Structural Organization of Virulence-Associated Plasmids of Yersinia pestis

Abstract

The complete nucleotide sequence and gene organization of the three virulence plasmids from Yersinia pestis KIM5 were determined. Plasmid pPCP1 (9,610 bp) has a GC content of 45.3% and encodes two previously known virulence factors, an associated protein, and a single copy of IS100. Plasmid pCD1 (70,504 bp) has a GC content of 44.8%. It is known to encode a number of essential virulence determinants, regulatory functions, and a multiprotein secretory system comprising the low-calcium response stimulation that is shared with the other two Yersinia species pathogenic for humans (Y. pseudotuberculosis and Y. enterocolitica). A new pseudogene, which occurs as an intact gene in the Y. enterocolitica and Y. pseudotuberculosis-derived analogues, was found in pCD1. It corresponds to that encoding the lipoprotein YlpA. Several intact and partial insertion sequences and/or transposons were also found in pCD1, as well as six putative structural genes with high homology to proteins of unknown function in other yersiniae. The sequences of the genes involved in the replication of pCD1 are highly homologous to those of the cognate plasmids in Y. pseudotuberculosis and Y. enterocolitica, but their localization within the plasmid differs markedly from those of the latter. Plasmid pMT1 (100,984 bp) has a GC content of 50.2%. It possesses two copies of IS100, which are located 25 kb apart and in opposite orientations. Adjacent to one of these IS100 inserts is a partial copy of IS285. A single copy of an IS200-like element (recently named IS1541) was also located in pMT1. In addition to 5 previously described genes, such as murine toxin, capsule antigen, capsule anchoring protein, etc., 30 homologues to genes of several bacterial species were found in this plasmid, and another 44 open reading frames without homology to any known or hypothetical protein in the databases were predicted.

Three species of Yersinia, Y. pestis, Y. enterocolitica, and Y. pseudotuberculosis, have been studied extensively because of their ability to cause disease in both humans and animals. These organisms are closely related at the genetic level, as demonstrated by DNA-DNA homology studies against Y. pestis that showed 83 and 23% homology to Y. pseudotuberculosis and Y. enterocolitica, respectively, under conditions of stringent reassociation (30). Nevertheless, the symptoms of disease caused by the three yersiniae are dramatically different, as are their mechanisms of transmission. Enteropathogenic Y. enterocolitica and Y. pseudotuberculosis are mainly food-borne pathogens causing infection of humans that is typically chronic and characterized by diarrhea, fever, and abdominal pain. On the other hand, Y. pestis causes bubonic plague, an acute lethal disease. Following infection of the dermis by flea bite, this organism disseminates to lymph nodes and then to favorite niches within the viscera, eventually promoting marked septicemia; lung involvement in humans may lead to highly infectious pneumonic plague (5, 37).

The observed major distinctions between chronic and acute disease reflect differences in mechanisms of transmission. The enteropathogenic yersiniae must survive in soil and water and then bypass the host gastrointestinal mucosa following ingestion, whereas Y. pestis remains within the closed and protected environment of its flea vector, thereby ensuring transmission by intradermal injection, a route that requires extensive dissemination to achieve favored visceral niches which support the bulk of replication in vivo (5, 20). The most striking genetic difference between Y. pestis and the enteropathogenic species in this regard is the presence in most but not all (14) strains of the former of two unique plasmids: pMT1 (60 to 110 kb) and pPCP1 (9.6 kb) (2, 12, 15, 25). Although some important virulence factors are encoded on pMT (murine toxin and F1 capsular antigen) (60) and pPCP1 (plasminogen activator), a complete catalog of genes present on these two plasmids is not yet available. The three species also share many additional processes that promote disease, as reflected by carriage of a common plasmid in which are clustered a large number of genes encoding virulence factors such as Yop proteins and the Yop protein secretion system, as well as salient regulatory and anti-inflammatory functions. The generic term “low-calcium response,” or Lcr plasmid, has been applied to this plasmid regardless of its origin; it is specifically termed pCD in Y. pestis, pCad or pIB in Y. pseudotuberculosis, and pYV in Y. enterocolitica (10, 22, 37).

Another major difference between Y. pestis and the enteropathogenic yersiniae is the presence in the former of as many as 30 copies of an insertion element termed IS100 both within the chromosome and on all three plasmids (42, 47). The existence of IS100, as well as additional insertion elements (15, 34, 40), accounts for loss of major chromosomal genes either by direct insertion (53) or by reciprocal recombination resulting in their deletion (13). In this publication, we report the entire nucleotide sequence of the three plasmids from Y. pestis KIM5. These sequences define for the first time a large number of genes homologous to those of several unrelated pathogenic bacteria, the presence of numerous insertion elements and transposons, a large number of open reading frames (ORFs) without known homology, and individual origins of replication. Additionally, we show detailed comparative analysis between the genes encoded by the newly sequenced pCD1 plasmid of Y. pestis and those of the analogous Lcr plasmids of Y. pseudotuberculosis and Y. enterocolitica.

MATERIALS AND METHODS

Bacterial strains and plasmids.

Bacterial strains used in this study are listed in Table 1. The three isogenic Y. pestis strains of KIM5 (6, 16) used to isolate the plasmids were obtained from Robert Brubaker (Michigan State University). The cells were grown in brain heart infusion medium at 28°C. Plasmid pMT1 was isolated from strain KIM5-D46 with a plasmid isolation kit (Qiagen, Santa Clarita, Calif.). Elution was achieved by using a heated buffer according to the manufacturer’s recommendations. Plasmid pCD1 was obtained from strain KIM5-D45 and purified by using a 0.7% agarose gel. Plasmid pPCP1 was isolated from strain KIM5-D1 by CsCl gradient ultracentrifugation.

TABLE 1.

Bacterial strains and plasmids

Strain	Relevant genotype
Y. pestis
KIM5-D1	pPCP1+ pmT1+ pCD1−
KIM5-D45	pPCP1− pMT1+ pCD1+
KIM5-D46	pPCP1− pMT1+ pCD1−
E. coli XL1-Blue	supE44 hsdR17 recA1 endA1 gyrA46 thi relA1 lacF′ [proAB+ lacIqlacZΔM15 Tn10 (Tetr)]

Plasmid subcloning.

Plasmid pPCP1 was subcloned into the BamHI site of pUC18. The resulting recombinant plasmid was subjected to random in vitro transposon bombing with a kit from Perkin-Elmer, Applied Biosystems Division (Foster City, Calif.). A total of 217 clones were sequenced by using dye terminator chemistry and primers SD118 and SD119 (11). M13 libraries were made by shearing purified pMT1 and pCD1 with a nebulizer (3). The instrument used was built at the technical development laboratory of the Center for Genetics in Medicine, Washington University School of Medicine, St. Louis, Mo. The ends of the resultant fragments were repaired with a mixture of T4 DNA polymerase and Klenow fragment as described by Martin-Gallardo et al. (27). Fragments ranging from 1 to 2 kb were ligated into the HincII site of M13mp18. The single-stranded templates for sequencing were isolated by a modified boiling method adapted for a 96-well format (26).

Sequence assembly and gap closure.

A combination of approaches was used to close the gaps between contigs, to obtain sequences from both strands, and to resolve problem regions or compressions. Reads for individual contigs were extended by asymmetrical PCR from end clones, and the PCR products were sequenced directly. Reaction conditions were those described in the AmpliTaq DNA Polymerase Kit from Perkin-Elmer/Roche Biosystems (Branchburg, N.J.). This process quickly joined a number of nearby contigs. A variation of this method involving purification and cloning of the PCR product before sequencing was especially useful for pCD1 because of the scarcity of original plasmid DNA and the high homology found in certain regions shared between pMT1 and pCD1. Regions between contigs were amplified by PCR by using the original plasmid template for pMT1 and M13 clones for pCD1. The PCR products were cloned into pGEM-Easy vector (Promega, Madison, Wisc.). The inserts (0.5 to 1.2 kb) were sequenced by using M13 (−21) and M13 reverse primers or random in vitro transposon bombing as described above. Areas containing compressions and other mobility artifacts were resequenced by using ABI-PRISM dye terminator or ABI BigDye terminator. All sequencing samples were run on ABI PRISM 373 or 377 sequencers.

Base calling and assembly of sequences were performed with the PHRED/PHRAP combination of software (developed by Phil Green, University of Washington). All plasmid sequences were subjected to quality control standards by using a program called Swedish, developed in-house, that automatically calculates error rates and ensures a cumulative error rate of less than 1 in 10,000 bases.

Annotation and analysis of sequences.

Sequences were searched against current protein and nucleotide databases (including those from recently sequenced microbial genomes) by using BLAST (1). Only homology scores of less than 10⁻¹² were considered in assigning homologue status during searches of the protein databases. The plasmid sequences were also analyzed by the GeneMark gene prediction program (4). The ORFs predicted by GeneMark were analyzed by MotifFinder (35) and Block (39), which looked for potential motifs and domains. A total of 169, 1,619, and 2,270 fragments were used to assemble pPCP1, pCD1, and pMT1, respectively. This represents an average redundancy of 7-, 9.2-, and 9-fold, respectively. The sequence was determined on both strands for >95% of each plasmid. Since all three plasmids contained at least one copy of IS100, we defined the start of IS100 as position 1 for each of the three plasmids. Only ORFs encoding peptides of more than 50 amino acids were analyzed.

Nucleotide sequence accession numbers.

The sequence of each plasmid was submitted to the GenBank database under accession no. AF053945 for pPCP1, AF053946 for pCD1, and AF053947 for pMT1.

RESULTS

Analysis of pPCP1.

The total length of plasmid pPCP1 is 9,610 bp. Its GC content is 45.3%. As previously described (47), a single copy of insertion element IS100 was found in this plasmid. Three known genes, the pesticin, pesticin immunity protein, and plasminogen activator genes, were located on the plasmid by BLAST searches (Fig. 1). No additional genes were found or predicted. A region between bp 3,119 and 3,899 was found to have high homology to the origin of replication and the immunity region of the ColE1 plasmid of Escherichia coli. It thus defined the origin of replication on pPCP1.

FIG. 1.

Structural organization of the 9,610-bp plasmid pPCP1 derived from Y. pestis KIM5. BLAST searches using the entire nucleotide sequence obtained in this work were performed to precisely localize potential new ORFs, insertion sequence elements, and the three previously described genes present in pPCP. The directions of transcription of these genes are indicated by the arrows. The single IS100 element was used to define position 1 of this plasmid. The characteristics of the genes and proteins involved are described in the text. The numbering above the line is the molecular size in base pairs.

Analysis of pCD1.

The total length of pCD1 is 70,504 bp, and its GC content is 44.8%. BLAST searches revealed numerous homologues to known virulence genes. Many intact or partial insertion sequences or transposons were found scattered throughout the plasmid, including IS100 and IS285. This phenomenon suggests an earlier transfer or “gathering” of virulence genes among the yersiniae and even among more distantly related organisms, mediated by transposition. Homologues to a large number of proteins previously described in plasmids derived from Y. pseudotuberculosis and Y. enterocolitica were identified by BLAST searches. Among these were Yop proteins, Yop translocation proteins, Yop protein chaperones, V antigen, and other proteins essential for virulence (Fig. 2 and Table 2). Two genes containing premature termination codons (pseudogenes) were found. One of these corresponded to the gene encoding the adhesin YadA, a virulence determinant of the enteropathogenic yersiniae, and the other corresponded to that encoding the lipoprotein YlpA. Both pseudogenes are characterized by frameshifts at the N termini of their putative products (a single deletion of a nucleotide at amino acid 80 in yadA and insertion of a single nucleotide at amino acid 32 in ylpA). Both frameshifts occur at a run of deoxyadenosine nucleotides in the sequence encoding two lysine residues. The ylpA gene is known to be carried by the pYV plasmid of Y. enterocolitica, where it encodes a typical lipoprotein signal peptide (8). The ylpA gene hybridizes with the pYV plasmid of Y. pseudotuberculosis, so that it appears to be conserved among the Yersinia species. ylpA also has significant homology at the protein level with the TraT protein encoded by plasmids pED208, R100, and F (>80% identity for all three) and 77% identity with the protein encoded by the virulence plasmid of Salmonella typhimurium. The yadA gene is known to be nonfunctional in Y. pestis (33, 48, 54); our sequencing results simply confirm this finding.

FIG. 2.

Physical map and genetic organization of pCD1. ORFs, insertion sequences, and other genetic elements were located in the map by using BLAST searches and GeneMark. ORFs and genes in the map are color coded according to function or unique characteristic, and their designations are placed either above or below the colored bars. The scale indicates the number of nucleotides measured from the start of the single IS100 found in this plasmid. Genes in the figures are located precisely in the map and drawn to scale directly from sequence annotation by using an in-house, UNIX-based annotation-rendering program. Genes positioned on top of each line are transcribed from left to right, whereas those placed below the line are encoded by the complementary strand. The two pseudogenes (ylpA and yadA) are represented by partially colored bars. The position of the origin of replication is marked as oriR. The characteristics of the genes, proteins, and sequences depicted are described in the text and in Table 2.

TABLE 2.

Localization and description of ORFs and noncoding elements in pCD1

ORF or noncoding element	Position no.	Size (amino acid residues)	Strand or direction	Description (homologue by BLAST)
ORFs
1	87–1109	340	Direct	Transposase
2	1109–1888	259	Direct	Transposase
3	1939–2343	134	Complement	Transposase
4	2379–2645	88	Complement	lcrS (Y. pseudotuberculosis)
5	3193–3540	115	Complement	lcrQ (Y. pseudotuberculosis) and yscM (Y. enterocolitica)
6	3765–4430	221	Complement	yscL (Y. enterocolitica)
7	4376–5005	209	Complement	yscK (Y. enterocolitica)
8	5005–5739	244	Complement	yscJ (Y. enterocolitica)
9	5746–6093	115	Complement	ycsI (Y. enterocolitica) and lcrO (Y. pseudotuberculosis)
10	6094–6591	165	Complement	yscH (Y. enterocolitica) and lcrP (Y. pseudotuberculosis)
11	6588–6935	115	Complement	yscG (Y. enterocolitica)
12	6937–7200	87	Complement	yscF (Y. enterocolitica)
13	7201–7401	66	Complement	yscE (Y. enterocolitica)
14	7398–8657	419	Complement	yscD (Y. enterocolitica)
15	8654–10477	607	Complement	yscC (Y. enterocolitica)
16a	10483–10896	137	Complement	yscB (Y. enterocolitica)
16b	11121–11220	32	Complement	yscA (Y. enterocolitica)
17	11299–12114	271	Complement	lcrF (virF) transcription factor
18	12238–12633	131	Complement	virG (Y. enterocolitica)
19	13209–14273	354	Complement	yscU (Y. enterocolitica)
20	14273–15058	261	Complement	yscT (Y. pseudotuberculosis)
21	15055–15321	87	Complement	yscS (Y. pseudotuberculosis)
22	15323–15976	217	Complement	yscR (Y. pestis)
23	15973–16896	307	Complement	yscQ (Y. pestis)
24	16893–18260	455	Complement	yscP (Y. pestis)
25	18260–18724	154	Complement	yscO (Y. pestis)
26	18721–20040	439	Complement	yscN (Yop secretion ATPase)
27	20238–21119	293	Direct	yopN (Y. pseudotuberculosis)
28	21100–21378	92	Direct	Y. pseudotuberculosis hypothetical protein
29	21365–21736	123	Direct	Y. pseudotuberculosis and Y. enterocolitica hypothetical protein
30	21733–22101	122	Direct	Y. enterocolitica hypothetical protein
31	22098–22442	114	Direct	Y. enterocolitica hypothetical protein
32	22429–24543	704	Direct	lcrD (Y. pseudotuberculosis and Y. enterocolitica)
33	24540–24980	144	Direct	lcrR (Y. pestis)
34	25022–25309	95	Direct	lcrG (Y. pestis)
35	25311–26291	326	Direct	lcrV (V antigen)
36	26304–26810	168	Direct	lcrH (sycD) (YopB and YopD chaperones)
37	26788–27993	401	Direct	yopB (Y. pseudotuberculosis and Y. enterocolitica)
38	28012–28932	306	Direct	yopD (Y. pseudotuberculosis and Y. enterocolitica)
39	29345–29512	55	Complement	Y. pestis hypothetical protein
40	29778–30038	87	Complement	Y. pseudotuberculosis transposase
41	30873–32102	409	Direct	yopM (Y. pestis)
42	32145–32444	99	Complement	Transposase
43	34860–35828	322	Complement	Y. enterocolitica hypothetical protein
44	36328–36876	182	Direct	yopK (Y. pseudotuberculosis) and yopQ (Y. enterocolitica)
45/46	37360–38110	36	Direct	ylpA pseudogene
47	38624–39016	130	Direct	Transposase
48	40080–41288	402	Direct	sopA (E. coli)
49	41417–42250	277	Direct	sopB (E. coli)
50	44186–44845	219	Complement	yopE (Y. pestis)
51	45039–45431	130	Direct	sycE (YopE chaperone)
52	45494–46123	309	Complement	Transposase
53	46241–47413	390	Direct	Transposase
54	47413–47844	143	Direct	Transposase
55	48188–48613	141	Direct	sycH (YopH chaperone) (Y. pseudotuberculosis and Y. enterocolitica)
56	49594–49860	88	Complement	Transposase
57	50911–51462	183	Complement	Transposon gamma-delta resolvase and tnpR (Y. enterocolitica)
58	51626–53941	771	Direct	Transposase and tnpA (Y. enterocolitica)
59	53938–54318	206	Complement	Transposase
60/61	54924–56227	50	Direct	yadA pseudogene
62	56488–56297	63	Direct	Hypothetical protein (Y. enterocolitica)
63	56928–57344	138	Direct	DNA helicase I (E. coli plasmid F)
64	58681–58929	82	Direct	Endonuclease (Y. enterocolitica)
65	59067–59321	84	Direct	repB (replication protein)
66	59618–60496	292	Direct	repA (replication protein)
67	63100–65298	732	Direct	ypkA (protein kinase) (Y. pseudotuberculosis and Y. enterocolitica)
68	65694–66557	287	Direct	yopJ (Y. enterocolitica)
69	67146–67649	167	Complement	Transposase
70	68243–69649	468	Direct	yopH (protein-tyrosine-phosphatase) (Y. pseudotuberculosis and Y. enterocolitica)
71	70502–70161	113	Complement	Transposase (partial)
Noncoding elements
	1–1954		Forward	IS100
	2655–1961		Reverse	Desulfovibrio vulgaris insertion sequence ISD1
	5716–4973		Reverse	IS285
	29501–30801		Forward	Y. pestis yopM gene, repeat region R1
	32118–34817		Forward	Y. pestis yopM gene, repeat regions R2 and R3
	36952–37179		Forward	Salmonella enteritidis insertion element IS1351 and Yersinia insertion element IS200
	38717–39225		Forward	Shigella dysenteriae insertion sequence IS911
	43032–43399		Forward	Erwinia herbicola IS1327
	46137–47226		Forward	Shigella sonnei insertion sequence IS640
	46137–45463		Reverse	Enterobacter agglomerans nifJ gene and insertion sequence IS2222
	49755–49521		Reverse	Rhizobium insertion element ISR1
	48647–48728		Forward	IS285 (partial)
	47600–47786		Forward	DNA IS100 (partial)
	53976–52230		Reverse	Y. enterocolitica DNA, partial Tn3 homologue
	60005–61324		Forward	E. coli resistance plasmid R100; replication incompatibility, and copy number regions
	70001–70145		Reverse	IS285 (partial)
	70504–70322		Reverse	S. enteritidis insertion element IS1351 and Yersinia insertion element IS200

Other homologues with diverse functions were found in this plasmid, including the genes encoding DNA helicase, DNA resolvase, and DNA replication proteins A and B (sopA and sopB). SopA and SopB function as plasmid partition proteins that ensure the stable and faithful inheritance of the F plasmid (24, 32). Finally, six putative proteins of unknown function that are homologous to those encoded by Lcr plasmids of enteropathogenic yersiniae were found (Table 2).

Analysis of pMT1.

The total length of pMT1 is 100,984 bp, and its GC content is 50.2%. This plasmid contains two copies of IS100 in opposite orientations. The start of one of the IS100 inserts was defined as position 1 of pMT1. The second IS100 copy was found to be located between positions 74592 and 76545. To rule out the possibility that sequences are incorrectly assembled due to the presence of these two identical IS100 elements, an extensive restriction digestion analysis was carried out on this plasmid. One copy of an IS200-like element (also known as IS1541) (34) was found (Table 3 and Fig. 3). Its orientation is the same as that of the second copy of IS100. Two putative transposases were found not far from the latter (positions 79222 to 80430 and 80899 to 81922), indicating the presence of an insertion element or a transposon at this location. BLAST searches showed that IS285 had the closest similarity to these sequences.

TABLE 3.

Localization and description of ORFs and noncoding elements in pMT1

ORF or noncoding element	Position no.	Size (amino acid residues)	Strand or direction	Description (homologue by BLAST)
ORFs
1	87–1109	340	Direct	Transposase
2	1109–1888	259	Direct	Transposase
3	2022–2618	198	Complement	Similar to hypothetical E. coli protein
4	2932–5514	860	Complement	Similar to tail fiber protein gp37
5	5901–6518	205	Complement	Unknown
6	6571–11049	1,492	Complement	Putative protein, similar to lambda host-specific protein J and Northern European squid neurofilament-like protein
7	11188–11739	183	Complement	Unknown
8	11763–12521	252	Complement	Similar to hypothetical Coxiella burnetii protein (motif: RGD cell attachment site)
9	12553–13251	232	Complement	Similar to phage lambda minor tail protein L
10	13341–13676	111	Complement	Unknown
11	13718–18295	1,525	Complement	Putative protein, similar to hypothetical Haemophilus influenzae protein HI1514
12	18653–18970	105	Complement	Unknown
13	19030–19776	248	Complement	Unknown
14	19851–20207	118	Complement	Unknown (Block hit: complement C1Q domain signature)
15	20236–20616	126	Complement	Unknown (Block hit: NSF attachment protein)
16	20700–21044	114	Complement	Unknown (Block hit: geminivirus AR1 coat protein)
17	21142–21975	277	Complement	Unknown (Block hit: histone H5 signature)
18	21975–22274	99	Complement	Unknown
19	22453–22884	143	Complement	Similar to glucan endo-1,3-β-d-glucosidase
20	23190–24065	291	Complement	Unknown
21	24092–24409	105	Complement	Unknown
22	24537–24977	146	Complement	Unknown
23	25015–26253	412	Complement	Unknown
24	26623–27831	402	Complement	Unknown (motif: ATP/GTP binding motif A [P loop])
25	27882–28523	213	Complement	Unknown
26	28719–28985	88	Complement	Unknown (motif: RGD cell attachment site)
27	28995–29885	296	Complement	Unknown
28	30138–30776	212	Complement	Similar to ABC transporter (motif: ATP/GTP binding motif A [P loop])
29	30773–31441	222	Complement	Unknown (motif: aminoacyl-RNA synthetase class II, signature 2)
30	31441–32121	226	Complement	Unknown
31	32285–33763	492	Direct	Unknown
32	33766–34044	92	Direct	Unknown
33	34269–34526	85	Direct	Unknown
34	34531–35058	175	Direct	Similar to S. typhimurium repressor of phase 1 flagellin gene
35	35382–36032	216	Direct	Unknown
36	36982–37464	160	Complement	Unknown
37	37669–37950	93	Complement	Unknown
38	38797–39384	195	Complement	Unknown
39	40132–42015	627	Complement	Unknown (motifs: ABC transporter, aminoacyl-tRNA synthetase class II signature 2, ATP/GTP binding motif A [P loop])
40	42278–43231	317	Complement	Unknown
41	45161–45385	74	Complement	Unknown
42	46550–47614	354	Direct	Similar to E. coli replication protein A (repA)
43	48184–48384	66	Complement	Unknown
44	48396–48656	86	Complement	Unknown
45	49291–49587	98	Complement	Unknown
46	50211–50996	261	Complement	Unknown
47	51336–52412	358	Complement	Similar to recA (motif: ATP/GTP binding motif A [P loop])
48	52681–53550	289	Complement	Similar to Streptococcus pneumoniae DNA Pol Ia
49	53686–54714	342	Complement	Unknown (motif: RGD cell attachment site) (Block hit: T. cruzi P2 protein signature)
50	54834–55265	143	Complement	Unknown
51	55810–56373	187	Direct	Unknown
52	56403–56846	147	Complement	Unknown
53	56843–60250	1135	Complement	Similar to DNA Pol III alpha subunit
54	60548–61783	411	Complement	Similar to CobS protein
55	61881–64235	784	Complement	Similar to CobT protein
56	67074–67588	175	Complement	F1 capsule antigen
57	67669–70170	833	Complement	F1 capsule anchoring protein
58	70195–70971	258	Complement	caf1M
59	71299–72204	301	Direct	caf1R
60	72719–73024	101	Complement	Unknown
61	73545–74582	347	Complement	Similar to DNA ligase
62	74658–75437	259	Complement	Transposase
63	75437–76459	340	Complement	Transposase
64	76861–77118	85	Complement	Unknown
65	77419–78048	209	Complement	Unknown
66	79222–80430	402	Direct	Transposase
67	80899–81921	340	Direct	Transposase
68	82120–82437	105	Complement	Similar to E. coli yhgA
69	82617–82988	123	Complement	Similar to hypothetical protein
70	83297–83752	151	Direct	Similar to E. coli hypothetical protein and reverse transcriptase-like protein
71	83997–85595	532	Complement	Murine toxin
72	87149–87340	63	Direct	Unknown
73	87363–88220	285	Complement	Putative protein, similar to E. coli yhgA
74	90045–91250	401	Direct	Similar to phage P7 parA
75	91247–92218	323	Direct	Similar to phage P7 parB
76	94893–95534	213	Direct	Similar to adenine DNA methyltransferase
77	95766–96185	139	Direct	Unknown (Block hit: NSF attachment site)
78	96239–97018	259	Direct	Unknown
79	97417–97923	168	Direct	Similar to antirestriction protein
80	98635–98865	76	Direct	Similar to E. coli hypothetical protein
81	98938–100947	669	Direct	Similar to Rhizobium meliloti protein and S. sonnei protein
Noncoding elements
	1–1954		Forward	IS100
	38040–38757		Reverse	IS200
	46565–47425		Forward	IncF plasmid RepFIB replicon
	74592–76545		Reverse	IS100

^aPol, polymerase. 

FIG. 3.

Physical map and genetic organization of pMT1. ORFs, genes, and other features displayed in the map are depicted as described in Fig. 2. The characteristics of all of the elements described in the map are defined in the text and in Table 3. The caf1A and caf1 genes located at about 70,000 nucleotides are incorrectly labeled calf1A and calf1, respectively.

Five previously sequenced genes were located: F1 capsular antigen, F1 capsule anchoring protein, its chaperone (Caf1M), the regulatory protein (Caf1R), and murine toxin. These genes are clustered in the region between positions 67669 and 85595 (Fig. 3). Twenty-six homologues (excluding transposases in transposons and insertion elements) were found by BLAST searches (Table 3). Three homologues are similar to the E. coli tail fiber protein and the lambda phage host-specific protein. Homology searches also revealed the presence of two possible new operons. One contains homologues to the phage p7 parA and parB genes, which are involved in plasmid partition (19). The other operon contains homologues to the genes cobS and cobT in Pseudomonas denitrificans. cobS and cobT were isolated as an independent cluster of the cobalamin biosynthetic genes, likely to be involved in cobalt insertion-mediating reactions and the transformation of precorrin-3 (7). The function of the Y. pestis homologue is not known.

An additional 44 ORFs were predicted by GeneMark. These ORFs have no homology with any known or hypothetical proteins currently in the databases. MotifFinder found 14 types of motifs (including transposase) in the PROSITE database. Several interesting motifs, including an ATP/GTP binding site (P loop), cell attachment site (RGD), ABC (ATP binding cassette) transporter, and sigma 54 interaction domain, were observed. Extensive experience with these searches suggests that, due to the problems inherent in the search algorithm, predicted phosphorylation, glycosylation, amidation, and myristoylation sites do not tend to have significant biological relevance; thus, such findings are not discussed here. However, a number of motifs identified by Block searches with potentially interesting and relevant biological functions included the Trypanosoma cruzi P2 protein signature, complement C1q domain signature, N-ethylmaleimide-sensitive factor (NSF) attachment site, and Rho family GDP dissociation inhibitor signature.

DISCUSSION

Plasmid pPCP1.

All structural genes identified during the sequencing of plasmid pPCP1 have been described previously (47, 55). The organization of the genes encoded in this plasmid was the same as that previously reported (Fig. 1). At the protein level, pesticin and pesticin immunity protein were found to be identical to those described in the databases. The predicted sequence of the plasminogen activator was identical to that described by McDonough and Falkow (28), even though their sequences were obtained from a different strain of Y. pestis (EV76). Two putative transposases were found within IS100. Their genes are transcribed in the same direction as the pesticin immunity protein and the plasminogen activator genes, while the pesticin gene is transcribed in the opposite direction. We did not find new ORFs larger than 50 amino acids in this plasmid. Replication of the plasmid is controlled by a mechanism highly homologous to the ColE1 replicon of E. coli. This is consistent with the fact that yersiniae and E. coli are taxonomically related.

Plasmid pCD1.

Plasmid pCD1 mediates the low-calcium response. Salient genes include those encoding the Yop proteins and their chaperones, secretory mediators, and regulatory genes. The Lcr plasmids are essential for virulence in all three species of Yersinia pathogenic for humans. The laboratories of Susan C. Straley, Hans Wolf-Watz, and Guy R. Cornelis were instrumental in defining the structures and functions of the Lcr plasmids in Y. pestis, Y. pseudotuberculosis, and Y. enterocolitica, respectively (10, 17, 21, 43, 44, 56, 57). As expected, all functional homologues to Yops and their related proteins were found on pCD1. The organization of the operons is highly conserved among the Lcr plasmids of the three pathogenic yersiniae (22, 37). However, a number of global structural differences can readily be discerned from the completed sequence of the Y. pestis Lcr plasmid. For example, in spite of the fact that the origins of replication in pCD1 and the corresponding pYV of Y. enterocolitica lie between the yadA and pkA genes in both plasmids, they map in entirely different positions within the plasmid (with respect, for example, to the yopBD and the yopM genes). YlpA maps near the origin of replication in the Y. enterocolitica plasmid but some 20 kb away in pCD1. ylpA, lcrVGRD, yopM, yopD, and a series of other homologous genes are transcribed in pCD1 in orientations opposite those in the Y. enterocolitica pYV plasmid. It is clear that in spite of the high degree of functional conservation observed among the virulence genes of all three plasmids, a number of rearrangements and internal translocations have taken place as the plasmids have proceeded to diverge and evolve (Fig. 4). It is interesting that more than 10 partial insertion sequences and other sequences homologous to those of diverse eubacteria (Salmonella, Erwinia, Rhizobium, and Desulfovibrio spp.) are scattered throughout the plasmid. This high proportion of insertions and mosaic sequences opens up the possibility that during evolution these virulence-associated genes were gathered from a diverse bacterial assemblage through transposition.

FIG. 4.

Diagram comparing the organization of selected genes and elements of the Lcr plasmid in yersiniae. Shown are circular maps of the pCD1 plasmid and of the homologous pYV and pCad plasmids derived from Y. enterocolitica and Y. pseudotuberculosis, respectively. The relative positions of selected loci with respect to the origin of replication of pCD1 are shown. Outer circle, pCD1; middle circle, pIB1 (Y. pseudotuberculosis); inner circle, pYVe O:9 (Y. enterocolitica). The nomenclature and approximate positions of genes in pYV and pIB1 are from Iriarte and Cornelis (22), Persson et al. (38), and Salyers and Whitt (51). The genes and sequence features of pCD1 and the corresponding regions in pYV and pIB1 are depicted in the same color to aid in their visualization (e.g., the repBA, oriR, and ypkA regions are presented in green, black, and red, respectively). Numbering inside the circles indicates the approximate sizes of the plasmids in nucleotides, measured from the start of their origins of replication. Arrows above each color segment representing a gene or gene group point to the direction of transcription.

A notable finding was the identification of a new pseudogene in pCD1 and the confirmation of a frameshift mutation in yadA. The genes yadA and ylpA, which are fully functional in Y. enterocolitica (37), have frameshift mutations that create premature termination of transcription in pCD1. Since yadA encodes an adhesin protein involved in attachment to epithelial cells, inactivation of this gene should represent no essential loss of function for Y. pestis pathogenesis. YlpA, on the other hand, is homologous to the TraT protein, encoded by the virulence plasmid of Salmonella (8). Since TraT is involved in serum resistance in Salmonella, YlpA is also likely to be involved in serum resistance in Y. enterocolitica. In Y. pestis, however, resistance to serum occurs independently of pCD1 as an evident function of lipopolysaccharide structure (41). It is nevertheless possible that a potential truncated protein could be translated from a downstream start codon in ylpA, leading to the production of a YlpA protein lacking the first 45 amino acids. Biochemical evidence, however, is not presently available to ascertain whether YlpA is absent or whether a truncated version of this protein is still expressed in Y. pestis. It is interesting that both of the frameshifts in these genes are caused by either deletion or insertion of a single deoxyadenosine nucleotide within a run of seven to eight deoxyadenosine nucleotides. Such stretches of redundancy are known to be hot spots for mutations and could be responsible for this phenomenon (49).

Although Y. pestis and Y. pseudotuberculosis are generally thought to be the most closely related species, we found that YopJ in pCD1 had higher homology to the Y. enterocolitica homologue (called YopP in this organism) than to its YopJ counterpart in Y. pseudotuberculosis. However, further inspection indicated that the YopJ protein from Y. pestis and its YopJ counterpart in Y. pseudotuberculosis are 99% identical (with a single amino acid difference) in the first 241 residues. After residue 241, the amino acid sequences differ markedly, due to an apparent change in the reading frame in the previously described Y. pseudotuberculosis yopJ gene. Our finding is entirely consistent with that reported by Mills et al. (29) during their studies of the Y. enterocolitica YopP. The apparent frameshift in Y. pseudotuberculosis has recently been shown to be a sequencing mistake (36). Since YopJ was first discovered in Y. pestis KIM by Straley and Bowmer (58), we have retained this terminology in pCD1.

In summary, most regions comprising the genetic material of plasmid pCD1 were identified as homologues of known or hypothetical proteins, or as occupied by insertion elements or transposons. Six putative proteins were found to have homologues in the databases, but their functions are unknown. Plasmid pCD1 contains very few large intergenic regions; its coding ratio is approximately 1 ORF per kb.

Plasmid pMT1.

The third plasmid and the largest, pMT1, is also the least studied of the Y. pestis plasmids. Although it was initially considered a cryptic plasmid, subsequent studies localized five important genes on this plasmid, encoding F1 capsular antigen, F1 capsule anchoring protein, Caf1M, Caf1R, and the plague murine toxin (45, 46). These genes are clustered in a region spanning approximately 18 kb of the entire plasmid. Interestingly, the DNA encoding this cluster of genes has a GC ratio of 45.8%, compared with 51.1% for the remainder of the plasmid. Such regions of atypical base composition have been found in several gram-negative and gram-positive organisms to be associated with what has been termed pathogenicity islands (18). These genetic elements, which cumulatively participate in pathogenicity, are likely acquired by genetic transfer among bacterial pathogens and sometimes contribute to differences in host specificity, tissue tropism, and disease manifestation (9). It is thus intriguing to conjecture if a plasmid such as pMT1, which readily integrates into the bacterial chromosome, may have arisen by a mechanism involving such a genetic mechanism.

Several ORFs (ORFs 34, 9, 4, and 6) were found to encode proteins that resemble E. coli flagellin or phage host-specific proteins (Table 3). Whether these proteins participate or aid in the actual biogenesis of pili or are involved in pathogenesis in Y. pestis is unknown. Interestingly, one putative protein is homologous to both phage p7 ParB and Shigella flexneri VirB protein or Shigella dysenteriae IpaR. VirB was implicated as a transcriptional activator of several invasion genes, and IpaR was found to induce apoptosis of macrophages (61). It is thus tempting to speculate that this gene may act as a virB or ipaA homologue. However, based on the following two observations, we suspect that this protein may instead function as a plasmid partition protein. First, although there is clear homology between this putative protein and VirB, BLAST searches showed a higher degree of homology with ParB. Second, whereas there are no other potential ipa genes in this plasmid, the next ORF upstream from the parB homologue is highly homologous to phage p7 parA. Thus, the region seems to constitute a parA parB operon containing the upstream element important for parA autoregulation and the parS site important for parB function. Thus, it is likely that ParA and ParB are fully functional as plasmid partition proteins. Plasmid pCD1, on the other hand, has at least two different and distinct proteins dedicated to plasmid partition which are homologous to the E. coli F plasmid-associated genes sopA and sopB (31). Thus, whereas in pMT1 the partition apparatus appears to resemble that of the p7 phage more closely, in pCD1 this apparatus is more akin to that of the F family of plasmids. The two plasmids have to use different partition systems in order to maintain the faithful inheritance of low-copy-number plasmids. Otherwise, the segregation of the plasmids would come disastrously close to random distribution.

One putative protein (ORF 19) is homologous to Bacillus circulans glucan endo-1,3-β-d-glucosidase (Table 3). Further work may show that it mediates an interaction between the organism and some polysaccharide moiety on the host cell surface. ORFs 18, 26, and 49 were found to contain an RGD (arginine-glycine-aspartate) cell attachment site. This protein sequence is a characteristic eukaryotic recognition motif which binds to cell surface integrins (50). It also has been found in an array of bacterial virulence factors, such as the Bordetella pertussis adherence factor filamentous hemagglutinin (FHA), pertactin, pertussis toxin, and BrkA. Studies have shown that the RGD sequence of FHA mimics that of the host cell (52). It binds to host cell CD11b/CD18, which mediates the uptake of the bacteria into macrophages without triggering an oxidative burst, thus protecting the bacteria (23). Although not all proteins containing RGD are involved in cell attachment, those containing properly presented RGD sequences have a strong potential for binding to host cell integrin or extracellular matrices. These proteins could thus be important candidates in adherence or in resistance to phagocytosis in Y. pestis. ORF 39 was found by MotifFinder to have an ABC transporter signature. ABC transporter superfamily members are found in both prokaryotes and eukaryotes, where they are involved in drug resistance and in the transport of substrates ranging from ions to large proteins (59). Thus, this ORF could have a role in the transport of substrates. In addition to the potential ORFs discussed above, some 40% of the ORFs predicted by GeneMark on plasmid pMT1 had neither homologous counterparts in the publicly available databases nor any manifestation of motif-associated features. Hence, we are presently unable to predict or speculate on the possible functions of these genes or if, in fact, any of these ORFs are translated into functional proteins. This issue is further complicated by the fact that pMT can integrate into the chromosome (46) and thus may contain copies of chromosomal genes required for normal vegetative functions.

The availability of the entire nucleotide sequence of these three plasmids should enable the global study of the gene complement encoded in them, as well as of the mechanism of expression that underlies their regulation. Such studies should help to elucidate the functions of presently unknown genes and should provide insight into the interplay of those virulence factors which are common to the three human pathogens and those that are unique to Y. pestis.