DNA Sequence and Comparison of Virulence Plasmids from Rhodococcus equi ATCC 33701 and 103

Abstract

The virulence plasmids of the equine virulent strains Rhodococcus equi ATCC 33701 and 103 were sequenced, and their genetic structure was analyzed. p33701 was 80,610 bp in length, and p103 was 1 bp shorter; their sequences were virtually identical. The plasmids contained 64 open reading frames (ORFs), 22 of which were homologous with genes of known function and 3 of which were homologous with putative genes of unknown function in other species. Putative functions were assigned to five ORFs based on protein family characteristics. The most striking feature of the virulence plasmids was the presence of a 27,536-bp pathogenicity island containing seven virulence-associated protein (vap) genes, including vapA. These vap genes have extensive homology to vapA, which encodes a thermoregulated and surface-expressed protein. The pathogenicity island contained a LysR family transcriptional regulator and a two-component response regulator upstream of six of the vap genes. The vap genes were present as a cluster of three (vapA, vapC, and vapD), as a pair (vapE and vapF), or individually (vapG; vapH). A region of extensive direct repeats of unknown function, possibly associated with thermoregulation, was present immediately upstream of the clustered and the paired genes but not the individual vap genes. There was extensive homology among the C-terminal halves of all vap genes but not generally among the N-terminal halves. The remainder of the plasmid consisted of a large region which appears to be associated with conjugation functions and a large region which appears to be associated with replication and partitioning functions.

Rhodococcus equi is an important pulmonary pathogen of foals and is increasingly isolated from pneumonic infections and other infections in human immunodeficiency virus (HIV)-infected patients (19, 33). Isolates from foals possess a large virulence plasmid, varying in size from 80 to 90 kb (45, 47, 49). Isolates lacking the plasmid are avirulent to foals (16, 51). Little is known about the function of the plasmid apart from its encoding a virulence-associated surface protein (VapA) (45, 49), the presence of a family of four vap genes (5), and the origin of replication (53). Infection with R. equi bacteria carrying the virulence plasmid may lead to immunomodulation in foals by causing failure to mount an effective Th1-based cellular immune response, but the basis of this effect is undefined (17). The expression of VapA is thermoregulated (≥34°C) and pH regulated (41, 42), so that in this respect the plasmid has similarities to the virulence plasmids of pathogenic Yersinia species, such as Yersinia pestis, and of Shigella species (11, 22, 30). The plasmid is of significant interest, since it is associated with survival of the bacterium inside macrophages (16, 21, 33). Understanding its structure and function may therefore yield insights not only into the basis of virulence of this organism but also into the mechanisms of macrophage survival of other facultative intracellular pathogens, including Mycobacterium tuberculosis, a fellow member of the mycolata.

To better understand the structure and function of the virulence plasmid, we have completely sequenced plasmids from two isolates of R. equi that were recovered from pneumonic foals, ATCC 33701 and 103. These strains have been used previously for laboratory and experimental infection studies (16, 51), and both are of the 85-kb type I (29, 46). We sequenced both plasmids on the assumption that a comparative approach would lead to a better understanding of the structure and function of the plasmid and of individual genes.

MATERIALS AND METHODS

Bacterial strains plasmid isolation.

The plasmids from R. equi ATCC 33701, a capsular serotype 1 prototype, and 103, a capsular serotype 6 isolate, were used; these are designated p33701 and p103. Escherichia coli strain XL1 Blue was used for plasmid or phage cloning.

DNA sequencing procedure.

For p33701, shotgun cloning was carried out. Plasmid DNA was extracted from bacterial cultures as described previously (45). Plasmid DNA was recovered using CsCl-ethidium bromide gradients to eliminate chromosomal contamination and then partially digested with Sau3AI. Partially digested plasmid DNA was fractionated by 10 to 40% sucrose gradient centrifugation, and fractions containing 1- to 2-kbp fragments were collected. These were ligated into the BamHI site of pUC19 or pBluescript (Stratagene, La Jolla, Calif.). Then, E. coli DH5α was electrotransformed using GenePulserTM (Bio-Rad). Each clone was sequenced in both directions using dye-terminator-labeled fluorescent cycle sequencing prism reagents and ABI 373A automated sequencers (Applied Biosystem Division, Perkin-Elmer). The sequence data was analyzed using SEQUENCHER version 3 (Gene Codes). From a total of 1,078 sequence data, 23 unique fragments (contigs) were obtained. To determine the position of each contig on p33701, EcoRI fragments b to i and HindIII fragments b to d of p33701 were cloned in pUC19, and both ends of each were sequenced. The positions of some contigs were determined from partial sequences of the EcoRI and HindIII fragments. The sequences of gap regions between contigs were determined by primer walking using pUC19 containing EcoRI and HindIII fragments as templates. For p103, sequencing was initially of a library of the seven smallest EcoRI fragments (c to i) cloned in pBluescript. The 27.5-kb and 15.4-kb EcoRI a and b fragments were subcloned from the cosmid vector pLARF1 after isolation from agarose using the Lambda DASH II cloning kit (Stratagene). Further subcloning of the a fragment into pBluescript was done after digestion with EcoRI, BamHI, or HindIII, and DNA was sequenced from the three fragments produced. Sequencing was from either end of the fragments from the T3 and T7 primers in pBluescript or Lambda DASH II and, in the case of the a and b fragments, by primer walking using sequence data available from some individual contigs of p33701. p103 was sequenced in both directions using dye-terminator-labeled fluorescent cycle sequencing Prism reagents and the ABI 377 (Perkin-Elmer) automated sequencers (Guelph Molecular Supercentre, University of Guelph, Guelph, Canada). After all the fragments were sequenced, the sequence data were assembled and analyzed using GeneRunner3.04 (Hastings Software, Hudson, N.Y.). Other sequences were aligned and compared using the Wisconsin package 10.0 from the Genetics Computer Group (GCG). Sequence data from the two plasmids were compared after alignment using GeneRunner. Where there were missing or discrepant base pairs between the two plasmids, the sequence obtained in these areas was reexamined and the area was resequenced if required.

DNA sequence analysis and annotation.

Open reading frames (ORFs) encoding at least 50 amino acids were identified using GeneRunner or GeneMark (24) programs to display start codons, stop codons, and codon usage statistics for each reading frame. The start codon farthest upstream was used to annotate the ORF start. Comparisons of the predicted protein sequence with protein sequences in the available databases and pairwise protein alignments were done using the program BLAST at the server of the National Center for Biotechnology Information at the National Library of Medicine (1). Known genes and putative functions were assigned for individual ORFs by inspection of the search output. Homologies were considered to be significant when at least 25% of the amino acids were identical for at least 50% of the protein in the database. When a putative protein did not show significant homology with any known proteins, we analyzed the upstream DNA for known ribosomal binding sites. Where relevant, multiple protein sequences were aligned by DNASTAR (DNASTAR Inc., Madison, Wis.) using the algorithm developed by Lipman et al. (23). DNASTAR was also used to construct a phylogeny tree. Direct and indirect repeats were also identified using DNASTAR. Protein sequences were also analyzed for functionally important motifs using PROSITE (3) and for functional domains using Pfam (38) and BLOCKS (20), using Internet-based protein database programs. For Fam and BLOCK, e values of ≤0.01 were regarded as significant, and for PROSITE normalized scores of ≥9.0 were considered significant.

GenBank nucleotide sequence accession numbers.

The annotated sequence of p33701 was submitted to the DNA Databank of Japan under the accession number AP001204, and the sequence of p103 was submitted to GenBank under accession number AF116907.

RESULTS AND DISCUSSION

General description.

p33701 (80,610 bp) and p103 (80,609 bp) are virtually identical plasmids. We describe here the main features of p33701, noting the few differences from p103. A genetic map of p33701 is shown in Fig. 1. Table 1 lists designated ORFs and their putative functions based on homology with genes of known function identified by BLAST search and by protein domain or motif database searches. Homology given in Table 1 is the percent amino acid homology between the virulence plasmid ORF and the entire gene in GenBank to which it is compared. The minor differences between the plasmids are described below.

FIG. 1.

Map of the R. equi virulence plasmid p33701. The circle shows orientation of the ORFs, with the direction of reading denoted by shading. Genes with a known or suspected putative function based on homology with genes of a known function in DNA databases are named (see Table 1).

TABLE 1.

ORFs in the virulence plasmid p33701 of R. equi ATCC 33701

Designation	Function or comments (gene designation)a	Reference organism (accession no.)b (% homology)	Orientation	Location (bp)	Number of amino acids
ORF1	Lysyl-tRNA synthetase (lsr2)	M. tuberculosis (Z95557) (63)	+	795–1154	119
ORF2	R. equi Vap protein (vapG)	R. equi (D21236)	−	2059–1541	172
ORF3	Unknown	E. coli hypothetical protein (U82598) (35)	−	4706–3951	251
ORF4	LysR family transcriptional regulator	B. subtilis (Z99124) (54)	+	4993–5829	278
ORF5	Bicyclomycin (sulfonamide) resistance protein	E. coli (P43531) (48)	+	6322–7230	302
ORF6	R. equi Vap protein (vapH)	R. equi (AF118813)	+	7367–7930	187
ORF7	Unknown		+	7982–8419	145
ORF8	Two-component response regulator	B. subtilis (P35163) (55)	+	8534–9292	252
ORF9	Unknown		+	10278–10505	75
ORF10	Unknown		+	10574–10759	61
ORF11	Unknown		+	10963–11325	120
ORF12	R. equi Vap protein (vapA)	R. equi (JC4072)	+	12546–13115	189
ORF13	Unknown		+	13356–13598	80
ORF14	R. equi Vap protein (vapC)	R. equi (AF118813)	+	13970–14494	174
ORF15	R. equi Vap protein (vapD)	R. equi (AF118814)	+	14835–15329	164
ORF16	Unknown		+	15897–16367	156
ORF17	Unknown		−	16906–16484	140
ORF18	Unknown		−	17856–17572	94
ORF19	R. equi Vap protein (vapE)	R. equi (AF118815)	+	19108–19728	206
ORF20	R. equi Vap protein (vapF)	R. equi (JC4072)	+	19874–20326	150
ORF21	Possible chorismate mutase, involved in phenylalanine biosynthesis (Pfam)	M. tuberculosis hypothetical protein (Z97193) (49)	−	21400–20927	157
ORF22	DNA invertase (invA)	Acetobacter pasteurianus (AF015307) (68)	+	22223–22801	192
ORF23	Unknown		+	23106–23630	174
ORF24	Unknown		+	23683–23994	103
ORF25	Unknown		−	24826–24388	145
ORF26	Putative methylase or helicase	Hypothetical methylase gene homologue, A. tumefaciens (AB016260) (41)	+	25511–35200	3,229
ORF27	Unknown	M. tuberculosis hypothetical protein (Z96072) (42)	+	36778–37767	329
ORF28	Unknown		+	38402–38869	155
ORF29	Unknown		−	39305–39006	99
ORF30	Conjugal transfer protein (traA)	Rhizobium sp. (AE000069) (19)	+	39644–43747	1,367
ORF31	Unknown		+	43797–44849	350
ORF32	Type 1 DNA topoisomerase (topA)	Salmonella enterica serovar Typhi (AAF14809) (38)	−	47409–44935	789
ORF33	Unknown	S. coelicolor hypothetical protein (AL049645) (26)	−	48102–47485	205
ORF34	Conjugative transfer gene complex (trsK)	S. coelicolor hypothetical protein (AL049645) (47); S. aureus trsK (L11998-12) (30)	−	50117–48105	670
ORF35	Unknown		−	50679–50134	181
ORF36	Peptidase, M23/M37 family, possible Gly-Gly endopeptidase (Pfam)	S. coelicolor (AL132662) (33)	−	52331–50691	546
ORF37	Unknown		−	52627–52328	99
ORF38	Unknown	S. coelicolor putative ATP-binding protein (AL049661) (23)	−	54399–52624	591
ORF39	Unknown	S. coelicolor putative integral membrane protein (AL049661) (34)	−	55914–54412	500
ORF40	Transfer gene complex (trbL)	Plasmid RP4 (M93696-12) (17)	−	55867–56911	318
ORF41	Unknown		+	57004–57417	137
ORF42	Unknown		−	58330–57491	279
ORF43	Unknown		−	58938–58327	203
ORF44	Unknown		−	59333–58935	132
ORF45	Unknown		+	59392–60114	240
ORF46	Unknown		−	61000–60233	255
ORF47	Unknown		−	62157–60997	386
ORF48	Unknown		−	63374–62223	383
ORF49	Possible biopterin-dependent aromatic amino acid hydroxylase (Pfam)		−	64669–63401	422
ORF50	Repressor protein (trbA)	Rhodococcus sp. plasmid (AF059700) (49)	+	65230–65676	148
ORF51	Chromosome-partitioning ATPase (parA)	Deinococcus radiodurans (AE001826) (41)	+	66207–66911	234
ORF52	Unknown	Unknown	+	66904–67386	160
ORF53	Unknown, putative phage excisionase	Mycobacteriophage TM4 gp82 (AF068845) (50)	+	67898–69046	382
ORF54	Unknown		+	69581–70498	305
ORF55	Unknown		+	70526–71326	266
ORF56	Regulatory protein (DNA binding helix-turn-helix protein) (Pfam)		+	71674–72045	123
ORF57	Unknown		+	72456–73454	332
ORF58	Unknown		+	74042–74563	173
ORF59	Unknown		+	74892–75185	97
ORF60	DnaB-like helicase protein (Pfam)		+	75258–75827	189
ORF61	DNA transposon resolvase (resA)	Corynebacterium glutamicum (AF121000) (34)	+	75874–76458	194
ORF62	Chromosome partitioning protein (parB)	Deinococcus radiodurans (AE001865) (28)	+	77971–79521	516
ORF63	Unknown		+	80202–80567	121
ORF64	Unknown		−	184–80267	175

^aORFs were assigned a putative function based on criteria outlined in Results and Discussion. (Pfam) indicates that the designated putative function is based on a protein domain search. 

^bHighest amino acid homology of the R. equi ORF with characterized genes from the organism listed (GenBank accession number). 

The majority (37, excluding the seven vap genes) of the 64 ORFs identified have unknown function, based on their lack of homology with genes of known function in GenBank or with protein domains or motifs in protein databases. Twenty-two have homology with genes of known function, and three are homologous with putative genes of unknown function. For five genes, putative functions were assigned based on protein domain or motif characteristics. Four ORFs are most closely related to genes identified for M. tuberculosis, two of which are hypothetical proteins identified through genome sequencing (7). Five ORFs are most closely related to genes identified in Streptomyces coelicolor, two of which are also hypothetical proteins identified through genome sequencing (34).

Despite the lack of homology to known genes of many of the ORFs, on the basis of genes of known function or of protein family characteristics, the plasmid appears to divide into three major regions: a virulence region, a conjugation region, and a replication and partitioning region. Further work is required to confirm these suggested divisions.

Pathogenicity island and virulence-associated protein genes.

A 27,536-bp region bounded by two transposon resolvases (ORF61 and ORF22) has the characteristics of a pathogenicity island because it is bounded by insertion elements, contains a block of apparently foreign genes (GC content, 60.8%, compared to 66.6% for the rest of the plasmid), and contains a tRNA gene (a lysyl-tRNA synthetase [lsr2] gene) (18, 26). This region contains vapA and six other related virulence-associated protein genes, discussed below. It is apparent that a Rhodococcus plasmid has at some stage acquired a pathogenicity island from a source of as yet unknown origin and that this region of the plasmid may be the virulence region. Based on homology with genes or protein families of known function and given the presence of this pathogenicity island, the remainder of the plasmid appears to be concerned with conjugation, replication, and partition functions, as discussed below. The discovery of a pathogenicity island is important, in part since it opens the way to more ready sequence analysis of the virulence region in other R. equi plasmids, including that of the intermediately virulent isolates (40, 43, 44).

The two ORFs with homology to transposon resolvases that define the boundaries of the pathogenicity island are not associated with other transposon-related genes, nor are there associated inverted repeat regions. ORF22 is actually most closely homologous to invertase genes but has significant homology to resolvases. Although invertases preferentially catalyze inversions, they may also at a low frequency mediate cointegration (31). The second transposon-related ORF, ORF61, has homology to resolvases. The surprising presence of a tRNA gene on a virulence plasmid suggests that the gene has been brought in with the pathogenicity island, a suggestion supported by the location of the gene in relation to the rest of the island. The significance of tRNA genes as insertion sites for pathogenicity islands is obscure (26). It may be relevant to immunity to R. equi in foals that human patients with leprosy recognize the Mycobacterium leprae Lsr2 protein by both a humoral and a cellular immune response (37).

The presence of seven vap genes was a striking feature of both plasmids. vapA has been previously described for each of the plasmids (36, 50), as have vapC, vapD, and vapE (5). Because a second intermediately virulent 22-kDa Vap protein has been described for large plasmids obtained from pig-derived isolates and some AIDS patient-derived isolates (43, 47), which we designate VapB, we have designated these additional vap genes as vapC to vapH. The seven vap genes are found over a span of 19,000 bp, in six instances in a positive direction. vapA, vapC, and vapD are clustered in close proximity to each other, and vapE and vapF are immediately adjacent to each other. vapG and vapH are present as individual genes, however, and vapG is unusual in that it is found on the negative strand. Interestingly, a series of 10 short oligonucleotides, varying in length from 12 to 26 bp, is present over a 1,056-nucleotide span from position 11484 to 12540, immediately upstream of the vapA–vapD cluster, and is directly repeated at 18034 to 19103, immediately in front of vapD and vapE. These direct repeats are not an ORF and are not found upstream of vapG and vapH. We speculate that these two regions are associated with the thermoregulation of vap gene expression. It may be relevant that large repeated sequences have been observed that are associated with the yopM gene on the Lcr virulence plasmid of Yersinia pestis, but analysis has shown no relation to regulation of gene expression (35).

The numbers of amino acids encoded by the vap genes are 150 (vapF), 164 (vapD), 172 (vapG), 174 (vapE), 187 (vapH), 189 (vapA), and 206 (vapE). Each vap gene is preceded by a ribosomal binding site, but only vapA is followed by a stem-loop structure. All except vapF have a signal peptidase I (A-X-A) site. Other than VapF, the Vap proteins showed extensive homology in their C-terminal halves (Fig. 2). Interestingly, VapF is not homologous in its C-terminal half but would have high homology in this region if there were not two frameshifts. A phylogenetic tree of the vap genes, including vapB obtained from an intermediately virulent R. equi strain, is shown in Fig. 3. It may be significant for the virulence function of Vap's that vapA and vapB are most closely related. The amino acid sequences encoded by the different vap genes on p33701 and p103 are identical, except for one amino acid difference in the product of vapA and an additional amino acid encoded at the start of vapF. VapA is surface expressed in R. equi (42). The presence of a family of vap genes that have the potential to be surface expressed and which lack homology in their N-terminal halves might suggest that the plasmid provides a system for antigenic variation, if these proteins were anchored to the cell wall via their C-terminal ends. Although differential expression of these genes within a host might result in antigenic variation, chronic persistent infection is not really a feature of the disease in foals. The identity of the vap genes on virulence plasmids from two unrelated R. equi strains of different serotypes (32) shows that the role of these genes within each plasmid is not the continuing production of antigenic variants. Rather than the usual arrangement in which surface expressed proteins of gram-positive bacteria are linked to the cell wall through the protein C-terminal end (28), it is thought that VapA is linked to the cell wall through an unusual lipid modification occurring at the N-terminal end (48), so that the conserved C-terminal region at least of VapA is presented on the surface of the organism. The processes by which different variants of the vap genes arose and the benefit of the presence of a family of such genes remain to be determined. Further work is required to investigate the expression of vap genes in vivo, their antigenicity, and their function in virulence. In this regard, the recent observation that R. equi bacteria expressing vapA in the absence of other virulence plasmid genes were avirulent for foals might be explained if a combination of Vap proteins (or other plasmid components) is required for virulence (16). Elements of the C-terminal end of vap gene nucleotides were present between vapA and vapC, but these gene fragments are not associated with an ORF.

FIG. 2.

Virulence-associated protein amino acid alignment of the seven Vap (VapA and VapC to VapH) proteins on p33701 and of VapB identified in an intermediately virulent R. equi isolate. Solid shading shows the consensus of the majority of amino acids and indicates the extensive homology in the C-terminal halves of these proteins. VapF has extensive homology in its C-terminal half, but this is out of frame with the other Vap's. Dash, no amino acid in this Vap protein.

FIG. 3.

Phylogenetic tree of Vap proteins from p33701 and of VapB identified in an intermediately virulent R. equi isolate.

The function of ORF5, which has extensive homology with an E. coli bicylomycin resistance protein gene, is unclear, since earlier studies have shown no difference in antimicrobial drug resistance between isogenic pairs with and without the plasmid (9). ORF21, which occurs immediately after vapF on the pathogenicity island, has 51% homology (amino acid identities and positives) with an M. tuberculosis hypothetical protein (7), supporting the value of sequencing the R. equi virulence plasmid, since it identifies a possible role for this protein in virulence in tuberculosis. Protein domain searches suggested that ORF21 may be a chorismate mutase, involved in phenylalanine biosynthesis. Apart from the vap genes, there are eleven ORFs in the pathogenicity island the functions of which are unknown or unsuspected.

A 178-bp repeat was present at 15674 to 15852 and 21317 to 21495; its function is unknown, but it is of interest that it occurs immediately after the vapA–vapC gene cluster and again at the end of the vapE and vapF cluster. A series of five short oligonucleotides, of 19, 46, 71, 54, and 74 bp in length, present at 36470 to 36744 outside the pathogenicity island are repeated within the pathogenicity island at 77499 to 77773.

Regulatory genes in the pathogenicity island.

Two ORFs with homology to previously described regulatory genes are present upstream of six of the seven vap genes in the virulence region. ORF4 has extensive homology with members of the LysR group of positive regulatory proteins, one of the characteristics of which is negative autoregulation of their own transcription. In Salmonella, the plasmid virulence (spv) genes are activated by the LysR-like transcription factor SpvR and by the alternative sigma factor RpoS in response to signals received during intracellular growth or as the bacteria enter the stationary growth phase. Chromosomally encoded DNA-binding proteins contribute to the control of spv expression (25). Since LysR proteins can be positive regulators of a regulon (6), it will be of interest to determine whether expression of the vap genes is regulated by this protein. It will also be interesting to examine the role of both plasmid and chromosomal genes in regulating the function of ORF4. ORF8 is a gene with extensive homology to the two-component regulator gene resD or phoP of Bacillus subtilis (52) and to many similar two-component regulator genes. The unusual feature of this regulator gene is that although it is homologous with the response regulator component of a two-component regulatory gene, no gene homologous to the sensor kinase component was identified. This characteristic is shared with a homologous glutamine synthetase glnR gene response regulator in S. coelicolor (52). If the response regulator is phosphorylated by a sensor kinase, the gene for this may be located on the chromosome. In B. subtilis, ResD activates global regulation of aerobic and anaerobic respiration (39). The position of this response regulator upstream of the vapA–vapD gene cluster suggests an important role for this regulator in controlling expression of vap genes. It may also be significant that the promoter of the mtrA gene of M. tuberculosis, a gene with which it shares significant homology, is activated on entry into macrophages (50).

Conjugation region.

A second region of the plasmid appears to be concerned with conjugation functions, a complex process which involves many genes. For example, the conjugation regions of plasmids pGO1 of Staphylococcus aureus and pMRC01 of Lactococcus lactis consist of 14 and 16 genes, respectively (12, 27). By analogy, we suggest that the conjugation regions of p33701 and p103 include ORFs 23 to 40 (23106 to 54412), thus including ORF39, which has homology to a Streptomyces integral membrane protein which would probably be involved in conjugation, and ORF40, which has homology to a gene in the transfer gene complex of plasmid RP4.

The largest ORF of the plasmid, ORF26, encodes a protein of 3,339 amino acids with 45% homology to a 1,693-amino-acid hypothetical methylase gene of Agrobacterium tumefaciens and 46% homology to a 2,231-amino-acid Helicobacter pylori hypothetical protein. Protein domain analysis of ORF26 identified a domain (amino acids 185 to 209) involved in transcriptional regulation, DNA repair, or chromatin unwinding and a domain (amino acids 2610 to 2692) characteristic of the conserved C-terminal domain of helicases. These functions are compatible with conjugation activities. Immediately following ORF26 (35544 to 36146) is a gene fragment with 43% homology to cell filamentation proteins induced by cyclic AMP and with homology to a hypothetical Rhizobium sp. gene. Another gene fragment that precedes ORF25 (24114 to 24326) has 48% homology with a transcriptional activating protein, WhiB2, in Mycobacterium and Streptomyces (8). ORF30 is most homologous to a gene (traA) encoding conjugal transfer proteins from a Rhizobium sp. and A. tumefaciens. ORF32 is homologous to topA, a gene encoding a putative DNA type 1 topoisomerase which controls the topology of DNA by transient cutting of one or both strands. Such topoisomerases are common in plasmids and have a role in DNA transfer during conjugation, although their precise function remains obscure (4, 15). ORF32 also has high homology (32%) with the TrsI protein of a plasmid found in L. lactis (12). Other ORFs in the conjugation region are homologous to hypothetical proteins from S. coelicolor (34). ORF34 has extensive homology with a hypothetical protein of S. coelicolor. Since it also has 30% amino acid homology to an S. aureus plasmid transfer complex protein, TrsK, we have designated it trsK. The Gly-Gly endopeptidase activity of ORF36 identified by protein domain analysis may be required for its lysozyme-like activity in hydrolyzing the polyglycine interpeptide bridge of peptidoglycan to assist the conjugation process.

Replication and partition functions.

The remainder of the plasmid (ORF41 to ORF60; 55,914 to 75,827) appears to be devoted to replication and partition (Table 1). It contains the putative ori, a region of two 15-bp direct repeats and a 32-base pyrimidine run preceded by a 9-bp purine run, present within ORF56, which was identified during the development of an R. equi-E. coli plasmid shuttle vector (53). The origin of replication of the R. equi virulence plasmid appears to be a novel type, since with one exception there is no homology with genes involved in replication. The exception is a gene (ORF50) most homologous with the trbA gene of a Rhodococcus sp. plasmid, pSOX (10). The Rhodococcus sp. trbA gene is homologous to the broad-host-range plasmid RK2 trbA gene, which produces a protein which represses both plasmid replication and conjugative transfer. ORFs with homology to parA and parB genes are found in this replication and partition region, although parB is unexpectedly present inside the pathogenicity island. ORF56 was identified by protein domain analysis as a helix-turn-helix regulatory protein. ORF60 has minor (5%) amino acid homology with mycobacterial dnaB, a DnaB-like helicase function supported by protein domain analysis. It seems likely that the other, unknown ORFs in this large region are therefore also concerned with replication and partitioning functions, but further work will be required to confirm this supposition.

Insertion and bacteriophage elements.

ORF53, which has significant homology to the mycobacteriophage TM4 gp82 protein, a protein of unknown function with homology to the putative excisionase protein gp 34.1 of mycobacteriophage D29 (13, 14), is present immediately adjacent to the origin of replication. An area with extensive homology to the mycobacteriophage TM4 gp17 protein (13, 14) occurs from position 51303 to 51671 within ORF36, an ORF homologous to an S. coelicolor putative peptidase and which, as described earlier, protein domain analysis identified as a protease with Gly-Gly endopeptidase activity. The only other bacteriophage element identified is a gene fragment with homology to bacteriophage T7 gene 7.7, present at 38,249 to 38,494. Apart from the pathogenicity island, the relative lack of transposable and bacteriophage elements in this plasmid is in contrast to the markedly mosaic nature of virulence plasmids in Y. pestis (22, 30).

Other elements.

A series of two oligonucleotides of 16 and 21 bp at 76,493 and 76,517 is repeated at 76,975 and 76,999, immediately after ORF61.

Differences between the two plasmids.

The product of ORF32 in p33701 was 825 amino acids in length, whereas in p103 it was 477 amino acids in length; both ORFs had the same start site. The shorter ORF32 in p103 more closely approximated the size of the homologous DNA topoisomerase of Salmonella enterica serovar Typhi. In addition to this difference, there are differences of seven base pair substitutions and six substitutions or deletions of 1 to 4 bp among the plasmids which do not affect the ORFs identified.

Conclusion.

The nucleotide sequences of two virulence plasmids isolated from foal virulent R. equi provide important new information about the development of virulence functions in this plasmid and potential insight into virulence of this organism for foals. It will be important to understand the role of the series of Vap proteins in virulence and the factors which control their expression. The latter include the role of DNA supercoiling and of histone-like molecules, such as H-NS of chromosomal origin, in thermoregulation (2) and pH regulation of the vap genes. Foal virulent isolates contain virulence plasmids which can be classified into five closely related types based on restriction digestion patterns (29, 46). The two plasmids sequenced in this study belong to the 85-kb type 1 group. The basis of the additional DNA present in the 87- and 90-kb types remains to be determined; it may be the result of insertion of mobile genetic elements. Comparison of the virulence region of the plasmid recovered from foal isolates with that of the intermediately virulent, vapB-containing plasmid isolated from pigs and some AIDS patients will be of interest, since this may explain host-adapted virulence.