Complete Sequence of pOZ176, a 500-Kilobase IncP-2 Plasmid Encoding IMP-9-Mediated Carbapenem Resistance, from Outbreak Isolate Pseudomonas aeruginosa 96

Abstract

Pseudomonas aeruginosa 96 (PA96) was isolated during a multicenter surveillance study in Guangzhou, China, in 2000. Whole-genome sequencing of this outbreak strain facilitated analysis of its IncP-2 carbapenem-resistant plasmid, pOZ176. The plasmid had a length of 500,839 bp and an average percent G+C content of 57%. Of the 618 predicted open reading frames, 65% encode hypothetical proteins. The pOZ176 backbone is not closely related to any plasmids thus far sequenced, but some similarity to pQBR103 of Pseudomonas fluorescens SBW25 was observed. Two multiresistant class 1 integrons and several insertion sequences were identified. The bla_IMP-9-carrying integron contained aacA4→bla_IMP-9→aacA4, flanked upstream by Tn21 tnpMRA and downstream by a complete tni operon of Tn402 and a mer module, named Tn6016. The second integron carried aacA4→catB8a→bla_OXA-10 and was flanked by Tn1403-like tnpRA and a sul1-type 3′ conserved sequence (3′-CS), named Tn6217. Other features include three resistance genes similar to those of Tn5, a tellurite resistance operon, and two pil operons. The replication and maintenance systems exhibit similarity to a genomic island of Ralstonia solanacearum GM1000. Codon usage analysis suggests the recent acquisition of bla_IMP-9. The origins of the integrons on pOZ176 indicated separate horizontal gene transfer events driven by antibiotic selection. The novel mosaic structure of pOZ176 suggests that it is derived from environmental bacteria.

INTRODUCTION

Pseudomonas aeruginosa is a medically important species that is intrinsically resistant to many antibiotics. Many resistance genes that are carried by R plasmids in P. aeruginosa have evolved with the clinical use of different classes of antibiotics (1). In the late 1980s, before the widespread use of third-generation cephalosporins, the common β-lactamases TEM-1/2 and PSE-2 were found to be encoded by R plasmids (2–5). In the late 1990s and early 2000s, shortly after the introduction of carbapenems into clinics, metallo-β-lactamase IMP (6–9) and VIM (10, 11) types emerged.

The first imipenem-resistant P. aeruginosa isolate (GN17203) carrying a transferable 47-kb plasmid (pMS350) encoding the bla_IMP-1 gene was isolated in 1988 and subsequently found to belong to the IncP-9 group (12). The bla_IMP-1 gene cassettes first emerged in class 3 integrons and are now more frequently found in sul1-type class 1 integrons with a 3′-CS that includes qacEΔ1 and sul1 (7). Most IMP, VIM, and GIM carbapenemase genes are associated with sul1-type class I integrons, but recently, carbapenemases have been found to occur on Tn402-type class 1 integrons with an intact tniRQBA module. Examples include VIM-1 in P. aeruginosa (13), VIM-2 in P. aeruginosa (14, 15) and in Pseudomonas putida (16), and IMP-8 in Klebsiella oxytoca (17), among others.

Plasmids found in Pseudomonas species are classified by incompatibility group (18). Compatible plasmids coexist stably in the same host, whereas incompatible plasmids either eliminate one another or recombine to form cointegrates (19). Thirteen incompatibility groups have been recognized in P. aeruginosa. Each group often possesses characteristic phenotypic properties (1). R plasmids of the P-1 group are promiscuous plasmids and have a broad host range, as reflected in their ability to be transferred into Escherichia coli from P. aeruginosa. Plasmids of Inc groups P-2, P-5, P-7, P-10, P-12, and P-13 have not been transferred to E. coli from P. aeruginosa by conjugation and are defined as having a narrow host range. So far, only a few IncP plasmids, such as IncP-1 (20, 21), IncP-6 (22), IncP-7 (23, 24), and IncP-9 (25), have been fully sequenced. IncP-2 plasmids are considered to be ubiquitous in the environment and universally resistant to tellurite (26). They are usually very large and single copy and are difficult to isolate due to DNA shearing from manipulation. They are among the most common plasmids encountered in clinical isolates of P. aeruginosa (1, 26). Thus far, no sequencing data are publicly available for plasmids of several of these incompatibility groups.

The multiresistant P. aeruginosa 96 (PA96) strain was isolated from a hospital in Guangzhou, China, during a multicenter surveillance study in 2000. The single bla_IMP-9-carrying plasmid in PA96 was characterized as an IncP-2 plasmid with an estimated size of 450 kb and named pOZ176 (9). The aim of this study was to use whole-genome sequencing of PA96 in order to understand the genetic context and resistance mechanisms of pOZ176, which encodes carbapenem resistance and has two class 1 integrons.

MATERIALS AND METHODS

Bacterial strain and plasmid.

The PA96 isolate employed in this study was a clinical specimen isolated from the sputum of an intensive care unit (ICU) patient who was suffering from a lower respiratory tract infection as a complication of posttraumatic neurosurgery. This isolate was part of a collection obtained during a multicenter surveillance study in Guangzhou, China, in 2000. Susceptibility testing of this isolate was initially performed in China in 2000 by using disk diffusion methods. Testing was repeated in 2005 by using the Etest (AB Biodisk, Solna, Sweden) methodology (Table 1). The preliminary results of plasmid isolation, conjugation, and incompatibility testing of this organism were reported previously (9). The plasmid, named pOZ176, in PA96 was very large (450 kb) and characterized as an IncP-2 plasmid by phenotypic methods (9). Given the difficulty of isolating the plasmid, we decided to sequence the whole PA96 genome, including the plasmid.

Table 1.

Susceptibilities of carbapenem-resistant PA96, its transconjugant 96T, and recipient strain P. aeruginosa NCTC 50814^a

Strain	MIC (mg/liter)
	IPM	MEM	CAZ	CTX	CRO	FEP	ATM	TIM	CPS	TZP	CIP	AMK	GEN
PA96	>32	>32*	256	256	256	256	4*	256	256	256*	0.50	16	256
96T	>32	>32*	256	ND	ND	256	4	ND	ND	32*	0.125	0.50	256
P. aeruginosa NCTC 50814b	0.50	0.125	2	ND	ND	0.50	0.50	ND	ND	0.5	0.0156	0.25	ND

^aThe tests were done by using the Etest gradient method. The values marked with asterisks indicate that there were discrepancies of one to two 2-fold dilutions between Etest results and the results of an agar dilution method described previously (9). IPM, imipenem; MEM, methicillin; CAZ, ceftazidime; CTX, cefotaxime; CRO, ceftriaxone; FEP, cefepime; ATM, aztreonam; TIM, ticarcillin-clavulanic acid; CPS, cefoperazone-sulbactam; TZP, piperacillin-tazobactam; CIP, ciprofloxacin; AMK, amikacin; GEN, gentamicin; ND, not done.

^bThe recipient P. aeruginosa NCTC 50814 (met his lys Rif^r).

Genome sequencing methods.

Total DNA was isolated from a culture of PA96 grown overnight in LB broth at 37°C by using a commercial DNA extraction kit (Qiagen, Toronto, Canada) and quantified by using a fluorometric, picogreen-based method (Qubit dsDNA BR assay; Life Technologies, Burlington, Canada). A library suitable for shotgun sequencing with the Roche GS-FLX/454 system (Roche, Basel, Switzerland) was prepared from 500 ng of total DNA. Pyrosequencing was performed by using Titanium chemistry and 2 of 4 slots of a picotiter plate. Sequencing generated a total of 430,000 reads and 179 million base pairs of data. De novo assembly with gsAssembler (Roche) generated 66 contigs ranging in size from 111 bp to 921 kb. Mauve (27)-based alignment of contigs revealed extensive synteny between PA96 sequences and available P. aeruginosa reference genomes of PAO1, PA14, LESB58, and PA7. However, five large contigs (26, 51, 58, 66, and 277 kb) did not align with chromosomal sequences. Notably, the average sequencing depth for these putative extrachromosomal contigs (22-fold) was slightly less than that observed for other contigs of similar sizes (28- to 30-fold).

Plasmid assembly and finishing methods.

Initial analysis of putative plasmid sequences indicated that the 5 contigs terminated in repetitive sequences that could not be resolved by gsAssembler. However, reanalysis resulted in assembly of three contigs (12,149, 210,023, and 277,894 bp) separated by three identical copies of a 1,156-bp, IS3-like element. Gap-closing strategies included designing primers in unique sequences, PCR amplification and sequencing to determine orientations, and confirming that the extrachromosomal contigs comprise a single giant plasmid. Tablet version 1.11.11.01 (28) was used to visualize the contigs and their reads to confirm the reliability and accuracy of the reads. Initial automatic gene calling and functional annotation were performed with BASys (29) and RAST (http://rast.nmpdr.org/) (30), and further editing and manual annotation were carried out by using Prodigal (31), GCG (version 11.1; Accelrys Inc., San Diego, CA), and Artemis (32). Additional software, including IS Finder (http://www-is.biotoul.fr/is.html) and various Softberry programs (http://linux1.softberry.com/berry.phtml), was used for analysis of specific plasmid genetic features, such as insertion sequence (IS) elements and pathogenicity islands. Further comparative analysis of the DNA sequence was performed by using ACT (32) and Mauve (27). CGview (33) was used for generation and visualization of the pOZ176 plasmid map. The codon frequency file of 5,596 genes of P. aeruginosa PAO1 (http://www.kazusa.or.jp/codon/cgi-bin/showcodon.cgi?species=208964) was used for correspondence analysis with the codon frequency files of antibiotic resistance genes and other genes of mobile elements. Evolutionary genetic analysis and phylogenetic tree construction for interesting genes/coding sequences (CDSs) were performed by using MEGA5 (34).

Nucleotide sequence accession numbers.

The complete sequence of pOZ176 has been submitted to GenBank under accession number KC543497. The integrons were previously submitted to GenBank under accession numbers AY033653 (Tn6016) and EU886981 (Tn6217).

RESULTS

General features of the pOZ176 sequence.

Plasmid pOZ176 has a length of 500,839 bp and contains 618 predicted coding regions with an average G+C content of 57%. Sixty-five percent of the open reading frames (ORFs) encode hypothetical proteins, as identified by RAST and later confirmed by manual annotation. Key features of the plasmid, such as the replication, stability, and transfer systems, show only weak similarity to other sequenced plasmids in GenBank. Many of the hypothetical proteins of pOZ176 exhibit similarity to open reading frames from plasmid pQBR103 (35) of the environmental bacterium Pseudomonas fluorescens SBW25 and/or to a putative genomic island of Pseudomonas syringae pv. maculicola strain ES4326, particularly in two blocks of the plasmid sequence (bp 65519 to 208437 and 358420 to 388001). The plasmid map of pOZ176 shows the genes and their locations (Fig. 1). The IncP-2 maintenance, replication, and transfer modules of this plasmid are indicated by a blue arc in Fig. 1. The plasmid includes two class 1 multiresistance integrons derived from different lineages and inserted in opposite orientations into pOZ176. Other interesting features of pOZ176 include a cluster of resistance genes similar to the center of Tn5, a tellurite resistance operon, two pil operons, and several mobile elements.

Fig 1.

The genome of pOZ176. The scale is indicated on the innermost circle. The second circle illustrates G+C skew in green (+) and purple (−), and circle 3 shows G+C content (deviation from the average) in light blue (+) and light green (−). The next two circles indicate positions of CDSs in minus (circle 4) and plus (circle 5) strands. The outermost circles indicate the positions of mobile elements, including the integron-containing Tn6016 and Tn6217, the IncP-2 tra-parBA-oriV-repA region, heavy metal resistance, two pil operons, and the two principal regions of weak similarity to pQBR103. CGview software was used to construct the genome map.

Antibiotic and heavy metal resistance genes and their mobile elements.

There are two class 1 multiresistance integrons from different lineages. The Tn402-like integron was named Tn6016 (bp 459885 to 451065; 8.8 kb). This bla_IMP-9-carrying class 1 integron contained three cassettes, aacA4→bla_IMP-9→aacA4, encoding resistance to aminoglycosides and carbapenems. Upstream of the cassettes were intI1 and inverted repeat IRi flanked by tnpMRA of transposon Tn21. Downstream of the cassettes, there was an almost complete tni operon of Tn402. The IRt end and the beginning of the tniA gene were truncated by the insertion of a Tn5044-like mercury resistance operon containing the merRTPFA genes. Its genes had more or less similarity to genes of this transposon. A second, partial mercury resistance element (merRTP) and a nearby transposase are present between the two integrons. This region is 99.9% identical to Tn5563 (deleted of its right end and resolvase) from pRA2 (36) and is also found in pUM505 (37) and a genomic island of P. aeruginosa PA7 (38).

The sul1-type integron, named Tn6217 (bp 403521 to 409381; 6.9 kb), carried three cassettes, aacA4→catB8a→bla_OXA-10, conferring resistance to aminoglycosides, chloramphenicol, and carbenicillin, respectively. Upstream of the cassettes were intI1 and IRi flanked by a Tn1403-like transposon having a tnpA that was 99% identical in amino acid sequence to that of Tn1403. Downstream of the cassettes, there was a sul1-type 3′-CS and an IS6100 transposase but no tni genes or an IRt end (Fig. 2). The 3′-CS is similar to that of Tn1403 (39), except that the sul1-orf5 region is intact and Tn5393 is absent. The three aacA4 genes in the two integrons were identical in nucleotide sequence.

Fig 2.

Two types of class 1 integrons identified in PA96, a Tn402-type integron in Tn6016 (A) and a sul1-type integron in Tn6217 (B). Arrow boxes show the genes and their orientations; each solid black oval indicates attI, and each open oval represents the attC of the preceding gene. tniAΔ represents a truncation: 87 bp at the beginning of the gene were disrupted by a mer operon, which is ∼90% identical to that of Tn5044. The small rectangular boxes represent the res site of the transposon, and the solid black boxes represent the 25-bp IRi and IRt sites. The IRt site of the integron in Tn6016 is missing; the small arrows represent the direction of promoters. HP, hypothetical protein; Drug^R, drug resistance.

Tn5-like resistance gene cluster.

An operon of three aminoglycoside resistance genes is found in pOZ176: an aminoglycoside phosphotransferase (neo, conferring resistance to neomycin) (bp 247073 to 246279; 264 amino acids [aa]), a ble bleomycin resistance gene (bp 246258 to 245878; 126 aa), and an str C-terminally truncated streptomycin 3′-phosphotransferase gene (bp 245839 to 245630). This operon is 94% similar to the center of Tn5 found in a genomic island of P. aeruginosa PA7 (38). The latter is identical to the center of Tn5 itself, and neither has any trace of the IS50 elements found at the ends of Tn5. A second putative bleomycin resistance gene is located 29 kb away (bp 275795 to 276220; 141 aa), and its product has 71% amino acid identity to a resistance protein with a glyoxalase/dioxygenase domain in Ralstonia eutropha H16.

Tellurite resistance operon.

The ter operon (bp 45490 to 55702) consists of terZ (bp 55702 to 55109; 197 aa), terA (bp 55112 to 53922; 396 aa), terB (bp 53874 to 53425; 149 aa), terC (integral membrane protein) (bp 53413 to 52379; 344 aa), terD (bp 52350 to 51772; 192 aa), terE (bp 51738 to 51160; 192 aa), and two other putative tellurite resistance genes (bp 45631 to 46134 [167 aa] and bp 46343 to 47602 [419 aa]), the former of which is C-terminally truncated. The operon is most similar to the ter operon in P. aeruginosa strain PACS171b clone fa1389 (40) but has three intervening ORFs between the first six and the last two genes. Another putative tellurite resistance protein (bp 33633 to 32383; 416 aa), which is located several kilobases away, has 90% amino acid identity to a protein from PACS171b.

Insertion sequences.

There are three copies of an IS3 family element [bp 276932 to 278163, bp 486230 to 487461, and base 1 plus bp 500839 to 499609 on the complementary strand] that are identical to each other and 98% identical to an IS3-like element in Pseudomonas putida S16. IS3 family elements encode OrfA and OrfB and can express a “complete” transposase (OrfAB) by translational frameshifting (41). There are another four copies of this IS3-like element in the chromosomal DNA of PA96. A different IS3 family element, in the 419-kb region, is 88% identical to an IS3-like element in the chromosome of Pseudomonas fluorescens SBW25. An ISPst3-like element is located at bp 192519 to 195123 and is 97% identical to an IS of a genomic island of P. aeruginosa PA7. A Tn3 family transposon composed of an ISPa40-like transposase and resolvase, followed by a methyl-accepting chemotaxis protein and a second, ISPa1328-like transposase, is found at bp 490984 to 497939 and is 99.9% identical to a mobile element from the chromosome of Pseudomonas stutzeri CCUG 29243 (42). There are three transposase pseudogenes, two IS5-like and one Tn1721-like.

Replication, partitioning, and transfer of pOZ176.

The plasmid replication initiator protein RepA (bp 252731 to 253588) is 285 aa in length. Members of this family of bacterial proteins are single-stranded DNA binding proteins involved in DNA replication, repair, and recombination. Highly similar RepA proteins are found in putative genomic islands of Azotobacter, Burkholderia, Stenotrophomonas, and Xanthomonas. These may be integrated replication origin regions of IncP-2 plasmids. A phylogenetic tree of the RepA protein and its closest homologs is shown in Fig. 3. Other close homologs include RepAs of genomic islands of P. aeruginosa PACS171b, 39016, PA2192, and PA7 (38, 40, 43, 44).

Fig 3.

Evolutionary relationships of the pOZ176 replication initiator protein (RepA) and its 12 best hits in GenBank. The inclusion criteria for the 12 best hits were an E value of 0 and percent amino acid identity of 99% to 91%. The evolutionary history was inferred by using the maximum likelihood method with MEGA5 software. The consensus tree after 1,000 bootstrap iterations is shown. The evolutionary distances were computed by using the JTT matrix-based method.

The proposed origin of vegetative replication, oriV, of pOZ176 (bp 253613 to 253676) was located just downstream of RepA; it was characterized by 3 copies of a 17-bp direct repeat with the sequence TCGTGCTATCAGGAGTA (iterons). The three copies are in the sequence TCGTGCTATCAGGAGTA(N7)TCGTGCTATCAGGAGTA(N6)TCGTGCTATCAGGAGTA. This 64-bp sequence is identical to one found in putative genomic islands of Azotobacter vinelandii DJ and Burkholderia ambifaria AMMD and highly similar to several others. It is 94% (60/64 bp) identical to a sequence of plasmid pBB1 of Cupriavidus necator. The partitioning protein ParA (bp 253830 to 254468; 212 aa) is 99% identical to its equivalent protein in Acidovorax sp. strain JS42. The putative partitioning protein ParB (bp 254465 to 254728; 87 aa) is 90% identical to a hypothetical protein of Burkholderia gladioli BSR3.

The adjacent conjugal transfer operon is about 17 kb and includes traF and virD2 (bp 254729 to 258231) and traG and trbBCDEJLFGI (bp 260752 to 274492). These genes have the highest similarity to putative genomic islands of Ralstonia solanacearum GMI1000 (45) and A. vinelandii DJ (46) and are also homologous to genomic islands of several strains of P. aeruginosa and to plasmid pBB1 of C. necator N-1. The operon is interrupted by genes encoding a major facilitator family transporter and its regulator, which have a distinct origin.

A second putative RepA protein (bp 179399 to 180586; 395 aa) was found to be located in another region of pOZ176. It is similar (54 to 55% amino acid identity) to putative RepA proteins of pQBR103 (an IncA/C plasmid of P. fluorescens SBW25) (35) and to a putative genomic island of P. syringae pv. maculicola strain ES4326. A second partitioning system of pOZ176 includes the parB gene product (bp 141885 to 143084; 399 aa), the parA gene product (bp 143084 to 143983; 299 aa), a hypothetical protein (bp 144158 to 145600; 510 aa), and another parB gene product (bp 145720 to 146580; 331 aa). These proteins are 36 to 48% identical to the partition proteins of pQBR103. Still another ParB family partitioning protein (bp 172006 to 173838; 610 aa) is 54 to 55% identical to distinct ParB family proteins in pQBR103 and P. syringae.

Pil operons.

Two pil operons have been identified in pOZ176. The first of the two (bp 109483 to 120876) is also found in pQBR103 and is a putative one due to the weak similarities for each protein in the operon. The second one (bp 147425 to 157795) very probably encodes a type IV pilus (T4aP), which modulates twitching motility by a mechanism of extension-retraction. The operon would be involved in pilus assembly similarly to the Pil-Chp system (47) and to other chemotaxis operons (Che). The total length of the operon is about 10.6 kb and encodes six proteins: the response regulator receiver protein CheY/PilG (bp 147425 to 147811; 128 aa); the chemotaxis signal transduction protein CheW/PilI (bp 147801 to 148289; 162 aa); a methyl-accepting chemotaxis sensory transducer protein, PilJ (bp 148301 to 150310, 669 aa); a putative CheR/PilK chemotaxis signaling protein (methyltransferase) (bp 150307 to 151098; 263 aa); a CheA/ChpA (PilL) signal transduction histidine kinase fusion protein (bp 151101 to 156866; 1,921 aa); and a CheB/ChpB protein-glutamate methylesterase (bp 156863 to 157795; 310 aa). The six proteins are most similar to those encoded by the same operon of pQBR103, with 42 to 60% amino acid identities.

DISCUSSION

Plasmid pOZ176 is the largest plasmid ever sequenced, and the first IncP-2 plasmid that has been completely sequenced, from P. aeruginosa. The 500,839-bp pOZ176 has little resemblance to other sequenced plasmids, and the percentage of genes encoding hypothetical proteins is high (65%). There are two blocks of weak homology to pQBR103 from P. fluorescens SBW25 (35). P. fluorescens is a common soil bacterium abundant on the surfaces of plant roots and leaves. The SBW25 strain was first isolated in 1989 from the leaf surface of a sugar beet plant grown in the United Kingdom. Plasmid pQBR103 was not present in the originally isolated strain but was acquired by SBW25 during a field release experiment. A genomic island of P. syringae pv. maculicola strain ES4326 (GenBank accession number AEAK00000000) is very similar to pQBR103 but has the same weak similarity to pOZ176. The replication, maintenance, and transfer region (see below) is distinct from pQBR103.

The two multiresistance class 1 integrons were derived from different lineages. The sul1-type class 1 integron is the most common type (>95%) found in clinical isolates and includes qacEΔ1 and sul1 downstream of the cassette array. The majority of metallo-β-lactamase genes occur in this type of integron. The incorporation of a sul1-type class 1 integron as part of Tn6217 into pOZ176 may have occurred first, possibly at the time when carbenicillin and chloramphenicol were more commonly used for treatment of infectious diseases. The bla_IMP-9-carrying Tn6016 in this study together with the bla_IMP-4-encoding Tn6017 from Citrobacter youngae (J. Xiong and P. H. Roy, unpublished data) were the first two transposons found to carry Tn402-type class 1 integrons with bla_IMP carbapenemase genes. This observation, together with recent reports of several Tn402-like integrons being associated with bla_VIM genes (48), represents a recent emergence of a distinct lineage of Tn402-like integrons, with complete tniABQR modules, originating from an ancestor never having had a sul1 gene. The three aacA4 genes found in two class 1 integrons encoding the amikacin-sensitive (Ak^s) gentamicin-resistant (Gm^r) version of AAC(6′)-IId with its DQSLA motif (49) were identical, suggesting multiple excisions and integrations catalyzed by integron integrase. The separate cluster of three aminoglycoside resistance genes (neo, ble, and str) is only 94% similar to that of PA7, whereas the latter is identical to that of Tn5. This suggests that there may be a family of such clusters and that Tn5 may have been formed by insertions of IS50 flanking the latter cluster.

Codon usage analysis suggests that most of the resistance genes carried by pOZ176 were not originally from P. aeruginosa but originated from other genera. Particularly, the codon usages of bla_IMP-9, bla_OXA-10, and qacEΔ1 are in strong contrast to those of the mobile element (intl1-tniRQBA) (50). The low G+C content of bla_IMP-9 suggests a very recent horizontal transfer from Firmicutes. Although our previous investigation showed that the dissemination of carbapenem resistance in the ICUs of four hospitals in Guangzhou was most likely to be mediated by conjugal transfer of the 500-kb plasmid, the other strains were not investigated in detail, and there remains the possibility of mobility of bla_IMP-9 by transposition and/or cassette excision and integration. Unlike bla_IMP-1/4 (8), bla_IMP-9 has never been found outside P. aeruginosa; this may be due to its transposon being defective, along with the narrow host spectrum of IncP-2 plasmids like pOZ176. It has been shown in vitro that pOZ176 could not be transferred into E. coli UB1637/R (9).

pOZ176 also has genes conferring resistance to heavy metals (e.g., mercury and tellurium). Resistance to tellurite is a key feature of IncP-2 plasmids in P. aeruginosa, as they are all resistant to tellurite (26). The mercury resistance transposon is related to Tn5044 from Xanthomonas campestris (51), Tn5046 from Pseudomonas sp. strain LS46-6 (52), and Tn5041 from another Pseudomonas sp. (53, 54). Since the mer operon is Tn5044-like rather than Tn21-like, and the tniA gene is truncated, it suggests that Tn21, with or without the potentially transposable tniABQR-type integron already in its res site, arrived in pOZ176 first, followed by the Tn5044-like element, whose 38-bp transposon end is found at the point of truncation of tniA. Certain mer loci but not others are found in multiresistant strains, and a majority of these loci are associated with integrons (55).

pOZ176 has acquired substantial numbers of mobile elements, suggesting a long history of DNA rearrangement. In addition to the three plasmid copies of the IS3-like insertion sequence, there are another four copies on the PA96 chromosome. The role of this element in the mobility of plasmid and of chromosomal genomic island DNAs remains an interesting question.

The plasmid maintenance region of pOZ176 is closely related to genomic islands of environmental species such as R. solanacearum GM1000 and A. vinelandii DJ, clinical isolates of Pseudomonas aeruginosa (PA14, PA7, NCGM2.S1, and PACS171b), as well as plasmid pBB1 of C. necator N-1. It contains a repA gene followed by three iterons, followed in turn by parA and parB genes. The partition systems of Pseudomonas plasmids often consist of two trans-acting proteins, ParA and ParB, and a cis-acting sequence. The cis-acting sites are termed centromere-like sequences because of the analogy to the mitotic apparatus of eukaryotes. ParA has an ATPase motif and is activated by ParB protein and DNA. parB encodes the protein that binds to the centromere-like sites. The centromere-like sequences usually are quite different among plasmid Inc groups and consist of various numbers of copies of iterons (56). The iterons of pOZ176 are of a unique type, with three 17-bp repeats with small (6- to 7-nucleotide [nt]) spacers. Identical or similar iterons occur in the genomic islands and in pBB1. The adjacent conjugal transfer genes are also observed in these elements. The occurrence of homologous replication/partition/transfer regions in genomic islands and in pBB1 suggests that these islands are chromosomally integrated parts of IncP-2 plasmids and that pBB1 is, like pOZ176, an IncP-2 plasmid. IncP-2 plasmids are known to be large, and pBB1 (1,499,175 bp) is among the largest.

The size of pOZ176, along with the comparable read depths between plasmid and chromosomal contigs, indicates that it is a single-copy plasmid that uses theta-type replication. Rep protein binds to the iterons to unwind the DNA at the replication origin and consequently introduce helicase and primase to initiate leading-strand synthesis (56).

The second parA-parB genes, with a distant repA gene, are weakly homologous to those of pQBR103 (and to those of a genomic island of P. syringae pv. maculicola). pQBR103 is an IncA/C plasmid, and the homologous genomic island sequence may represent another example of an integrated plasmid. These two homologs are much more similar to each other than to pOZ176. It is probable that the two independent replication and partition systems were introduced into pOZ176 at different times and that, given the results of incompatibility testing (9), only the IncP-2 maintenance system is active in pOZ176.

In summary, sequencing and analysis of the multidrug-resistant 500-kb plasmid pOZ176 has revealed numerous genetic features that could contribute to the virulence and transmission of the plasmid in the PA96 outbreak strain. This is the first IncP-2 plasmid from P. aeruginosa that has been sequenced and annotated and the first plasmid carrying a bla_IMP gene to be completely sequenced. Codon usage analysis suggests a recent acquisition, from an origin distant from P. aeruginosa, of bla_IMP as part of a Tn402-like integron. The presence of multiple mobile elements, including a pair of class 1 integrons from distinct lineages, indicates that horizontal gene transfer events have had a key role in the acquisition of antibiotic resistance and the evolution of pOZ176. The dual replication and partition systems of pOZ176 underline the mosaic nature of the plasmid, whose IncP-2 replication region is found in genomic islands of several species of environmental origin. Whole-genome sequencing has proven to be useful for large plasmids that are difficult to isolate.