Vancomycin Resistance in Enterococcus faecium from the Dallas, Texas, Area Is Conferred Predominantly on pRUM-Like Plasmids

ABSTRACT Vancomycin-resistant E. faecium (VREfm) is a significant public health concern because of limited treatment options. Genomic surveillance can be used to monitor VREfm transmission and evolution. Genomic analysis of VREfm has not been reported for the Dallas/Fort Worth/Arlington, TX, area, which is currently the 4th largest metropolitan area in the United States. Our study aimed to address this gap in knowledge by analyzing the genomes of 46 VREfm strains and 1 vancomycin-sensitive comparator collected during routine fecal surveillance of high-risk patients upon admission to a Dallas, TX, hospital system (August to October 2015). Thirty-one complete and 16 draft genome sequences were generated. The closed VREfm genomes possessed up to 12 extrachromosomal elements each. Overall, 251 closed putative plasmid sequences assigned to previously described and newly defined rep family types were obtained. Phylogenetic analysis identified 10 different sequence types (STs) among the isolates, with the most prevalent being ST17 and ST18. Strikingly, all but three of the VREfm isolates encoded vanA-type vancomycin resistance within Tn1546-like elements on a pRUM-like (rep17) plasmid backbone. Relative to a previously reported typing scheme for the vanA-carrying Tn1546, new variants of the Tn1546 were identified that harbored a combination of 7 insertion sequences (IS), including 3 novel IS elements reported here (ISEfa16, ISEfa17, and ISEfa18). We conclude that pRUM-like plasmids are important vectors for vancomycin resistance in the Dallas, TX, area and should be a focus of plasmid surveillance efforts. IMPORTANCE Vancomycin is an antibiotic used to treat infections caused by multidrug-resistant Gram-positive bacteria. Vancomycin resistance is common in clinical isolates of the Gram-positive pathogen Enterococcus faecium. Among E. faecium strains, vancomycin resistance genes can be disseminated by plasmids with different host ranges and transfer efficiencies. Surveillance of resistance plasmids is critical to understanding antibiotic resistance transmission. This study analyzed the genome sequences of VREfm isolates collected from the Dallas, TX, area, with particular focus on the mobile elements associated with vancomycin resistance genes. We found that a single plasmid family, the pRUM-like family, was associated with vancomycin resistance in the majority of isolates sampled. Our work suggests that the pRUM-like plasmids should continue to be studied to understand their mechanisms of maintenance, transmission, and evolution in VREfm.

and widespread occurrence of vancomycin resistance among infection isolates of E. faecium (4).
Phylogenetic analysis of E. faecium strains revealed the presence of two major clades, referred to as clades A and B, with some evidence for subclades A1 and A2 within clade A (5,6). Clade B E. faecium strains are mostly commensals of the healthy human gut. The multidrug-resistant E. faecium strains responsible for hospital outbreaks typically belong to clade A1, whereas animal gut commensals typically belong to clade A2 (6). Multilocus sequence typing (MLST) predates whole-genome-based phylogenetic analyses and has been used globally to classify E. faecium isolates by nucleotide sequence variations occurring in 7 housekeeping genes (7). Isolates are assigned to different sequence types (STs) based on the allelic variants. Most reported hospital outbreaks of VREfm have emerged from a single genetic lineage, referred to as clonal complex 17 (CC17), that was founded by ST17 (8).
Vancomycin resistance genes are carried within mobile elements in VREfm and can be horizontally disseminated (9)(10)(11). VanA-type resistance commonly occurs among VREfm strains, and VanA-type resistance genes are typically carried within Tn1546 (12). Tn1546 may be chromosomally integrated or plasmid borne. Different plasmid backbones carrying Tn1546, including Inc18, pRUM, and pLG1, have been reported worldwide (for examples, see references 10 and 13). Structural variations of Tn1546 due to the presence of insertion sequences (IS), including IS1251, IS1216, IS1485, and ISEf1, have also been reported. These previous studies characterized Tn1546 structural variation by PCR mapping and then sequencing of the overlapping fragments (10,13,14). However, in the absence of a completely closed genome assembly, conclusive links between vanA, Tn1546 variants, and specific plasmid backbones in VREfm can be difficult to achieve.
Surveillance of antibiotic resistance elements is important because changes in their host range and transfer frequency could evolve, which could impact clinical care. Utilization of whole-genome sequencing for epidemiological study provides an opportunity for multilevel analysis that includes detection of novel mobile genetic elements and monitoring DNA sequence changes in resistance gene vectors. The advent of long-read next-generation sequencing (NGS) by Pacific Biosciences and Oxford Nanopore Technologies (ONT) has made it possible to generate genome assemblies of a large number of bacterial isolates in a cost-effective manner. The combination of long (ONT in this study) and short (typically Illumina) reads generates high-quality, completely closed genome assemblies with fully assembled mobile genetic elements (15).
In this study, we analyzed phylogenetic relationships and plasmid diversity among a previously reported collection of VREfm isolates from the Dallas, TX, area (16). These isolates were recovered from rectal surveillance of high-risk patients upon admission to a Dallas hospital system. These represent colonization reservoirs from which future hospital outbreaks could emerge. Some results of our study were expected, including the predominance of CC17/clade A1 among the VREfm isolates. Other results highlight important areas for future study. Specifically, we note the high number of extrachromosomal elements in VREfm, with some isolates possessing 12 unique elements independent from the chromosome. How these multiple elements are coordinated and maintained in VREfm, and what they contribute to VREfm physiology, are largely unknown and are important topics for future investigation. Moreover, pRUM-like plasmids are the dominant carriers of vancomycin resistance genes among these VREfm isolates. This plasmid family should continue to be investigated in laboratory experiments to understand its maintenance and transmission mechanisms.

RESULTS
ST17 and ST18 VREfm isolates were prevalent in the collection. Genome sequencing and hybrid assembly were performed for 47 E. faecium isolates with Oxford Nanopore MinION and Illumina technologies. Thirty-one closed and 16 draft genome assemblies were generated (see Data Set S1A in the supplemental material). In our previous analysis of these isolates using PCR, all isolates were found to be vanA positive and vanB negative, with the exception of the isolate 163-1, which had neither vanA or vanB and had a vancomycin MIC of 2 mg/mL by broth microdilution (16). We included 163-1 in this study for comparative purposes to evaluate its relationship to the vanA-positive isolates in the collection. After recovery from Spectra VRE rectal surveillance cultures, all 46 vanA-positive isolates sequenced in this study grew on agar supplemented with 256 mg/mL of vancomycin. Subsequent broth microdilution assays performed over a year later after genome sequencing and freezer restocking reconfirmed vancomycin resistance for all isolates except for 9-2 and 154-1, which had vancomycin MICs of #4 mg/mL. Likely, loss of vancomycin resistance elements during laboratory culture led to these phenotypes; however, this was not investigated further in this study.
Phylogenetic analysis based on alignment of 1,706 core genes allowed us to assess the diversity of the VREfm isolates (Fig. 1). Two reference genomes, E. faecium 1,231,501 (5) and E. faecium ATCC 700221 (17), were included as representatives of E. faecium clades A2 and A1, respectively. E. faecium 1,231,501 is a human bloodstream isolate from the United States and is susceptible to vancomycin. E. faecium ATCC 700221 is a U.S. isolate and VanA-type VREfm control strain used in antimicrobial susceptibility testing. As expected, the Dallas VREfm isolates clustered with the clade A1 reference, E. faecium ATCC 700221.
Ten different sequence types (STs), all belonging to clonal complex 17 (CC17), were identified in the collection ( Fig. 1 and 2). Isolates belonging to ST17 and ST18 were the most prevalent in the collection, comprising 50% of the isolates, which clustered together in the core genome tree ( Fig. 1 and 2). Six STs (ST18, ST664, ST1703, ST736, ST80, and ST612) are double-locus variants of ST17. ST262 and ST412 are triple-locus variants and ST1383 is a single-locus variant of ST17 (Data Set S1D). ST1703, a double-locus variant of ST17 (pstS allele 20; novel gyd allele 70), is a novel ST in the MLST database and comprised 11% of the collection. We conclude that E. faecium isolates of multiple different STs, all within CC17, colonized the Dallas patients sampled at the time of the study.
We performed an additional analysis to determine how closely related the E. faecium strains in our collection are. All-versus-all average nucleotide identity (ANI) was calculated with all E. faecium assemblies as input (Data Set S1E). Dallas E. faecium isolates possess $98.98% ANI, consistent with the core gene phylogeny in Fig. 1. In most cases, strains isolated from the same patient were identical or nearly so. For example, the three isolates from patient 124 shared 99.99% ANI. Similar observations were made for isolates from patients 100, 16, 53, 71, and 9. Exceptions were patients 51 and 93, who each were colonized with different E. faecium STs (Fig. 1), also reflected in the ANI analysis of the isolates (#99.4% ANI).
Large numbers of plasmids were present in the Dallas isolates. The plasmid content of the Dallas isolates was analyzed by an in silico detection method utilizing the conserved nucleotide sequences of replication initiation (rep) genes (18). Each contig of ,2 Mbp in size in a closed genome assembly was considered a putative plasmid. Of the 259 closed circular putative plasmids, 141 were assigned to 10 known rep families described in the PlasmidFinder database. The remaining 118 putative plasmids were categorized into 9 novel plasmid groups (pMI1 to pMI9) and 8 other elements (discussed in the next section).
The Dallas isolates harbored variable numbers of putative plasmids (n = 3 to 12) of sizes ranging from 1.9 to 294 kb ( Fig. 1; Data Set S1F). The most common elements were a megaplasmid (126 to 294 kb) and a pRUM-like plasmid (17 to 63 kb), which were each present in all isolates except 144-1, which lacked the megaplasmid. Both the reference strains (E. faecium 1,231,501 and ATCC 700221) harbored the megaplasmid, but 1,231,501 lacked the pRUM-like plasmid. Each of the megaplasmids harbored the rep gene repUS15, encoding a RepA_N protein motif. The pRUM-like plasmid harbored rep17 or both rep17 and rep18, which we refer to as a mosaic plasmid. An analysis of plasmid replicon enrichment among specific STs is presented in Text S1 and Data Set S1G.
Uncategorized plasmids were sorted into 9 plasmid groups, pMI1 to pMI9. PlasmidFinder (18) did not identify rep genes in a total of 118 circular entities. Of these 118 circular entities, 110 were established to be plasmids based on the presence of rep genes not currently represented in the PlasmidFinder database (see Materials and Methods). The remaining 8 circular entities were designated putative excision products of genomic islands (Text S1; Data Set S1H and I), including a 5-kb circular fragment of Tn1546 in isolate 111.
The 110 plasmids not classifiable by PlasmidFinder were sorted into 9 plasmid groups, pMI1 to pMI9, based on phylogenetic analyses of the rep genes ( Fig. S1; Data Set S1J). The nucleotide sequences of the rep genes and their corresponding amino acid sequences shared .95% identity within each pMIX plasmid group, except for pMI4 ( Table 1). The clustering of the rep gene sequences belonging to the pMI4 plasmid group supported their inclusion in a single group, despite lower pairwise sequence identities (Fig. S1). A total of 6 Rep protein families (RepA_N, Rep_3, Rep_1, Rep_trans, Rol_rep_N, and Rep_2) were encoded by the rep genes of the 9 pMIX groups (Data Set S1J). Specifically, the rep genes of pMI1 and pMI3 encoded protein families RepA_N and Rep_1, respectively. The rep genes of pMI2, pMI5, pMI6, and pMI7 encoded the protein family Rep_3. The rep genes of pMI4 and pMI8 encoded protein families Rep_trans and Rep_2, respectively. The rep of pMI9 was the only exception, encoding two Rep protein families, Rep_trans and Rol_rep_N ( Table 1).
Each of the Dallas isolates harbored 1 to 5 of the pMIX plasmid groups, except isolate 83 (ST17), which had none of them ( Fig. 1). Neither of the two reference strains harbored any of the pMIX plasmid groups. pMI1 was the most frequently occurring pMIX plasmid group, identified in 75% of the isolates (Table 1). pMI8 and pMI9 were the rarest ones, found only in isolates 51_1 (ST612) and 5 (ST18), respectively. The pMIX plasmid groups were not enriched in any specific ST. We observed that plasmids within a very narrow size range (standard deviation, 0.2 to 2.3) were grouped together by this rep typing scheme (with the exception of pMI1), although plasmid size was not taken into consideration in assigning plasmid groups.
We used PLSDB (19,20) to compare representative rep sequences from the pMIX plasmid groups with existing plasmids in the database (as of January 2023). PLSDB compiles plasmid sequences and their associated metadata, including PlasmidFinder and other analyses. This was done as an additional check on our analysis, as ours was initially performed in 2018-2019. We reasoned that the pMIX typing scheme would be most useful if it encompassed other previously untypeable plasmids in the database.  We applied strict thresholds of 90% identity and 90% query coverage to filter BLASTn hits to the database. The results support pMI1, pMI3, pMI5, pMI6, and pMI8 as newly defined rep types that include other previously untyped Enterococcus plasmids from around the world (Data Set S1K). The only hit to pMI9 rep was the Dallas isolate from which it was defined. In this PLSDB analysis, pMI2 was assigned to rep11a, and pMI4 was assigned to rep14b. The pMI7 analysis was more complicated, with one group of identical hits (100% identity and query coverage) for many plasmids with no known rep type assigned and another group of hits with 100% identity but lower (98%) coverage that were assigned to rep17.
All of the Dallas isolates, except for 163-1, carried VanA-type vancomycin resistance genes. The vanA genes possessed 99.9% nucleotide and 99.7% amino acid sequence identity with the reference vanA sequence (GenBank accession number AAA65956.1 (Data Set S1L). The reference sequence is encoded within the transposon Tn1546 (GenBank accession number M97297), originally described for the human fecal isolate E. faecium BM4147, collected from France in 1988 (21) (Fig. S2). A classical structure of the vanA gene cluster, consisting of vanHAX and vanYZ carried within Tn1546 elements, was observed among the Dallas VREfm isolates, except for isolate 17-1, which lacked the genes vanY and vanZ. The Tn1546 structure varied among the isolates in terms of insertion site (chromosome versus plasmid), IS elements, and point mutations, discussed further below.
Tn1546 was borne on pRUM-like plasmids in most isolates. The complete Tn1546 was carried by the pRUM-like plasmid in 42 of 46 vanA-positive isolates. A representative pRUM-like plasmid from our collection is shown in Fig. 4. Isolate 111 did not carry the entire Tn1546 in the pRUM-like plasmid; rather, it carried vanRS in a 17-kb pRUM-like plasmid, while vanHAX was carried within a separate 5-kb circular element (te111_5kb) (Data Set S1H). Tn1546 was chromosomally borne in 3 isolates, 53-1, 53-2, and 91, each belonging to ST17, with all three strains also harboring a pRUM-like plasmid lacking Tn1546 (Table 2).
We categorized the 43 pRUM-like plasmids encoding Tn1546 or a fragment of Tn1546 (for isolate 111) into four groups based on the presence or absence of replication and stability modules on their backbones (Table 2; Fig. 5), per a previously described typing scheme (14). For the plasmids in our collection, two toxin-antitoxin (TA) systems were observed in different distributions, axe-txe (TA axe-txe ) and a relE-parE family toxin associated with an uncharacterized antitoxin component (TA relE ) ( Table 2). Both of the TA systems were actively expressed in two representative VREfm isolates (Fig. S3). Plasmid groups. (i) Group 1 plasmids (n = 16). Group 1 plasmids possess rep17, rep18, and both TA axe-txe and TA relE . All the group 1 plasmids carried rep18b (GenBank accession number CP018068), except isolate 113, which carried rep18a (GenBank accession number AB158402). The plasmids had a narrow size range (44 to 48 kb) and were present in VREfm isolates from ST17, ST18, ST1383, and ST80.
(ii) Group 2 plasmids (n = 17). Group 2 plasmids possess rep17 and both TA axe-txe and TA relE . The group 2 plasmids were classified further based on size and the genetic background of the isolates carrying them. Group 2a plasmids (29 to 45 kb) were carried by the reference strain ATCC 700221 and isolates belonging to ST18 (5, 66, 97-1, 124-1, a The presence of the gene is indicated by shaded boxes. The pRUM-like plasmid from isolate 113 harbored rep18a (reference gene; GenBank accession number AB158402) and is symbolized as "a" in the corresponding box. The rep18 gene for the rest of the resistance plasmids was rep18b (reference gene; GenBank accession number CP018068). TA, toxin-antitoxin.
(iv) Group 4 plasmids (n = 5). Group 4 plasmids possess rep17 and no known stability modules and have a size range of 17 to 35 kb. These plasmids were carried by isolates belonging to the novel genetic background ST1703 (111, 137, 154-1, and 158) and by a single ST412 isolate (17-1).
We further analyzed the pRUM-like plasmids in our collection using two approaches: a core genome phylogeny based on sequence variation in 7 pRUM-like plasmid core genes (rep17, IS1216, vanS, and genes for ImpB/MucB/SamB family protein, hypothetical protein, plasmid partitioning protein, and replication control protein PrgN) (Fig. S5) and an analysis of ANI ( Fig. S6; Data Set S1N). The pRUM-like plasmids form two clusters in the phylogenetic tree (Fig. S5), which correspond to the presence/absence of TA axe-txe . The pairwise ANI of the pRUM-like plasmids ranged from 93.9 to 100%. ANI analysis identified well-defined clusters of plasmids with high percent ANI that correspond to plasmid groups 2b, 3, and 4, while groups 1 and 2a were intermixed (Fig. S6). This is also consistent with the presence/ absence of TA axe-txe being an important discriminator for grouping of pRUM-like plasmids.
Finally, we compared pRUM-like/rep17 plasmids from the Dallas collection with other rep17 plasmids that have previously been published. PLSDB was queried to obtain circular plasmids assigned to genus Enterococcus and to rep17. A strict cutoff of $99.9% pairwise ANI was applied, which identified a cluster of closely related pRUMlike/rep17 plasmids from different sources (Fig. 6). These plasmids have $99.9% ANI, are from geographically distinct regions (United States [this study], France [23], India [24], and Australia [25,26]), and are present in different E. faecium STs. The most closely related are the pRUM-like/rep17 plasmid from Dallas isolate 16-1 and the rep17 plasmid from the VREfm strain 15-307.1. These plasmids have identical gene contents and differ by single nucleotide polymorphisms (SNPs) and deletions/insertions at 10 sites. VREfm 15-307.1 was isolated in France in 2015 from a rectal swab of a hospitalized patient, which is similar to the collection of Dallas 16-1 in 2015.
Dallas VREfm isolates harbored 17 variants of Tn1546. The Tn1546 region of the Dallas isolates varied from the typical structure, which consists of ORF1 (transposase), ORF2 (resolvase), and the van genes flanked by IR L (inverted repeat, left) and IR R (inverted repeat, right) (Fig. 7). A total of 17 different sequence variations in the Tn1546 were observed among 45 of the Dallas isolates ( Fig. 7; Table 3). These 17 variations were classified using a nomenclature system developed previously (13) based on the presence of IS elements, point mutations, and deletions relative to the reference, Tn1546. VREfm isolates belonging to the same Tn1546 group shared 99.8% to 100% nucleotide sequence identity in the Tn1546 region. Isolate 111 was excluded from this grouping since it harbored a fragmented Tn1546 occurring in a pRUM-like plasmid and another circular entity (Data Set S1H). The Tn1546 variation was mostly associated with the distribution of 7 different IS elements (IS1251, IS1216, ISEfa16, ISEfa5, ISEfa17, IS256, and ISEfa18), three of which were novel to the Dallas isolates. Most of the Tn1546 groups described here are novel relative to those previously reported in the literature, except for group BC2, which was reported previously in Poland (13). The most predominant Tn1546 groups were BC9 (n = 14) and J (n = 6). The BC9 variety was observed in ST17 (n = 9) and ST18 (n = 5) isolates. The reference strain, ATCC 700221, harbored the C2 structural variant of Tn1546 that was described previously (13) but was not present in the Dallas VREfm isolates. The presence of different pRUM groups and Tn1546 types across the phylogeny of E. faecium isolates in this study is presented in Fig. S7.
Specific structural variations of the Dallas Tn1546 elements are described further in Text S1.
Potential VREfm patient-patient transmission. We used our genome data to look for identical E. faecium clones colonizing different patients. This would be suggestive  of prior transmission of VREfm between patients in our area. We found one possible instance supported by our genomic data. E. faecium isolates from patients 55 and 52 were of identical ST (Fig. 1), the same pRUM plasmid size and group (Table 2), and the same Tn1546 group (and were the only members of the group in this study) (Table 3) and possess 99.99% ANI overall (Data Set S1E). Their pRUM plasmids possess .99.99% ANI (Data Set S1N), differing by one SNP. However, the total genome sizes for these two strains differ. The isolate from patient 55 has a closed genome that is 1,026 bp longer than the draft genome of the patient 52 isolate (Data Set S1A).

DISCUSSION
This study provided a snapshot of fecal VREfm colonizing hospitalized patients in Dallas, TX, from August to October 2015. VREfm isolates from previously described and new STs were detected. The most prevalent VREfm isolates, comprising 50% of those analyzed, belonged to ST17 and ST18. These STs were previously reported to be responsible for outbreaks in countries including Portugal (14), Spain (27), Ireland (28), Denmark (29), Columbia (30), Brazil (31), Canada (32). The other 8 STs identified among our collection (ST1383, ST612, ST412, ST80, ST1703, ST664, ST736, and ST262), were single, double, or triple locus variants of ST17. ST1703 is a novel sequence type described in this study. ST182, which caused an outbreak in San Antonio, TX, in the 1990s (33), was not detected in our collection.
All the plasmid-borne Tn1546 elements in Dallas isolates were carried within pRUMlike plasmids that were clustered into 4 subgroups (1, 2a, 2b, 3, and 4), defined by the presence of the replication module rep17 and a combination of other replication and stability modules. The patients from whom the isolates in this study were collected resided in different areas of Dallas and in unconnected facilities, including nursing homes and personal residences. Therefore, the presence of only one type of plasmid backbone carrying vanA among these patients was striking. Our data are consistent with another recent analysis of VREfm from the United States. Chilambi et al. (34) analyzed gastrointestinal and blood VREfm isolates from pediatric patients at St. Jude's Children's Research Hospital over a 10-year period. VanA-type VREfm was isolated from 23 of 24 patients analyzed, and for all 23, Tn1546 was carried on a rep17 element. Clearly, these plasmids are significant vectors for vancomycin resistance in E. faecium, as proposed by Freitas et al. (10) in a multilevel analysis of VREfm. In our study, multiple predicted antibiotic resistance genes were detected on these plasmids, in addition to vancomycin resistance genes. The pRUM-like plasmids should be a focus of plasmid surveillance. Resistance genes may be consolidating on this specific backbone, as opposed to the many other plasmids detected in VREfm in this study.
We detected carriage of up to 12 extrachromosomal elements in the VREfm isolates in our study. This is consistent with the comprehensive work of Arredondo-Alonso et al., who conducted a large-scale analysis of the E. faecium plasmidome (35). The authors determined that E. faecium isolates from hospitalized patients possess the largest plasmidome in terms of plasmid number and total base pairs. The striking number of extrachromosomal elements among VREfm isolates warrants further investigation into their maintenance, transmission, and functions.

MATERIALS AND METHODS
Bacterial strains analyzed in this study. The isolates used in this study (Data Set S1A) were cultured from surveillance rectal swabs of high-risk patients on hospital admission (August 2015 to October 2015). Their clinical collection and initial characterization were previously reported (16) and are briefly  In our previous study of these isolates, presumptive VREfm isolates were confirmed by ddl typing and for the presence of vanA or vanB genes by PCR (16,36). All isolates used in this study, with the exception of 163-1, initially grew on brain heart infusion (BHI) agar supplemented with 256 mg/mL of vancomycin. The vancomycin MIC of isolate 163-1 was previously determined to be 2 mg/mL by broth microdilution (16). The vancomycin MICs of 10 additional isolates were previously determined to be .256 mg/mL by broth microdilution (16). The vancomycin MICs of all other isolates were determined in this study using the same broth microdilution method as previously reported (16).
Genome sequencing and assembly. Methods for DNA isolation, MinION sequencing, Illumina sequencing, and hybrid genome assembly are described in Text S1.
Phylogenetic analyses and MLST. Genome assemblies were annotated with PROKKA (v 1.12) (37). A sequence alignment was generated for 1,706 core genes by Roary (v 3.12) (22) using PRANK where the paralogs were not split. A phylogenetic tree was built by RAxML (v 8.2.12) (38) with rapid bootstrapping (1,000 inferences) using the GTR1GAMMA model and subsequent ML search on the core gene alignment where the outgroup was specified to be the reference genome E. faecium 1,231,501 (GenBank accession number NZ_ACAY00000000). The phylogenetic tree was visualized with iTOL (v 4) (39). MLST was determined by the E. faecium MLST database (https://pubmlst.org/efaecium/) (7,40). The novel sequence type ST1703 and novel gyd allele number (allele 70) were assigned for the 5 VREfm isolates 111, 121, 137, 154-1, and 158 by request from the database.
Plasmid detection and typing. Each contig of ,2 Mbp in a closed genome assembly was considered a putative plasmid. In silico detection and typing of plasmids were performed with PlasmidFinder (v 2.1) (18). Putative plasmids in closed genome assemblies that were not classifiable by PlasmidFinder were analyzed further (see below).
Definition of pMIX plasmid groups. Putative plasmid sequences that were unclassifiable by PlasmidFinder were analyzed as follows. Presumptive rep genes were identified from plasmid annotations and analyzed by NCBI conserved domain analysis. Confirmed rep genes were those that encoded protein domains that matched known Rep protein families with E values of e 210 to e 278 . The Rep sequences were also scanned by HMMER (v 2.41.1) (42) to analyze Pfam domains. Multiple-sequence alignment of the nucleotide and predicted amino acid sequences of rep genes was performed with MUSCLE (v 3.8.424) where the "group sequences by similarity" option was selected. A neighbor-joining tree was built based on the sequence alignment using the "Tamura Nei" genetic distance model (43). The tree was visually scrutinized in order to sort the plasmids into 9 groups harboring rep genes with at least 95% identity in both nucleotide and predicted amino acid sequence (except for pMI4; see Results). Each unique rep group of plasmids was given a name of format pMIX, where the "X" is a numeric number from 1 to 9. A representative sequence from each pMIX rep group is in Data Set S1B. Representative sequences were compared to PLSDB v 2021_06_23_v2 (19,20) using BLASTn with 90% identity and query coverage thresholds.
Eight closed circular DNA elements of various sizes (3 to 72 kb) could not be classified by PlasmidFinder or by the pMIX rep typing scheme described above. Analysis of these elements is described in Text S1.
ANI analysis. All-versus-all average nucleotide identity (ANI) was calculated using ANIclustermap v 1.1.0 at default parameters with either all E. faecium assemblies or rep17/pRUM plasmids as input.
Transcriptional activity of predicted TA systems. Transcriptional activity of the toxin-antitoxin (TA) systems, TA axe-txe and TA relE , was determined by reverse transcription-quantitative PCR (RT-qPCR) for the VREfm isolates 1 and 5, described in Text S1. The primer sequences used for qPCR are provided in Data Set S1C.
Tn1546 analysis. The Tn1546 nucleotide sequence described for E. faecium BM4147 (GenBank accession number M97297) was used as a reference to identify variations occurring in the Tn1546 elements identified in this study. The Tn1546 elements were classified using a previously described nomenclature (13). The presence of insertion sequence elements (IS elements) was indicated with a one-letter code as follows: IS1216, B; IS1251, C; ISEfa5, D; IS256, J; ISEfa16, K; ISEfa17, L; and ISEfa18, M. The combination of IS elements within the transposon is described by a two-or three-letter code; e.g., group BC possesses one each of IS1216 and IS1251. Arabic numerals, following the alphabet code, indicate differences due to point mutations or different IS element insertion sites. For the group BC, a previous study (13) described BC1 to BC5, each with specific insertion sites of IS elements. Any novel insertion sites identified in the present study were therefore numbered from BC6 forward. The novel ISEfa16, ISEfa17, and ISEfa18 were identified in this study and registered in the ISFinder public database (45).
Accession number(s). DNA sequences from this study have been deposited under BioProject PRJNA682584. Individual accession numbers for all sequence files are provided in Data Set S1A.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. DATA SET S1, XLSX file, 0.2 MB. TEXT S1, PDF file, 0.1 MB.