Microevolutionary Events Involving Narrow Host Plasmids Influences Local Fixation of Vancomycin-Resistance in Enterococcus Populations

Vancomycin-resistance in enterococci (VRE) is associated with isolates within ST18, ST17, ST78 Enterococcus faecium (Efm) and ST6 Enterococcus faecalis (Efs) human adapted lineages. Despite of its global spread, vancomycin resistance rates in enterococcal populations greatly vary temporally and geographically. Portugal is one of the European countries where Tn1546 (vanA) is consistently found in a variety of environments. A comprehensive multi-hierarchical analysis of VRE isolates (75 Efm and 29 Efs) from Portuguese hospitals and aquatic surroundings (1996–2008) was performed to clarify the local dynamics of VRE. Clonal relatedness was established by PFGE and MLST while plasmid characterization comprised the analysis of known relaxases, rep initiator proteins and toxin-antitoxin systems (TA) by PCR-based typing schemes, RFLP comparison, hybridization and sequencing. Tn1546 variants were characterized by PCR overlapping/sequencing. Intra- and inter-hospital dissemination of Efm ST18, ST132 and ST280 and Efs ST6 clones, carrying rolling-circle (pEFNP1/pRI1) and theta-replicating (pCIZ2-like, Inc18, pHTβ-like, two pRUM-variants, pLG1-like, and pheromone-responsive) plasmids was documented. Tn1546 variants, mostly containing ISEf1 or IS1216, were located on plasmids (30–150 kb) with a high degree of mosaicism and heterogeneous RFLP patterns that seem to have resulted from the interplay between broad host Inc18 plasmids (pIP501, pRE25, pEF1), and narrow host RepA_N plasmids (pRUM, pAD1-like). TAs of Inc18 (ω-ε-ζ) and pRUM (Axe-Txe) plasmids were infrequently detected. Some plasmid chimeras were persistently recovered over years from different clonal lineages. This work represents the first multi-hierarchical analysis of VRE, revealing a frequent recombinatorial diversification of a limited number of interacting clonal backgrounds, plasmids and transposons at local scale. These interactions provide a continuous process of parapatric clonalization driving a full exploration of the local adaptive landscape, which might assure long-term maintenance of resistant clones and eventually fixation of Tn1546 in particular geographic areas.


Introduction
Since its first description in the late 80's, vancomycin-resistant enterococci (VRE) have been increasingly reported worldwide, but presenting remarkable geographical and temporal differences in local rates (http://www.cddep.org/ResistanceMap/bug-drug/ EFa-VC) [1][2][3]. Vancomycin-resistant Enterococcus faecium (VREfm) became endemic in most North American hospitals since the mid 909s [1,2,[4][5][6] while their overall occurrence in Europe remained low until recently, when VRE nosocomial outbreaks started to be increasingly reported in some European countries (Annual Report of the European Antimicrobial Resistance Surveillance Network, EARS-Net, 2009) [1,3,7,8]. Despite E. faecium (Efm) being less frequently found than Enterococcus faecalis (Efs) in clinical isolates, it is far more frequently resistant to vancomycin, one of the last-line intravenous antibiotic resources for therapy. However, although the rate of vancomycin-resistant E. faecalis (VREfs) has remained low, they are steadily increasing in both the US and in EU countries (http://www.cddep.org/ResistanceMap/bug-drug/ EFe-VC) [3].
Recent analysis of enterococcal populations in the clinical setting depicts a rugged epidemiological profile, with successive waves of isolates causing infections, which belong to specific lineages of E. faecium (ST17, ST18 and ST78, previously considered within the same clonal complex (CC) 17), and E. faecalis (ST6, ST40) [19][20][21]. However, regional differences in the rates of VRE cannot be only explained by clonal replacement dynamics as suggested for other pathogens [22][23][24].
The aim of this study was to address the dynamics of vancomycin resistance among enterococci in Portugal, one of the developed countries with higher rates of both VREfm (21-23%) and VREfs (1.8-4.1%) (www.earss.rivm.nl; http://www. cddep.org/ResistanceMap/bug-drug/EFe-VC), and where VanA is prevalent over VanB [3,[25][26][27], by analyzing the clonal and plasmid backgrounds influencing the spread and persistence of Tn1546. Our study suggests that clonalization, the local selection of distinct clonal variants giving rise to durable bacterial lineages, might result and be modified by the local spread and recombinatorial dynamics of mobile genetic elements, thus providing new clues about the local multi-hierarchical evolutionary biology of vancomycin resistance.

Local dynamic landscape drives the spread and fixation of vancomycin resistance in Portuguese hospitals
We have determined that the enterococcal population from the Portuguese hospitals is formed by an ensemble of MLST/PFGE clones. Efm isolates fit in three out of six phylogenomic groups recently established by using Bayesian Analysis of Population Structure (BAPS), namely BAPS groups 2, 3 and 5 [19] (Figure 1). Most of the isolates cluster into the predominant BAPS group 3 [subgroup 3-3 comprising main human lineages ST18 (ST18 and ST132) and ST17 (ST16); and subgroup 3-1 comprising ST280], and the BAPS group 2 (including ST80 and ST656/ST78 lineage, ST5/CC5, ST190/CC9), which have been previously associated with isolates from humans and both animals and humans, respectively [10,19,25,[28][29][30]. A number of clones cluster in the small Efm BAPS group 5 (ST366, ST367, ST369), which seems to comprise mosaic genomes [19]. Isolates of Efs belong to ST6/ CC2, ST30, ST55, ST117, and ST159 lineages although, to the date of this publication, Efs population has not been clustered in different BAPS groups. Among all them, isolates within ST18 Efm and ST6 Efs lineages were predominant, in line with the intra-and interhospital spread of particular highly transmissible Efm and Efs clones recovered in Portuguese hospitals since the late 90s [22,25,31,32]. While ST6 Efs was widely disseminated in all hospitals analyzed in this country [26], specific Efm lineages were overrepresented in Coimbra (ST18) and Oporto (ST132, a single locus variant, SLV, of ST18). Strains belonging to ST18 (showing PFGE types H70 and H78), ST132 (PFGE type H88) and ST280 (with PFGE types 71 and H100) were spread in different hospitals ( Figure 1 and Figure 2).
It is worthwhile to note the possible relatedness between isolates of different STs (Figure 1 and 2). They include some isolates linked to BAPS 3-3 subgroup as ST18, ST80, ST125, ST132, ST368, ST369, all SLVs of each other, with PFGE patterns differing in less than 8 bands difference. Similarly, strains identified as ST280 and ST391, both linked to BAPS group 3-1, showed related PFGE patterns despite being trilocus variants (# 8 bands difference).
vanA-Tn1546 is located on highly transferable mosaic plasmids involving narrow host pRUM and pAD1 derivatives The plasmid content of the isolates studied appears in Figure 2. Efm isolates carried a variable number of plasmids (n = 1-6) which contained specific sequences of different families including rollingcircle plasmids (RCR) related to pRI1 and small theta plasmids related to pCIZ2, RepA_N (pRUM-like, pLG1), pHTb (present in all ST132 isolates), and Inc18 (pRE25 and pEF1-related). All Efs contained RCR plasmids and pheromone responsive-plasmids.
vanA-Tn1546 was located on plasmids ranging from 30 to 150 kb, successfully transferred by conjugation in 95% (n = 71/75) of Efm and 97% (n = 28/29) of Efs, with a variable frequency (10 21 -10 28 ). Transferable plasmids were identified as members of pRUM and Inc18 families or were mosaic plasmids of pRUM, Inc18 and pheromone plasmids (see sections below). Although some of these mosaic plasmids were detected in both Efm and Efs hosts, species-specific plasmid variants were predominant.
We have classified the enterococcal plasmids according to the content in rep/rel/TA systems, and RFLP profiles (Table 1, Figure 2). For the better interpretation of the results, we should keep in mind that members of the most common plasmid families classified in this and other studies as Inc18-like (pRE25, pIP501, pVEF1, pVEF2, pVEF3, pIP816, pEF1, pWZ909) or pRUM-like (pRUM, p5753cB, pS177) exhibit a high degree of modular dissociability or propensity for independent variation and shuffling, and may contain multiple replicons or be devoid of conjugation systems, thus making it very difficult to establish an accurate classification and to trace the origin of certain elements [9,[33][34][35][36][37][38][39][40]. See Clewell et al. for a comprehensive updated revision of enterococcal plasmids [9]. In the following sections we will describe vancomycin resistant plasmids of Efm and Efs.

i)
pRUM derivatives (rep 17.2/pRUM-like +rel 6/pEF1 ) of variable size (30-120 kb) were detected since the mid 90 s from a diversity of clonal backgrounds. pRUM plasmids showing different ClaI-digested DNA RFLP patterns were identified carrying a whole copy of Tn1546 (RFLP_1, RFLP_2, RFLP_20, 30-80 kb), IS1216::Tn1546 (RFLP_8-12, 40-120 kb) or ISEf1::Tn1546 (RFLP_3-7, RFLP_13, 50-95 kb). Despite the heterogeneity of plasmid profiles, RFLP_3-6 or RFLP_8-10 shared a variable number of common bands that suggest a relationship among them (see Table 1 and Figure 3 for details about relationships among plasmids). pRUM-like plasmids exhibiting distinct RFLP profiles and carrying different transposon variants were isolated in early and recent isolates of different clonal backgrounds ( Figure 2  HST, Hospital São Teotó nio; HPH, Hospital Pedro Hispano; CHCB, Centro Hospitalar da Cova da Beira; HVR, Hospital S. Pedro; SW, sewage wastewaters; UW, urban wastewaters; R, river; ND, not determined; NI, not identified; UK, unknown. a The distribution of the different isolates is shown by BAPS subgroups as described [19]. b PFGE types shown in bold represented widespread clones in Portuguese hospitals and/or aquatic surroundings over years. c Most Efm isolates expressed resistance to vancomycin, teicoplanin, erythromycin, ampicillin, ciprofloxacin (92-100%) and to a lesser extent to high levels of kanamycin (65%), gentamicin (41%), streptomycin and tetracycline (28% each). While acm was identified in different CC17 and non-CC17 lineages (76%), esp was detected in CC17 isolates (35%, ST132 and its SLVs ST368, ST369) and hyl was sporadically found (9%, ST18, ST125, ST132, SLVs of each other, and ST280 isolates) [25]. Efs isolates (mostly ST6) showed resistance to vancomycin, teicoplanin, erythromycin, ciprofloxacin, high levels of gentamicin and kanamycin (82-100%), tetracycline and chloramphenicol (65% each) and high levels of streptomycin (46%), and mostly contained gelE and agg (.90%), cyl (82%) and esp (46%) [26]. d Tn1546 designation is based on the results obtained by a PCR assay described by Woodford et al. consisting on the amplification of overlapped fragments covering the whole Tn1546 [68]. Fragments of unexpected length were further analysed by sequencing (this study) [27]. e The total rep/rel/TA content of isolates is represented according to its location on plasmids of different size ranges. Rep (normal cells), rel (cells with dots) and TA (cells with diagonal stripes) genes belonging to the same plasmid are represented with the same color and that belonging to the same plasmid family with the same range of colors. The content of VanA plasmids including rep, rel, and TA genes is indicated according to the plasmid type in which they were identified, as well as by the numeric nomenclature used by Jensen et al. [72] [72]. The name of the most representative plasmid of the family is also represented for a better followup of the results (e.g. rep 17/pRUM , rep17 from pRUM and related plasmids p5753cB and pS177; rep 1/pIP501 rep1 linked to Inc18 plasmids as pIP501, pIP816 and pRE25; rep 9/pAD1 , rep9 linked to pCF10, pAD1, pTEF1, pTEF2, pBEE99, pMG2200; rep 14/pRI1-like , rep14 associated with RCR plasmids pEFNP1, pJS42 and/or pRI1; rep 18a/pEF418 , rep18 from pEF418; and rep 22/pHTb , rep of both pHTb and pMG1 plasmids). We further specified the name of different plasmids associated with a given group if necessary. For example, it results helpful for Inc18 family given the number of plasmids containing the same rep gene. These plasmids are increasingly identified among isolates of different origins (e.g. rep 2/pRE25/pEF1 for designing rep2, as rep and rel modules of pEF1, a plasmid originally identified in olives [35], seems to be widely present in all Efm clinical isolates) . Sequencing identified the different variants within these families (see text). Rep 18b , rep 18c and rep 20 were not included in Jensen's scheme [72] and the numbers were assigned in this paper following that numeration (rep 18b/pB82 , rep from pB82; rep 18c/pCIZ2 , rep from pCIZ2; rep 20 Figure 4). These results illustrate the possibility of efficient intraclonal and intraplasmid diversification of Tn1546::IS1216 variants. Acquisition of a vanA-Inc18

E. faecalis vanA plasmids
The vanA Efs plasmids were Inc18-pheromone-responsive mosaics, further classified in four main types on the basis of their RFLP patterns (RFLP_27-30), rep-rel/TA content, and replicase sequences. These plasmids have been documented in different Portuguese hospitals since the mid 90 s [26].
Plasmids showing highly related patterns designated as RFLP_27 (carrying ISEf1-Tn1546) or RFLP_28 (carrying IS1216-Tn1546) were recovered from both Efs (ST6, ST55, ST159) and Efm (ST80, ST132). However, despite the similarity of their RFLP patterns, they differed in the rep/rel/TA content and transposon variant content (Table 1). Conversely, the finding of an ST117 Efs isolate from Oporto with two different vanA plasmids of 150 kb and 300 kb indicates acquisition and further recombination of widespread pRUM-vanA plasmids from Efm with narrow host pheromone responsive plasmids of Efs.
The observed differences in transposon variants and plasmid modules reflect frequent rearrangements during transfer of plasmids between Efs and/or Efm clonal backgrounds and also highlight the connectivity of these enterococcal populations resulting in the acquisition and generation of plasmids with enhanced host range.

Fixation of vanA-Tn1546 variants is associated with plasmid connectivity
Tn1546 backbones were classified in three main groups corresponding to Tn1546 with no insertion sequences (''type A'' and ''type D'') and variants containing ISEf1 (5 types) or IS1216 (11 types) at different locations of the Tn1546 backbone ( Figure 4). Variants with a single copy of ISEf1 within the vanX-vanY region at nt 9044 were located on early (1996-1997) Efm plasmids identified as Inc18 and pRUM lacking Axe-Txe, and also on early (1996) Efs Inc18-pAD1 mosaics. Some of them were isolated from strains for more than one decade, which can be explained by their successful long-term recovered clonal and plasmid backgrounds.
Variants containing IS1216 were mostly located on Inc18 plasmids or on mosaic plasmids Inc18-pRUM or Inc18-pAD1. Most variants contained the IS1216 at 8839nt of the transposon (PP13, PP17, PP20, PP23, PP30) similarly to other Tn1546 variants previously described in Europe [42]. Some of them also harboured different insertions corresponding to unknown sequences (X, PP23) or RCR plasmid sequences (PP10) [43] suggesting frequent recombination between acquired genes/ plasmids and housekeeping Efm and Efs plasmids (Figure 4). Tn1546 type D was specifically linked to megaplasmids from CC5 Efm from swine of different continents (Figure 2, Figure 4).
The presence of early plasmids carrying Tn1546 belonging to different families suggests independent acquisitions of the transposon by pRUM and Inc18 plasmids, which would have been acquired by diverse Efm and Efs populations. Local fixation would be influenced by connectivity of plasmid and population backgrounds enabling further evolvability of transposon variants.

Discussion
This paper shows the local dynamics of Tn1546-vanA among Enterococci is shaped by horizontal genetic transfer of pRUM and Inc18 plasmids and by recombination-driven evolution of them within and between Efs and Efm clones. The clonal diversity reported in this study has also been observed in areas where the spread of VRE has been documented [44]. Recent retrospective analysis of enterococcal populations suggests that the temporal evolution of the population biology of Enterococci is driven by a succession of epidemic waves of enterococcal human specific lineages, Efm ST78 and Efs ST6 emerging in the last decade at global scale similarly to that reported for other pathogens [19,23,24]. In Portugal, the population structure of VRE analysed in this study comprises isolates of main human Efm lineages, ST18 (ST18, ST132) being much more abundant than ST17 (represented by a single isolate of early ST16 lineage) [31], or ST78 (represented by sporadic ST80 and ST656, the first one linked to early VRE outbreaks) [25,29]. It is worthwhile highlighting the recent detection of isolates of another Efm lineage in hospitals of the Oporto area (http://www.mlst.net) as ST117 Efm (ST78 lineage), which would reflect the increasing trend of isolates belonging to the ST78 lineage at international level. However, regional differences in the rates of VRE cannot be fully explained by clonal replacement dynamics since similar enterococcal clones appear widely distributed in areas with high and low rates of VRE (Tedim AP et al., unpublished data). Instead, local conditions, including type and density of hosts, antibiotic usage, and transmission facilities, may influence regional differences in the proportions of VRE, as suggested by mathematical modelling studies on local trends of antibiotic resistance [45,46]. Clones can locally evolve by variation, drift and short-distance migration, leading to changes in colonization ability, pathogenicity or even host range, the fittest clonal variants being able to facilitate the spread of antibiotic resistance [23,[47][48][49][50]. The observed clonal heterogeneity of the predominant ST18 lineage which comprises particular ST18 and ST132 strains widespread in different cities, highlights the role of certain efficiently transmissible clones in the designated with numbers corresponding to different plasmid types [9] (Francia et al, unpublished data): Rel 1 , pJS42, pRI1; Rel 2 , rel from p200B, pCIZ2 and/or pB82 plasmids; Rel 3 , pRUM; Rel 5 , rel from pAD1, pTEF1, pAM373 and the pathogenicity island of V583; Rel 6 , pEF1; Rel 8 , pHTb and pMG1; Rel 9 , pCF10. Toxin-antitoxin systems included Axe-Txe from pRUM, v-e-f from Inc18 plasmids and par from pAD1. Genes hybridizing in the same band as vanA plasmids appear in bold rectangles. doi:10.1371/journal.pone.0060589.g002  [51]. One remarkable fact is the similarity among PFGE patterns of isolates with different STs. Given the high content of plasmids and transposons of the isolates studied, and the frequent rearrangements identified among Efm and/or Efs isolates [21], chromosomal transfer can not be discarded. Recent phylogenomic analysis based on the degree of admixture among a diversity of isolates studied suggests that recombination is restricted to isolates within specific BAPS groups [19]. Most plasmids coding for vancomycin resistance are found in similar clonal backgrounds. This observation suggests that recombination does occur within isolates of similar BAPS groups as recently described [19]. However, the observed mosaicism and enhanced host range of particular plasmid variants indicates the existence of an unexpectedly high degree of connectivity between phylogenetically distant enterococcal populations and/or in bacterial genetic exchange communities integrating enterococci. Broad host and narrow host plasmids carrying vancomycin resistance would have a high ''betweenness centrality'', which is a pivotal index in network theory useful for measuring the load placed on the given node in the network as well as the node's importance to the network than just connectivity [52]. A recent in silico network analysis of all plasmid sequences available at the GenBank databases confirms very high˝betweenness values for some Inc18 plasmids as pVEF3 (an Inc18 derivative highly spread among Efm from animals in Europe) [13,37], and also for a pheromone-responsive plasmid pTEF1 (a plasmid recovered from ST6_Efs strain V583, highly related to the ST6 described in this work) [53] (unpublished data). Other plasmids with a high degree of modular dissociability, would be pRUM-like elements, which may enhance their complexity resulting in new configurations with enhanced betweenness. It is tempting to suggest that plasmid variability has contributed to intra-clonal diversification both in Efm and Efs, giving rise to a local wealth of clonal variants able to fully explore the local adaptive landscape. In fact, this and other studies demonstrate that selected variants of Inc18, pAD1, and pRUM plasmids can determine differences in the dynamics of VRE in different areas, further influencing the plasmid host range and the selection of specific clones within human adapted lineages. Examples of widespread plasmid variants of Inc18 or pRUM plasmids coding for vancomycin resistance have been reported recently. They included Inc18 widespread among Efm poultry isolates from Europe [13] or among Efs clinical isolates from the USA, the last one being able to transfer Tn1546 to S. aureus [15]; and mosaics of pRUM variants containing Axe-Txe and Inc18 from humans in different continents (Freitas AR et al. unpublished data). The identification of chimeric pRUM-Inc18 plasmids containing rep/rel/TA of Inc18 sequences and Tn1546 variants widely observed in poultry, hospitals and hospital sewage in the Oporto area reflects genetic exchanges between enterococci from different origins and highlights the need to enforce barriers to avoid the spread of multidrug resistance human pathogens to the environment and viceversa.
In this scenario, the genetic context of Tn1546 seems to greatly influence the evolvability of the transposon and explains the high diversity of variants found in this and other studies [1,27,42,54]. The frequent presence of insertions in the backbone of Tn1546 and the abundance of IS1216 and ISEf1 in enterococcal genomes [9,55] makes homoplasic evolution of Tn1546 in different backgrounds possible. However, other IS (IS1251, IS1542, IS1476, IS19 and IS1485) linked to different plasmid and clonal backgrounds [9,40] have been identified at different sites of Tn1546, thus suggesting that chance and selection are responsible to differences in variants collected in different areas. The widespread of Inc18 plasmids with a common origin in Europe [13,56] indicates local fixation of Tn1546 influenced by a founder effect and further connectivity of plasmid and population backgrounds enabling further evolvability of transposon variants as reported in this study.
Our results suggest that VRE spread is facilitated by selected clones of different lineages through strong interactive processes of clonalization and plasmid diversification that might occur at local scales. Despite the maintenance of significant gene flow, a sympatric, or more probably, parapatric bacterial clonalization process (when diverging populations share a common or neighbouring environment), might contribute to the formation of temporary genetic mosaics and the preservation of ecologically important genomic traits [57]. Such micro-evolutionary process will result in an array of clonal complexes forming a population structure able to exploit the local spatio-temporal patch heterogeneities [58]. Note that exploitation of connected microenvironments should accelerate evolution of antibiotic resistance [59]. The expected result of such a successful population structure is the local persistence of antibiotic resistant clones, and eventually the local fixation [60] of vancomycin-resistance [46].
In summary, this study highlights the relevance of studying the local microecology of genes, elements, lineages and populations to decipher the robustness of the trans-hierarchical networks connecting these evolutionary elements in order to describe and predict the local evolvability of vancomycin-resistance [61]. Traditional surveillance studies are one-off cross sectional surveys focused on single traits as epidemic strains, genes or mobile genetic elements over limited periods of time which only gives one shot view that precludes addressing the long-term dynamics of antibiotic resistance. The more comprehensive approach described in this study is needed for understanding in depth the evolution of complexity in multihierarchical systems as those involved in the  Part of the isolates analyzed in this work corresponds to strains from previous surveillance studies [25][26][27]62]; this paper constitutes the first description of isolates obtained during 2007 and 2008. Contemporary Portuguese VRE isolates of animal origin were used for comparative analysis of lateral transfer events [10].

Genetic context of Tn1546
Characterization of Tn1546 backbone was determined by amplification of overlapping transposon fragments and further sequencing of PCR products [27,68]. We have accomplished the analysis for the isolates not studied in previous surveys and have interpreted the resulting transposon diversity (this study) [27], under the light of the plasmid and clonal backgrounds identified in this geographical area.

Plasmid analysis
Isolates (n = 62 Efm and n = 13 Efs) representing the clonal diversity observed in both species were selected for plasmid characterization (Table 1, Figure 2). The content and size of plasmids from transconjugants obtained by filter mating were determined by using either the technique described by Barton et al. (plasmids .10 kb) or the alkaline lysis extraction method of Kado & Liu (plasmids ,10 kb) [51,69,70]. Classification of E. faecium plasmids was based on the presence of specific modules for replication (rep-initiator proteins), mobilization (relaxases) and stability (toxin-antitoxin systems). Relaxases (rel) were sought by a multiplex-PCR-based relaxase typing method which differentiates relaxases of the MOBQ, MOBP, MOBC and MOBV families related to 27 known plasmids [9,71] (Francia MV, unpublished data). Replication initiator proteins (rep) were investigated by amplification of 24 replicons, which allows discriminating among DNA sequences from more than 100 published Gram-positive plasmids [9,72]. Designation of rep sequences pointed out the plasmid type in which they were initially identified, as well as the numeric nomenclature originally used by Jensen et al. (Figure 2s footnote) [72]. Toxin-antitoxin systems (TA) previously identified among streptococci and enterococci (Axe-Txe, v-e-f par, mazEF) or Gram-negative bacteria (relBE) were detected by PCR [73]. PCR products were sequenced in order to confirm the specificity of the method and to analyze similarities with other wellcharacterized plasmids. Genomic location of the Tn1546 and the rel/rep/TA sequences was determined by hybridization of vanA and rel/rep/TA specific probes obtained by PCR from DNA from reference plasmids with S1 or I-CeuI digested genomic DNA from representative strains [51,69]. Structural relationship between plasmids of similar size was established by comparison of their RFLP patterns obtained after digestion with different restriction enzymes (EcoRI, HindIII and ClaI; see Figure 3). Plasmid DNA was obtained by using a modified protocol based on the alkaline lysis method described by Handwerger et al. [74] consisting of increasing two-fold the volume of lysozyme, SDS/ NaOH and acetate potassium solutions, extending the incubation period in potassium acetate solution for at least three hours, precipitating the supernatant obtained after extraction with phenol-chloroform using ethanol-acetate potassium solution (2:0.1 vol/vol) at 25uC for at least 2 hours, and resuspending final DNA pellets in 30 ml of water for further enzyme digestion analysis.

Molecular techniques
Southern blot DNA transfer and hybridization were performed by standard procedures [75]. The vanA and rep/rel/TA/bac probes used in the hybridization assays were generated by PCR using well known positive controls as template DNA. Labelling and detection were carried out using Gene Images Alkphos Direct Labelling system kit, following the manufacturer's instructions (Amersham GB/GE Healthcare Life Sciences UK Limited). PFGE was performed as described previously [76] using the Figure 4. Genetic maps of Tn1546 variants. Tn1546 variants are represented as previously described by Novais et al. [27] although grouped differently and specific types have been further explored (PP10, PP30): Tn1546 prototype A corresponds to the original sequence described by Arthur et al. [77] and D corresponds to Tn1546 variants from animals. Tn1546 variants with ISEf1 within vanX-vanY intergenic region (PP2a, PP4, PP5, PP9, PP24) and Tn1546 variants with IS1216 insertions at different positions (PP10, PP2b, PP13, PP15, PP16, PP17, PP20, PP23, PP27, PP30, X) are represented. The positions of genes and open reading frames and the direction of transcription are depicted with open arrows. IS elements are represented by triangles; other sequences are designated by rectangles. DNA insertions are represented highlighting the first nucleotide upstream and downstream from the insertion sites whenever known. Deletions are indicated by dots and discontinuous lines indicate sequences that were not characterized. ( a ) DNA sequence with homology to ORF3 (unknown protein product) and ORF1 (replication protein) of pEFNP1 plasmid (GenBank accession number AB038522). ( b ) DNA sequence with no match to any sequence available in GenBank. (*) PP23 was identified in an isolate susceptible to teicoplanin; this variant contained an insertion in the vanY gene that would affect the transcription of vanZ and it might explain the susceptibility to this glycopeptide as previously reported [27]. (**) PP30 was identified in an ST78 isolate susceptible to both glycopeptides (MIC against vancomycin and teicoplanin of 4 mg/L) carrying vanA-Tn1546. This variant contained alterations within the vanS-vanH intergenic region (an IS1216 insertion), which is involved in the expression and regulation of the resistance to vancomycin, and it constitutes the first description of a vanA isolate phenotypically susceptible to vancomycin in Portugal. doi:10.1371/journal.pone.0060589.g004 following conditions: switch time of 5 s to 25 s for 6 h, followed by 30 s to 45 s for 18 h (S1 nuclease); 5 s to 30 s for 22 h, 14uC, and 6 V/cm 2 (I-CeuI) and 1 s to 20 s for 26 h, 14uC, and 6 V/cm 2 (SmaI).