Complete nucleotide sequence of pRS218, a large virulence plasmid, that augments pathogenic potential of meningitis-associated Escherichia coli strain RS218

Background Escherichia coli is the most predominant Gram-negative bacterial pathogen associated with neonatal meningitis. Previous studies indicated that the prototypic neonatal meningitis E. coli (NMEC) strain RS218 (O18:K1:H7) harbors one large plasmid. Objectives of the present study were to analyze the complete nucleotide sequence of this large plasmid (pRS218) and its contribution to NMEC pathogenesis using in vitro and in vivo models of neonatal meningitis. Results The plasmid is 114,231 bp in size, belongs to the incompatibility group FIB/IIA (IncFIB/IIA), and contains a genetic load region that encodes several virulence and fitness traits such as enterotoxicity, iron acquisition and copper tolerance. The nucleotide sequence of pRS218 showed a 41- 46% similarity to other neonatal meningitis-causing E. coli (NMEC) plasmids and remarkable nucleotide sequence similarity (up to 100%) to large virulence plasmids of E. coli associated with acute cystitis. Some genes located on pRS218 were overly represented by NMEC strains compared to fecal E. coli isolated from healthy individuals. The plasmid-cured strain was significantly attenuated relative to the RS218 wild-type strain as determined in vitro by invasion potential to human cerebral microvascular endothelial cells and in vivo by mortalities, histopathological lesions in the brain tissue, and bacterial recovery from the cerebrospinal fluid of infected rat pups. Conclusions The pRS218 is an IncFIB/IIA plasmid which shares a remarkable nucleotide sequence similarity to large plasmids of E. coli associated with cystitis. Both in vitro and in vivo experiments indicated that pRS218 plays an important role in NMEC pathogenesis.


Background
Neonatal meningitis (NM) and sepsis is the third most common disease in neonates that accounts for approximately 393,000 deaths per year worldwide [1]. Escherichia coli has been identified as the most predominant Gram-negative pathogen associated with NM [2][3][4][5]. Despite advanced antimicrobial therapy and supportive care, mortality and morbidity rates of NM due to neonatal meningitis-associated E. coli (NMEC) continue to be as high as 30-50% [6]. Other than high mortality, adverse consequences such as mental retardation, vision loss or impairment, hearing impairment and speech impediment of NM in surviving neonates are also a major medical concern [7,8].
Plasticity of E. coli genomes has led to the development of different pathovars of E. coli each of which is associated with a particular form of animal and/or human disease [9,10]. Genomic plasticity of E. coli is mainly due to the acquisition of 'genomic islands' through horizontal gene transfer by means of plasmids, phages and insertion sequences (IS) [9]. Of these elements, bacterial plasmids are self-replicating extra-chromosomal genetic materials which have the potential to transmit a variety of phenotypic characteristics among the same or different species of bacteria [9][10][11]. These phenotypic characteristics include novel metabolic capabilities, antibiotic resistance, heavy metal tolerance, virulence traits that are important for bacterial adherence, invasion and survival in host tissues [10,11]. Plasmid that encodes such phenotypic characteristics may provide competitive advantages to the bacterium for survival and adaptation to novel niches.
Many virulence associated plasmids have been identified in pathogenic E. coli [10,[12][13][14]. A vast majority of these plasmids belong to IncF compatibility group. Structurally, IncF plasmids consist of a conserved region common to all IncF plasmids which encodes conjugal transfer proteins, replication proteins and plasmid stability proteins and a 'genetic load region' or a variable region that encodes various virulence and fitness traits. A recent study that analyzed over 40 completed genomic sequences of IncF plasmids of E. coli revealed that these plasmids have evolved as virulence plasmids by acquiring novel virulence traits to their 'genetic load regions' through IS-mediated site specific recombination [10]. Also, comparative genomic analysis of virulence plasmids in each pathovar of E. coli has shown that these genetic load regions encode virulence traits that are essential for and specific to their respective pathotype [10]. These data suggest that acquisition of plasmid-encoded genes may play a significant role in the emergence of pathogens and different pathotypes of E. coli.
Although many virulence-associated plasmids in various intestinal pathogenic E. coli have been sequenced and studied, only a few virulence plasmids associated with each pathotype of extra-intestinal pathogenic E. coli (ExPEC) causing human infection have been sequenced [10]. For example, at the time of preparing this manuscript, only two plasmid sequences from NMEC strains were available in the public domain [14,15]. These two strains represent two of three major serogroups of E. coli (O18, O45 and O7) that have been implicated in NM; pECOS88 from E. coli S88 (O45:K1) and pEC10A-D from E. coli CE10 (O7: K1). Despite the fact that the NMEC prototypic strain RS218 belonging to O18 serogroup is the most commonly used E. coli strain to study NMEC pathogenesis since 1980's, its genomic sequence including the plasmid, has not been reported [16]. It has been documented that the NMEC RS218 strain harbors a large plasmid and similar sized plasmids have been observed in other NMEC and avian pathogenic E. coli (APEC) belonging to the O18 serogroup [17,18].
Therefore, the objectives of the present study were to: (i) analyze the nucleotide sequence of pRS218 and its genetic and evolutionary relationship with virulence-associated plasmids in other pathogenic E. coli, (ii) analyze the distribution of pRS218 genes among NMEC, and (iii) evaluate the contribution of pRS218 to NMEC pathogenesis by comparing the virulence of plasmid-cured and wild-type strains in vitro and in vivo.

General properties of pRS218
Initial de novo assembly of short reads generated with Ion Torrent PGM technology identified 26 plasmid contigs ranging from 253 to 7,521 bp in length. These contigs were aligned to the reference plasmid sequence pUTI89 of uropathogenic E. coli strain UTI89 which was selected as the reference according to the sequence similarity of contigs (>90%). Complete sequence of pRS218 revealed that it is a circular plasmid of 114,231 bp in size with a G + C content of 51.02% ( Figure 1). A total of one hundred and sixty open reading frames (ORFs) were annotated including IncFIB and FIIA replicons. Based on the blast analysis, nearly one third of the ORFs (n = 51) represents the genes involved in plasmid replication and conjugal transfer, along with 20 and 7 genes encoding mobile genetic elements (MGEs) and products involved in DNA repair, respectively. Of the remaining ORFs, 59 encode unknown or hypothetical proteins, and 23 represent genes previously characterized in other bacteria. The plasmid does not harbor any antibiotic resistance genes that may provide a selective advantage in the face of antibiotic therapy. Genetic load region of the pRS218 encodes several virulence-and fitness-associated genes which have been reported in other bacteria ( Table 1 Figure 2) [23]. Analysis of the repA1 sequence of FIIA replicon of 24 IncFIB/IIA plasmids in pathogenic E. coli revealed three main lineages of virulence plasmids ( Figure 3). All NMEC virulence plasmids were clustered into one lineage based on the repA1 sequence suggesting a common origin. Interestingly, pRS218 showed an identical origin with several virulence plasmids of E. coli causing cystitis (pUTI89 and pEC14_114), pECSF1 of the commensal phylogenetic group B2 E. coli strain SE15 and pCE10A of NMEC strain CE10.
Genes of pRS218 are overly represented in NMEC strains compared to fecal E. coli Plasmid profiling revealed 27 of 53 (51%) of NMEC strains examined in the study harbored a plasmid similar in size to pRS218 (130-100 kb) (Table 2). Furthermore, PCR analysis revealed that a vast majority of pRS218associated genes tested (n = 59) were overly represented (n = 52) among NMEC strains as compared to commensal E. coli (Table 3).

Plasmid-cured strain demonstrated a marked attenuation in vitro and in vivo
To analyze the virulence potential of pRS218, the plasmid was cured from the wild type strain by mutating stbA Figure 1 Graphical circular map of pRS218. From outside to the center: ORFs in forward strand, ORFs in reverse strand, and GC skew. Plasmid genes are color coded as follows: Blue, conjugal transfer genes; Green, virulence or fitness-associated genes; Orange, plasmid replication genes; Red, IS elements; Black, plasmid stability genes; Light blue, hypothetical and putative genes. In the GC skew lime indicates the areas where the GC skew above average (51%) and purple indicates the areas below average. Putative Na-translocating NADH dehydrogenase Na + efflux (NrqC subunit) [19] pRS218_013 High affinity Fe 2+ permease Iron acquisition [13] pRS218_014 High affinity Fe 2+ periplasmic component Iron acquisition [13] pRS218_015 High affinity Fe 2+ protein, membrane component Iron acquisition [13] pRS218_016 High affinity Fe 2+ binding protein, permease Iron acquisition [13] pRS218_017 Putative ABC transport sys, permease Type I secretion [13] pRS218_018 Putative ABC transport system, ATP-binding Type I secretion [13] pRS218_019 TonB-dependent heme/hemoglobin receptor Iron acquisition [13] pRS218_039 SenB Enterotoxin in EIEC/Shigella [26] pRS218_001 Putative GTP binding protein, YihA Cell signaling and membrane ruffling [20] pRS218_190 Hemolysin expression modulating protein Thermo-osmotic regulation of hly [13] pRS218_007 Suppressor for copper sensitivity ScsC/ScsD Copper tolerance [25] pRS218_022 Glucose-1-phosphatase Virulence regulator [21] pRS218_116 Lytic transglycosylases Cell division [22] followed by 10% SDS treatment. Curing of plasmid was confirmed by the absence of the plasmid in the purified plasmid preparation and the absence of 5 selected genes of pRS218 by PCR in a crude DNA extract made from the plasmid-cured strain (RS218 cured) . Figures 4A and B show the plasmid profiles and PCR amplification results of wild-type RS218 (wtRS218) and plasmid-cured RS218 (RS218 cured ). No difference was observed in the growth rates between wtRS218 and RS218 cured ( Figure 4C). Virulence potential of pRS218 was determined by comparing  RS218 cured with wtRS218 based on their ability to invade human cerebral microvascular endothelial (hCMEC/D3) cells in vitro and to cause septicemia, meningitis and mortality in vivo in a rat pup model of neonatal meningitis. In vitro invasion assays using hCMEC/D3 cells revealed a significant attenuation (p < 0.05) of RS218 cured (relative invasion 38 ± 9.6%) as compared to the wild type strain (100%) ( Figure 5A). Furthermore, invasiveness was restored after complementation of RS218 cured strain with the pRS218 indicating its contribution to NMEC pathogenesis. Similar results were observed with the in vivo experiments as well. Although fewer pups died within 24 hrs post-infection in the groups infected with RS218 cured as compared to the groups infected with wtRS218 and RS218 compl , there was no statistically significant difference in mortality rates between the three groups ( Figure 5B). No mortalities were detected in the negative control group treated with PBS or E. coli DH5α. In groups infected with wtRS218 or RS218 compl , 84-87% of rat pups that survived 24 hrs post-infection showed septicemia, whereas in groups treated with RS218 cured strain, only 33% had septicemia. In all three groups the number of bacteria in the blood was too numerous to count (>1.5-2.8 *10 4 CFU/ml). Also, E. coli were re-isolated from CSF collected from 84-87% of rat pups infected with wtRS218 or RS218 compl whereas only 29% CSF samples collected from rat pups infected with RS218 cured strain contained E. coli suggesting a role of pRS218 in translocation of bacteria through the blood brain barrier (BBB) to cause meningitis. Similarly, histopathological evaluation of brain tissue from the rat pups inoculated with wRS218 or RS218 compl strains demonstrated lesions consistent with meningitis ( Figure 6). The bacterial loads in CSF were 4.57 + 3.02*10 3 in rat pups infected with wtRS218 strain and 3.77 + 2.24*10 3 in rat pups infected with RS218 cured strain.

Discussion
Virulence plasmids in bacterial pathogens have been shown to play a major role in pathogenesis of many bacterial diseases [10,12,24,25]. In pathogenic E. coli, virulenceassociated large plasmids that are required to establish distinct disease phenotypes have been characterized using   in vitro and in vivo studies [10,[12][13][14]17,25]. Recently, it has been suggested that the plasmids may play a role in NMEC pathogenesis since most of the NMEC strains harbor plasmid-associated genes as compared to commensal E. coli [26]. Escherichia coli RS218 which was isolated from CSF of a neonate with meningitis in 1974 is considered as the prototype strain of NMEC. This strain has been used in the studies since then to identify the virulence traits that are particularly involved in NMEC pathogenesis [16]. Here, we determined and analyzed the complete nucleotide sequence of pRS218, a large plasmid of E. coli RS218, and studied its contribution to the NMEC pathogenesis. The pRS218 sequence revealed a backbone typical to IncFIB/IIA-like plasmids in other pathogenic E. coli which possess both repA and repA1 replicons [10]. In addition to the replication proteins, the constant region of the plasmid encodes proteins involving conjugal transfer (Tra locus) and plasmid stability/inheritance. The tra locus comprises 34.9 kb region containing 34 tra genes from traM to finO similar to F-like plasmids of E.coli and R100 plasmid of Shigella [27]. The plasmid SOS inhibition protein (PsiAB), plasmid stabilizing proteins StbAB and CcdAB, toxin-antitoxin proteins involved in post segregation killing are also present in the constant region that confers stability and inheritance of the plasmid in progeny cells. Parallel to these findings, we have observed that the curing of pRS218 is very difficult with chemical methods such as ethidium bromide and SDS treatment alone. Therefore, we mutated the stbA gene which has been identified as an essential gene for stable inheritance of IncF plasmids to achieve successful curing of pRS218 from E. coli RS218.
Genetic load region or the variable region of the pRS218 contains IS elements, virulence-associated genes, and several putative and hypothetical genes. The pRS218 contains 20 IS elements belonging to twelve different types. Previous studies have shown that IS-mediated recombination might play a major role in acquiring novel genes into plasmids thereby allowing the plasmid to act as a "pathogenicity island precursor" [10,12,14]. Interestingly, IS elements of pRS218 are located upstream or downstream of virulence/fitness-associated genes in genetic load regions providing further evidence for such speculation (Figure 1). Types of virulence or fitness genes in the genetic load region of pRS218 are depicted in Table 1 and are mainly located upstream and downstream of IncFIB replicon. Upstream to the IncFIB replicon, are the secreted copper-sensitivity suppressor proteins C and D (scsC and scsD). Copper is an essential trace element required for bacterial growth and it acts as a toxic compound if available in excess. Antibacterial properties of ionic copper have been studied and used in hospital settings to prevent nosocomial infections [28]. The pRS218 encoded scsC and scsD are identical to copper suppressor proteins in the genomic island GI-DT12 of Salmonella enterica subsp. enterica serovar Typhimurium str. T000240 which have been studied in relation to conferring copper resistance in recombinant E. coli carrying GI-DT12 providing a fitness advantage to the pathogen [29]. Additionally, this region encodes several iron acquisition proteins, hemoglobin receptors and a putative ABC transporter which may be involved in the survival of bacteria in an iron-limited milieu inside the host. Furthermore, pRS218 also encodes an enterotoxin called SenB, which has been found in enteroinvasive E. coli and Shigella spp and accounts for 50% of their enterotoxic activities [30]. Interestingly, nucleotide blasting of senB sequence reveled that it is also present in the genomes of E. coli CE10 and the Citrobacter koseri which are associated with meningitis in newborns. Moreover, senB is located just downstream of cjr operon which is an iron-and temperature-regulated operon expressed only during the pathogenic process of E. coli suggesting that senB may be involved in NMEC pathogenesis [30]. A recent study reported that mutation of cjr area of pUTI89 (which is >99% similar to pRS218) significantly decreased bacterial invasion and intra-cellular bacterial community (IBC) formation in infected bladders [12]. However, the association of pRS218-encoded traits such as SenB in NMEC penetration of the intestinal epithelium and iron acquisition systems in NMEC survival within the host are yet to be identified. Other than these putative virulenceassociated genes, many hypothetical proteins of unknown functions are present both upstream and downstream of IncFIB replicon.
Furthermore, we screened 59 pRS218 genes among 53 NMEC strains and fecal E. coli strains isolated from healthy individuals. A vast majority of pRS218-associated genes tested were overly represented among NMEC strains as compared to commensal E. coli (Table 3) suggesting a relationship between the presence of pRS218 genes and the NMEC pathotype. These overly represented genes included several hypothetical proteins and virulenceassociated genes present in pRS218 such as copper sensitivity, iron acquisition, ABC transporter components, traJ and senB.
We also analyzed the sequence similarity and the evolutionary relationship of pRS218 with other NMEC plasmids, namely pECOS88 and pCE10A, and some other IncFIB/IIA plasmids of pathogenic E. coli (Figures 2 and  3). The pRS218 showed a remarkable sequence similarity to four plasmids found in E. coli associated with acute cystitis (pUTI89, pEC14_114, p1ESCUM, and pUMN146) and a plasmid present in an enteroinvasive E. coli (pECSF1) (Figure 2). The differences detected among pRS218, pUT89, pEC14_114 and pUMN146 revealed only SNPs and insertion of a tetABCD antibiotic resistance cassette in p1ESCUM and pECSF1 ( Figure 2 and Table 4). However, the nucleotide sequence of pRS218 showed a marked difference from those of two NMEC plasmid sequences currently available in the public domain. For example, pECOS88 shares similarity only with tra locus, repA and repA1 regions of pRS218 revealing that the genetic load regions of these plasmids harbor different putative virulence and hypothetical genes to those of pRS218. Compared to pECOS88, pCE10A plasmid showed a relatively higher nucleotide sequence similarity to pRS218 genetic load region containing the copper resistanceassociated genes (scsDC), cjrABC and senB. However, pCE10A lacks the tra locus thereby making the plasmid incapable of conjugal transfer.
Among many capsular types of E. coli, K1 is the most common type associated with NM and according to previous studies, approximately 80% of NMEC possessed a K1 capsule [4,5]. Neonates acquire E. coli K1 mainly from the urogenital microflora of the mother. Although there are no studies done on the mechanisms that facilitate the vaginal epithelial colonization and survival of the NMEC strains in the urogenitary tract of women, it has been well documented that cystitis causing E. coli can survive and persist inside bladder epithelial cells as IBCs which is a dormant stage that becomes activated and shed when the immunity of the host is suppressed as is the case during pregnancy [26]. The same study has also indicated that the pUTI89 plasmid is essential for filamentation of IBCs which is the first event of reactivation of E. coli from the dormant state. A high degree of sequence similarity of pRS218 to other cystitis-associated plasmids and their close evolutionary relationship suggest that E. coli RS218 Figure 5 Evaluation of virulence potential of pRS218 in vitro and in vivo. A, Involvement of pRS218 in invasion of hCMEC cells. B, Comparison of mortality, septicemia and meningitis among the groups of rat pups infected with wtRS218, RS218 cured , RS218 compl . ** denotes statistical significance and * denotes no statistical significance. might use the same strategy to survive in the urogenitary tract. However, the ability of E. coli RS218 to invade bladder epithelial cells and to survive within the urogenitary tract remains to be investigated.
Pathogenesis of NMEC meningitis involves three main sequential events that are governed by the virulence potential of bacteria. These include initial colonization and invasion of gastrointestinal tract, survival and multiplication in blood, and invasion of BBB [5]. We examined the pathogenic potential of pRS218 to penetrate BBB in vitro and in vivo using hCMEC/D3 cells and a rat pup model of neonatal meningitis, respectively. Curing of Figure 6 Histopathological evaluation of brain tissue from rat pups. Five-day-old rat pups were infected by the IP route with wtRS218, RS218 compl , RS218 cured , E. coli DH5α or PBS. Pups that survived were euthanized 24 hrs post-infection, and the brains were excised, embedded in formalin, sectioned in paraffin, and stained with haematoxylin and eosin. A-F: meningitic lesions observed in pups infected with wtRS218 (A and B) or RS218 compl (C, D, E, and F). Arrows indicate rod-shaped bacteria in meninges and brain tissue (black), neutrophilic infiltration/neutrophilia (blue), and cerebral edema (orange). G to I: normal histology of brain tissue from pups inoculated with RS218 cured (G), PBS (H) or DH5α (I). pRS218 from E. coli RS218 did not show any effect on the growth rate revealing that differences observed between wild type and plasmid cured strains during in vitro and in vivo studies were not due to the differences in their growth rates ( Figure 4C). It is believed that the high level of septicemia is a prerequisite for the penetration of BBB by NMECs to establish neonatal meningitis [4]. We observed a higher incidence of septicemia among the rat pups infected with wtRS218 strain (84%) than the RS218 cured strain indicating that plasmid-encoded genes might be involved in developing septicemia. Iron is a major limiting factor that restricts the survival and multiplication of bacteria inside the host. The genetic load region of pRS218 encodes several high affinity iron acquisition proteins, hemolysin modulation factor and hemoglobin receptor which may be involved in iron acquisition. Interestingly, these genes were highly prevalent in NMEC strains as compared to fecal E. coli (Table 3). Furthermore, in vitro and in vivo study results clearly demonstrated that RS218 cured strain is far less capable of invading epithelial and endothelial cells as well as establishing meningitis in neonatal rat pups as compared to its wild type strain, suggesting that pRS218 might play a role in NMEC pathogenesis. The traJ which is present in pRS218 has been previously identified as a potential virulence trait in NMEC by signature-tagged mutagenesis and in vitro endothelial invasion assays [31]. The mutation of traJ was shown to be attenuated in terms of invasive ability to penetrate the BBB. However, more than 50% of the NMEC strains used in this study did not possess traJ even though the gene was more prevalent in NMEC than in fecal E. coli (Table 3). The present study demonstrated that the curing of pRS218 offered a greater attenuation to RS218 strain than did the mutation of traJ alone suggesting that addtionalpRS218 genes other than traJ might be involved in NMEC pathogenesis. Interestingly, as shown in Table 3, pRS218 carries several genes that encode hypothetical proteins which are also more prevalent in NMEC than in fecal commensal E. coli. Most gene prevalence studies carried out to identify potential virulence markers of NMEC have used already known virulence genes of other ExPEC and only a limited number of studies have attempted to explore novel traits that might be helpful in defining the NMEC pathotype [5,26,32]. Therefore, future studies aimed at delineating the mechanistic aspects of hypothetical proteins encoded by pRS218 and are more commonly occurring in NMEC than in fecal commensal E. coli may help to close the knowledge gaps pertaining to our understanding of NMEC pathogenesis.
Although RS218 cured strain was significantly attenuated in terms of in vitro (on the basis of bacterial invasion) and in vivo (on the basis of reduction in bacterial counts in CSF and blood from infected rat pups) assays as compared to the wild type strain, it was not completely avirulent. This finding suggests that the full virulence of E. coli RS218 requires both chromosomal and plasmidlocated genes. Further studies including in depth analysis of RS218 chromosome will advance our understanding of NMEC pathogenesis.

Conclusions
Incomplete understanding of NMEC pathogenesis is a major hindrance that has been identified and pointed out by many scientists particularly in relation to formulation of novel therapeutic and prevention strategies for neonatal meningitis. The plasmid pRS218 in NMEC RS218 strain belongs to IncFIB/IIA subset of virulence plasmids in pathogenic E. coli. These plasmids harbor many virulence traits that are required for bacterial survival inside the host. The nucleotide sequence of pRS218 showed a greater similarity to the plasmids of E. coli associated with acute cystitis than the plasmids from NMEC. However, the prevalence of pRS218 virulence-related genes was significantly higher in NMEC strains tested than fecal commensal E. coli. We have also demonstrated that the pRS218 is involved in NMEC pathogenesis using both in vivo and in vitro experiments. Future studies on pRS218 transcriptome analysis, identification of plasmidlocated genes responsible for current observations and indepth analysis of E. coli RS218 whole genome will likely broaden our knowledge of NMEC pathogenesis.

Bacterial strains and media
The prototype NMEC strain E. coli RS218 (O18: H7: K1) and NMEC strain EC10 (O7: K1) were kindly provided by Dr. James Johnson (Department of Medicine, University of Minnesota, Minneapolis, MN). Both E. coli RS218 and EC10 strains have been isolated from cerebrospinal fluid of neonates diagnosed with bacterial meningitis (15). A total of 51 NMEC strains which were isolated from neonatal meningitis cases were also obtained from Dr. K.S. Kim (School of Medicine, John Hopkins University, Baltimore, MD) and 49 fecal E. coli strains isolated from feces of healthy individuals were obtained from the E. coli Reference Center (Pennsylvania State University, University Park, PA). All E. coli were stored in Luria Bertani broth (LB) at -80°C until further use. Bacteria were grown in MacConkey agar or LB broth. All bacteriologic media were purchased from Becton, Dickinson and Company (BD), Sparks, MD.

Plasmid isolation, sequencing, assembly and annotation
Sequencing of pRS218 was performed as a part of a project that sequenced the whole genome of E. coli RS218. The genomic DNA including the plasmid DNA was extracted using phenol-chloroform method as described previously [33]. The DNA preparation was further cleaned using Genomic Tips (Qiagen, Valencia, CA) [33]. Whole genome sequencing was performed using Ion Torrent PGM Technology (Life Technologies, Carlsbad, CA) at the Genomics Core Facility (Pennsylvania State University, University Park, PA). After initial de novo assembly of short reads using SeqManNGen 10 (DNASTAR Inc, Madison, WI), plasmid contigs were identified using BLAST algorithm (Blastn; www.ncbi.nlm.nih.gov) and subsequently aligned to the sequence of the reference plasmid, pUTI89 [GenBank:CP000244]. Gap closure was performed using primer walking into the gaps with the LongRange PCR Kit (Qiagen). The complete sequence of the plasmid was annotated using Rapid Annotation using Subsystem Technology (RAST) [34].

Analysis of plasmid profiles of NMEC strains
Extraction of large plasmids from NMEC strains was performed using an alkaline lysis method described previously [33]. In brief, 1 ml of overnight culture of each E. coli strain was subjected to alkaline lysis using 10% sodium hydroxide followed by phenol-chloroform extraction of plasmid DNA. Plasmid profiles of NMEC strains were evaluated by electrophoresis on a 0.7% agarose gel containing 0.5 μg/ml ethidium bromide.
Evaluation of prevalence of selected pRS218 genes in other NMEC and fecal E. coli Specific polymerase chain reactions (PCRs) were performed to determine the presence of selected gene coding regions (n = 59) of pRS218 in other NMEC and fecal E. coli strains. Primers were designed using the Primer 3.0 web tool (http://bioinfo.ut.ee/primer3-0.4.0/) ( Table 5). PCR amplifications were performed using crude DNA extracted by the rapid boiling method [38]. The PCR mixture contained 1 U of Taq polymerase (Qiagen), 1× Taq polymerase buffer, 3.5 mM MgCl 2 , 125 μM each deoxynucleotide triphosphate (dNTP) and150 nM each primer pair. PCR conditions were as follows: 1 cycle of 95°C for 1min, followed by 30 cycles of 95°C for 30 s, 55°C for 30 s, and 72°C for 1.5 min, and a final extension at 72°C for 10 min. Amplicons were visualized on a 1.5% agarose gel containing 0.5 μg/ml ethidium bromide.
The plasmid complementation was performed using conjugation as described previously [41]. The main obstacle for complementation was the absence of an antibiotic resistance marker in pRS218 which could have been used for subsequent selection. Therefore, pRS218 was first tagged with cat using the one step inactivation method [39]. Briefly, the cat was amplified using pKD3 plasmid and primers consisted of 36 nucleotides extensions at 5′ and 3′ ends of a putative noncoding region of pRS218 located between base pairs 591 and 831 in the plasmid sequence (Forward primer 5′-CGC CTT CGC GTT GCT CAG TTG TCC AAC CCC GGA AAC GTG TAG GCT GGA GCT GCT TC-3′ and reverse primer 5′-CTC CTC AAT ACT CAA ACA GGG ATC GTT TCG CAG AGG ACA TAT GAA TAT CCT CCT TAG-3′). Purified PCR

Bacterial growth curve
Bacteria were grown in LB broth at 37°C with shaking overnight. Cultures were diluted to 1:100 with LB broth, tissue culture medium or M9 medium with 10 μg/ml niacin and incubated at 37°C with shaking. Optical density at 600 nm (OD600) was taken in triplicate for every 20 min for 6 hrs. The OD values from each time point were averaged and graphed to obtain a growth curve.

In vitro invasion assay
Invasion assays were performed using hCMEC/D3 cells provided by Dr. Weksler B, Cornell University, NY. The hCMEC/D3 cells were grown in endothelial basal medium (Lonza, Walkersville, MD) containing 5% fetal bovine serum (PAA The Cell Culture Company, Piscataway, NJ), 1.4 μM hydrocortisone (Sigma-Aldrich, St. Louis, MO.), 5 μg.ml −1 acid ascorbic (Sigma), 1% chemically defined lipid concentrate (Gibco, Carlsbad, CA), 10 mM HEPES (PAA The Cell Culture Company), and 1 ng.ml −1 human basic fibroblast growth factor (Sigma), The invasion assay was performed as described previously [32]. Briefly, endothelial cells were seeded at about 1 × 10 5 cells per well in 12-well tissue culture plates (Corning Life Sciences, Manassas, VA.) coated with rat collagen (R&D Systems, Trevigen, Gaithersburg, MD) and incubated at 37°C with 5% CO 2 in a humid chamber. Once the monolayer was confluent, it was washed with phosphate-buffered saline (PBS, pH 7) and incubated with the cell culture medium containing bacteria at a multiplicity of infection (MOI) of 100 for 2 hrs at 37°C with 5% CO 2 to allow cellular invasion [32]. The extracellular bacteria were eliminated by incubation of the monolayers with a culture medium containing gentamicin (100 μg/ml) for 1 h. The monolayers were washed three times with PBS and lysed with 0.1% Triton X-100. The intracellular bacteria that were released during cell lysis were enumerated by plating on LB agar plates. Invasion frequencies were calculated by dividing the number of invaded bacteria by the initial inoculum and expressed as a percentage relative to the invasion frequency of wtRS218. The assays were performed three times in triplicate and student's t test was used to compare the groups.

Neonatal rat meningitis model
Five-day-old Sprague-Dawley out-bred rat pups (n = 10) were used in each experimental group. Rat pups were injected with approximately 200 CFU (range160 to 210 CFU) of E. coli (wtRS218 and RS218 cured ) by the intraperitoneal route. For the negative control group, PBS was injected intraperitoneally. Mortalities of rat pups in each group were monitored for 24 hrs post-inoculation. The pups that survived were euthanized 24 hrs postinoculation to collect blood, cerebrospinal fluid (CSF) and brain tissues. For bacterial enumeration, blood was collected by intra-cardiac puncture and plated on MacConkey agar to detect septicemia. Cerebrospinal fluid was collected by cisternal puncture, and plated on MacConkey agar to demonstrate meningitis. Brain tissues collected from each group were fixed in 10% neutral-buffered formalin, routinely processed for histopathology, stained with haematoxylin-eosin, and examined