Genomic Analysis of Vancomycin-Resistant Staphylococcus aureus Isolates from the 3rd Case Identified in the United States Reveals Chromosomal Integration of the vanA Locus

High-level vancomycin-resistant Staphylococcus aureus (VRSA) began to emerge in the United States in 2002 and has since then been reported worldwide. Our study reports the closed genome sequences of multiple VRSA isolates obtained in 2004 from a single patient in New York State. ABSTRACT Vancomycin-resistant Staphylococcus aureus (VRSA) is a human pathogen of significant public health concern. Although the genome sequences of individual VRSA isolates have been published over the years, very little is known about the genetic changes of VRSA within a patient over time. A total of 11 VRSA, 3 vancomycin-resistant enterococci (VRE), and 4 methicillin-resistant S. aureus (MRSA) isolates, collected over a period of 4.5 months in 2004 from a patient in a long-term-care facility in New York State, were sequenced. A combination of long- and short-read sequencing technologies was used to obtain closed assemblies for chromosomes and plasmids. Our results indicate that a VRSA isolate emerged as the result of the transfer of a multidrug resistance plasmid from a coinfecting VRE to an MRSA isolate. The plasmid then integrated into the chromosome via homologous recombination mediated between two regions derived from remnants of transposon Tn5405. Once integrated, the plasmid underwent further reorganization in one isolate, while two others lost the staphylococcal cassette chromosome mec element (SCCmec) determinant that confers methicillin-resistance. The results presented here explain how a few recombination events can lead to multiple pulsed-field gel electrophoresis (PFGE) patterns that could be mistaken for vastly different strains. A vanA gene cluster that is located on a multidrug resistance plasmid that is integrated into the chromosome could result in the continuous propagation of resistance, even in the absence of selective pressure from antibiotics. The genome comparison presented here sheds light on the emergence and evolution of VRSA within a single patient that will enhance our understanding VRSA genetics. IMPORTANCE High-level vancomycin-resistant Staphylococcus aureus (VRSA) began to emerge in the United States in 2002 and has since then been reported worldwide. Our study reports the closed genome sequences of multiple VRSA isolates obtained in 2004 from a single patient in New York State. Our results show that the vanA resistance locus is located on a mosaic plasmid that confers resistance to multiple antibiotics. In some isolates, this plasmid integrated into the chromosome via homologous recombination between two ant(6)-sat4-aph(3′) antibiotic resistance loci. This is, to our knowledge, the first report of a chromosomal vanA locus in VRSA; the effect of this integration event on MIC values and plasmid stability in the absence of antibiotic selection remains poorly understood. These findings highlight the need for a better understanding of the genetics of the vanA locus and plasmid maintenance in S. aureus to address the increase of vancomycin resistance in the health care setting.

explain why the vancomycin resistance phenotype was stable in one isolate but not the other (16,17). Since these studies included only one isolate per patient and did not produce closed genome assemblies, they were unable to show if and how the genomes changed over time. To gain a better understanding of the genetics underlying vancomycin resistance in S. aureus over a period of 4.5 months, we used a combination of long-and short-read sequencing to reconstruct the genomes of 18 clinical isolates obtained from the patient's urine, rectum, and nephrostomy tube and one isolate from the patient's roommate's nares. Although the isolates were obtained from the same patient as those described in the Weigel study, they are not the same isolates (12). In the present study, one vancomycinsusceptible Enterococcus faecalis (VSE), three VRE, four MRSA, two vancomycin-resistant and methicillin-susceptible S. aureus (VRMSSA), and nine vancomycin-and methicillin-resistant S. aureus (VMRSA) isolates underwent a detailed sequence analysis. The latter two groups will be jointly referred to as VRSA in this report.
Our results indicate that a VRSA isolate emerged as the result of the transfer of a multidrug resistance plasmid from a coinfecting VRE to a MRSA isolate. The plasmid then integrated into the chromosome via homologous recombination mediated between two regions derived from remnants of transposon Tn5405. Once integrated, the plasmid underwent further reorganization in one isolate, while two others lost the staphylococcal cassette chromosome mec element (SCCmec) determinant that confers methicillin resistance. Variant calling showed that all 11 VRSA from the patient differed only by a few single-nucleotide polymorphisms (SNPs), indicating that they are all closely related. The results presented here explain how a few recombination events can lead to multiple PFGE patterns that could be mistaken for vastly different strains. A vanA gene cluster that is located on a multidrug resistance plasmid that is integrated into the chromosome could result in the continuous propagation of resistance, even in the absence of selective pressure from antibiotics.

RESULTS
The VRSA isolate was firmly established in the primary patient but did not spread to other patients or staff in the long-term-care facility. A total of 650 specimens were collected in 2004 from the patient, her roommate, and facility staff as part of this investigation. The only vancomycin-resistant S. aureus isolates obtained were from the patient, which include the two methicillin-susceptible and nine methicillin-resistant isolates shown in Table 1. A MRSA isolate collected from the patient's roommate, MRSA-WC000, was included in the present study for comparison.
The patient's last vancomycin exposure had been in October of 2003. Despite this, and repeated antibiotic treatment, the patient remained VRSA-positive until at least 18 August 2004. Characterization of bacterial isolates from the patient included species identification, real-time PCR, and vancomycin Etest (Table 1). PFGE analysis grouped the 11 VRSA strains into 6 different SmaI restriction patterns.
Hybrid genome assemblies demonstrate that the vanA locus is present on the chromosome in some VRSA isolates. WGS was used to provide additional insights into the genetics underlying vancomycin resistance in these VRSA isolates. Short Illumina and long MinION reads were subjected to hybrid assembly with Unicycler ( Table 2). The program produced circular assemblies for all DNA molecules in all isolates, except for two linear contigs of 52 and 216 kb in isolate VRE-WC084. These likely represent one or two plasmids that could not be circularized by the Unicycler program.
The hybrid assembly of long and short reads also revealed what de novo assembly using short reads only had failed to detect: pWC79 was present as a free plasmid in three of the VRSA isolates but integrated into the chromosome in the remaining eight VRSA isolates.
Mobile genetic elements make the patient's VRSA and VRE isolates highly multidrug resistant. A search for antibiotic resistance genes revealed that the E. faecalis isolates VRE-WC031 and VSE-WC032 both carry the erm(B), ant(6)-Ia, and aph(39)-IIIa genes, which confer resistance to macrolides, streptomycin, and amikacin/kanamycin, respectively (Table 3). While these three genes are located on the chromosome in VSE-WC032, they are present on plasmid pWC85 in VRE-WC031. VRE-WC031 also harbors a 90.9 kb plasmid, pWC91, that carries vanA as part of the vanRSHAXYZ locus, which confers resistance to vancomycin. The same plasmid also contains a copy of Tn4001, which is a composite transposon consisting of two genes, a N-acetyltransferase and a bifunctional aac(69)-aph (20) gene, that are flanked by IS256 elements on either side (20). While the aac(69)-aph (20) gene is generally involved in aminoglycoside resistance, it is an inactive pseudogene in this and all other isolates, except for VMRSA-WC062, VMRSA-WC082, VMRSA-WC102, VMRSA-WC123, and VRSA-WC113.
The chromosomes of all S. aureus isolates carry the mph(C) and msr(A) genes, conferring resistance to macrolides, and two copies of the ant(6)-sat4-aph(39) locus, referred to as L1 and L2 from here on. Except for VRMSSA-WC111 and VRMSSA-WC113, which are methicillin susceptible, all S. aureus isolates possess the mecA gene, which confers methicillin resistance. The blaZ beta-lactam resistance gene, which is located on plasmid pN315, is present in all S. aureus isolates in this study.
All VRSA isolates possess a copy of plasmid pWC79 (Fig. 1, Table 3), which carries the vanA locus in addition to erm(B), tet(S), and aac(69)-aph (20), the latter being a pseudogene in most isolates as described above. Plasmid pWC79 also carries a copy of the ant(6)-sat4-aph(39) locus, designated L3. Unlike the two chromosomal loci L1 and L2, locus L3 includes an intact copy of the ant(6)-Ia gene.
In three of the isolates, VMRSA-WC052, VMRSA-WC083, and VMRSA-WC102, pWC79 is circular, which is consistent with an independently replicating plasmid. In the remaining VRSA isolates, pWC79 is integrated into the chromosome ( Table 3).
The Tn1546 element that carries the vanA locus in the VRSA and VRE lacks the genes for transposition. A comparison of the 9.8 kb region that includes the vanA operon among the different VRE and VRSA isolates shows that all share the same locus, regardless of species. Compared to Tn1546 (6), the left inverted repeat, the transposase, and part of the resolvase has been replaced in all isolates by an IS1216 family transposase (Fig. 2). The missing features were found nowhere else in the VMRSA-WC052 genome, suggesting that they are permanently lost. Another difference between the vanA locus in Tn1546 and the vanA loci described in this study is that in all 14 VRE and VRSA isolates, the vanRSHAXYZ locus is interrupted by an ISL3 family transposase that inserted between the vanS and vanH genes. This transposase is similar to the IS1251 transposase located in the intergenic region between vanS and vanH in some E. faecium isolates (21) but differs from it by seven amino acids.
Aligning the vanA loci with progressiveMauve (22) showed 100% nucleotide identity a Hybrid assembly with Unicycler was done using the settings "conservative," "normal," and "bold." The mode that produced the best results for each isolate is shown. b (mecA) and (vanA) indicate the location of these loci, where applicable. c Patient's roommate. d Denotes a pseudogene.
for the 9.8 kb sequence (Fig. 2) in all VRSA and VRE isolates. However, two isolates stood out from the rest when five additional genes located 39 to vanZ were included in the alignment. The region in VRE-WC031 differed by 5 SNPs and 10 indels from all other isolates, while 5 SNPs and 7 indels made the region in isolate VRE-WC084 unique (Fig. 2). Since the 13.5 kb region is identical in VRE-WC072 and all VRSA isolates, it makes this E. faecium isolate the most plausible source for transmitting vanA to S. aureus. The mosaic, multidrug resistance plasmid pWC79 from VRSA shares sequence elements with plasmids from Enterococcus faecium and Enterococcus faecalis. Previous reports found that the vanA resistance locus was transferred via the Tn1546 transposon from an enterococcal plasmid to a staphylococcal plasmid (2). In the present study, Tn1546 lacks the genes for transposition, and we did not detect any staphylococcal plasmids that  Fig. 2. Outer ring, genes in the sense orientation; inner ring, genes in antisense orientation; red, antibiotic resistance genes; blue, transposase genes; green, replicons and the axe/txe toxin antitoxin system; gray, other genes.
a Plasmid pWC79 can occur as a free plasmid or integrated into the chromosome. Two ant(6)-sat4-aph(39)-III loci, L1 and L2, are located on the chromosome of all S. aureus isolates and a third, L3, on pWC79. L29 and L39 denote the corresponding loci after pWC79 integration into the chromosome. b Some genes, such as ant(6)-Ia, aph(39)-IIIa, or aac(69)-aph(20) might be only pseudogenes in some isolates or loci. c The gene is located on the chromosome due to the chromosomal integration of plasmid pWC79.
could have served as a recipient for Tn1546 or the vanA locus. Strains VRE-WC031, VRE-WC072, and VRE-WC084 carry various plasmids or contigs that include replicons and genes that are also found in pWC79; however, none of the enterococcal isolates available for analysis harbor a copy of pWC79. Plasmid pWC84 from E. faecium VRE-WC072 has a vanA locus identical to that found in pWC79 and has an almost identical complement of antibiotic resistance genes, making it the most likely predecessor of pWC79 (Fig. 3). However, pWC84 lacks the Tn4001 transposon with the aac(69)-aph (20) gene. E. faecalis VRE-WC031 contains two plasmids of interest: pWC85 carries the ant(6)-Ia, sat4, aph(39)-IIIa, and erm(B) genes (Table 3), while pWC91 carries a unique vanA locus and a copy of the aac(69)-aph (20) gene that is  [6]) of Tn1546 has been replaced by an IS6-like element IS1216 family transposase. As in Tn1546, the locus includes the vanRSHAXYZ genes and the right inverted repeat, but unlike Tn1546, the locus also includes an ISL3 family transposase. The locus is identical in all VRSA isolates and in VRE-WC072, while VRE-WC031 and VRE-084 differ by 5 SNPs each (indicated by the isolate name and the number of SNPs in parentheses). HTH, helix-turn-helix transcriptional regulator; AbrB, AbrB family transcriptional regulator; RelE, type II toxin-antitoxin system RelE/ParE family toxin; rec, recombinase family protein.

FIG 3
Comparison between pWC79 (78.6 kb) from VMRSA-WC052, pWC91 (90.9 kb) from VRE-WC031, and pWC84 (83.6 kb) from VRE-WC072. Regions of the same color are shared among the plasmids; regions that are depicted inside the plasmid are in the antisense orientation relative to pWC79. Genes of interest are indicated for reference; an asterisk (*) denotes that the gene is a pseudogene.
Longitudinal Isolation of VRSA with Chromosomal vanA Microbiology Spectrum also found in pWC79. All three plasmids, pWC79, pWC84, and pWC91, have regions of DNA in common, such as rep2, rep17 (which is a pseudogene in pWC91), and the axe/ txe locus. In addition, pWC84 (E. faecium) and pWC91 (E. faecalis) share two regions of DNA that are absent from pWC79. These findings support the notion that pWC79 is a mosaic plasmid with origins in at least two of the VRE isolates, VRE-WC031 and VRE-WC072. This hypothesis is further supported by the high prevalence of insertion sequences and transposons in pWC79: 23 (24%) of the 96 genes in pWC79 encode transposases (Fig. 1), including ten IS1216 family transposases, two IS256 family transposases, one IS1182 family transposase, one ISL3 family transposase, and two IS30 family transposases. The ability of these mobile genetic elements (MGEs) to mediate recombination events has been well documented (23) and will be described in greater detail below.
VRSA strains lack a functional SauI restriction-modification system. The SauI type I restriction-modification system has been shown to limit the uptake of foreign DNA through conjugation or transduction in S. aureus (24). In the present study, the R subunit that is responsible for restriction of foreign DNA is a pseudogene due to a C nucleotide deletion at position 1381 in all S. aureus isolates from the patient. Only an isolate from the patient's roommate, MRSA-WC000, contains a functional R subunit, which would make it capable of restricting incoming foreign DNA.
Understanding how VRSA isolates with very different PFGE profiles arose within a few months from a single ancestor. PFGE analysis of VRSA isolates from the patient in the current (Table 1) and a previous study (12) showed a number of different PFGE patterns, suggesting that multiple strains of S. aureus had become independently resistant to vancomycin. However, an alternative hypothesis is that a single MRSA isolate acquired the vanA locus and then diverged through genome rearrangements, rather than point mutations, giving rise to the different PFGE patterns. This hypothesis is supported by the following findings.
Second, genome comparisons identified the inversion of entire regions that are flanked by inverted repeats. One involves Tn4001, a composite transposon with two IS256 elements that form indirect repeats (20). The orientation of Tn4001 in pWC79 is reversed in VMRSA-WC083 and VMRSA-WC102 compared to all other VRSA isolates, suggesting that either the transposon inserted into the same position in two different orientations or that it was inverted through homologous recombination between the two IS256 indirect repeats after the insertion.
Third, another inversion event involves a chromosomal region of 24 kb that includes the msr(A) and mph(C) resistance genes. The region is in the msr(A)-mph(C) orientation in all S. aureus isolates except for VMRSA-WC102, MRSA-WC000, and MRSA-WC090, where it is in the opposite orientation. The entire region is flanked by the two ant(6)-sat4-aph(39) loci L1 and L2, which are oriented as indirect repeats (Fig. 4 and 5).
The fourth finding is that two of the eight isolates with a chromosomal vanA locus, VRMSSA-WC111 and VRMSSA-WC113, have a methicillin-susceptible phenotype and lack the SCCmec locus that contains the mecA gene. This suggests that either MRSA and MSSA isolates independently acquired the vanA locus or one or more VMRSA isolates lost the SCCmec locus after obtaining vanA, e.g., to compensate for the added fitness cost of becoming vancomycin resistant (25).
Finally, mapping each isolate's Illumina reads to the genome sequence of VMRSA-WC052 as a common reference shows that none of the isolates differ by more than nine single nucleotide polymorphisms (SNP) from the reference. The only exception is MRSA-WC000, which was isolated from the patient's roommate and which differs by 33 SNPs from the closest isolate, MRSA-WC090 (Fig. 6). Most isolate pairs from the patient differ by 2 or fewer SNPs, and no pair differs by more than 11 SNPs. Even the VRMSSA isolates differ by only 5 to 7 SNPs from the nearest VMRSA isolates, which is consistent with the hypothesis that the SCCmec locus was lost in these isolates. The integration of pWC79, excision of SCCmec, and inversion of two genomic regions show how MGEs increase genome plasticity, either through transposition or by providing repeat regions for homologous recombination. This genetic mobility can explain the different PFGE patterns described earlier (Table 1) (12). The SNP analysis supports the hypothesis that a single isolate diverged after acquiring the vanA gene.
Integration of pWC79 into the chromosome of S. aureus was mediated through homologous recombination between two ant(6)-sat4-aph(39) loci. The VRSA isolates described here are noteworthy because the majority contain an integrated vanA locus. This was of interest because all other VRSA previously documented in the United States reportedly contained a vanA locus that was located on a plasmid.
All VRSA isolates in this study possess three ant(6)-sat4-aph(39) loci, two of which, L1 and L2, are located on the chromosome, while the third, L3, is located on plasmid pWC79. In VRSA isolates with a chromosomal vanA locus, the entire plasmid is integrated through homologous recombination between L2 and L3 (Fig. 4). The presence of an additional gene, orfB, in both L2 and L3 increases the size of homologous DNA relative to L1, which lacks orfB (Fig. 5).
The ant(6)-sat4-aph(39) locus has been described as part of the composite transposon Tn5405 (Fig. 5). While the chromosomal L2 locus includes genes found near the 39 end of the transposon, locus L3 on the plasmid includes the 59 half of Tn5405 (Fig. 5). After integration of pWC79 into the chromosome, the combination of L2 and L3 restores the complete Tn5405 sequence, except for a second IS1182 that is missing at the 39 end. Fig. 5 shows how the three ant(6)-sat4-aph(39) loci are oriented relative to each other after the integration of pWC79, resulting in an inverted and a direct repeat. The inverted repeat formed by L1 and L2 is thought to have mediated the inversion of the 24 kb region containing the msr(A) and mph(C) genes in some of the isolates, while the direct repeat formed between L2 and L3 is the result of the integration of pWC79. Isolate VMRSA-WC082 was very distinct from the other isolates because it appears that the pWC79 plasmid underwent several recombination events after integrating into the chromosome. One feature that sets VMRSA-WC082 apart from the other isolates is an additional copy of an IS257 transposase. Based on the model depicted in Fig. 7, an  (39) loci that result in an indirect repeat (IR) and a direct repeat (DR) in S. aureus isolates with chromosomal integration of the pWC79 plasmid (with the exception of VMRSA-WC082). [1], the mph(C)containing region that is inverted in some of the S. aureus isolates; [2], the remaining sequence of pWC79.

FIG 6
Minimum spanning tree. Illumina reads were mapped to VMRSA-WC052 as a common reference. Numbers indicate the number of single nucleotide polymorphisms (SNP) between a pair of isolates. Lines connect isolate pairs that are most similar to each other. Isolates in the same bubble have no differentiating SNPs. Green, MRSA isolates; blue, VMRSA isolates with pWC79 as a free plasmid; orange and magenta, VMRSA and VRMSSA isolates, respectively, with chromosomally integrated pWC79. An asterisk (*) denotes the isolate from the patient's roommate. inverted repeat formed by two IS257 elements led to the first inversion event, followed by a second inversion mediated by the two ant(6)-sat4-aph(39) loci, L29 and L39, and a third inversion between two IS1216 elements. The genetic evidence provided here implicates the two ant(6)-sat4-aph (39) loci, as well as other repeat regions from MGEs, in large-scale DNA rearrangements.

DISCUSSION
Antibiotic resistance, especially in S. aureus, is a significant cause of prolonged morbidity and increased mortality (26). Here, we present a detailed bacteriological and genetic analysis of 11 VRSA isolates collected from a single patient over the course of 4.5 months while residing at a long-term-care facility in New York State. At the time of collection in 2004, this was only the third time that high-level VRSA isolates had been reported in the United States, and it is, to our knowledge, the only longitudinal study (11). Two of the initial VRSA isolates from the patient, termed VRSA-3a and VRSA-3b, had been sequenced previously using either Illumina, 454 FLX, or Ion PGM system sequencing technologies (16,17). While these short-read sequencing technologies yield valuable data, they are only able to produce fragmented genome assemblies that do not provide information on how the contigs are ordered and oriented relative to each other. Using a combination of short Illumina reads and long Oxford Nanopore Technologies (ONT) reads, we reconstructed closed, circular chromosomes and plasmids for almost all isolates. Only the assembly for isolate VRE-WC084 resulted in a circular chromosome, a circular plasmid, and two contigs of 52 and 216 kb, respectively, which might be derived from one or two plasmids.
Comparing the genomes of the various VRSA isolates showed evidence for largescale recombination events in some isolates, including the inversion of a 24 kb region and the integration of a 79 kb plasmid, pWC79, into the chromosome. The latter is remarkable because pWC79 harbors the vanA locus that is responsible for high-level vancomycin resistance. This represents, to our knowledge, the first description of the vanA locus residing on the chromosome of S. aureus. Previous studies had noted that the Tn1546 transposon that harbors vanA always resides on a plasmid in S. aureus, even though the chromosome would make a bigger target for transposition (16,17). However, a direct transposition of Tn1546 seems to be highly unlikely for the isolates described in the present study since the genes and sequences required for transposition, such as the left inverted repeat, the transposase, and the resolvase, are either missing from the genome or truncated. In this case, instead of transposing Tn1546, we provide evidence that the entire plasmid integrated into the chromosome through homologous recombination between two repeat regions. This could explain why some bacteria maintained a stable vancomycin-resistant phenotype even in the absence of antibiotic selection (12). While the chromosomal location of vanA in some of the S. aureus isolates described here appears to be a first, there have been reports of plasmids similar to pWC79 integrating into the chromosome. Schwarz et al. (10) showed that plasmid pRE25, which carries the rep2 (Inc18) replicon, erm(B), ant(6)-Ia, sat4, and aph(39)-IIIa, among other genes, transferred from enterococci to other species via conjugation and then integrated into the chromosome. From there, the plasmid excised again and transferred into a new host (10).
In the present study, pWC79 was present in all VRSA isolates, either as a free plasmid or integrated into the chromosome. However, while regions similar to those in pWC79 were also found in plasmids and contigs from VRE isolated from the same patient, none of the VRE isolates available for this study contained an exact copy of pWC79. In an earlier study that used a different set of isolates collected from the same patient, Weigel et al. reported isolating plasmid pLW2547 from E. faecium VRE2547 and plasmid pLW595 from S. aureus VRSA595 (12). Both plasmids had virtually identical BglII restriction patterns and both carried the tet(S) and vanA genes, suggesting that VRE2547 was the source of the vanA gene in VRSA595. Plasmid pWC79 from the current study also harbors the tet(S) and vanA genes, and the predicted BglII restriction pattern matches that of pLW2547 and pLW595 (data not shown), suggesting that pWC79 might be similar or identical to pLW2547 and pLW595. Unfortunately, neither strain VRE2547 nor VRSA595 was available for sequencing in the present study.
Comparing pWC79's sequence to that of the plasmids present in the available VRE isolates indicates that pWC79's origins can be traced back to mosaic plasmids that can be found in E. faecalis and E. faecium. The genes for vanA, erm(B), ant(6)-Ia, sat4, aph(39)-IIIa, and tet(S) are present in pWC84 from E. faecium, while the vanA and aph(20)-aac(69) genes were found in pWC91 from E. faecalis. This finding highlights that a multispecies biofilm, such as the patient's nephrostomy tube, can serve as a site for interspecies gene exchange.
The data presented here indicate a central role for the ant(6)-sat4-aph(39) locus in the evolution of the NY VRSA isolates. Previous studies had noted the presence of these genes in the context of antibiotic resistance (12,17). However, comparing the genomes of multiple MRSA, VMRSA, and VRMSSA isolates revealed the presence of multiple ant(6)-sat4-aph (39) loci. The locus is present in two copies on the chromosome, and a third copy is located on pWC79; in combination, they serve as repeat regions for homologous recombination. One resulted in the inversion of a 24 kb region in VMRSA-WC102, MRSA-WC000, and MRSA-WC090, another facilitated the integration of pWC79 into the chromosome in some VRSA isolates, and a third resulted in an inversion that affected a part of the chromosomally integrated pWC79 in VMRSA-WC082. The ant(6)-sat4-aph(39) locus has been associated with the composite transposon Tn5405 and numerous plasmids (10,(27)(28)(29)(30)(31), which could explain the finding of multiple copies in one cell. Interestingly, various parts of Tn5405 were present on the chromosome and on the plasmid, and integration of the plasmid into the chromosome nearly restored Tn5405, except for a missing second IS1182 element. While the first chromosomal locus only consisted of ant(6)-Ia, sat4, and aph(39)-IIIa, the other two loci had a fourth gene in common, orfB. This larger region of homology would better facilitate homologous recombination and might explain pWC79's preference for the second ant(6)-sat4-aph(39) locus as an insertion site.
A question of interest was to determine if multiple VRSA strains emerged independently at various time points in the patient. The presence of VRSA isolates with multiple PFGE patterns, strains with different antibiotic resistance gene content (12,17), and the existence of methicillin-resistant and methicillin-susceptible VRSA isolates in the present study suggested that more than one S. aureus isolate could have acquired pWC79 from a VRE isolate. However, the paucity of SNPs and evidence of various recombination events suggests an alternative hypothesis. The inversion of entire genomic regions and the integration and excision of pWC79 could explain variations in PFGE patterns. The loss of the SCCmec locus that confers methicillin resistance would explain the simultaneous finding of VMRSA and VRMSSA isolates (32,33). Périchon and Courvalin (34) demonstrated the incompatibility of high-level methicillin and vancomycin resistance in S. aureus. This could be the driving force for losing either the SCCmec or the vanA locus and might explain, at least in part, why some isolates have a lower than expected vancomycin MIC (Table 1). Another reason for the unexpectedly low MIC values could be that glycopeptide induction of the vanA locus exerts a high fitness cost on VRSA isolates. (35) This was shown to result in an extended lag phase and slower rate of exponential growth, which could give the appearance that the bacteria are not highly resistant when measured by the standard Etest.
Cells that contain plasmids with antibiotic resistance genes often pay a fitness cost in the absence of antibiotic selection (36)(37)(38), which allows faster-growing, plasmidfree cells to expand within the population. During periods of antibiotic selection, resistant cells can then rapidly replace the susceptible cells that were killed by the antibiotic. This repeated selection for different subpopulations would explain why VMRSA, VRMSSA, and MRSA were isolated from the same patient over a period of time and why these isolates differ by only a few SNPs as described below.
The results of the SNP analysis indicate that the two most divergent S. aureus isolates from the patient, a VMRSA and a VRMSSA strain, differ by only 11 SNPs. By comparison, MRSA isolates from the patient and a MRSA isolate from the roommate differ by 33 or more SNPs. These results indicate that only one isolate obtained the vanA gene from a coinfecting VRE isolate and then diverged through various recombination events that include inversions of the genome, loss of SCCmec, and the integration of the vanA plasmid pWC79 into the chromosome. Interestingly, while the roommate's MRSA isolate contained a functional SauI type I restriction-modification system, the VRSA isolates from the patient had apparently lost that ability due to a mutation in the R subunit. If and how this could have facilitated the horizontal gene transfer of pWC79 into these isolates remains to be determined.
While colonization with MRSA and VRE strains is not uncommon among long-term care patients (39) and those with indwelling medical devices, it is surprising that more VRSA strains have not been detected in the United States. Having multiple VRSA isolates from the same patient that are spread out over several months provided a unique opportunity for comparisons and an improved understanding of the microbiology and the potential for vanA plasmids to incorporate into the genomes of S. aureus strains. WGS analysis combining long and short sequencing reads provided a better understanding of the development and maintenance of VRSA from this NY patient and elucidated the genetics of these antibiotic-resistant organisms. Continued vigilance detecting antibiotic resistance and assessing transmission through infection control efforts as described in this investigation will positively impact public health in the future.

MATERIALS AND METHODS
Bacterial isolates. Following the initial isolation of a VRSA isolate from a urine culture in March of 2004, numerous specimens were collected during the subsequent months from the patient, her roommate of many years, and nursing home staff members. Clinical isolates were cultured on Trypticase soy agar plates with 5% sheep's blood (Wadsworth Center Tissue and Media Core). Plates were examined for colonies resembling S. aureus; identification of S. aureus was confirmed by Gram stain, catalase, coagulase, and nuc gene detection by real-time PCR. Methicillin and vancomycin resistance was confirmed by MIC testing with oxacillin and vancomycin Etest (AB Biodisk, Solna, Sweden), according to interpretive criteria of the Clinical and Laboratory Standards Institute (CLSI). Vancomycin resistance was confirmed by Etest after revival of isolates from freezer stocks. Real-time PCR was used to screen for the vanA, vanB, vanD, and mecA genes. For pulsed-field gel electrophoresis (PFGE) analysis, genomic DNA was digested with SmaI restriction endonuclease, DNA fragments were separated, and PFGE patterns were analyzed and compared using the BioNumerics software package (Applied Math, Kortrijk, Belgium).
Illumina short-read sequencing. Single colonies were selected for sequencing based on the date of isolation, antibiotic resistance profile, PFGE pattern, and colony morphology to get a broad spectrum of isolates. Isolates were subcultured from frozen stocks twice on Trypticase soy agar with 5% sheep's blood and incubated overnight at 37°C. For genomic DNA isolation, samples were preprocessed by inoculating a 1-mL loopful of growth from an isolated colony into 180 mL enzymatic lysis buffer (1 M Tris-HCl, pH 8.0, 0.5 M EDTA, and Triton X-100) and 1 mL of lysostaphin (1 mg/mL). Samples were incubated at 56°C for 30 min at 350 rpm in a MultiTherm shaker (Benchmark Scientific), followed by the addition of 25 mL of proteinase K (Qiagen) and a second incubation for 30 min at 56°C. Finally, 4 mL of RNase A (100 mg/mL) was added to each sample and incubated at room temperature for 3 min. All samples were then processed using the DNeasy blood and tissue kit (Qiagen) according to the manufacturer's instructions. Nucleic acid was eluted into 10 mM Tris, quantitated using the Qubit double-stranded DNA (dsDNA) high-sensitivity (HS) assay kit (Life Technologies) on a Qubit 4.0 fluorometer (Life Technologies), and submitted to the WC Applied Genomics Technologies Core for WGS using the Illumina MiSeq platform. The Nextera XT protocol (Illumina) was used to generate sample libraries that produced pairedend reads of 250 bp in size as described previously (40).
Oxford Nanopore long-read sequencing. Isolates were subcultured from frozen stocks twice on Trypticase soy agar with 5% sheep's blood. For genomic DNA extraction, colonies were resuspended in 1.5 mL sterile water to a McFarland concentration of 3.0 and harvested as pellets. The pellets were resuspended in 100 mL STET buffer (8% sucrose, 50 mM Tris-HCl, pH 8.0, 50 mM EDTA, pH 8.0, 5% Triton X-100), 5 mL lysozyme (100 mg/mL, Sigma-Aldrich), and 10 mL lysostaphin (5 mg/mL, Sigma-Aldrich). High-molecularweight genomic DNA was extracted using the Nanobind CBB Big DNA kit (Circulomics) following the Grampositive extraction protocol. Genomic DNA was quantified using a Qubit 4.0 fluorometer (Thermo Fisher Scientific). The quality of the genomic DNA was assessed using the TapeStation system (Agilent).
MinION sequencing libraries were prepared from 1.5 mg of input DNA. Genomic DNA was sheared in 50 mL total volume in Covaris G tubes using an Eppendorf 5425 centrifuge at 6,000 rpm. Sequencing libraries were prepared according to the manufacturer's instructions (Oxford Nanopore) and multiplexed using 1D Native barcoding kits (EXP-NBD104, EXP-NBD114) followed by a ligation and sequencing kit (SQK-LSK109). to repair any small-scale errors. Unicycler was run in all three available assembly modes (conservative, normal, bold); the mode providing the best result for each isolate is shown in Table 2.
As quality control, reads were mapped to the assembly created by Unicycler using BWA MEM (44,45). Default options were used for the Illumina reads, while option "-x ont2d" was used for the ONT reads. The resulting BAM files were analyzed with Pilon (46), whose output was visualized with the Integrative Genome Browser (version 9.1.10) (47) to identify regions of concern. DNA fragments were identified as assembly artifacts and removed from the genome if they contained any regions of zero coverage according to the corresponding BAM file. These fragments were always of very low coverage, less than 5 kb in size, and duplications of chromosomal regions, and all but one were linear molecules. At most, six artifacts, totaling 8.7 kb, were removed from an assembly, although most were free of artifacts.
PlasmidFinder (version 2.1) was used to detect plasmid replicon sequences for Gram-positive bacteria in all genomes (48). Genomes were annotated with the stand-alone version of NCBI's Prokaryotic Genome Annotation Pipeline (PGAP) (49). Where necessary, circular DNA molecules were rearranged to begin either with the dnaA gene or a plasmid replication initiation protein. ABRicate, using the NCBI database, was used to find antibiotic resistance genes (50,51). In cases of discrepancies between the PGAP and ABRicate annotations, the PGAP annotations were used.
The program progressiveMauve was used for genome sequence comparisons (22). For single nucleotide polymorphism (SNP) analysis, Illumina sequence reads from MRSA and VRSA isolates were mapped with BWA to reference VMRSA-WC052 to perform a SNP analysis with FreeBayes (27), which is part of the LegioCluster pipeline (52). The analysis included plasmids and other mobile genetic elements; however, questionable or low-quality SNP calls were removed with vcffilter (53) if the SNP quality assigned by FreeBayes was less than 20, the read depth was less than 10, the alternate allele quality sum was less than 20, the count of full observations of an alternate haplotype exceeded 10% of all counts, or the reads supporting a SNP call were limited to only one of two strands.