHODS

Bacterial strains and plasmids.

The bacterial strains used in this study are listed in Table 1. The 50-kb virulence plasmid pKDSC50 was isolated from serovar Choleraesuis strain 2N-3, an isogenic derivative of RF-1 cured of a 6.7-kb cryptic plasmid (35). The large virulence plasmids were prepared from Salmonella strains grown overnight at 37°C in Luria-Bertani medium and obtained by the method of Kado and Liu (34).

TABLE 1.

Bacterial strains used in this study

Serovar and strain	Description	Reference or sourcea
Choleraesuis
RF-1	pKDSC50, 6.7-kb plasmid	35
2N-3	derivative of RF-1 cured of 6.7-kb plasmid	35
TG21	Human isolate	35
TG22	Human isolate	35
AH1	Swine isolate	NIAH
AH2	Swine isolate	NIAH
AH3	Swine isolate	NIAH
AH4	Swine isolate	NIAH
AH5	Swine isolate	NIAH
AH6	Swine isolate	NIAH
Typhimurium
SL1344		30
14028s		ATCC
LT2		Laboratory collection
Enteritidis
S72		NIID
1305-96		NIID
1305-97		NIID
1305-98		NIID
2871-97		NIID
Dublin
Lane		18
215K		Laboratory collection
Gallinarum
E56		NIID
E57		NIID
Pullorum
S190		NIID
S191		NIID

^aATCC, American Type Culture Collection; NIAH, National Institute of Animal Health, Ibaraki, Japan. NIID, National Institute of Infectious Diseases, Tokyo, Japan. 

Subcloning for sequencing and DNA sequence.

Library construction for DNA sequencing was based on the previously established restriction map of pKDSC50 (36). DNA fragments generated with the restriction endonucleases EcoRI and SalI were cloned into Escherichia coli DH5α (Gibco BRL, Grand Island, N.Y.) using the sequencing vector pBluescript II SK(+) (Stratagene, Heidelberg, Germany). Series of nested deletions were generated from each clone. Purified pBluescript II SK(+) templates were sequenced using cycle-sequencing reactions with fluorescein isothiocyanate-labeled forward and reverse primers (Amersham-Pharmacia Biotech, Piscataway, N.J.). Gaps including regions between EcoRI, SalI, and EcoRI and SalI fragments in the pKDSC50 molecule were amplified by PCR using the original plasmid template for pKDSC50. The PCR products were cloned into a pBluescript II SK(+) vector. To resolve the ambiguity in the sequence of pKDSC50, 15 sequence data were obtained by the primer-walking method using fluorescein isothiocyanate-labeled synthetic oligonucleotides designated from contig ends according to the method described above. All sequence samples were run on a DSQ-2000L sequencer (Shimadzu, Kyoto, Japan).

DNA sequence analysis and annotation.

A 6.9-kb DNA sequence of pKDSC50, which contains IS630 (accession no. D10689) and spvRABC regions (accession no. E03417), was previously published by our group (44–46). This sequence was combined with sequences determined in this study to obtain a simple circular sequence of pKDSC50. Open reading frames (ORFs) were initially identified using GENETYX-Mac software (version 10.1) and a BLAST database of putative genes. For subsequent analysis, each ORF was compared to the current nonredundant protein database of the National Center for Biotechnology Information by using BLAST software through the Internet. Only ORFs encoding peptides of more than 50 amino acids were analyzed.

PCR and DNA-DNA hybridization.

The primers for PCR amplification of genes between repB and repC on the large virulence plasmids of Salmonella serovars are listed in Table 2. Dot blot hybridization was carried out using the standard protocol. In the Southern hybridization test, approximately 100 ng of Sau3AI-digested plasmid DNA of each Salmonella strain was blotted onto a GeneScreen Plus membrane (NEN Life Science Products, Boston, Mass.) using a Bio-dot microfiltration apparatus (BioRad Laboratories, Hercules, Calif.). The blots were prehybridized in hybridization buffer containing 0.5 M Na₂HPO₄ (pH 7.2), 1 mM EDTA, and 7% sodium dodecyl sulfate at 60°C for 1 h and then hybridized overnight at 60°C with probe DNA that was made by PCR with gene-specific primers (Table 2) using pLT2, the large virulence plasmid of serovar Typhimurium strain LT2, as the template and fluorescein labeled using a random prime labeling and detection system (Amersham-Pharmacia Biotech). Hybridization reactions were detected with horseradish peroxidase-conjugated rabbit antifluorescein antibody and Western Blot Chemiluminescence Reagent Plus (NEN Life Science Products) and used to expose X-ray film.

TABLE 2.

Oligonucleotide primers used for PCR amplification

Gene	Primer	Orientation	Sequence
spvR	SPVR-1	Forward	5′-GATTTCTTGATTAATAAAAAATTAA-3′
	SPVR-2	Reverse	5′-TCAGAAGGTGGACTGTTTCA-3′
spvC	SPVC-1	Forward	5′-CCCATAAATAGGCCTAATCT-3′
	SPVC-2	Reverse	5′-TTACTCTGTCATCAAACGAT-3′
pefA	PEFA-1	Forward	5′-TTCCATTATTGCACTGGGTG-3′
	PEFA-2	Reverse	5′-AAGCCACTGCGAAAGATGCC-3′
pefB	PEFB-1	Forward	5′-ATGATGCTGAACAGAAAAGATGCTG-3′
	PEFB-2	Reverse	5′-AATAATAAACAACCATCTGCGCAGC-3′
pefC	PEFC-1	Forward	5′-GACAGTATTATGTCGACGTC-3′
	PEFC-2	Reverse	5′-CCCATCCAGAACATGCCGTT-3′
pefD	PEFD-1	Forward	5′-GATGAAGTGGGGACTGGTGTCCCTG-3′
	PEFD-2	Reverse	5′-TTCAGCGTGTAGTCCTGGGTGCCG-3′
orf5	ORF5-1	Forward	5′-GACGAAGTGAAATCAGGTGG-3′
	ORF5-2	Reverse	5′-CTGTGGGTAAACCACATCAA-3′
rck	RCK-1-1	Forward	5′-CTGACACCCATTCCGTGT-3′
	RCK-1-2	Reverse	5′-GTAACCGACACCAACGTT-3′

Nucleotide sequence accession numbers.

The nucleotide sequence data reported in this paper will appear in the DDBJ/EMBL/GenBank nucleotide sequence databases under accession no. AB040415 and AB041905.

RESULTS AND DISCUSSION

General overview.

Large virulence plasmid pKDSC50 from serovar Choleraesuis strain RF-1 was isolated, purified, and finally cloned into the E. coli sequencing vector. Subsequently, the total nucleotide sequence of pKDSC50 was determined. The entire DNA sequence of pKDSC50 consists of 49,503 bp forming a circular plasmid. The nucleotide sequence from bp −501 to bp 6417, which contains IS630 and the spvRABC, was previously published (44–46). pKDSC50 contains 48 identified ORFs, which were searched against a nonredundant protein database summarized in Table 3. Of the 48 putative ORFs, 28 (58%) were known to be encoded by other virulence plasmids of Salmonella; 9 (19%) were homologous to genes of other conjugative plasmids, including F and R plasmids; and only 1 (2%) was an insertion sequence. Four (8%) ORFs which were predicted to encode truncated proteins were unrelated to Salmonella. The remaining six (13%) ORFs had no regions of significant homology to protein sequences in the current database. In addition to these ORFs, four noncoding elements were found on pKDSC50 (Fig. 1 and Table 3).

TABLE 3.

ORFs and noncoding elements of pKDSC50

ORF element	Gene	Position (bp)a	Product size (aa)b	Homologue found by BLAST	Identity/similarity or homology (%)	Accession no.
ORF1		c486–552	345	IS630		BAA01531
ORF2	spvR	1161–2054	297	SpvR		S30896
ORF3	spvA	2565–3332	255	SpvA		P17449
ORF4	spvB	3515–5290	591	SpvB		P17450
ORF5	spvC	5571–6296	241	SpvC		P15805
ORF6	spvD	6556–7206	216	SpvD	99/100	P24420
ORF7	orfE	7815–8165	116	OrfE	100/100	S24394
ORF8		c8670–9215	181
ORF9		c9441–9911	156
ORF10		c9982–10488	168
ORF11		c10632–11183	183	Putative ORF	99/99 (truncated)	AAA27204
ORF12	rsd	c11192–11974	260	Rsd	100/100	A38114
ORF13		c12009–12530	173	Putative ORF1	100/100	AAA27203
ORF14		c12527–12817	96	Putative ORF2	100/100	AAA27202
ORF15	ccdB	c12819–13124	101	CcdB	82/99	P05703
ORF16		14053–14541	162	ORF1	90/96	CAB46352
ORF17	repA	15786–16802	339	Replication protein (RepA)	95/96	AAB53036
ORF18		c17305–17691	128	LO140 (bacteriophage 933W)	48/89 (truncated)	BAA84356
ORF19	pefB	18397–18699	100	PefB	100/100	S32887
ORF20	pefA	18974–19492	172	PefA	89/99	S32888
ORF21	pefC	19719–22127	802	PefC	98/99	P37868
ORF22	pefD	22129–22812	227	PefD	98/100	AAC36961
ORF23	repA	c23470–24339	289	Replication protein (RepA)	99/100	AAB53039
ORF24	tap	c24332–24409	25	Tap	100/100	AAB53038
ORF25	copB	c24659–24904	81	CopB	100/100	AAB53037
ORF26		25077–25571	164
ORF27	finO	c25817–26380	187	FinO	75/97	P08315
ORF28		c26652–27176	174	TraX	74/96 (truncated)	P27710
ORF29		c28038–28475	145	TraI	78/93 (truncated)	BAA7885
ORF30		c28609–29691	360	TraI	84/98 (truncated)	BAA7885
ORF31		c29783–30265	160	TrbH	73/92 (truncated)	P19381
ORF32		c30384–32564	726	TraD	85/96 (truncated)	AAC44181
ORF33	traT	c32935–33666	243	TraT	91/98	P32885
ORF34		c34208–34819	203	TraG	46/90 (truncated)	P33790
ORF35	samA	35211–35633	140	SamA	100/100	P23831
ORF36	samB	35633–36907	424	SamB	99/100	P23832
ORF37	parB	c37093–37962	289	ParB	99/99	M97752
ORF38	parA	c38202–39167	321	ParA	100/100	M97752
ORF39		39579–40523	314	35.3-kDa protein	100/100	P37415
ORF40		c41178–41594	138
ORF41		c41697–42113	138	15.2-kDa protein	100/100	P37414
ORF42	tlpA	c42199–43314	371	TlpA	99/100	A44122
ORF43		c43588–44067	159
ORF44	mig-5	44715–45455	246	Mig-5	99/100	AAB80735
ORF45	rlgA	45662–46222	186	RlgA	98/99	AAC04696
ORF46		46242–47330	362	ORF3 of Pseudomonas sp. strain TW3	78/95 (truncated)	AAF23987
ORF47		47377–47871	164	ORF3 of Pseudomonas sp. strain TW3	58/90 (truncated)	AAF23987
ORF48		47873–48829	318	ORF2 of Pseudomonas sp. strain TW3	76/97 (truncated)	AAF23986
Noncoding		13453–14087		crs	80	X66934
		23135–23285		oriR	100	U64797
		36888–36974		parS	98	M97752
		40534–40587		incR	100	M97752

^aPositions beginning with the letter c represent ORFs in the minus strand. 

^baa, amino acids. 

FIG. 1.

Map of the whole pKDSC50 plasmid. The circle shows ORFs with their orientations denoted by their positions; ORFs outside the ring have a clockwise orientation: and those inside the ring have a counterclockwise orientation. ORFs encoding virulence-associated factors are indicated by red boxes; ORFs encoding proteins related to replication and plasmid maintenance functions are indicated by blue boxes; ORFs encoding proteins homologous to the tra operon are indicated by green boxes; and the IS element is indicated by a yellow box. For the nomenclature of the ORFs, see Table 3.

Virulence-associated genes.

A BLAST search revealed several homologues to known virulence-associated genes which localized to the large plasmid derived from other Salmonella serovars. Among these were Spv proteins, Pef proteins, and other proteins essential for virulence, such as Mig-5, a carbonic anhydrase that is required for systemic infection in the mouse (60), and another possible virulence-associated regulatory protein, TlpA, which functions as a thermometer by regulating its own transcription according to temperature (32).

Serovar Typhimurium Pef fimbriae are encoded by the pef locus, which is composed of the pefB, pefA, pefC, pefD, orf5, orf6, pefI, and orf7 genes (Fig. 2). The predicted amino acid sequence of these proteins suggests that PefA is the major fimbrial subunit; the PefC and PefD proteins are the outer membrane usher and the periplasmic chaperone, respectively; PefB, Orf5, and Orf6 are the minor fimbrial subunits; and PefI and Orf7 are regulatory proteins (19). On the pKDSC50 plasmid, a 6.9-kb region downstream of the pefD gene was completely deleted (4, 19) (Fig. 2 and 3). It contained a part of the pef operon and the srg (SdiA-regulated gene) region, including srgA, a homolog of dsbA (disulfide bond isomerase); srgB; rck (resistance to complement killing); and srgC, a homolog of the AraC family of transcriptional regulators. This incomplete pef operon suggests that serovar Choleraesuis cannot express functional Pef fimbriae. Thus, we further determined and compared the genetic organization of the second virulence-associated region, the pef-rck region, among different Salmonella serovars. All clinical and laboratory Salmonella strains, including three serovar Typhimurium strains, nine serovar Choleraesuis strains, four serovar Enteritidis strains, two serovar Dublin strains, two serovar Gallinarum strains, and two serovar Pullorum strains, were confirmed for the presence of virulence plasmids by PCR using two different primers specific for the spvR and spvC genes. To determine whether the pef and rck genes are localized on the five different serotype plasmids, we amplified the internal region of the genes by PCR (Fig. 2). All virulence plasmids from serovar Typhimurium carried the pefB, pefA, pefC, pefD, orf5, and rck genes, whereas these genes were all absent in all of the plasmids from serovars Dublin, Gallinarum, and Pullorum. Like pKDSC50 from serovar Choleraesuis strain RF-1, only the pefB, pefA, pefC, and pefD genes were detected in all of the virulence plasmids from other serovar Choleraesuis strains. Furthermore, plasmids from serovar Enteritidis strains carried pefB, pefA, pefC, pefD, and rck but not orf5. Complementary Southern blot hybridizations were performed by using gene-specific probes, with almost the same results (Fig. 2). The restricted distribution of the pef-rck region among the virulence plasmids strongly suggests that this region was recently introduced into the virulence plasmid, probably by horizontal transfer.

FIG. 2.

Distribution of the pef and rck genes among selective Salmonella serovars. (A) The genetic organization of the pef operon on pLT2, the 94-kb virulence plasmid of serovar Typhimurium (19) is shown. Gene-specific PCR products obtained by using a set of primers listed in Table 2 are indicated as thin lines with each DNA size amplified. (B) The pef and rck genes were detected by PCR and DNA-DNA hybridization (Blot). In Southern dot hybridizations, fluorescein-labeled pefB, pefA, pefC, pefD, orf5, and rck gene-specific PCR products from pLT2 of serovar Typhimurium were used as probes. Note that PCR and Southern blot analyses yielded identical results, except in the case of the pefC gene of the virulence plasmid from serovar Choleraesuis strain AH1, which was PCR negative and Southern blot positive.

FIG. 3.

Comparison of genetic organization of the pef region from Salmonella serovars Choleraesuis, Typhimurium, and Enteritidis. On pKDSC50, a 6,880-bp DNA sequence downstream of pefD on pLT2 is completely deleted. On pS72, the virulence plasmid of serovar Enteritidis, orf5 was truncated and orf6 was replaced with a 1,296-bp DNA fragment unrelated to the pef operon (see text for details). The identities of amino acid sequences with the pef-, srg-, and rck-encoded proteins are given as percentages.

To further analyze this region, we determined the nucleotide sequence of a 14.7-kb segment of the 60-kb virulence plasmid of serovar Enteritidis and compared its genetic organization to the corresponding region of the virulence plasmid from serovar Typhimurium (Fig. 3). DNA sequencing revealed that the 1.3-kb region corresponding to bp 7285 to 8545, which contains orf6 in the serovar Typhimurium plasmid, was replaced with an unrelated 1,296-bp DNA sequence which contains orf6e (accession no. U66901). In addition, Orf5 consisted of 95 amino acids corresponding to the N-terminal half of the 185-amino-acid protein predicted to be encoded by orf5 of the serovar Typhimurium plasmid. Since insertional inactivation in orf5 affects the surface presentation of Pef fimbriae in the E. coli host (19), it is likely that serovar Enteritidis cannot produce complete Pef fimbriae. Thus, these results indicate that the functional pef operon is conserved in only the serovar Typhimurium plasmid. Minor mutations were also found in dlpA (srgA homolog) (CAA63987) (51), srgB, and srgC of the serovar Enteritidis plasmid.

Interestingly, the amino acid sequence deduced to be ORF18 of pKDSC50, which was found just 0.5 kb upstream of repA of the repC region and 0.7 kb downstream of the pef operon and was shown to have homology to hypothetical protein L0140 (accession no. BAA84356) of Shiga toxin 2-converting bacteriophage 933W from enterohemorrhagic E. coli (EHEC) strain O157:H7, was highly conserved on both the virulence plasmids of serovars Choleraesuis and Enteritidis but absent in the serovar Typhimurium plasmid (Fig. 3), indicating more recent acquisition.

Replication and plasmid maintenance.

Three potential plasmid replication regions were found, one resembling the RepFIIA replicon, another resembling the RepFIB replicon, and the third containing genes involved in stable maintenance in the host. The first replication region showed very high homology to the repB (RepFIIA) region of serovar Enteritidis plasmid pFM82139 (accession no. U64796) (52) in the translated ORFs (99% identical to RepA, 100% identical to Tap and CopB). Tap is necessary for the translation of repA through translational coupling (9), and CopB is involved in plasmid copy number control. The same sequence showed high similarity to the RepFIIA replicon of plasmids pB171, the enteropathogenic E. coli (EPEC) adherence factor plasmid of EPEC strain B171 (O111:NM) (accession no. AB024946) (59); pO157 of EHEC strain O157:H7 (AB011549 and AF074613) (10, 41); pYVe227 of Yersinia enterocolitica (AF102990); pCD1 of Y. pestis (AF074612) (31); and R100 (NR1) (NC002134). The second replication region exhibited 95% identity (96% similarity) to plasmid pFM82139 of serovar Enteritidis; 97% identity (100% similarity) to the RepFIB replicon of pB171, the EPEC adherence factor plasmid of EPEC; 91% identity (98% similarity) to pO157 of EHEC; and 62% identity (93% similarity) to pMT1 of Y. pestis (AF074611) (40) (Fig. 4). Outside of the encoding region of the RepFIB replicon, repeats B through J, which contained the consensus sequence 5′-ANATAAGCTGTAGNNNGNAAA-3′, were also found on plasmid pKDSC50. Both the RepFIIA and RepFIB replication regions of serovars Typhimurium and Enteritidis, named repB and repC, respectively, have been proven to be functional replicons (51, 58). The third replication region contains genes necessary for stable maintenance of the plasmid. pKDSC50 contains the par region of the serovar Typhimurium virulence plasmid consisting of four loci, incR, parA, parB, and parS, which are required for incompatibility and partition (11), between bp 36888 and bp 40587. Another region composed of rsd, the trans-acting resolvase gene, and crs, the cis-acting resolution site, which encodes a multimer resolution function involved in plasmid stability (38), was found. In addition to these loci, the samAB operon was located downstream of the parAB operon (D90202), which has been shown to be involved in the mediation of UV mutagenesis (48). ORF15, which was located in the rsd-crs region, showed high similarity (99%) to the ccdB genes of the F plasmid (P05703). The ccd operon, which is responsible for postsegregational killing of segregant cells, consisted of two genes, ccdA and ccdB, on the F plasmid, but the corresponding region on the pKDSC50 plasmid did not contain a ccdA homologue, indicating that the ccd system of pKDSC50 is unlikely to be functional.

FIG. 4.

Primary structure of the RepFIIA and RepFIB regions. Comparison of the repB (RepFIIA) (A) and repC (RepFIB) (B) regions of pKDSC50 and the corresponding regions of other plasmids from entropathogenic bacteria. These regions included pFM82139 of serovar Enteritidis (accession no. U64796), the pB171 plasmid of EPEC (AB024946), and the pO157 plasmid of EHEC O157:H7 (AB011549), as well as the pMT1 (AF074611) and pCD1 (AF074612) plasmids of Y. pestis, the pYVe227 plasmid of Y. enterocolitica (AF102990), and the R100 (NR1) plasmid (NC_002134). ORFs are indicated by arrows under the DNA of plasmid pKDSC50. Nucleotide positions are indicated on the line. Homologous amino acid sequences of these plasmids are shown by thick lines with the percentages of similarity and identity under the arrow, indicating homologous regions in the nucleotide position in each sequence.

Transfer region.

Although some virulence plasmids of serovar Typhimurium are mobilized by conjugation under inducing conditions (3), almost all of the Salmonella virulence plasmids, including pKDSC50, are known to be nonconjugative (50). Genes from bp 25817 to 34819 in pKDSC50 were found to be homologous to the tra region genes from the F and R100 (NR1) plasmids, which included finO, traX, traI, trbH, traD, traT, and traG. Only two genes, finO (75% identical and 97% similar to the R100 plasmid) and traT (91% identical and 98% similar to the R100 plasmid), were complete; all of the other genes were truncated. In addition, the region downstream of the traG gene, which contains the tra genes required for DNA transfer, was completely deleted. This suggests that pKDSC50 is defective in DNA transfer by conjugation.

IS element.

pKDSC50 contains a single copy of IS630. However, the nucleotide sequence from bp 45662 to bp 48829 was highly homologous to the sequence downstream of the ntn operon (AF043544), which is involved in the catabolism of 4-nitrotoluene and toluene in Pseudomonas sp. strain TW3. ORF45 (RlgA) has 75% similarity to ORF4, which is, in turn, similar to Xanthomonas campestris transposase gene tnpR. Recently, it has been reported that RlgA is a member of the resolvase family of site-specific recombinases and that mutation in rlgA on the virulence plasmid from serovar Typhimurium did not affect the stability of the plasmid (43). ORF46 and ORF47 correspond to one-half of the N-terminal part and one-third of the C-terminal part of ORF3, respectively; ORF3 is similar to the Deinococcus radiodurans putative transposase. ORF48 showed strong homology (97% similarity) to ORF2, which is similar to a Pseudomonas syringae protein of unknown function. These observations raise the possibility that a DNA element of plasmid pKDSC50 was acquired from Pseudomonas through horizontal transfer of a mobile element.

Base distribution.

In addition to the presence of sequences related to different mobile genetic elements, the mosaic nature of pKDSC50 was further indicated by the results of a base composition analysis of the plasmid. Although the average G+C content of the plasmid was 52.1%, which is close to the value of the Salmonella chromosome, a more detailed analysis revealed that the region of the plasmid containing the spv operon had a significantly different G+C content (45.7%) than the surrounding region of the DNA (Fig. 5). Moreover, the G+C content of the predicted coding regions ranged from 39 to 70%, suggesting that development of the pKDSC50 plasmid may occur through the acquisition of DNA fragments from various microorganisms whose DNAs have a lower or higher G+C content.

FIG. 5.

G+C content of pKDSC50 with a window of 100 bp across. The line indicates 50% G+C content, and selected ORFs are shown as hatched boxes drawn to the correct scale. The graph was generated using the GENETYX-Mac program (Software Development Co., Ltd., Tokyo, Japan).

Conclusions.

The complete sequence of pKDSC50, the 50-kb virulence plasmid of serovar Choleraesuis, reveals a plasmid size unique to each Salmonella serovar. Comparison of the nucleotide sequences determined for the 93,939 bp of pLT2, the virulence plasmid of serovar Typhimurium strain LT2, obtained from the Salmonella genome database (Genome Sequencing Center, Washington University), and the 49,503 bp of pKDSC50 showed that 47 out of the 48 ORFs of pKDSC50 are highly homologous to the corresponding ORFs of pLT2 and that both plasmids contain the genes in the same order. Furthermore, two large deletions downstream of the pef region and most of the tra region were present on pKDSC50 (Fig. 6). This suggests that the virulence plasmid of serovar Choleraesuis is a variant of the virulence plasmid of serovar Typhimurium and was generated by deletion events that occurred during their divergence from a common predecessor. In addition, all genes of the 60-kb virulence plasmid of serovar Enteritidis thus far detected, and their genetic orders, are identical to these plasmids (13, 52), suggesting that the virulence plasmids from serovars Typhimurium, Choleraesuis, and Enteritidis share a common ancestry.

FIG. 6.

Linear genetic maps of the virulence plasmids pLT2 of Salmonella serovar Typhimurium and pKDSC50 of serovar Choleraesuis. The nucleotide sequence of pLT2 was obtained from the Salmonella genome database (http://genome.wustl.edu/gcs/). Areas of pKDSC50 that are not present in plasmid pLT2 are shown as open bars. The nucleotide positions of the deletions in the pKDSC50 sequence and the corresponding positions in pLT2 are indicated. The positions of genes are shown above the plasmids.

The differing distributions of the virulence-associated genes among virulence plasmids from Salmonella serovars Typhimurium, Enteritidis, and Choleraesuis suggest that the present genes might confer an advantage for adaptation to and infection of the respective hosts. For example, a number of Salmonella strains harbor several different kinds of adhesive structures that mediate attachment to host surfaces. Phylogenic studies have revealed that the genes encoding mannose-sensitive type 1 Fim fimbriae (55) and the thin aggregative Agf fimbriae (16), both of which are also found in commensal E. coli, are present in most Salmonella isolates. In contrast, the prevalence of the Salmonella-specific fimbrial operons, which include the operons encoding long polar Lpf fimbriae (7), plasmid-encoded Pef fimbriae (7), and serovar Enteritidis Sef fimbriae (14, 57), were limited to a small number of Salmonella serovars. In this study, we have shown that the complete pef operon is conserved only in the virulence plasmid of serovar Typhimurium and not in those of serovars Choleraesuis and Enteritidis. The expression of serovar-specific fimbriae may confer a selective advantage by facilitating bacterial attachment, resulting in a possible relationship to bacterial host adaptation. Furthermore, serovars Typhimurium and Enteritidis, but not Choleraesuis, carry the rck gene on their virulence plasmids and this gene encodes a protein that confers high levels of host serum resistance on the bacteria (26, 27). Therefore, it is possible that these two serovars, which have a wider range of host specificity than serovar Choleraesuis, have increased chances of survival during infection of murine and nonmurine hosts.