Comparative Genomics of Multidrug Resistance-Encoding IncA/C Plasmids from Commensal and Pathogenic Escherichia coli from Multiple Animal Sources

Incompatibility group A/C (IncA/C) plasmids have received recent attention for their broad host range and ability to confer resistance to multiple antimicrobial agents. Due to the potential spread of multidrug resistance (MDR) phenotypes from foodborne pathogens to human pathogens, the dissemination of these plasmids represents a public health risk. In this study, four animal-source IncA/C plasmids isolated from Escherichia coli were sequenced and analyzed, including isolates from commercial dairy cows, pigs and turkeys in the U.S. and Chile. These plasmids were initially selected because they either contained the floR and tetA genes encoding for florfenicol and tetracycline resistance, respectively, and/or the blaCMY-2 gene encoding for extended spectrum β-lactamase resistance. Overall, sequence analysis revealed that each of the four plasmids retained a remarkably stable and conserved backbone sequence, with differences observed primarily within their accessory regions, which presumably have evolved via horizontal gene transfer events involving multiple modules. Comparison of these plasmids with other available IncA/C plasmid sequences further defined the core and accessory elements of these plasmids in E. coli and Salmonella. Our results suggest that the bla CMY-2 plasmid lineage appears to have derived from an ancestral IncA/C plasmid type harboring floR-tetAR-strAB and Tn21-like accessory modules. Evidence is mounting that IncA/C plasmids are widespread among enteric bacteria of production animals and these emergent plasmids have flexibility in their acquisition of MDR-encoding modules, necessitating further study to understand the evolutionary mechanisms involved in their dissemination and stability in bacterial populations.


Introduction
The use of antimicrobial agents in agriculture has been scrutinized over the past two decades because of their potential detrimental effects on animal and human health. Although the administration of antibacterial agents is an effective means to control bacterial infections, the use of antibiotics in agriculture is not limited to disease treatment and control; they are also used to prevent disease and to promote growth. Such use is postulated to facilitate the emergence of multidrug resistant bacteria isolated from animal sources (e.g., non-typhoidal Salmonella spp., Escherichia coli and other food-borne pathogens), and the dissemination of their multidrug resistance (MDR)-encoding determinants to other susceptible bacteria through horizontal gene transfer. The dissemination of MDR via conjugative plasmids can potentially limit future therapeutic options for treating infections in animals and humans [1,2,3,4,5,6,7].
Horizontal transfer of individual or arrays of resistance genes occurs mainly through the acquisition of conjugative plasmids, integrons, and transposons in enteric bacteria. Bacterial plasmids are self-replicating, extrachromosomal replicons, and as such they are key agents of genetic change in microbial populations. Besides conferring resistance to antibiotics, naturally occurring plasmids promote the spread of a variety of traits, including resistance to mercury and other heavy metals, virulence, fitness, and the metabolism of unusual compounds [6,7,8,9,10,11,12]. In recent years, there has been growing interest in the study of plasmids belonging to the IncA/C incompatibility group, mainly because of their ability to confer resistance to a diverse group of antimicrobial agents and their broad host range. IncA/C plasmids have been identified in numerous bacterial species, including Aeromonas hydrophila [13,14], Yersinia pestis [15,16], Photobacterium damselae subsp. Piscicida [17,18], Klebsiella pneumoniae [19], Vibrio cholera [20,21], E. coli [22], A. salmonicida [1], and S. enterica [8,22]. Analysis of the completed sequences of these plasmids has revealed that, with the exception of accessory components containing resistance-encoding elements, they were virtually identical to one another [14,16,22]. Among the genes identified within the IncA/C plasmid accessory regions are those encoding for resistance to tetracycline (tetA), chloramphenicol/florfenicol (floR), streptomy-cin/spectinomycin (aadA2), sulfonamides (sul1 and sul2), and extended-spectrum b-lactamases (bla CMY-2 ). In addition, the recent epidemic emergence of strains containing the bla NDM-1 metallobeta-lactamase gene, which are resistant to all antibiotic options in humans, has been associated with the IncA/C plasmid [23].
We recently completed the sequence of an IncA/C plasmid from E. coli isolated from a dairy cow in Illinois. This plasmid, approximately 165 kb in size, shared strong similarities with IncA/ C plasmids isolated from human-source Salmonella, suggesting recent movements of this plasmid type among a variety of enteric populations [22]. The widespread distribution of IncA/C plasmids among E. coli and Salmonella necessitates studying their genetic repertoire and similarities with plasmids from other bacterial populations in order to fully understand their emergence and evolution in these species. Therefore, the aim of this study was to analyze genetic differences in several IncA/C plasmids from E. coli recovered from differing production animal sources and geographical locations using comparative plasmid sequencing and analysis.

Sequence overview
Four plasmids were sequenced in this study, including the resequencing of pAR060302, previously isolated from a florfenicol-resistant E. coli commensal isolate from a US dairy cow [22]. The remaining three plasmids sequenced were from a commensal E. coli strain from a dairy cow in Chile (pPG010208), an avian pathogenic E. coli strain from a turkey in the USA with colibacillosis (p199061_160), and a porcine enterotoxigenic E. coli strain from a pig in the USA with post-weaning diarrhea (pUMNK88_161). All were sequenced using high-throughput Roche 454 DNA sequencing. These plasmids were isolated from farms in different geographical locations in the U.S. and Chile (Table 1). Single contiguous sequences with at least 15-fold coverage were obtained for each plasmid sequenced using draft assembly and PCR-based gap closure. BLAST analysis of the completed nucleotide sequences confirmed that they all belonged to the IncA/C incompatibility group based upon analysis of the predicted replicon. The sequence of pAR060302 was identical to the previous sequence generated via Sanger sequencing [22]. The plasmids varied in size from 135 to 165 kb, and with the exception of the accessory regions (see below), their backbone sequences were highly conserved (.99% nucleotide sequence similarity) and syntenic. Of the predicted open reading frames, approximately 40% were of unknown function. The predicted proteins with known function were primarily associated with resistance to antibiotics and heavy metals, conjugative transfer, and replication (Table S1). sul2-containing accessory region As described above, the sequenced plasmids differed primarily in their accessory regions. These regions mainly included insertion sequences and transposases, class 1 integrons, antibiotic resistance determinants, and heavy metal detoxification proteins. Analysis of these regions revealed the presence of several accessory gene clusters implicated in resistance to multiple antimicrobial agents. One of these regions occurs between repA and a putative conjugative transfer region (designated Tra1), and is a 16-kb module containing floR-tetA-strAB-sul2, encoding resistance to phenicols, tetracyclines, aminoglycosides, and sulfonamides ( Fig. 1). This sul2-containing region is identical in plasmids pAR060302, pUMNK88_161, and p199061_160, with the exception of a truncated strB gene in p199061_160. Also, located upstream of the floR gene were two copies of IS26, two ORFs encoding unknown functions, and an ISCR2 element [24]. This region was also present in pPG010208; however, the mph2 and mel genes encoding for macrolide resistance and an additional IS26 element are also located upstream of this region (Fig. 1).

Conjugative transfer and bla CMY-2 -containing regions
Within the backbone of all of the sequenced IncA/C plasmids are two putative conjugative transfer-associated regions designated Tra1 and Tra2. The Tra1 region consists of 22 ORFs, including 9 conserved hypothetical proteins, all located in a single gene cluster (Fig. 2). In bla CMY-2 -containing plasmids, one or more copies of bla CMY-2 is inserted within the Tra1 region in different locations. Among our sequenced plasmids, all except pPG010208 contained a bla CMY-2 insertion within Tra1. pAR060302 and pUMNK88_161 contained the insertion downstream of traA. In addition to bla CMY-2 , this accessory module contains genes with homology to the blc, sugE, and dsbC genes as previously described [22]. Also, the insertion sequence ISEcp1, belonging to the IS1380 family, exists upstream of the bla CMY-2 gene in all cases. This gene varies in size from 948 bp in the swine-source E. coli plasmid (pUMNK88_161) to 1,262 bp in the avian-and bovine-source E. coli isolates' plasmids (p199061_160 and pAR060302) (Fig. 2). In 199061_160, a region containing traLEKBV upstream of the insertion is absent.

Class 1 integron-containing accessory region
A third accessory module exists in most sequenced IncA/C plasmids, separating a cluster of hypothetical genes and the Tra2 region. This region typically contains a Tn21-like class 1 integron structure with multiple antibiotic resistance gene cassettes and mercury resistance genes. This structure is absent from pPG010208, but present in the three other plasmids from this study inserted in identical locations. In pAR060302 and p199061_160, identical class 1 integrons are present that contain aminoglycoside resistance genes, aadA and aacC; heat-shock chaperones groSEL; the qacEdelta1 and the sul1 genes; and mercury resistance genes merDBAPTR. The integron region is flanked by IS4321 elements similar to that previously described for Tn21 [25]. In pUMNK88_161, the Tn21-like structure is identical to that of plasmids pAR060302 and p199061_160, except that the gene cassette region in pUMNK88_161 contains cmlA, encoding resistance to chloramphenicol, and aadA2, encoding aminoglycoside resistance.

Transcriptional regulators
The recent influx of available genome sequences in the public database has improved the ability to effectively annotate possible functions to predicted proteins based on inferred sequence similarity. Previously, most of the predicted proteins of the IncA/C plasmid were hypothetical proteins. Upon re-annotation of these sequences, six predicted transcriptional regulators have been identified on the IncA/C plasmid backbone. These include proteins with similarity to families HU-beta, H-NS, Xre, LysR, and LuxR. In the case of the H-NS-like and HU-like proteins, these represent novel orthologs with only 81% sequence similarity (HU-beta) and 52% sequence similarity (H-NS) to their closest matches (Fig. 3). Also, these proteins do not share any significant similarity with the previously described plasmid-encoded H-NS proteins Pmr and Sfh from IncP-7 and IncH plasmids [26,27].

G+C content
The G+C contents of each plasmid sequenced here were analyzed and compared to the archetypic IncA/C plasmid pRA1   Figure  S1). Local G+C content varied from approximately 28% to 73% across the different plasmids, with two regions of high G+C content observed (,60%). The first corresponded to the accessory module containing the floR, tetA and sul2 genes, which is absent in pRA1. The other high G+C content fragment corresponds to the Tn21-like accessory regions, present on p199061_160, pUMNK88_161, and pAR060302. Additionally, two low G+C content regions were observed (,30%) corresponding to the genes that confer resistance to macrolides and the conjugative transfer Tra1 region containing bla CMY-2 . These regions of low or high G+C content were generally flanked by IS elements or inverted repeats.
Comparison of all sequenced IncA/C plasmids Linear maps were constructed for twelve completed IncA/C plasmids, including those from this study and from A. hydrophila (pRA1), Y. ruckeri (pYR1), P. damselae (pP91278 and pP99-018), Y. pestis (pIP1202), E. coli (peH4H), and S. enterica (pAM04528) (Figs. 4 and 5) [22]. As previously determined, the core backbones of these  plasmids are all highly syntenic with no genetic rearrangements [14,22]. The plasmids were grouped into those lacking (Fig. 4) or possessing (Fig. 5) the bla CMY-2 insertion. Eleven of twelve plasmids had sul2 in an accessory region between repA and Tra1. All of the bla CMY-2 plasmids had an identical sul2 accessory region structure containing floR-tetAR-strBA-sul2. The sul2 regions from plasmids pPG010208, pIP1202, pP99-018, pP91278, and pYR1 all differed from the bla CMY-2 plasmids and each other. The bla CMY-2 insertions within Tra1 varied with regard to insertion location, copy number, and genetic layout. p199061_160, pUMNK88_161, and pAR060302 all contained a single bla CMY-2 insertion downstream of traA, with the deletion of several Tra1 genes from p199061_160. peH4H contained duplicate bla CMY-2 insertions in Tra1 with multiple truncations of the Tra1 region. pAM04528 and pSN254 both contained adjacent and inverted copies of the bla CMY-2 region downstream of traA.
The third accessory region site lies upstream of the Tra2 region, and generally involves integron-like elements (Figures 4 and 5). In the bla CMY-2 plasmids, all of these insertions involve intact Tn21 or remnants thereof, inserted into identical sites. All of these regions contain a mercury resistance operon. In all bla CMY-2 plasmids except pAM04528, a class 1 integron is upstream of mer and varies between plasmids only with regard to gene cassette content. In pAM04528, mer is present but the class 1 integron appears to have been deleted. Among the non-bla CMY-2 plasmids, three (pP91278, pP99-018, and pPG010208) do not have an accessory element upstream of Tra2. One plasmid, pIP1202, contains a Tn21-like structure like the bla CMY-2 plasmids except that it is inverted. pRA1 and pYR1 contain insertions upstream of Tra2 that do not involve Tn21. In pRA1, a sul2 element with tetAR is present, and in pYR1 Tn10 and strAB are present.

Discussion
It is increasingly evident that IncA/C plasmids have emerged among populations of human and animal enteric bacteria, particularly E. coli, Salmonella, and Klebsiella spp. To better understand the genetic structure of these plasmids in production animal E. coli populations, we sequenced three plasmids from E. coli that had been isolated from a healthy commercial dairy cow, a diseased pig, and a diseased turkey, and re-sequenced a fourth plasmid from a commercial dairy cow. These strains and plasmids were selected for their ability to confer resistance to florfenicol. DNA sequencing confirmed that all four plasmids belong to the IncA/C group. Plasmids from this group have been isolated and described previously from a variety of Proteobacteria from animals, soil, and water. These include both pathogens and commensal bacteria [8,14,22,23,28]. Despite the fact that the IncA/C plasmids sequenced thus far were isolated from different geographical locations and diverse sources, the growing collection of IncA/C plasmid sequences all share a remarkably conserved backbone with varying accessory elements collectively encoding for a broad spectrum of antimicrobial resistance. The E. coliand Salmonella-source IncA/C plasmids sequenced thus far are quite similar in that they all have similar accessory regions, including the sul2-containing, bla CMY-2 -containing (except pPG010208), and class 1 integron-containing modules. Temporal screening of historical Salmonella Newport and E. coli from humans and animals dating from 1940 to date suggests that IncA/C plasmids emerged after 1980 [29,30], which is suggestive that a transfer event might have occurred prior to this time resulting in the introduction of a prototype IncA/C plasmid into these populations. However, plasmid and resistance phenotypes of IncA/C plasmids in recently isolated E. coli and Salmonella suggest that a variety of IncA/C Figure 5. Linear maps of sequenced IncA/C plasmids possessing bla CMY-2 . Core sequences are colored red (IncA/C replicon and hypothetical genes), blue (Tra1 region), and green (hypothetical genes and Tra2 region). Pink boxes depict the sul2-containing regions. Green boxes depict bla CMY-plasmid variants exist in these populations with differing resistance phenotypes and genetic content [31,32,33,34]. Therefore, it is unclear is these variants have arisen from recombinational events while in these species, or if multiple plasmid introductions have occurred.
The comparison of IncA/C plasmids from different production animals and geographic locations provides further evidence that the bla CMY-2 plasmids represent a unique IncA/C lineage that appears to be quite successful among bacterial populations, since they have been increasingly isolated and identified [35,36,37,38,39,40,41,42]. Analyses of this lineage suggests that its basic structure, in addition to the IncA/C conserved components among all sequenced plasmids, also includes a sul2 module containing floR-tetAR-strAB and a Tn21-like module. However, while the sul2 module appears to be stably maintained among this lineage, the bla CMY-2 and Tn21-like regions appear to be in constant flux. The ISEcp1 element is associated with all copies of IncA/C-encoded bla CMY-2 . ISEcp1, like ISCR elements, is involved in 'one-ended transposition' and has the ability to mobilize itself and adjacent resistance-associated genes [43]. Mobilization and duplication of beta-lactamase genes mediated by ISEcp1-like elements are well described in multiple bacterial species [44]. This helps to explain the variable duplication of the bla CMY-2 -ISEcp1 module throughout the Tra1 region on IncA/C plasmids. The floR and sul2 genes have been associated with ISCR2 in numerous other genetic contexts [24], suggesting that ISCR2 was involved in the introduction of the floR-tetA-strAB-sul2 element that is conserved in this lineage of IncA/C plasmids. In the class 1 integron region, most sequenced plasmids contain an ISCR16 element adjacent to the groESL genes, as previously described [24,45,46]. Overall, the bla CMY-2 -containing IncA/C plasmids are remarkable in that they contain at least three 'integration hotspots' for the acquisition of accessory genetic modules; they contain multiple means of acquiring these elements, including gene cassette acquisition via integrons, classical ISmediated acquisition via IS26 elements; and 'one-ended' acquisition via ISEcp1 and ISCR elements; and are of apparent broad host range [47].
The backbone of the IncA/C plasmid contains a number of putative DNA binding transcriptional regulators classified as nucleoid-associated proteins (NAPs). Such proteins are named for their ability to fold chromosomal DNA and form the nucleoid within the bacterial cell, and are well studied and also known for their immense regulatory properties. NAPs are categorized into several groups, including Fis, H-NS, HU, IHF, and Lrp [48]. H-NS is known to bind to A+T-rich regions and acts as a global transcriptional repressor; HU is also a global regulator that binds to DNA non-specifically. While the most-studied NAPs are those encoded on the bacterial chromosome, a number of plasmids have also been shown to possess NAPs. The first plasmid type identified with an H-NS NAP homolog, Sfr, was the IncH plasmid R27 from S. enterica serovar Typhimurium [49]. The effects of this plasmid and its Dsfh mutant were studied in S. Typhimurium. Interestingly, when pR27 was introduced into S. Typhimurium a limited number of chromosomal genes were differentially expressed, but the introduction of pR27Dsfh resulted in a nearly 4-fold increase in the number of chromosomal genes affected [49]. Furthermore, the Dsfh mutation greatly increased the fitness cost of carrying pR27 to the bacterial host. These observations were termed ''stealth functions'' elicited by such plasmid-encoded NAPs for their ability to silence the effects of pR27 on the host chromosome. Follow-up chromatin immunoprecipitation (ChIP) studies found that Sfh acts to bind to regions within the H-NS regulatory network and thus minimizes the effects of pR27 acquisition on the host chromosome regulatory networks [27]. A second plasmid-encoded H-NS-like protein with stealth function (Pmr) was identified on IncP-7 plasmids [26]. It is possible that the nucleoid-associated proteins encoded on IncA/C plasmids could elicit similar and immense effects on the transcriptional regulatory networks of their hosts, resulting in decreases in fitness costs and increases in host range associated with this plasmid group. However, the H-NS-and HUlike proteins identified on IncA/C plasmids are novel and only share low amino acid sequence similarity with their closest matches (Fig. 3); therefore, the roles of these proteins in such activities would need to be experimentally characterized. Certainly, further studies involving the biological mechanisms by which IncA/C plasmids succeed in various hosts are warranted, given their immense dissemination and association with pan-resistance.
A potential predecessor to the bla CMY-2 lineage of IncA/C plasmids is hinted at by the sequence of pPG010208 from a Chilean bovine-source E. coli isolate, which contains the sul2 region identical to bla CMY-2 plasmids but lacks bla CMY-2 itself and lacks a Tn21-like accessory region. This plasmid has additionally acquired the mel and mph-2 genes surrounded by two IS26 copies upstream of ISCR2, present only on pPG010208 as compared to other sequenced plasmids. These genes confer resistance to macrolides, which are mainly active against Gram-positive bacteria and are considered the drug of choice for group A streptococcal and pneumococcal infections when penicillin cannot be used [50]. Similar genetic structures to this have been described on plasmid pMUR050, isolated from an E. coli strain from a diarrheagenic pig [51] and on the pCTX-M3, a highly conjugative plasmid responsible for the dissemination of bla CTX genes in clinical populations of the family Enterobacteriaceae in Poland [52]. The differences observed between the Chilean isolate plasmid and other sequenced plasmids from U.S. isolates could represent an ''isolation by distance'' scenario, where differing local pressures could affect the acquisition of accessory elements in these plasmids. Ceftiofur is used frequently in the dairy industry of Chile, including on the farm where this isolate was obtained, but in our experience resistance to third generation cephalosporins and the bla CMY-2 gene encoding this ability are rarely identified among Chilean E. coli isolates. The discrepancies observed between pPG010208 and other sequenced IncA/C plasmids are not fully understood from an evolutionary and selective pressure standpoint, and deserve further study.
In addition to the accessory elements, we detected differences on the conjugal transfer system of the sequenced plasmids. In the non-bla CMY-2 plasmids, their Tra1 and Tra2 regions were generally complete and intact. However, the mosaic nature of the bla CMY-2 insertions and their duplications within the Tra1 region resulted in apparent disruptions of this region. For example, p199061_160 lacks of a 4-kb segment that includes the traEKBVA genes, which is present in pUMNK88_161 and pAR060302. Also, the bla CMY-2 insertion in several of these plasmids disrupts the traA and traC genes. Poole et al. studied the conjugative transferability of IncA/C plasmids containing or lacking the bla CMY-2 gene in Salmonella, concluding that plasmids encoding bla CMY-2 were rarely transferred compared with higher conjugation efficiencies where bla CMY-2 was absent [37]. Call et al. also reported the failure of self-conjugation for some of the IncA/C plasmids [22]. They reported that the failure of transferability of some of the IncA/C plasmids in their study was due to differences of the tra genes localized with bla CMY-2 . Others have noted that bla CMY-2 insertions do not necessarily affect the conjugative ability of IncA/C plasmids [31]. Thus, possible transfer deficiencies conferred through bla CMY-2 acquisition, the role of co-residing plasmids in decreasing its fitness cost and increasing its conjugative frequency, and dissection of the Tra1 and Tra2 regions in conjugative transfer still need to be studied.
A key question pertaining to multidrug resistance encoded by IncA/C plasmids is their maintenance in bacterial populations in the absence of selective pressures. Third-generation cephalosporins such as ceftriaxone and ceftiofur have important applications to both human and animal health [53]. Various genes encode for proteins that confer reduced susceptibility to these antimicrobials, and bla CMY-2 is commonly responsible for the resistance to these antimicrobial agents in the U.S. [22,35,37,39,54,55]. Potential risks have been identified due to the possible co-selection of the bla CMY-2 through the use of florfenicol in food animal production. Compounding this scenario is the presence of multiple means of selection, including antibiotics and heavy metals, on many IncA/C plasmids. It was recently demonstrated the long-term maintenance of IncA/C plasmids might require selective pressure, which contrasts the apparent success of this plasmid type in a variety of environmental niches including possible non-selective environments [56]. This underscores the need to elucidate the selective pressures that drive the success of this plasmid type in Enterobacteriaceae.
The presence of the cmlA gene on the class 1 integrons was another trait that differed between the IncA/C plasmid sequences. This gene was detected only on the swine-source E. coli plasmid, pUMNK88_161 ( Figure 1). These results agree with the study by Bischoff et al., who found that 48 of the 90 E. coli isolates from swine production in Oklahoma exhibited resistance to chloramphenicol and 47 of these isolates possessed the cmlA gene [41]. This gene encodes a putative efflux pump that confers resistance to chloramphenicol, which has been banned in the U.S. since 1985. Thus, the presence of this gene on IncA/C plasmids is an example of an additional selection mechanisms for its widespread dissemination in commercial pig hosts and persistence in the absence of the particular selective pressures, and is aggravated by the fact that the use of any antimicrobial encoded by the IncA/C plasmid can potentially co-select for a number of additional phenotypes.
Previous studies of the IncA/C plasmids suggest that these plasmids probably did share a recent common ancestor. Fricke et al. sequenced pRA1 [14], considered the first member of the IncA/C group of MDR plasmids to be fully described [57]. This plasmid showed a reduced antimicrobial resistance spectrum, which the authors attributed to a probable minimal selective pressure. The authors proposed an evolutionary model in which each ''IncA/C plasmid diverged from a common ancestor through a specific process of stepwise integration events of horizontally acquired resistance gene arrays'' [14]. It appears that the bla CMY-2 plasmid lineage is such an example, where its emergence resulted from initial acquisitions of its sul2 module, bla CMY-2 module, and Tn21. Further evolution of this plasmid lineage and other IncA/C lineages seems to be rapidly occurring, as recent reports have identified the New Delhi Metallo-b-Lactamase (NDM-1) occurring on or with IncA/C plasmids [23,58]. Again, the underlying mechanisms driving the evolution and emergence of such IncA/C plasmid variants is unclear, but will likely present great challenges to human and animal health.
In conclusion, variants of broad-host-range IncA/C plasmids have emerged in a variety of bacterial species. The association of MDR with integrons, complex transposons, and ISCR elements, all on a conjugative plasmid, infers the possibility of dissemination among clinical isolates that creates opportunities for the rapid emergence of multidrug resistant bacterial clones. Strains harboring these plasmids serve as a reservoir for antibiotic resistance genes, the further spread of which could likely limit therapeutic options. Based upon the recent analyses of IncA/C plasmids revealing their genetic components and dissemination among E. coli and Salmonella of humans and production animals, future studies are essential to determine the specific mechanisms of acquisition, persistence, and dissemination of these plasmids among bacterial populations.

Bacterial isolates
All strains used in this study are listed in Table 1. Isolates were collected in previous studies by the investigators [9,22]. The research in these studies complied with all relevant animal use federal guidelines and institutional policies. The strains were selected because they all harbored a large plasmid and exhibited resistance to ceftriaxone, florfenicol or tetracycline. All strains were cultured at 37uC in Tryptone soy agar (TSA) and stored in 40% glycerol at 280uC.

Plasmid isolation and sequencing
Single colonies were inoculated into 100 mL LB broth and grown overnight at 37uC with shaking. Plasmid isolation was performed using Plasmid Midi Kit (Qiagen Inc., Valencia, CA). After purification, plasmid DNA was resuspended in sterile water and detected by electrophoresis on 0.8% agarose gels at 4uC. Ten micrograms of purified plasmid DNA was sequenced at Biomedical Genomic Center at the University of Minnesota using the Roche 454 GS-Titanium sequencing platform (454 Life Sciences, Branford, CT).

Assembling and annotation
For each strain, sequencing reads were assembled de novo using SeqMan software from DNAStar (Lasergene, Madison, WI). Assembled contigs were then mapped to a reference genome (FJ621588) for arrangement of the contiguous sequences in their most likely orientation. The final gaps were closed using standard PCR. Open Reading Frames (ORFs) in the plasmids sequences were identified using GeneQuest from DNAStar (Lasergene, Madison, WI), and ORF Finder (http://www.ncbi.nlm.nih.gov/ projects/gorf/), followed by complete manual inspection. Translated ORFs were then compared to known protein sequences using BLAST [59]. Those with greater than 80% homology with database protein sequences were considered matches, while hypothetical proteins with greater than 80% amino acid sequence identity to one or more previously published proteins were classified as conserved hypothetical proteins. Insertion sequences and repetitive elements were identified using IS FINDER (http:// www-is.biotoul.fr/). Finished sequences are deposited in Genbank under accession numbers HQ023864 (pAR060302), HQ023861 (pPG010208), HQ023862 (pUMNK88_161), and HQ023863 (p199061_160).

Comparative genomics
Following annotation, the assembled nucleotide sequences were analyzed to other plasmid sequences using BlastN [60]. Through this analysis, comparative linear maps (http://www.iayork.com/ XPlasMap/) were created. The G+C content of each plasmid was analyzed using ARTEMIS software [61].

Antimicrobial susceptibility testing
Wild type strains were subjected to disk diffusion to determine the susceptibilities of isolates to the following drugs: streptomycin, tetracycline, florfenicol, chloramphenicol, ampicillin, and sulfisoxazole. Figure S1 Sliding G+C contents of each plasmid sequenced in this study. A = region containing floR, tetA and sul2 genes, B = the Tn21-like accessory regions, C = genes that confer resistance to macrolides, and D = the conjugative transfer region together with bla CMY-2 gene. (TIF)