Twelve years of genomic surveillance of vancomycin-resistant Enterococcus faecium: emergence of linear vanA and bacteriocin-carrying plasmids challenging infection control Authors: Ana C Almeida-Santos, Ana P Tedim, Bárbara Duarte, Ana P Pereira, Luís M Silva, Júlio Teixeira, Ana P Castro, Louise Roer, Anette M Hammerum, Henrik Hasman, Teresa M Coque, Carla Novais, Ana R Freitas, Luísa Peixe, ESCMID Study Group for Epidemiological Markers (ESGEM), and ESCMID Study Group on Mobile Elements and Plasmids (ESGMAP) Published: 2026 ABSTRACT AbstractBackgroundThe epidemiology of vancomycin-resistant Enterococcus faecium (VREfm) varies across different countries, with a steady global increase. In Portugal, however, epidemiological data on clinical VREfm have been scarce since the early 2000s. This long-term study investigates VREfm isolates from human infections collected at a Porto hospital between 2010 and 2021.MethodsTwo hundred VREfm isolates, mostly urinary (39%) were characterized by antimicrobial susceptibility testing to 8 antibiotics, chlorhexidine susceptibility, and PCR-based detection of vancomycin-resistance genes, virulence markers, plasmid replicases, and the bac43/T8 gene. Whole-genome sequencing, by Illumina, was used to assess clonal diversity (MLST, cgMLST, SNP phylogeny) and genomic content of antimicrobial resistance (AMR) genes, bacteriocins (76 genes) and putative virulence markers (35 genes). The plasmidome size and replicase initiation proteins of selected isolates was further improved by incorporating nanopore sequencing which enabled hybrid assemblies.ResultsAll isolates were multidrug resistant; 98% carried vanA, while two (1%) were resistant to linezolid (G2576T mutation). Chlorhexidine susceptibility remained stably low over time (MICs: 2–4 mg/L). The population was polyclonal, with a shift from ST18-like lineages to ST80 and ST117 dominance. While ARGs and virulence markers showed no clear association with clonal waves, bacteriocin profiles did, with bac43 becoming increasingly prevalent. ST117-CT24 emerged as the most persistent clone. The plasmidome comprised stable Rep3-like mobilizable plasmids carrying bacteriocins (bac43, bacAS5), RepA_N mega-plasmids harboring virulence/AMR/bacteriocins, and highly plastic medium-to-large vanA plasmids with diverse replicase initiation proteins. Strikingly, linear vanA plasmids (repUS78_pZY2) appeared in the most recent isolates, paralleling findings in vancomycin-variable E. faecium from the same hospital and VREfm from other countries.ConclusionsOur findings reveal dynamic clonal shifts, novel plasmid architectures, and the key role of bacteriocins in shaping clonal success in a WHO priority pathogen. Furthermore, we highlight the need for AMR surveillance frameworks to consider factors beyond conventional prevalence metrics and core-genome comparisons. Integrating intra-species genomic heterogeneity and non-traditional evolutionary indicators will be essential to more accurately predict, track, and ultimately mitigate the dissemination of multidrug-resistant human pathogens.Supplementary InformationThe online version contains supplementary material available at 10.1186/s13073-026-01656-4.Keywords: Vancomycin resistance, Enterococcus faecium, Linear plasmids, vanA, Bacteriocins, Bac43, Genomics, Surveillance Background Enterococcus faecium, a typical commensal of the gastrointestinal tract (GIT) in humans and various animals, has emerged as a relevant opportunistic pathogen [1]. It ranks among the top five causes of bloodstream infections (BSIs) [2], posing a major threat to healthcare systems, a concern further exacerbated by the global rise of antimicrobial resistance (AMR) [3]. In Portugal, the most recent estimates (2022–2023) reported 2,599 healthcare-associated infections (HAIs), with E. faecium responsible for 4% of cases, closely aligning with the EU/EEA average of 3.9% [4]. Of particular concern is vancomycin-resistant E. faecium (VREfm), which is recognized by the World Health Organization as a high-priority pathogen due to its limited treatment options [5]. The VREfm landscape is highly heterogeneous in Europe, with the weighted average percentage ranging from less than 1% to over 50% and showing an overall upward trend since 2018 [6]. In Portugal, the weighted average percentage has been declining since 2007, although with a slight increase of 2.5% observed since 2018 [6]. VREfm causes a variety of infections, such as urinary tract infections, wound infections, endocarditis, and BSIs, posing a significant risk to critically ill or immunosuppressed patients, those on long-term antimicrobial therapy, and individuals with prolonged hospital stays [7, 8]. Once MDR E. faecium, and especially VREfm, colonizes the GIT in such vulnerable populations, it disrupts the commensal microbiota, promoting its own overgrowth and facilitating the spread of other multidrug-resistant organisms [9, 10]. This overgrowth increases the VREfm GIT burden and their release into the environment, where they can persist on contaminated surfaces for years, enabling further transmission [7, 10]. The success of VREfm, primarily clustering within the hospital-adapted E. faecium clade A1, is driven by its exceptional adaptability. Over time, they have accumulated ARGs, virulence factors, and specialized metabolic pathways, enhancing both its survival and pathogenicity [11–13]. This adaptability is fueled by its dynamic pan-genome and its ability to acquire plasmids and other mobile genetic elements (MGEs) [14, 15], which facilitate the rapid spread of AMR and virulence traits. Although several studies have explored the epidemiology and dynamics of VREfm within healthcare settings, few have examined other sub-hierarchical units to follow gene dynamics, especially in recent years. While the global dominance of certain clones is well recognized, the bacterial factors driving their success and the ongoing clonal shifts within healthcare settings and across countries pose unanswered questions that advanced sequencing technologies are only starting to address. This knowledge gap is particularly evident in Portugal, where studies on clinical VREfm have been scarce since the early 2000s. To address this, we aimed to analyze the VREfm epidemiology within a single hospital over a 12-year period, while investigating their content in adaptive traits such as antimicrobial resistance genes, virulence factors and bacteriocin genes to assess their potential role in population shifts. This comprehensive approach offers valuable insights into the epidemiology, evolution, and clinical impact of VREfm, guiding the development of more effective infection control strategies. Methods Study design and bacterial collection A total of 200 clinical VREfm isolates recovered over a 12-year period (2010–2021) from a tertiary care hospital in northern Portugal, were characterized in this study. This central reference hospital, located in downtown Porto, has approximately 800 beds and serves an estimated population of three million residents. It has routinely provided VREfm isolates for characterization and longitudinal analysis since 2001 [16]; however, no isolates were available from 2005 to 2009. All samples were collected by hospital staff as part of routine clinical practice, without any interference with patients’ standard care. Samples were obtained from inpatients across five hospital wards, and routine microbiological analyses, including identification and antimicrobial susceptibility testing, were performed at the hospital. These data and isolates were subsequently transferred to our laboratory. Furthermore, the isolates analyzed in this study represent all available VREfm received between 2010 and 2021, corresponding to approximately half of the total VREfm cases identified at the hospital during this period (Table 1). All 200 isolates underwent antibiotic susceptibility testing and PCR-based screening. From this collection, subsets of isolates were selected for specific analyses. A total of 63 isolates, representing different infection types and phenotypic/genotypic profiles, were used to evaluate response to chlorhexidine (CHX), while 34 isolates were subjected to multilocus sequence typing (MLST). Additionally, 58 representative isolates were selected for whole-genome sequencing (WGS) based on phenotypic, genotypic, clinical, and epidemiological parameters, including year of isolation, hospital ward, and infection type. Table 1Characteristics of VREfm isolates by year: sequence types, sources, linezolid susceptibility and chlorhexidine toleranceYearNº isolatesResistant isolates (%)aSTsProductLZD MICs mg/LCHX MICs mg/LCHX MBCS mg/L20101626.2%17, 78, 564Urine, peritoneal fluid, blood, pus, exudate, cerebrospinal fluid1–42–44–1620111320.6%78, 117, 125, 412, 564Urine, peritoneal fluid, blood, pus, exudate, tissue1–42–41–420121416.1%18, 78, 80, 117, 280, 412, 1184, 2073Urine, peritoneal fluid, blood, pus, exudate1–42–42–420131931.3%78, 117, 203, 564, 780Urine, peritoneal fluid, blood, pus, bile, drainage fluid1–42–42–420143231.1%78, 117Urine, peritoneal fluid, blood, pus, exudate, tissue, bile, drainage fluid1–41–21–420153145.4%80, 117Urine, peritoneal fluid, blood, pus, exudate, bile, drainage fluid, pus from abscess2–41–42–420162231.4%80, 118Urine, peritoneal fluid, blood, exudate, bile, tissue, drainage fluid, pus from abscess, prosthesis2–42–42–420171817.8%80, 117, 192, 280Urine, blood, pus, drainage fluid, tissue, prosthesis1–41–41–42018910.9%17, 117, 1478, 2073Urine, peritoneal fluid, blood, bile, drainage fluid1–42–41–42019817.8%117, 494, 2073Urine, peritoneal fluid, blood, tissue2–41–41–420201013.6%17, 80, 117Urine, peritoneal fluid, blood, tissue, drainage fluid2–41–42–42021813.9%80, 494Urine, blood, pus, drainage fluid, synovial fluid2–41–42–4MICs minimum inhibitory concentrations, MBCs minimum bactericidal concentrations, LZD linezolid, STs sequence typesaThe percentage of resistant isolates was provided by the hospital and calculated as the proportion of vancomycin-resistant E. faecium among the total number of E. faecium isolates collected each year Characteristics of VREfm isolates by year: sequence types, sources, linezolid susceptibility and chlorhexidine tolerance MICs minimum inhibitory concentrations, MBCs minimum bactericidal concentrations, LZD linezolid, STs sequence types aThe percentage of resistant isolates was provided by the hospital and calculated as the proportion of vancomycin-resistant E. faecium among the total number of E. faecium isolates collected each year Phenotypic assays Antimicrobial susceptibility testing was performed by disk diffusion against eight antibiotics (ampicillin, vancomycin, teicoplanin, erythromycin, tetracycline, ciprofloxacin and high-level gentamicin), following the guidelines of European Committee on Antimicrobial Susceptibility Testing (EUCAST) [17] or Clinical & Laboratory Standards Institute (CLSI) guidelines [18]. Given that disk diffusion does not accurately determine susceptibility against linezolid, minimum inhibitory concentration (MIC) values were determined by broth microdilution and interpreted according to EUCAST guidelines, including the evaluation of MIC50 (minimum concentration that inhibits the growth of 50% of the isolates) and MIC90 (minimum concentration that inhibits the growth of 90% of the isolates) [17]. The response to chlorhexidine was performed by determining MICs and minimum bactericidal concentrations (MBCs) [19]. Molecular techniques PCR assays were conducted in the 200 VREfm to confirm species identification, detect most prevalent vancomycin resistance genes (vanA and vanB), and screen for key virulence markers, plasmid replicases (rep) genes, and bacteriocin bac43 (previously associated with hospital/outbreak E. faecium populations) [12, 20]. Searched virulence markers included cell-wall-anchored proteins (esp, sgrA), microbial surface components recognizing adhesive matrix molecules (acm, ecbA), and others enhancing colonization (hylEfm, orf1481, ptsD) [21]. IS16 driving genomic plasticity was also searched. Plasmid replicases commonly associated with vancomycin resistance in E. faecium [RepA_N family: rep17_pRUM (GenBank accession no.: AF507977), rep20pLG1 (similar to repUS15pNB2354p1; GenBank accession no.:CP004064), and Inc18 family: rep1_pSM19035 (GenBank accession no.:AY357120), rep2_pRE25 (GenBank accession no.:X92945) were also investigated, as previously described [22]. Whole genome sequencing: short- and long-read sequencing data analysis Genomic DNA was extracted using the Wizard Genomic DNA Purification kit (Promega Corporation, Madison, WI, USA) according to the manufacturer’s instructions. WGS sequencing was performed using an Illumina NovaSeq platform (2 × 150 bp), according to standard Illumina protocols performed at Eurofins Genomics (Ebersberg, Germany). The assembled genomes (Unicycler, v0.5.0) [23] were screened for ARGs, AMR point mutations, plasmid content and MLST by using in silico genomic tools (ResFinder v.4.6, PlasmidFinder v.2.1, and MLST v.2.0 tools, respectively) available at the Center for Genomic Epidemiology (CGE; https://www.genomicepidemiology.org/services). Furthermore, mutations associated with reduced daptomycin susceptibility were assessed using AMRFinderPlus v3.12 [24]. Putative virulence markers (PVMs) were screened using VirulenceFinder v.2.0 from the CGE, including 35 known PVMs [25] and eight novel, unpublished markers (lysM1-lysM4, prpA, tirE1, tiE2 and capD) [26]. Raw sequence reads in FASTQ format were analyzed to detect the presence of the enterococcal surface protein gene of E. faecium, espfm, due to its diversity in both size and structure. Additionally, bacteriocin genes were screened using a custom database comprising 76 bacteriocin genes from Bacillota [12]. Clonal relationship was established by core genome MLST (cgMLST) using Ridom SeqSphere + v.10.0 (Münster, Germany; https://www.ridom.de/seqsphere/). Clusters were defined as isolates differing by ≤ 20 alleles [27]. Additionally, a single nucleotide polymorphism (SNP)-based analysis was performed employing the CSI Phylogeny tool from CGE (https://cge.food.dtu.dk/services/CSIPhylogeny/), using the reference genome E. faecium DO (NC_017960.1). Putative transmission was inferred when isolates differed by ≤ 10 SNPs, as previously described by Higgs et al. (2022) [28]. Microreact web application was used to visualize both trees and metadata (https://microreact.org/). Of the 58 VREfm sequenced by Illumina, 19 were selected for additional nanopore sequencing using the MinION platform (Oxford Nanopore Technologies) at Plasmidsaurus (USA). These isolates represent distinct STs or complex types (CTs) within the same ST, ensuring diversity across sample types, and balanced year-wise representation. Hybrid assemblies were generated by combining Illumina short reads and nanopore long-read data using Unicycler (v0.5.0) [23]. Analysis of the VREfm plasmidome The plasmid content of each isolate, with available hybrid assembly data (n = 19; 18 vanA and 1 vanB), was characterized by integrating mapping/annotation results from Geneious Prime with analyses using CGE tools, focusing on ARGs, virulence factors, plasmid replicases, and bacteriocins, as outlined in the previous section. Mapping and annotation of plasmids were performed with Geneious Prime (v.2025.0.3) and PlasmidFinder (80% coverage, 80% identity). For plasmids in which the initial screening did not detect rep genes, an additional search was conducted using PlasmidFinder, applying relaxed thresholds of 20% coverage and 50% identity. The following rep genes found were analyzed using BLASTx against the NCBI database to determine their plasmid family based on protein domain similarity. Once the replication families were established, the rep genes were further compared to all rep genes from the CGE PlasmidFinder database within the same family, using Clustal Omega, to assess whether they belonged to a new subfamily (> 80% coverage; < 80% identity) or represented a novel sequence within a previously known subfamily (> 80% coverage; > 80% identity). Finally, BLASTn was used to determine whether each rep gene had been previously described in a plasmid within the NCBI database or was unique to our plasmid collection. The nomenclature assigned to the rep genes follows the PlasmidFinder classification system. Typing of vanA-carrying plasmids Mapping and annotation of vanA plasmids was performed as described above for the VREfm plasmidome. Structural comparison of plasmids was performed using Pling (v2.0) [29], applying a containment threshold of 0.3 to identify potential transmission events. Briefly, complete plasmid sequences obtained from hybrid assemblies served as input. Pairwise containment distances and the minimum number of rearrangements or indel events were calculated using the DCJ-Indel method (default 4). A plasmid relatedness network was then generated, with plasmids grouped into communities (containment threshold ≤ 0.3) and subcommunities (based on DCJ-Indel distance ≤ 4) based on Pling’s integrated community detection algorithm. Statistical analysis A Chi-squared test was conducted using Excel (v.16.95) to assess associations (P ≤ 0.05) between clonality (STs) and ARGs, PVMs, and bacteriocin content. Results Epidemiology of VREfm in the hospital studied A total of 200 non-duplicate VREfm isolates associated with human clinical infections were identified in the hospital under study, between 2010 and 2021. Among the 200 VREfm, 57.5% were from female patients and 41.5% from male patients (Fig. 1). Patient ages ranged from nine to 95 years, with an average of 64 ± 17 years. The age distribution (9–95 years) was clearly skewed toward older individuals [median: 66.5 (IQR 55–75) years], with 126/198 patients (63%) aged ≥ 60 years. The isolates were predominantly collected from medical- (45%) and surgical wards (37%), with smaller contributions from the intensive care unit (8%), emergency services (7%), and transplant units (5%) (Fig. 1). Urine samples represented the largest proportion of isolates (39%) and were significantly associated with female patients (P < 0.01), followed by peritoneal fluid (14%), blood (12%), drainage fluid (10%), and pus (8%). The remaining 17% originated from a variety of other clinical specimens. Clinical data of patients infected by VREfm. Data on sex and age were missing for two samples and information about the ward of origin and specimen type was unavailable for another sample Phenotypic analysis: patterns of antibiotic resistance and chlorhexidine susceptibility over time All VREfm isolates were multidrug-resistant, consistently exhibiting resistance to vancomycin, ampicillin, ciprofloxacin, and erythromycin, and all but four VanB isolates were resistant to teicoplanin. To a lesser extent, we detected resistance to high-levels of gentamicin (35%), tetracycline (26%) and linezolid (1%; MIC50 = 4 mg/L and MIC90 = 4 mg/L). The two VREfm resistant to linezolid expressed an MIC of 8 mg/L and were isolated in 2014. The CHX MICs of the selected representative isolates ranged from 1 to 4 mg/L, with an MIC50 and MIC90 of 2 and 4 mg/L, respectively (Additional file 1: Fig. S1). MBCs for CHX exhibited a wider range, from 2 to 16 mg/L with MBC99 recorded at 16 mg/L. No clear temporal or clonal association was observed for CHX MICs or MBCs throughout the study period (Additional file 1: Fig. S1). Genotypic analysis: vancomycin resistance, virulence, plasmid replicases and bac43 genes The majority of VREfm isolates carried the vanA gene (98%), whereas only four isolates (2%) harbored the vanB gene. The vanB-carrying isolates were identified in 2011 (n = 1), 2012 (n = 2), and 2017 (n = 1). Our initial molecular screening of hospital-associated PVMs revealed that all isolates carried at least four of the nine genes analyzed, while 7% harbored the complete set. The most prevalent PVMs were ptsD (99%), orf1481 (97%), acm (93%), espfm (87%), ecbA-2 (82%), and sagrA (82%) (Additional file 1: Fig. S2A). IS16 was present in all isolates (100%). In contrast, ecbA (43%) and hylEfm (24%) were the least frequently detected but remained present across different clones throughout the study period. Although we screened for both ecbA and ecbA-2 to capture the maximum variability, 12 isolates (6%) still lacked both gene variants. No consistent temporal trends or patterns were observed in the prevalence of the targeted PVMs over the years (Additional file 1: Fig. S2B). Eighty-four percent of the isolates harbored at least one of the targeted plasmid replicases, with rep17-pRUM being the most prevalent (57%), followed by rep2_Inc18 (46%), rep1_Inc18 (31%), and rep20_pLG1 (19%) (Additional file 1: Fig. S3A). The rep17_pRUM was the predominant replicase up until 2017, except for 2012, when it was surpassed by rep1_Inc18 (Additional file 1: Fig. S3B). Its prevalence declined steadily until 2015, showed a resurgence in 2016, and then experienced a sharp decline through 2020, ultimately being surpassed by both rep1_Inc18 and rep2_Inc18 (Additional file 1: Fig. S3B). The rep20_pLG1 was no longer detected after 2017. Remarkably, 15% of the isolates did not carry any of the replicases tested, usually dominant in clinical VREfm, suggesting the presence of other circulating variants or novel ones that may only be detectable through sequencing (see sections below). bac43 was found in a total of 81% of isolates, increasing in recent years from 44% in 2010 to 100% in 2021 (Additional file 1: Fig. S4). VREfm clonality: intertwined clonal waves and dominance of ST80 and ST117 Most isolates (82%) belonged to the clonal lineage ST78, with the most prevalent STs being ST117 (36%), ST80 (18%), and ST78 (12%), followed by different STs found in ≤ 5% of the isolates (Fig. 2A). The clonal evolution of VREfm exhibited dynamic shifts over time (Fig. 2A). In 2010, only three STs were present (ST78, ST564, ST17), and diversity peaked in 2011–2012 with the introduction of other STs, including ST117. This was followed by a decline in ST diversity, with ST78 disappearing by 2014 and ST117 peaking in 2013. From 2015 onward, ST80 became dominant, particularly in 2015, 2016, 2017 and 2021. Later years saw the introduction of other STs (e.g., ST192, ST1478, ST2073), alongside the sporadic reappearance of previously identified STs as ST17 (Fig. 2A). There was a clear shift from early dominance of ST78 to the later predominance of ST117 and ST80, reflecting the dynamic nature of VREfm clonal evolution in this setting. Of note, ST1478 is a pstS-null variant of ST117. A Global Sequence types (STs) distribution between 2010 and 2021. B STs and the correspondent complex types (CT) per year. STs flagged with the symbol (✣) belong to the clonal lineage ST78-like Core genome MLST analysis identified 28 distinct CTs (Fig. 2B), grouped into 10 clusters (≤ 20 allelic differences), and 13 singletons (Fig. 3A). ST117 isolates were distributed across two clusters and four were singletons, with the largest cluster comprising 20 isolates, primarily belonging to ST117-CT24 (60%) (Fig. 3A). ST80 isolates were dispersed across seven clusters (four singletons), scattered throughout the spanning tree (Fig. 3A). Most isolates clustered across different time periods, with only a few grouping within the same or consecutive years (Fig. 3B). Most dominant types were typically replaced by other types within 2–3 years, except for the ST117-CT24 clone, which persisted over four years (Fig. 3B). The last three years of the study (2019–2021) were marked by the introduction of several novel CTs from ST17, ST80 and ST117 clusters. To further investigate genomic diversity, we complemented our cgMLST analysis with SNP-based analysis which revealed a broad variation ranging from 2 to 3,653 SNPs. Notably, 19 pairwise isolate comparisons showed ≤ 10 SNP differences, involving 23 unique isolates (39.7% of the sequenced isolates), which formed 10 independent transmission clusters. These clusters included four of the 16 STs and nine of the 28 CTs, typically comprising isolates collected within the same year. Putative transmissions occurred most frequently between surgical wards, although inter-ward transmissions were also observed. The largest suspected transmission event involved the ST117-CT24 vanA cluster over a span of four years, affecting nine patients, three of whom had stayed in surgical wards. During the only instance when this clone was identified in the same ward, the isolates were collected three weeks apart. Minimum spanning tree of Portuguese VREfm (n = 58; 2010–2021) using cgMLST in Ridom SeqSphere + (v.10.0). Isolates are colored A) according to complex types (CT) based on the cgMLST nomenclature or B) according to year of collection. Groups were defined with a maximum of 20 allelic differences (shading in grey) Resistome, virulome and bacteriocin profiles The sequenced VREfm carried between 4–11 acquired ARGs, conferring resistance to aminoglycosides, macrolides-lincosamides-streptogramin B (MLSB), lincosamides-pleuromutilins-streptogramin A, tetracyclines, vancomycin (vanHAX, vanHBX), chloramphenicol and trimethoprim, with variable homology against reference strains (Fig. 4; Additional file 2: Table S1). Interestingly, most isolates carrying tet(M) and/or tet(L), displayed resistance to tetracycline, although three remained susceptible despite harboring such genes. Additionally, lsa(E) was restricted to early ST78 isolates from 2010 to 2012, while cat was exclusively detected in ST117 isolates across different years (2012–2015 and 2017) (Fig. 4; Additional file 2: Table S1). The same 19 pbp5 mutations, linked to pbp5 clade A1R and ampicillin resistance, were identified in all sequenced VREfm isolates, together with consistent mutations in gyrA and parC (Additional file 2: Table S1). Linezolid resistance was identified in two ST117-CT24 from 2014 and linked to the G2576T mutation in the 23S rRNA gene (Additional file 2: Table S1). Regarding daptomycin-associated mutations, 86% of the sequenced isolates harbored at least one mutation (rpoC_T641K, rpoB_H486Y, liaR_S19F, liaR_E75K dltC_S63C, or liaS_T120A). Isolates lacking such mutations were predominantly from earlier years and mainly belonged to ST78. The type and number of mutations varied by sequence type, with all ST117 isolates but one carrying three mutations (rpoC_T641K, liaR_S19F/E75K, and dltC_S63C). The liaS_T120A mutation was detected exclusively in two ST80 isolates (Additional file 2: Table S1). Whole-genome SNP-based phylogenetic tree of 58 VREfm genomes from this study. Phylogenetic tree construction was done using CSIPhylogeny tool from CGE (v.1.4.0) while metadata integration and visualization were carried out using the Microreact web application (https://microreact.org/). E. faecium DO was used as reference genome (Biosample accession number: SAMN00002237). The scale bar indicates the number of substitutions per site Using a recently available E. faecium and E. lactis virulence database and eight novel PVMs (n = 35), VREfm isolates carried 26–34 PVMs but none containing all of them. All harbored a consistent set of PVMs, including surface-exposed cell wall-anchored proteins (acm, lysM4 and ecbA) biofilm, adhesion and colonization, (sagA, and lysM2), the PGC-2 cluster (fms14, fms13, fms14), and general stress proteins (gls20, gls33, glsB) (Fig. 4; Additional file 2: Table S2). The remaining PVMs were variably detected without clear clonal or temporal association. While most were associated with hospital gene variants, community-associated E. faecium gene variants were also identified, with complete pili gene clusters observed in most isolates. The sagA community-associated variant was restricted to ST80-CT9087 isolates from 2016, whereas those ST80-CT9087 from 2015 and 2017 carried the hospital-associated variant (Fig. 4; Additional file 2: Table S2). E. lactis gene variants were limited to fms20, fms21 and ccpA (Fig. 4; Additional file 2: Table S2). The hylEfm frequency increased over time, while the hospital-associated ecbA variant was predominantly detected in ST117 isolates (p < 0.05). Hospital variants of tirE1 and tirE2 were exclusively detected in isolates from 2010–2012 (ST78-CT221, ST78-CT230 and ST412-CT258 vanA) and 2017 (ST192-CT26 vanB) (Fig. 4; Additional file 2: Table S2). Ten different bacteriocin genes were identified, with VREfm isolates carrying 1–6 genes each (Additional file 2: Table S3). Although the entire gene entA was universally present, only eight isolates (14%) contained the complete enterocin A locus. These included four early isolates, ST78/CT221 (n = 2; 2010) and ST78/CT230 (n = 2; 2011–2012) and four more recent ones, ST17/CT300 (n = 1; 2020), ST80/CT9089 (n = 2; 2020–2021), and ST80/CT6676 (n = 1; 2021). bac43 and bacAS9 were present in 79% and 43% of isolates, respectively (Additional file 2: Table S3). ST78 isolates harbored the highest average number (n = 5) of bacteriocin genes (Additional file 2: Table S3). Notably, bacAS9 was exclusive to ST117 (present in 89% of ST117 isolates and all ST117-CT24), including its pstS-negative variant (Additional file 2: Table S3). As noted by PCR, bac43 gene prevalence increased, with all bac43-positive isolates consistently carrying the rep11a_pB82 plasmid replicase (Additional file 2: Table S3). A total of 20 bacteriocin profiles were identified. While identical or highly similar profiles were mostly observed within the same STs, some identical profiles were also detected across different STs (Additional file 2: Table S4). For instance, ST117 isolates exhibited seven distinct bacteriocin gene combinations, with the most prevalent profile, found in 70% of isolates, comprising entA (most with an incomplete locus), bac43, and bacAS9 (Additional file 2: Table S4). This combination was not limited to a single clone but was shared across multiple ST117 clonal types (Additional file 2: Table S4). Of note, the remaining two ST117-CT24 isolates exhibited a similar bacteriocin profile but lacked bac43. VREfm plasmidome The 18 VREfm isolates analyzed by hybrid sequencing contained 113 plasmids (4–9 per isolate). Plasmid sizes ranged from 1,928 to 241,743 bp. Mapping and annotation identified 51 reps from six plasmid families, including 15 novel reps undetected by the initial PlasmidFinder search (Additional file 2: Table S5). rep gene families comprised RepA_N (37/113 plasmids), Rep3 (34/113), Inc18 (25/113), Rep_trans (20/113), Rep2 (3/113) and Rep1 (1/113). Notably, 21% (24/113) of plasmids were mosaics, incorporating up to two or three different rep families (Additional file 2: Table S5). Plasmids carrying Tn1546 ranged from 30 to 200 kb and were associated with different plasmid families (see section below). Interestingly, 11 out of 18 isolates carried linear plasmids (all isolated after 2016), six of which contained the Tn1546 transposon (see section below). All linear plasmids shared a Rep3-family RIP (repUS78_pZY2, repUS80_pSRR24 or rep41_pCEP232) (Additional file 2: Table S5). All but one VREfm carried small plasmids (2–18 kb) of the Rep3 and Rep_trans plasmid families. Rep_trans small plasmids (2–4 kb), predominantly found in ST117 isolates, mainly carried rep14a and rep14b, showing homology to the pEFNP1 and pRI1/pEfm755686-vanA plasmids, respectively (Additional file 2: Table S5). Rep3-like plasmids (n = 13; 5–10 kb), and occasionally larger mosaic plasmids (n = 4; 18–58 kb), carried bac43. While smaller plasmids were mostly cryptic only carrying bac43, two of the larger ones additionally harbored ARGs [erm(B) and aph(3′)-III] but no additional virulence or metabolic genes. All these plasmids encoded the rep11a_pB82 plasmid replicase (Additional file 2: Table S5). In turn, those Rep3-like carrying the bacAS5 contained the plasmid replicase rep18a_p200B (Additional file 2: Table S5). Plasmids of medium size (20–90 kb) exhibited the greatest diversity in reps and frequently carried ARGs such as erm(B), aph(3')-III, ant(6)-Ia, aac(6')-aph(2''), cat or Inu(B). The simultaneous presence of erm(B), aph(3')-III, ant(6)-Ia genes was exclusive to pRUM-like plasmid derivatives (Additional file 2: Table S5). All mega-plasmids (155–250 kb) consistently carried RepA_N-family RIPs (repUS15_pNB2354p1 or repUS15_DO3), sharing 99.9% and 98.6% nucleotide identity, respectively, with the well-known pLG1 megaplasmid. Most megaplasmids contained virulence genes (fms20, fms21, hylEfm), aac(6')-aph(2''), and the bacteriocin bacAS11 or bacAS12 genes (Additional file 2: Table S5). A similar plasmidome was observed in the sequenced VanB-carrying isolate, which harbored Tn1549:vanB on the chromosome. Mapping and annotation of vanA plasmids identified a total of 14 reps, including seven novel ones that were not detected in the initial PlasmidFinder search (see below; Additional file 2: Table S5). Tn1546::vanA plasmids belonged to either Rep3, RepA_N, Inc18 families or were mosaics containing RIPs of these three plasmid families and/or Rep_trans (Additional file 2: Table S5). The vanA-plasmids were grouped into two main communities (threshold ≤ 0.3) using the Pling tool (Fig. 5). The largest community included all 12 circular plasmids (Fig. 5A), while the other consisted exclusively of linear plasmids (Fig. 5B). The largest community was further subdivided into one sub-community, representing plasmid clusters with recent evolutionary histories and evidence of potential transmission events and one singleton: Plasmid communities of (A) circular and (B) linear plasmids, with communities defined based on plasmid containment distances (threshold = 0.3). Plasmid sub-communities (double-cut-join-indel distances ≤ 4) are distinguished by the inner node colors within each plasmid network, while STs are represented by the outer ring colors of the nodes The sub-community included 11 plasmids ranging from 38 to 62 kb, along with a 188 kb megaplasmid, recovered from ST78, ST80, ST117, and ST2073 isolates collected between 2010 and 2019 from different hospital wards and clinical sources (Fig. 5A). These plasmids harbored replicases from the RepA_N family or exhibited a mosaic structure (Additional file 2: Table S5). Within this group, we identified four novel replicases: rep14a_pACA-S-CCP502.3 and rep14b_pACA-S-CCP522.4 (both belonging to the RepA_N family), and repUS82_p4_03A17012 and repUS82_p4_03A17012 (from the Rep_trans family). Notably, rep14b_pACA-S-CCP522.4 shared 98.53% and 98.45% nucleotide identity with plasmid replicases previously described in isolates from Italy (accession no. OR262468) and Taiwan (accession no. CP131807), respectively. In contrast, repUS82_p4_03A17012 showed 69% nucleotide identity with pRUM and was detected in the majority of VREfm isolates (61%). In silico analysis further revealed that repUS82_p4_03A17012 is present in approximately 100 E. faecium genomes available in GenBank (half associated with infection, half with colonization) across various countries and exclusively appearing after 2000. These plasmids encoded various combinations of five different ARGs [aph(3')-III, ant(6)-Ia), aac(6')-aph(2''), aph(2'')-Ia and erm(B)] (Additional file 2: Table S5). The sub-community included 11 plasmids ranging from 38 to 62 kb, along with a 188 kb megaplasmid, recovered from ST78, ST80, ST117, and ST2073 isolates collected between 2010 and 2019 from different hospital wards and clinical sources (Fig. 5A). These plasmids harbored replicases from the RepA_N family or exhibited a mosaic structure (Additional file 2: Table S5). Within this group, we identified four novel replicases: rep14a_pACA-S-CCP502.3 and rep14b_pACA-S-CCP522.4 (both belonging to the RepA_N family), and repUS82_p4_03A17012 and repUS82_p4_03A17012 (from the Rep_trans family). Notably, rep14b_pACA-S-CCP522.4 shared 98.53% and 98.45% nucleotide identity with plasmid replicases previously described in isolates from Italy (accession no. OR262468) and Taiwan (accession no. CP131807), respectively. In contrast, repUS82_p4_03A17012 showed 69% nucleotide identity with pRUM and was detected in the majority of VREfm isolates (61%). In silico analysis further revealed that repUS82_p4_03A17012 is present in approximately 100 E. faecium genomes available in GenBank (half associated with infection, half with colonization) across various countries and exclusively appearing after 2000. These plasmids encoded various combinations of five different ARGs [aph(3')-III, ant(6)-Ia), aac(6')-aph(2''), aph(2'')-Ia and erm(B)] (Additional file 2: Table S5). A single 52 kb plasmid recovered from an ST117 isolate in 2012 (Fig. 5A; Additional file 2: Table S5). This plasmid carried only Inc18 family replicases (rep2_pRE25 and rep1_pSM19035) and encoded the ARGs erm(B), aph(3')-III, and ant(6)-Ia (Additional file 2: Table S5). A single 52 kb plasmid recovered from an ST117 isolate in 2012 (Fig. 5A; Additional file 2: Table S5). This plasmid carried only Inc18 family replicases (rep2_pRE25 and rep1_pSM19035) and encoded the ARGs erm(B), aph(3')-III, and ant(6)-Ia (Additional file 2: Table S5). The smaller community of linear plasmids was formed by six distinct plasmids, with only two forming a sub-community (closely related) (Fig. 5B). They were recovered from different clones (ST80, ST117, ST494), collected in 2019 and 2021, originating from various hospital wards and biological sources. These plasmids ranged in size from 96 to 119 kb and exclusively harbored replicases from the Rep3 plasmid family, including repUS78_pZY2 and repUS80_pSRR24, which were present in all isolates, and rep41_PR02395-7p2, found in the majority (4/6) of them. (Additional file 2: Table S5). Notably, no ARGs were identified within this group (Additional file 2: Table S5). Discussion This study is one of the few long-term analyses, and the first in Portugal, to investigate E. faecium dynamics from diverse human infections over an extended period at a single center. It highlights the interplay between clonal and plasmid expansion and the dynamic of VanA plasmids. Moreover, the local expansion of ST80 and ST117 underscores their global success and suggests that bacteriocins may play a key role in VREfm lineage replacement, alongside the emergence of vanA-carrying linear plasmids. The predominance of VanA-Efm observed in our study is similar to that of other European countries as France, Italy, and Belgium [13] and contrast with reports from Germany, The Netherlands, Poland, Denmark, and Norway, where vanB has gradually replaced vanA [13]. Consistent with previous studies [22, 30], all VREfm from this collection exhibited multidrug resistance. The steady rise in antibacterials for systemic use in the hospital sector, including linezolid, in Portugal, positions our country among the top five EU/EEA nations with the highest in-hospital consumption of antibiotics classified by the WHO as ‘Reserve’ agents [31]. Despite this, only two ST117-CT24 isolates in our dataset were resistant to linezolid, contrasting with the rising prevalence of linezolid-resistant infections reported in several countries [32, 33], driven by increased global use [34, 35]. Additionally, we identified mutations previously associated with decreased susceptibility to daptomycin. Daptomycin resistance in E. faecium is most commonly linked to mutations in liaFSR, which were highly prevalent among our isolates, whereas cls mutations, also frequently implicated, were not detected. Although some studies suggest that a single liaFSR mutation (e.g., liaS Gly92Asp) may confer intermediate resistance, and that additional cls mutations (e.g., Asn13Ser) are associated with high-level resistance (MIC ≥ 64 µg/mL), none of our isolates carried these specific variants, limiting interpretation in the absence of MIC data [36–38]. We also observed mutations in rpoC and dltC, which, while not considered primary resistance determinants, have been proposed to play secondary or compensatory roles, often occurring alongside liaFSR and other resistance-associated mutations [39]. Chlorhexidine MIC and MBC values remained well below in-use concentrations, yet sub-inhibitory exposure can decreased the biocide susceptibility [40] and promote horizontal gene transfer of clinically-relevant ARGs as vanA, poxtA or optrA [19]. Clinical VREfm exhibited a polyclonal structure, reflecting a global trend marked by the rise of dominant clones as ST80 and ST117, which were previously rare or absent [41–44]. Similar shifts have been reported, over the last decade, in countries as Germany [41], Ireland [42], China [43], and Denmark [44]. ST80 has been implicated in outbreaks in both China [43] and Denmark [44] while vancomycin-susceptible ST117 E. faecium became endemic in regions where VREfm rates are low [45]. In our study, ST117 was not detected prior to 2010 but subsequently established itself as a successful and persistent lineage (especially ST117-CT24). This lineage has driven significant outbreaks across Europe, including a key role in Germany's nationwide VREfm spread during the 1990 s [41], as well as in the Netherlands (2014, 2017) [46], Austria (2014) [47], Norway (2013–2014) [48] and Denmark (2015–2022) [49], highlighting its adaptability and transmission efficiency. Transmission efficiency of ST117 was particularly evident, with most putative clusters involving this lineage across both same-ward and inter-ward transmissions. However, ST117's higher baseline prevalence in our collection may contribute to this overrepresentation. Notably, several isolates fell outside transmission clusters, likely reflecting independent importations from community or other healthcare facilities, or undetected gastrointestinal colonization within the hospital. VREfm isolates consistently exhibited a high number and diversity of PVMs across years and clones, without a clear correlation to clonal waves. Only a few PVMs, including tirE1 and tirE2, were found in a small number of isolates, showing inconsistent geographical and temporal patterns. In Denmark, these PVMs appeared exclusively in emerging vanB clones (2019–2022; ST80/CT2406) [26], while in our collection, they were primarily linked to early vanA clones from 2010–2012 (ST78 and ST412), with a single detection in a vanB isolate from 2017 (ST192/CT26). More comprehensive studies are needed to draw definitive conclusions once these genes are integrated into the CGE virulence database. No specific gene or combination of known AMR or PVMs clearly explained an advantage in the hospital environment or the observed shifts in STs and CTs over the study period. However, distinct bacteriocin profiles were linked to specific VREfm clones. The majority of ST117-CT24 exhibited a unique bacteriocin gene profile (entA, bac43, bacAS9), detected exclusively from 2011 and 2013–2019. Although enterocin A is widespread in E. faecium [12, 50], most isolates lack the complete locus, indicating it may be non-functional, as a recent study suggests [20]. Bacteriocin 43 exhibits proven antimicrobial activity against Listeria monocytogenes and different Enterococcus species including E. faecium [51], which may provide a competitive advantage in the dysbiotic gut of hospitalized patients. This suggests it could be a highly specialized adaptive trait acquired over time by specific VREfm strains [12]. In fact, the increasing prevalence of bac43 over time, especially since the early 2000 s in our and other countries [22], is notable, as confirmed by recent studies [20, 46, 52]. Lisotto et al. identified bac43 in a VanB-ST117-CT24 hospital outbreak strains in the Netherlands (2017), noting differences in accessory genes related to prophages and bacteriocin content, while virulence and ARGs profiles remained similar [46]. Recent U.S. studies on VREfm population dynamics over a six-year period also revealed a rising prevalence of bac43/T8. In 2024, Mills et al. (2025) reported a shift from ST17 to ST78-related clones over a 20-year period (2002–2022) [52], linking bac43/T8 to increased prevalence and suggesting that emerging strains inhibit historical ones. Similarly, Garretto et al. (2024) found bac43 frequency rose from 8% (2016) to 91% (2021), coinciding with the dominance of ST117 and ST80 [20]. While no causal relationship was established, bac43 may act as a selfish plasmid, ensuring its survival [20]. Our study supports the idea that bac43, located on small mobilizable Rep3-like plasmids (pB82-like, 6–8 kb), may play a role in enhancing colonization within healthcare settings. While the crucial role of plasmids in shaping the emergence of successful E. faecium clones is undeniable [14, 22], few studies have conducted an in-depth analysis of the plasmidome in E. faecium [14, 22, 45]. E. faecium from hospitalized patients harbor a significantly larger and more complex plasmidome compared to other sources, suggesting a key role in source or age specificity [14]. The pan-plasmidome of E. faecium has been proposed as a driving factor in its evolution into a prominent nosocomial pathogen [14], with a successful plasmidome largely restricted to clinical contexts [22, 30]. Our findings indicate no single dominant plasmidome configuration among VREfm isolates, pointing to multiple pathways for plasmid acquisition and frequent recombination in the hospital environment. Nonetheless, the VREfm plasmidome includes stable mobilizable plasmids, carrying bacteriocins, and RepA_N megaplasmids harboring virulence and ARGs, while vanA genes reside on medium-to-large plasmids with high plasticity and rep variation. These insights underscore the need for ongoing surveillance to track not only plasmids linked to clinically relevant genes but also small mobilizable plasmids with potentially important bacteriocins. Previous studies identified vanA-type vancomycin resistance predominantly on derivatives of pRUM, pLG1, and Inc18 plasmids [22, 53]. While pRUM and Inc18 types remained detectable in our collection, they declined in favor of apparently novel replicases, such as repUS82_p4_03A17012 and repUS78_pZY2. Notably, medium-sized circular mosaic plasmids combining RepA_N, Inc18, Rep3 and/or Rep_trans replicases were increasingly replaced after 2019 by large-to-mega linear plasmids carrying Rep3 family replicases, linked to different CTs. Until 2019, the ST117-CT24 clone carried diverse vanA plasmids (e.g., Inc18 or RepA_N-Inc18/RepA_N-Rep3 mosaics). However, from 2019–2021, a specific linear plasmid type expanded across ST80, ST117, and ST494. These linear plasmids also emerged in vancomycin-variable enterococci from the same hospital [54], carrying a functional Tn1546 transposon with truncated genes (e.g., transposase, resolvase, vanZ). Similar linear plasmids have been reported globally, including Japan [55, 56], India [57], Ireland [42], France [58], South Korea [59] and Latvia [58]. Although reports of linear plasmids in E. faecium are still limited, emerging evidence highlights their increasing role in spreading resistance genes [57], particularly vanA. These plasmids seem to exhibit remarkable plasticity, horizontal self-transmissibility [58], and potential advantages in nutrient utilization [60]. Nonetheless, their overall impact on bacterial fitness and resistance evolution remains to be fully understood. It is important to note that we cannot definitively conclude whether the replicases identified in this study were introduced recently. Previous PCR schemes lacked the sensitivity to detect them, and even whole-genome sequencing failed to identify some replicases when using the PlasmidFinder tool. For instance, while our PCR scheme showed a decline in the pLG1 megaplasmid, all megaplasmids identified were linked to US15 replicases from the RepA_N family, showing > 98% nucleotide identity to pLG1. Despite some limitations, including the absence of clinical vancomycin-susceptible and colonizing isolates (either VSEfm or VREfm) and the sequencing of only a subset of the available isolates, and the lack of consistent data on individual antibiotic exposures (e.g., prior or ongoing vancomycin or other antibiotic treatments), the long duration of our study at a single institution provides a robust view of E. faecium population dynamics and reflects broader regional trends. Although these gaps may limit the completeness of genomic diversity captured, the temporal breadth of the sequenced isolates allows us to identify the major circulating lineages and their evolutionary trajectories. Importantly, the use of long-read sequencing was essential for resolving complex genomic structures, such as the linear plasmids identified in this study, which would likely have remained undetected with short-read sequencing alone. Moreover, this is one of the few studies that provides an exhaustive characterization of the contemporary VREfm plasmidome, extending well beyond the traditional focus on vancomycin-resistance plasmids. Overall, our findings of highly plastic, recombinogenic plasmids, including linear vanA plasmids and multi-replicon structures, fits within recent evolutionary insights showing that, while most pre-antibiotic era plasmids remained genetically stable and lacked resistance determinants, a minority evolved into the complex, recombinogenic plasmids now driving the global dissemination of AMR [61]. These findings reinforce that contemporary VREfm plasmids represent the modern, high-risk stage of plasmid evolution shaped by human antibiotic use. Conclusions This study highlights how plasmids contribute to the genomic diversity and adaptive potential of VREfm lineages. The emergence and persistence of clones such as ST80 and ST117, alongside the rise of bacteriocin-carrying and linear plasmids, suggests that the interplay between clonal evolution and mobile genetic elements shapes the epidemiology of VREfm. More broadly, these findings underscore the value of integrating plasmidome and accessory genome analyses alongside conventional strain typing and resistance profiling into AMR surveillance and infection control strategies. Considering the intra-species heterogeneity of E. faecium, where only a subset of specialized lineages drives most AMR acquisition and transmission, is essential for accurately identifying high-risk clones and transmission routes. This approach should also incorporate non-conventional indicators, such as resistance genes, plasmids, and other mobile genetic elements that play a critical role in the spread of resistance. From a translational perspective, bridging genomic data with clinical surveillance will be crucial for designing adaptive infection control strategies. Continuous genomic-based monitoring, judicious antimicrobial stewardship, and coordinated data sharing across institutions will enhance our ability to anticipate and contain the spread of highly adaptive VREfm lineages within healthcare networks. Supplementary Information Additional file 1: Supplementary figures. Figures S1–S4. Fig S1A) Average minimum inhibitory concentrations (MIC) and minimum bactericidal concentrations (MBC) of chlorhexidine in a representative subset of isolates (n=63). B) Distribution of MICs and MBCs by year and sequence type. Fig.S2A Overall percentage of virulence markers detected by PCR. B) Annual distribution of virulence markers detected by PCR. Fig. S3 A) Overall percentage of plasmid replicases [rep17_pRUM (GenBank accession no.: AF507977) , rep20_pLG1 (GenBank accession no.:CP004064), and rep1_pSM19035 (GenBank accession no.:AY357120) and rep2_pRE25 (GenBank accession no.:X92945), by PCR. B) Temporal dynamics of the same plasmid replicases detected by PCR. Fig. S4. Distribution of the bacteriocin 43 gene over a 12-year period. Additional file 2: Supplementary tables. Tables S1–S5. Table S1. Acquired genes and chromosomal mutations conferring antibiotic resistance in VREfm strains. Table S2. Virulence genes present in the VREfm strains. Table S3. Plasmid replicases and bacteriocin genes present in the VREfm strains. Table S4. Bacteriocin profiles present in the VREfm strains. Table S5. Plasmidome of the sequenced VREfm. Acknowledgements Not applicable Abbreviations Antimicrobial Resistance Antibiotic Resistance Genes Bacteriocin 43 Blood Stream Infections Chlorhexidine Core genome Multilocus Sequence Typing Complex Type Healthcare-Associated Infections Minimum Bactericidal Concentrations Mobile Genetic Elements Minimum Inhibitory Concentrations Multi Locus Sequence Typing Plasmid Replicases Single Nucleotide Polymorphisms Sequence Type Vancomycin Resistant Enterococcus faecium Whole-Genome Sequencing Authors’ contributions **A. C. A-S:** Writing original draft, Writing – review & editing, Investigation, Methodology, Conceptualization, Validation, Software, Data Curation. **A.P. T.:** Methodology, Software, Formal analysis, Writing – review & editing. **B. D.:** Investigation, Methodology. **A.P. P.:** Investigation, Methodology, **L. M. S.:** Resources **, J. T.:** Resources **, A. P. C.:** Resources **, L. R.:** Methodology, Writing – review & editing **, A. M. H.:** Writing – review & editing **, H. H.:** Writing – review & editing **, T.M.C.:** Writing – review & editing **, C. N.:** Writing – review & editing, Conceptualization, Supervision, **A. R. F.:** Writing – review & editing, Methodology, Conceptualization, Validation, Project administration, Supervision, Resources, Funding acquisition, **L.P.:** Writing – review & editing, Conceptualization, Supervision, Resources, Funding acquisition. All authors read and approved the final manuscript. Funding This work was supported by FCT—Fundação para a Ciência e Tecnologia, I.P., in the scope of the project UID/04378/2025 (DOI identifier: 10.54499/UID/04378/2025), and UID/PRR/04378/2025 (DOI identifier: 10.54499/UID/PRR/04378/2025), of the Research Unit on Applied Molecular Biosciences—UCIBIO and the project LA/P/0140/2020 (DOI identifier 10.54499/LA/P/0140/2020) of the Associate Laboratory Institute for Health and Bioeconomy—i4HB. Ana C. Almeida-Santos acknowledges her PhD fellowship (10.54499/UI/BD/151317/2021) from FCT/MCTES through national funds. Data availability The data generated in this study were made available in the NCBI database at https://www.ncbi.nlm.nih.gov/bioproject/PRJNA1229488, under the BioProject “PRJNA1229488” [62]. Assembly accession numbers are listed in Additional file 2: Table S1. Declarations Ethics approval and consent to participate This study was approved by the Ethics Committee of Centro Hospitalar do Porto (Comissão de Ética da ULS de Santo António, ULSSA), Portugal. The requirement for informed consent was waived by the ethics committee, as the study used samples collected as part of routine clinical practice, with no additional procedures performed for research purposes, and all data were analyzed in anonymized form. All procedures were conducted in accordance with the ethical standards of the institutional research committee and the Declaration of Helsinki. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Ana R. Freitas, Email: ana.freitas@iucs.cespu.pt Luísa Peixe, Email: lpeixe@ff.up.pt. References Associated Data Supplementary Materials Additional file 1: Supplementary figures. Figures S1–S4. Fig S1A) Average minimum inhibitory concentrations (MIC) and minimum bactericidal concentrations (MBC) of chlorhexidine in a representative subset of isolates (n=63). B) Distribution of MICs and MBCs by year and sequence type. Fig.S2A Overall percentage of virulence markers detected by PCR. B) Annual distribution of virulence markers detected by PCR. Fig. S3 A) Overall percentage of plasmid replicases [rep17_pRUM (GenBank accession no.: AF507977) , rep20_pLG1 (GenBank accession no.:CP004064), and rep1_pSM19035 (GenBank accession no.:AY357120) and rep2_pRE25 (GenBank accession no.:X92945), by PCR. B) Temporal dynamics of the same plasmid replicases detected by PCR. Fig. S4. Distribution of the bacteriocin 43 gene over a 12-year period. Additional file 2: Supplementary tables. Tables S1–S5. Table S1. Acquired genes and chromosomal mutations conferring antibiotic resistance in VREfm strains. Table S2. Virulence genes present in the VREfm strains. Table S3. Plasmid replicases and bacteriocin genes present in the VREfm strains. Table S4. Bacteriocin profiles present in the VREfm strains. Table S5. Plasmidome of the sequenced VREfm. Data Availability Statement The data generated in this study were made available in the NCBI database at https://www.ncbi.nlm.nih.gov/bioproject/PRJNA1229488, under the BioProject “PRJNA1229488” [62]. Assembly accession numbers are listed in Additional file 2: Table S1.