Intra- and Interspecies Genomic Transfer of the Enterococcus faecalis Pathogenicity Island

Enterococci are the third leading cause of hospital associated infections and have gained increased importance due to their fast adaptation to the clinical environment by acquisition of antibiotic resistance and pathogenicity traits. Enterococcus faecalis harbours a pathogenicity island (PAI) of 153 kb containing several virulence factors including the enterococcal surface protein (esp). Until now only internal fragments of the PAI or larger chromosomal regions containing it have been transfered. Here we demonstrate precise excision, circularization and horizontal transfer of the entire PAI element from the chromosome of E. faecalis strain UW3114. This PAI (ca. 200 kb) contained some deletions and insertions as compared to the PAI of the reference strain MMH594, transferred precisely and integrated site-specifically into the chromosome of E. faecalis (intergenic region) and Enterococcus faecium (tRNAlys). The internal PAI structure was maintained after transfer. We assessed phenotypic changes accompanying acquisition of the PAI and expression of some of its determinants. The esp gene is expressed on the surface of donor and both transconjugants. Biofilm formation and cytolytic activity were enhanced in E. faecalis transconjugants after acquisition of the PAI. No differences in pathogenicity of E. faecalis were detected using a mouse bacteraemia and a mouse peritonitis models (tail vein and intraperitoneal injection). A 66 kb conjugative pheromone-responsive plasmid encoding erm(B) (pLG2) that was transferred in parallel with the PAI was sequenced. pLG2 is a pheromone responsive plasmid that probably promotes the PAI horizontal transfer, encodes antibiotic resistance features and contains complete replication and conjugation modules of enterococcal origin in a mosaic-like composition. The E. faecalis PAI can undergo precise intra- and interspecies transfer probably with the help of conjugative elements like conjugative resistance plasmids, supporting the role of horizontal gene transfer and antibiotic selective pressure in the successful establishment of certain enterococci as nosocomial pathogens.


Introduction
Enterococci are increasingly important nosocomial pathogens, known for being highly recombinant and for possessing several antibiotic resistance traits.These characteristics provide them with the capacity to adapt to clinical environments and modify their pathogenic properties thereby representing the possibility of acquisition and transfer of these genes from and to other pathogens.The Enterococcus faecalis Pathogenicity island (PAI) (153 kb) was first described by Shankar et al..It encodes several pathogenicity factors, among them the enterococcal surface protein (esp) conferring increased biofilm and colonization, a cytolysin with haemolytic, cytolytic and antibacterial activity, the aggregation substance, a bile acid hydrolase, surface proteins and general stress proteins [1].The E. faecalis PAI is widely distributed among isolates of different origins, clonal types and complexes, and it probably evolved by modular gain and loss of internal gene clusters [2].Contrasting the situation in Gram-negatives, where PAI structures are not modular and quite conserved, the E. faecalis PAI shows a highly variable gene content in isolates of different origins and regions [3,4,2,5].The mechanisms underlying the mobilization and transfer of PAIs are still not well understood, neither is the evolution of PAIs within a certain bacterial species.The E. faecalis PAI in MMH594 has a higher G + C content compared to the chromosome, contains phage-related integrase and excisionase genes, and 10 bp Direct Repeats (DR) at the flanking ends [1] that resemble the attL and attR of integrated DNA.These characteristics suggest that the PAI can be mobilized as an ICE [6,7].For ICEs mobilization, the excisionase-mediated homologous recombination between attL and attR produces excision and formation of circular intermediates (carrying attB), while a recombinase (integrase) catalyzes integration by recombination between attB and a chromosomal target sequence attP generating the junction sequences attL and attR [8].
The spontaneous excision from the chromosome and formation of circular intermediates are supposed to be prerequisites for the horizontal transfer of PAIs and have been observed in the PAIs of Vibrio [9,10], Salmonella [11], Yersinia pseudotuberculosis [12], Shigella flexneri and Escherichia coli [13,14].
Horizontal transfer of PAIs into recipient strains facilitated by phages has been described in VPI-1 [15] and SaPI-1 [16].Similarly, plasmids harbouring regions of three SaPIs, of the E. coli O26 locus of enterocyte effacement (LEE O26) and of shePAI demonstrated PAI integration into the chromosome of recipient strains [17,18,12], while SGI1 and the E. coli HPI have been shown to transfer with the help of plasmids [11,19].So far, only the Y. tuberculosis HPI has been observed to transfer spontaneously in a liquid medium without the help of a phages or plasmids [20].Previous attempts to transfer precisely the E. faecalis PAI have failed; only parts of the PAI or larger chromosomal elements containing the PAI have been observed to transfer [21,6,7].Here we confirm for the first time precise excision, circularization and horizontal transfer of the entire E. faecalis PAI and integration into the E. faecium and E. faecalis chromosomes in parallel with the conjugative transfer of the erm(B)encoding pheromone-plasmid pLG2.The acquisition of the PAI modulated certain pathogenic properties such as esp expression, cytolytic activity and biofilm formation.These findings support the role of horizontal gene transfer in the evolution of enterococci and transformation from commensal to pathogenic variants under antibiotic selective pressure.

Transfer of the E. faecalis PAI
The esp gene was transferred from E. faecalis donor strain UW3114 into E. faecalis and E. faecium recipients by filter mating conjugation along with the transfer of plasmid-encoded erythromycin resistance.
PCR analysis of overlapping regions covering the entire PAI (153.57kb as in reference strain MMH594) demonstrated that the transferred PAI of donor strain UW3114 was different to the PAI in the reference strain MMH594.However, no difference was observed between the donor and the transconjugants, indicating that the PAI did not undergo any changes during horizontal transfer (Figure 1, Table 1).Approximately 90.6 kb of the reference PAI -MMH594 were present in the PAI of UW3114 including the aggregation substance gene, the complete cytolysin operon and the entire esp gene.Long PCR bridging the gaps of the putative absent regions confirmed deletions located in regions: 2b-2c, 4b-5a and 9 (Figure 1, Table S1).No regular or long PCR amplification was possible that closed the gaps in regions 6a-6b and 7b-8b, suggesting that elements too large to be amplified were present (Figure 1).PCR at the flanking ends of the PAI confirmed exact presence of the PAI ends integrated into the chromosome of E. faecalis and E. faecium (see below).
The size of the transferred PAI element and its chromosomal integration were determined by Pulse field gel electrophoresis (PFGE) and Southern hybridization.SmaI -PFGE analysis revealed that a single restriction fragment in each transconjugant strain was enlarged and hybridised to the esp probe (Figure 2).The fragment size shift appeared to be ca.200 kb (calculated sizes were 206 kb in E. faecalis and 193 kb in E. faecium) and corresponded to the total size of the PAI transferred element.Given that only 90.6 kb of the reference PAI were confirmed to be present and that additional DNA seemed to be integrated within certain regions (see above), it is likely that additional ca.119.4 kb DNA were inserted within the PAI (Figure 1).S1-nuclease analysis demonstrated that the E. faecalis and E. faecium transconjugants acquired a ca.66 kb plasmid (pLG2) (Figure 2), which did not hybridize to the esp probe, but to the selection marker erm(B).I-CeuI macrorestriction analysis confirmed localization of esp on a chromosomal band in both transconjugant strains (data not presented).

Integration site
The integration site of the E. faecalis PAI in the chromosome of E. faecalis strain OG1RF (NZ ABPI01000001:1) is at coordinates 374617:374626 (Figure 3).This integration site was found by comparing the OG1RF genome to the chromosomal region spanning the E. faecalis PAI in the genome of E. faecalis strain V583 (gbDAE016830:1D) (PAI coordinates, 445126:583433).PCR amplification of the integration site of the PAI prior to its insertion confirmed that it is intact in the recipient strain OG1RF and occupied by the PAI in the transconjugant OG1RFxUW3114 T-12.The region where the PAI integrated into the E. faecium transconjugant 64/3xUW3114 T-10 genome was identified to be a tRNAlys gene, which is located in the E. faecium U0317 contig00182 (gbDABSW 01000168:1D) at coordinates 47388:47460 (Figure 3).PCR amplification confirmed that the identified integration site tRNAlys was intact in the recipient strain 64/3 and occupied by the E. faecalis PAI in the transconjugant 64/ 3xUW3114 T-1 (Figure 4).
BLAST comparisons of the E. faecalis and the E. faecium regions where the E. faecalis PAI integrated revealed that they share 85% similarity at the nucleotide level along a 94 and 93 bp region respectively.A 10 bp (AATTCTCAGT) sequence (resembling attB) was found to be present at the PAI integration site of the E. faecium and E. faecalis recipient strains.PCR and sequencing of the flanking ends of the PAI and chromosomal regions spanning it demonstrated the exact integration of the PAI and confirmed the presence of the two 10 bp DR (resembling attl and attR).
The region where the E. faecium putative esp PAI is located in E. faecium (E.faecium U0317 Contig 00248 (gbDABSW 01000226:1D) [22]) was amplified by PCR and confirmed to be free in the recipient (64/3) and transconjugant (64/3xUW3114 T-10) strains, indicating that neither the E. faecium nor the E. faecalis PAI are integrated in this chromosomal region in the E. faecium donor and transconjugant strains.

Excision and circularization
Presence of a circular intermediate formed from the precisely excised PAI element could be detected by nested PCR in the donor strain UW3114.Sequencing of these amplicons confirmed that the PAI can precisely excise at its flanking ends and that these ends join together, resulting in circularization of the PAI element (See Fig. 4).The precisely excised circular intermediate carried the specific 10 bp sequence (AATTCTCAGT) resembling attP.Accordingly, excision of the PAI from the chromosome could be confirmed by amplification of the free chromosomal region that contained the PAI in the E. faecalis donor strain UW3114.Excision from the chromosome could also be detected only after two rounds of nested PCR.In addition to precise excision and circularization of the PAI (only detected in donor UW3114), imprecise PAI excision and circularization was also observed.PCR and sequencing demonstrated that different fragments of the PAI can remain in the chromosome after excision, or chromosomal fragments can be excised together with the PAI.The following two scenarios of imprecise excision were the most frequently detected: (a) the fragment 1:223 bp of PAI remains in the chromosome after PAI excision and (b) the fragment 1:1704 bp of PAI remains in the chromosome after PAI excision (Figure 4).
Case (a) was detected in donor strain UW3114, while case (b) was seen in reference strain MMH594, donor strain UW3114 and transconjugants 64/3xUW3114 T-10 and OG1RFxUW3114 T-12.These events of imprecise excision could be triggered by the similarity between internal regions of the PAI at this positions and the chromosomal region flanking downstream the PAI.

Phenotypic changes after PAI acquisition
The acquisition of the PAI changed some phenotypic properties of the E. faecium and E. faecalis strains (Table 2).Presence of Esp protein on the cell surface, biofilm formation, cytolytic activity and growth in vitro were assessed.Pathogenicity and in vivo survival of the E. faecalis strains was compared using mouse bacteraemia and peritonitis models.
Cell surface esp expression.cTransmission immunoelectron microscopy displayed the Esp protein associated with the cell wall of the donor and transconjugant strains but not on the surfaces of the recipients (Figure 5).FACS analysis revealed no Esp on the cell surface of the recipient strains.Enhanced expression in the E. faecalis transconjugant strain OG1RFxUW3114 T-12 (p,0.05) and in the E. faecium transconjugant 64/3xUW3114 T-10 (p,0.05),compared to their corresponding recipient strains, was confirmed using one way ANOVA and Tukey's post-test.The E. faecium transconjugant 64/3xUW3114 T-10 showed a slightly lower esp expression than the E. faecalis strain (p.0.05) (Figure 6).All strains had a lower esp expression at 21uC in comparison to that at 37uC (p = 0.0076), as confirmed using a two way ANOVA comparisson (Figure 6).
Biofilm formation.The E. faecalis transconjugant strain OG1RFxUW3114 T-12 showed a significant 2-fold increase in biofilm forming capacity compared to its homologous strain, the recipient OG1RF (p,0.05).In the E. faecium strains both the recipient and the transconjugants showed a similar level of low biofilm formation (p.0.05) (Figure 7).Statistical significance was evaluated using One way ANOVA with Tukey's post test.
Cytolysin/haemolysin. Both the reference (MMH594) and the donor (UW3114) strains showed haemolytic activity while none of the recipient strains were haemolytic.The E. faecalis transconjugant developed a strong beta-haemolysis after PAI acquisition, while the PAI positive E. faecium transconjugant remained non-haemolytic.Since the cytolysin operon can also be plasmid encoded, haemolysis was also tested in transconjugants carrying the erm(B) plasmid pLG2 but lacking the PAI (see Table 2) and were confirmed non-haemolytic.Furthermore, sequencing of pLG2 did not reveal the presence of any haemolysin or cytolysin genes (see below).Consequently the acquired haemolytic activity of E. faecalis transconjugants could be linked exclusively to the cytolysin operon present within the PAI.
Animal experiments.Mouse bacteraemia and peritonitis models were used to compare the pathogenic potential of E. faecalis isogenic strains differing only by presence of the PAI.The bacterial loads in blood and kidneys after 24 h of infection were compared.Kruskal-Wallis analysis indicated that the distributions of the data did not differ significantly, excepting in the bacterial counts of kidneys in the peritonitis model (Peritonitis: blood p = 0.3794, kidney = 0.0327; Bacteraemia: blood p = 0.094, kidney p = 0.088).However Dunn's multiple comparisson post-test indicated no significant differences when comparing the bacterial counts of each strain, meaning that the pathogenic potential of the PAI-positive and PAI negative transconjugant strains was the same (Figure 8).
In vitro Growth.Growth of the recipient strains and the transconjugant strains lacking and carrying the PAI was compared in order to determine differences in growth as a consequence of the altered genome size (ca.266 kb larger genome in PAI-positive strains).There was no significant difference between the growth of the recipient and the transconjugant strains both in E. faecium (p = 0.9909) as in E. faecalis (p = 0.9329), as demonstrated by one way ANOVA and Tukey's post test analysis (Figure 9).

Plasmid sequencing
The plasmid pLG2 was sequenced in order to investigate the presence of gene clusters that might be involved in conjugation and could have supported the horizontal transfer of the E. faecalis PAI.The sequences of pLG2 can be found in GeneBank under the accession numbers HQ426650 to HQ426665.454-sequencing of pLG2 yielded a total of 46 contigs.The largest contig was 33,447 bp long; 17 contigs were larger than 500 bp and covered 53.2 kb.The total number of aligned bases was 62.6 kb, very close to the 66 kb calculated plasmid size.Analysis of the plasmid sequences showed that pLG2 does not correspond entirely to any previously described enterococcal plasmid.Similarities to fragments of different enterococcal plasmids (pAD1, pTEF1, pTEF2, pAM373, pVEF1, pVEF2, pEF1, pIP816, pRE25) were discovered, especially at the replication and mobilization related regions suggesting that pLG2 has a modular structure.Complete replication and mobilization modules including an oriT suggest that pLG2 is a conjugative plasmid, theoretically capable of triggering the transfer of the E. faecalis PAI.The finding of a pheromone binding protein and the structure of the replication module suggested that pLG2 is a pheromone responding plasmid.Determinants encoding antibiotic resistance to erythromycin, streptomycin and streptothricin were also located within the plasmid sequence (Table 3).
We identified a replicase gene, repR-pLG2 (pLG2-0001) that shared 100% identity at the nucleotide level to the replicase genes of E. faecium plasmids p5753cB (GQ900487.1),pIP816 (AM932524.1), Presence of the entire PAI was investigated based on reference strain MMH594 [1]) in the donor and transconjugant strains PAI overlapping regions were arbitrarly chosen to allow long-PCR amplification as shown in Table S1 and Figure 1.Recipient strains were also tested and were negative for all PCRs.* A product of unexpected size was amplified.ins an insertion within this region(s) is suspected.del a deletion within this region(s) was confirmed.doi:10.1371/journal.pone.0016720.t001 Table 1.Cont.homologous region described in pRE25.This region contains a putative replication origin oriR, up to five DnaA boxes and several inverted repeats [23].With regard to the segregation machinery, we identified a putative PrgP-PrgO partitioning system, upstream and inverted to the replicase gene.No ORF that exhibits homology to a relaxase could be identified.The replication and segregation elements identified in pLG2 are listed in Table 3.
All the elements necessary for plasmid mobilization in Grampositives were identified in pLG2 as can be seen in Table 3.The region between coordinates 2783:1091 of contig 8 shared 98% similarity at the nucleotide level along 1.6 kb of the 4.8 kb mobilization region of pEF1 (e = 0.0).A resolvase (pLG2-0012) and a ''Filamentation induced by cAMP protein-Fic'' (pLG2-0013), homologous to pEF1 MobC are encoded in this region.
Within the same contig, an origin of transfer (oriT) was identified based on the 99% similarity (expect = 0.0) to the mobilization region of pAD1 where an oriT has been previously identified between ORF53 and ORF57 (AF343837) [24].Other genes identified that are related to plasmid mobilization are listed in Table 3.
Other mobilization related genes of pLG2 include a putative single strand binding protein gene (pLG2-0026) that could act as a coupling protein that binds the DNA for substrate presentation during transfer, two peptidase putative genes (pLG2-0058 and  pLG2-0059), which might degrade the bacterial cell wall for the formation of a transfer canal, and two putative resolvase genes (pLG2-0012 and 38) that could be involved in plasmid segregation by resolving plasmid dimers.

Discussion
The E. faecalis PAI is flanked by 10 bp DR, and the first two genes encoded within the PAI are a putative phage-related integrase and an excisionase [1], suggesting that it can excise from and integrate into the chromosome by homologous recombination and can be transferred as a single entity [6].Our results demonstrate for the first time that the entire E. faecalis PAI is capable of precise excision, circularization and horizontal intraand interspecies transfer.We describe chromosome-to-chromosome transfer and site-specific integration of the E. faecalis PAI into the chromosome of E. faecalis and E. faecium.
The E. faecalis strain used as donor, UW3114, had been successfully used in previous studies to demonstrate the intraspecies transfer of the esp gene into E. faecalis [21], however, transfer of further fragments of the PAI could not be investigated in detail because recipient strain E. faecalis JH2-2 already possesses parts of the PAI.Coburn et al. have described excision of a 27,744 bp internal fragment of the PAI, circularization, integration into a pTEF1-like plasmid and transfer into an E. faecalis recipient strain [6].Recently Manson et al. have shown that the E. faecalis PAI can transfer into E. faecalis, not precisely but as part of larger chromosomal fragments.The horizontal transfer of such regions requires the help of either pTEF1 or pTEF2 pheromoneresponsive plasmids and is recA-dependent, while the integrase and excisionase genes present in the PAI were not required [7].However, a role of the integrase and excisionase genes for the precise excision and integration of the PAI cannot be ruled out.
van Shaik et al. have recently indicated that some hospital strains of E. faecium carry the esp gene in a PAI-like structure different from the E. faecalis PAI.The E. faecium esp PAI can be transferred horizontally among E. faecium and contains a 10 kb fragment identical to a gene cluster in the E. faecalis PAI that suggests a recent intra-genus transfer or acquisition of the PAIs from a third source [22].
Our analysis of the PAI structure in the donor and the transconjugants demonstrated that it did not undergo any changes during transfer.Compared to the originally described PAI of strain MMH594, the transferred PAI structure contained some deletions and insertions (Figure 1).
The differences observed along the transferred PAI compared to the prototype PAI of strain MMH594 agrees with the previous observation that the gene content of the E. faecalis PAI is highly variable and PAI subtypes are widely spread [2,5,3].The clusterlike variability in gene composition and presence of sequences related to mobile genetic elements within the PAI suggest that it has evolved by initial spread of some core elements, and later recombination, acquisition, and loss of gene clusters [2,3].
The E. faecalis PAI is integrated into an non-coding region flanked by an ORF specifying a hypothetical protein with no predictable function and a putative oxidoreductase [1].The E. faecalis PAI of the E. faecalis donor strain UW3114 was located into the same region and it also integrated into this region in the recipient strain OG1RF, suggesting a high site-specific integration.In E. faecium transconjugant strain 64/3xUW3114 T-10 the E. faecalis PAI integrated into a tRNAlys gene.This integration site of the E. faecalis PAI in E. faecium is distinct from the region where the esp -encoding PAI of E. faecium (AY322150) is integrated in E. faecium [25,22].
The regions of integration of the E. faecalis PAI in E. faecium and in E. faecalis shared 85% similarity at the nucleotide level along a ca.95 bp region.This homology presents the possibility that larger regions act as the recognition sequences for PAI integration, although the 10 bp sequence (attP-like) should be the effective site for homologous recombination, since it resulted in DR (attL-and attR-like) at both ends after PAI chromosomal integration.It seems evident that the horizontal transfer event reported here resembles that of ICEs.
It is known that due to the short stretches of DNA sequence similarity necessary to initiate recombination events, circular DNA molecules are more likely to be transferred and acquired, making them the most promiscuous recombinogenic states of all DNA [26].On the other hand circularized intermediates of other PAIs have been seen and associated with their transfer [27]   from the chromosome and for the formation of circular intermediates.
Besides confirming that the PAI can precisely excise and circularize (only observed in the donor strain UW3114), imprecise excision and circularization were also detected (Fig. 4) apparently due to homology between internal regions of the PAI and flanking chromosomal regions.The circular intermediates of the PAI lacking the integrase and excisionase genes might be unable to integrate/excise in a recipient strain.
The fact that precise excision and circularization could only be detected in the donor strains but not in the reference strain MMH594 might explain why previous attempts to transfer the entire PAI element from this strain have not been successful [6].
Esp protein was detected on the surface of all PAI-positive strains, although FACS analysis revealed a lower esp expression in the E. faecium transconjugant than in E. faecalis (Figure 6).It is possible that the expression of genes encoded within the PAI are affected by regulatory factors inherent to the cell or speciesspecific.esp expression was diminished when cells were grown at 21uC, as previously described for esp in E. faecium [28].Temperature-dependent regulation of expression is an ecological feature for regulation of pathogenicity factors also observed in other pathogens that alternate between an environmental reservoir and a mammalian host [28].Our results show that the temperature-dependent expression of esp is also a feature of E. faecalis (Figure 6).
The acquisition of the E. faecalis PAI increased in vitro biofilm formation in E. faecalis.However, the acquisition of E. faecalis PAI was not sufficient to develop a biofilm forming phenotype in E. faecium.The nearly absent biofilm formation of the recipient strain, together with the low Esp on the cell surface, might explain the low biofilm formation in the E. faecium PAI-positive strain.Additionally, it is well-known that E. faecium has low biofilm forming capacity compared to E. faecalis [29,30], and it has been demonstrated that although esp can increase the biofilm forming capacity in E. faecium and E. faecalis [31,32,28,33], it is neither sufficient nor absolutely necessary for biofilm formation [34,35].
The cytolysin/haemolysin cyl operon encodes a bacterial toxin expressed by some strains of E. faecalis which in addition to mediating lysis of erythrocytes, also possesses antibacterial activity toward a broad range of Gram-positive bacteria [36].The cyl operon is either encoded within large, pheromone-responsive plasmids or on the chromosome within the E. faecalis PAI [37].Our PCR results showed that the complete and intact cyl operon was present in the transferred E. faecalis PAI element.However, a strong (beta) haemolytic activity could only be detected in E. faecalis strains.To our knowledge this is the first description of transfer of the cyl operon and cyl genes to E. faecium.Previous studies have reported complete absence of cyl genes (n = 271) among E. faecium [38] as well as absence of both cyl genes and haemolytic activity (n = 21) [39].Vancanneyt et al. have reported beta haemolytic activity in 5 E. faecium isolates (n = 78) but no link could be established between haemolytic activity and gene presence [40].

It remains unclear what prevents the expression or activity of the E. faecalis cytolysin/haemolysin in E. faecium.
Transfer of the PAI and expression of pathogenicity-associated factors encoded within it prompted us to compare the pathogenicity of the E. faecalis isogenic strains carrying and lacking the PAI.Several pathogenicity factors encoded within the E. faecalis PAI are demonstrated to contribute to the virulence of E. faecalis in different animal models.These include Esp, cytolysin/haemolysin, and aggregation substance (AS), Asc10 [47,37].The mouse bacteraemia and peritonitis models did not show a significant difference between isogenic PAI-positive and PAI-negative E. faecalis strains (Figure 8).The supposed increase in pathogenicity in PAI-positive transconjugants could be compensated by an impaired growth associated with the enlarged genome size [48], altough the observed diminihsed growth of the E. faecalis PAIpositive transconjugant was not significant (Fig. 9).However, enterococci exhibit virulence through mechanisms such as adhesion and biofilm formation, translocation through eukaryotic cell layers, or release of proinflammatory molecules, and the absence of an observable effect in the models used does not exclude an increased pathogenicity of the transconjugants.
pLG2, the erm(B) -plasmid transferred along with the PAI, carries all the elements necessary for replication including repR-pLG2 and oriR and segregation locus (PrgP-PrgO).Although a replicase and an origin of replication homologous to pEF1 are present in pLG2, no relaxase could be identified, as has also been described in other pheromone-responsive plasmids (i.e.pAD1, pCF10 and pAM373).Ruiz Barba et al. suggest that pEF1 replicates using the oriR and repR using the h-type replication mechanism [23], which could also be true for pLG2 given the homology of its replicase to that of pEF1.However, the mechanism of replication of pheromone-responsive plasmids is still unknown [49].The PrgP-PrgO partition system of pLG2 (segE locus) constitutes a putative ParAB-like partition system [50] and is putatively involved in replication as it has been observed in pCF10 and pRE25 [51,52].The two genes are upstream and inverted with respect to repR-pLG2 and constitute another characteristic of pheromone-responsive plasmids.A similar organization of these partition protein genes, normally known as repBC, has been previously described in pCF10, pAD1, pPD1, pAM373 and pRE25 [51].
The conjugative transfer of Gram-positive plasmids requires elements that provide contact between the mating cells, the formation of a canal to cross the cell wall, and the transfer of the plasmid DNA.Homologues to all these elements were found in the DNA sequence of pLG2.
One of the mobilization-related regions of pLG2 is homologous to that of bacteriocin-encoding plasmid pEF1 and the other to that of the RepA_N-family plasmid pAD1.Meanwhile, the replicase gene identified was a RepR, which is unrelated to the RepA_Nplasmid family.This suggests a modular composition of pLG2, which is a common trait of enterococcal and other plasmids [53,54].Phylogenetic analyses have shown that the repA replication gene and repBC loci are separable modules that do not need to be closely associated [49].
A recent report of Manson et al. describes that pTEF1 and pTEF2 plasmid transfer functions (i.e.oriT, relaxase and TraG) were essential for the transfer of chromosomal DNA among E. faecalis, and its integration into the recipient's chromosome was recA dependent.BLAST comparisons revealed that pLG2 is different pTEF1 and pTEF2.However, the mobilization region bearing the oriT in pLG2, as well as that of pTEF1 is homologous to the oriT region of pAD1.Despite this congruence, they contain different replication regions, and neither a relaxase nor TraG were detected in pLG2 sequences.
Although not formally proven in the present study by functional knockout experiments, it seems that the pLG2 plasmid resembling conjugative E. faecalis pheromone plasmids can support horizontal PAI transfer by providing the conjugative modules necessary for cell-to-cell transfer.

Materials and Methods
E. faecalis strain UW3114 harbouring a variant of the E. faecalis PAI which contains the esp gene, and an erythromycin resistance plasmid (pLG2), was used as donor.E. faecalis strain OG1RF, and E. faecium strain 64/3 lacking the E. faecalis PAI were used as recipients.Filter mating experiments were done as described elsewhere [55].Transconjugants were selected on Brain Hearth Infusion (BHI) agar (Difco Labs., Detroit, USA) containing the selective marker for the donor (Erythromycin 10mg/L) and those for the recipient (rifampicin 30mg/L and fusidic acid 20mg/L).Presence of the esp gene in the transconjugants, tested by PCR, was indicative of horizontal transfer of the PAI.The strains used are listed in Table 2.

PCR and long template PCR
The presence of the esp gene was tested by PCR using the primers esp1 and esp2 as previously described [21] using Ilustra PuReTaq Ready-To-Go PCR Beads (GE Healthcare Europe GmbH).Primers binding outside the integration site of the PAI PAI164 and PAI167UW3114 or PAI164Efm or PAI167Efm were used to amplify the unoccupied integration site in the E. faecalis and E. faecium recipient strains.Primers EF2 and ER1 binding inside the PAI (reading outwards the PAI) were used with primers binding outside of the PAI (reading into the PAI) to demonstrate the occupation of the integration site by the PAI in the transconjugants (Table 4) (Figure 4).Presence of the whole E. faecalis PAI, based on the sequence of the E. faecalis PAI from strain MMH594 [1], was tested by amplification of overlapping regions of ca. 10 kb using the Expand long template PCR kit (Roche Biochemicals, Mannheim, Germany) under the conditions recommended by the manufacturer.
Regions yielding unexpected size or no PCR amplification product were further analysed by regular PCR of each predicted ORF present within these regions (Primers are listed in Table S2, Table S1).If PCR results suggested that some regions were absent, first regular and then Long template PCRs were performed that would bridge the resulting gap having always confirmed that the binding sites of the primers used were present.
All PCR products were completely or partially sequenced according to the manufacturer recommendations for cycle sequencing of PCR products (Applied Biosystems Germany, Darmstadt, Germany).
Excision and circularization of the E. faecalis PAI were tested by two-stage nested PCR.The first PCR was done using the Expand long template PCR kit, with 5min.elongation time.Nested PCR was done using Ilustra PuReTaq Ready-To-Go PCR Beads.
Presence of circular intermediates was investigated with primers TSP2 59 and TSP2 39 in the first PCR.For the nested PCR, 1ml of PCR product was used for amplification of an inner region thereof using primers TSP3 39 and TSP3 59.To investigate the excision of the PAI from the chromosome of E. faecalis, primers PAI164 and PAI167 UW3114 were used for the first round of PCR and primers PAI164nested and PAI167nested for the second (Figure 4 and Table 4).

PFGE and Southern hybridization
Preparation of samples for subsequent macrorestriction analysis was done as previously described [55].PFGE and analysis of the resulting band patterns were done as described elsewhere [56].A CHEFF III apparatus (BIO-RAD, Munich, Germany) was used.The sizes of the resolved macrorestriction fragments were predicted according to an external size standard (SmaI -digested Staphylococcus aureus NCTC8325) and calculated using BioNumerics v. 6.0 software (Applied Maths, Sint-Martens-Latem, Belgium) [57].Southern hybridization and immunological detection were done using DIG system kits and CDP-Star detection following the manufacturer's recommendations (Roche).Labelled probes were generated using DIG-labelled dUTP (Roche Biochemicals, Mannheim, Germany) using primers esp1 and esp2 for esp probes, and primers ermB1 and ermB2 for erm(B) probes.
I-CeuI macrorestriction.I-CeuI cuts inside ribosomal operons and thus linearizes chromosomal DNA.Chromosomal location of esp was analysed by Southern hybridization of I-CeuI digested genomic DNA resolved in PFGE as previously described [58].
S1-nuclease macrorestriction.Plasmid content of the strains and possible plasmid localization of esp and erm(B) were analysed by S1-nuclease treatment and subsequent Southern hybridization analysis [59,60].Briefly, genomic DNA was digested with 14U of S1 nuclease (Takara, Bio Inc., Shiga, Japan)for 15 minutes at 37uC.DNA was separated in PFGE using ramped pulse field times as follows: 5-35 seconds for 22 hours at 14uC.This method is based on the specific digestion of single-stranded DNA or single-stranded regions of double-stranded DNA allowing the linearization of plasmids and resolution in PFGE as detectable bands on a faint genomic background [59].

Genomic walking
The integration site of the PAI in E. faecium was determined using a DNA Walking Speedup Premix kit (Seegene, Seoul, Korea).Specific primers: TSP1 59, TSP2 59 and TSP3 59, tagging the 59end of the PAI and TSP1 39, TSP2 39, TSP3 39 tagging the 39 end of E. faecalis PAI (Table 4 and Figure 4) were designed and used according to the recommendations of the manufacturer.2).
Cell surface expression of esp.Transmission electron microscopy (TEM) and Fluorescence activated cell sorting (FACS) were used for detection of Esp on the bacterial surface as previously described [28,61].
Esp expressed on the cell surface was visualized after immunologic labeling with gold particles.Briefly 200-mesh Formvar-carbon-coated copper grids were floated on drops of bacterial suspensions (1|10 9 CFU/ml) for 10 min.. Grids were washed and blocked with 1% bovine serum albumin in PBS (PBSb).Esp was labelled by floating the grids for 1 hour on drops containing a 1/100 dilution of anti-Esp rabbit immune serum in PBSb.Antibodies were labelled with protein A-Gold (10 nm) in PBS.Grids were washed four times, fixed, and bacteria were stained with 1.8% methylcellulose (25 centipoises -Sigma-Aldrich) and 0.4% uranyl acetate (pH 4) and subsequently air dried for 10 min.. Grids were examined using a Jeol 1010 transmission electron microscope (Jeol-Europe, Amsterdam, The Netherlands) [28].
For FACS aanalysis 1|10 8 bacterial cells were incubated for 30 minutes in 50 ml RPMI 1640 supplemented with 0.05% human serum albumin (HAS) containing 1:100 anti Esp-rabbit immune serum.After washing, cells were incubated for additional 30 min. in 50ml RPMI 1640-HAS containing 1:50 goat anti-rabbit flourescein isothiocyanate (Sigma-Aldrich, Saint Louis, USA).Bacteria were washed and resuspended in 300 ml RPMI 1640-HSA before analyses in a FACSCalibur (BD, Alphen aan den Rijn, The Netherlands).All measurements were performed with the same apparatus using the same parameters.The data were normalized for bacterial size, and experiments were performed in triplicate.The mean fluorescence (channel 1) was used as a measure for cell surface-associated Esp.Samples incubated without anti-Esp rabbit immune serum were used as conjugate controls [28].
Biofilm formation.In vitro biofilm formation was evaluated on polystyrene microtiter plates following the methodology described previously [33] with some modifications.Briefly, bacteria were grown on blood agar plates and resuspended in Tryptic Soy broth supplemented with 0.25% glucose to a concentration of 5|10 8 CFU/ml.100ml bacterial suspension was added in triplicate per well using Flat bottomed 96-well polystyrene microtiter plates (Greiner Bio-one, Germany and Corning Inc., NY, USA).After incubation at 37uC for 24 hours, bacterial suspensions were removed and the wells were washed three times with 200ml of PBS.The plates were dried for 1 h at 60uC and stained with 100ml of 0.2% Gram's crystal violet solution (Merk, Darmstadt, Germany) for 15 minutes at room temperature.The dye was removed and the wells washed again with 200ml PBS three times.The plates were dried for 10 min at 60uC.The dye was resuspended for 30 minutes by adding 100 ml of ethanol 100% into each well.Absorbance at 595 nm was measured with a Sunrise ELISA reader (Tecan Group,Maennedorf, Switzerland).The experiment was repeated 3 times.

Figure 1 .
Figure 1.The E. faecalis PAI of strain UW3114 that was horizontally transfered.The structure was investigated based on the E. faecalis PAI of strain MMH594 (gbDAF 454824:1D), by long template PCR, regular PCR, sequencing and Southern hybridizations.The long template PCR-regions are indicated by horizontal lines (1 to 9) (Table S1).Regions present are shown in black and sum up to ca. 90.6 kb.White regions are absent.Grey regions are absent and hold additional insertions.Only representative ORFs from the regions present are shown.doi:10.1371/journal.pone.0016720.g001

Figure 3 .
Figure 3. Chromosomal region where the E. faecalis PAI integrated in E. faecium as in strain U0317 gbDABSW01000168:1D (top) and E. faecalis OG1RF ref DNZ A BPI00000000D (bottom).The vertical white arrow points at the integration site of the E. faecalis PAI.The detailed nucleotide sequence shows the 10 bp region of recognition and integration of the PAI as underlined bold text.doi:10.1371/journal.pone.0016720.g003

Figure 4 .
Figure 4. Representation of the E. faecalis PAI showing the primers used in this study for esp detection and investigation of PAI excision, circularization and integration (not to scale).Arrowheads represent the primers: white arrowheads are primers used to investigate excision from the chromosome, and grey arrowheads are primers used to detect circular intermediates.Solid black region represents chromosomal DNA, grey region represents the PAI.The junction-nucleotide sequences after PAI integration in the chromosome of E. faecalis are shown in detail; the DR appear in underlined bold text.Dashed lines indicate joining of PAI regions during formation of circular intermediates; A and B and precise excision indicate the different events of excision and circularization of the PAI that were confirmed in this study.doi:10.1371/journal.pone.0016720.g004 [9-10] [17].These observations prompted us to test for the excision of the PAI

Figure 6 .Figure 7 .
Figure 6.Differential expression of cell wall-associated Esp at 376C and 216C measured by flow cytometry (FACS) of anti-Esp labelled cells.Strains UW3114 (donor) and MMH594 (reference) are shown for comparison.64/3 and OG1RF are the recipient strains.Strains 64/3 T-10 (64/3xUW3114 T-10) and OG1RF T-12 (OG1RFxUW3114 T-12) are the transconjugants that acquired the E. faecalis PAI from strain UW3114 (See also Table 2).Notice the enhancement of esp expression in the transconjugant strains compared to the corresponding recipient (p,0.05), and decreased esp expression at 21uC(p = 0.0076).Control is a conjugate control.The average of the mean values from two different experiments are presented, error bars denote standard deviation.doi:10.1371/journal.pone.0016720.g006

Table 1 .
Results of Long template PCRs and regular PCRs screening for presence of the entire E. faecalis PAI.

Table 3 .
Annotation of complete identified ORF in the sequencing contigs of pLG2.