Genetic Analysis of a High-Level Vancomycin-Resistant Isolate of Staphylococcus aureus

See allHide authors and affiliations

Science  28 Nov 2003:
Vol. 302, Issue 5650, pp. 1569-1571
DOI: 10.1126/science.1090956


Vancomycin is usually reserved for treatment of serious infections, including those caused by multidrug-resistant Staphylococcus aureus. A clinical isolate of S. aureus with high-level resistance to vancomycin (minimal inhibitory concentration = 1024 μg/ml) was isolated in June 2002. This isolate harbored a 57.9-kilobase multiresistance conjugative plasmid within which Tn1546 (vanA) was integrated. Additional elements on the plasmid encoded resistance to trimethoprim (dfrA), β-lactams (blaZ), aminoglycosides (aacA-aphD), and disinfectants (qacC). Genetic analyses suggest that the long-anticipated transfer of vancomycin resistance to a methicillin-resistant S. aureus occurred in vivo by interspecies transfer of Tn1546 from a co-isolate of Enterococcus faecalis.

Staphylococcus aureus, a major cause of potentially life-threatening infections acquired in health care settings and in the community, has developed resistance to most classes of antimicrobial agents soon after their introduction into clinical use. In the late 1960s, few options were available for treatment of infections caused by strains resistant to penicillins, macrolides, aminoglycosides, and tetracyclines, a situation that led to widespread use of methicillin and other semisynthetic penicillins. However, the acquisition of genes encoding an additional penicillin-binding protein, PBP-2a, resulted in the emergence of methicillin-resistant S. aureus (MRSA) in many health care institutions around the world (1). Currently, glycopeptides such as vancomycin provide effective therapy against most strains of multidrug-resistant S. aureus, including MRSA (2).

Resistance to vancomycin, first reported in 1988 for an isolate of Enterococcus faecium (3), is associated with one of several gene clusters, which are classified as vanA to vanG on the basis of the resulting phenotypic characteristics of the strain. The VanA phenotype, characterized by inducible, high-level resistance to vancomycin and teicoplanin, is mediated by Tn1546 or closely related mobile genetic elements. The target of vancomycin is the carboxy terminal D-alanyl-D-alanine (D-ala-D-ala) of the disaccharide pentapeptide cell wall precursor, which is translocated by a lipid carrier to the outer surface of the cytoplasmic membrane. Enzymes encoded on the transposon replace D-ala-D-ala with a depsipeptide, D-alanyl-D-lactate (D-ala-D-lac). The affinity of vancomycin for the depsipeptide is decreased by a factor of 1000 when compared with the normal cell wall precursor terminating in D-ala-D-ala (4). The potential for transfer of vancomycin resistance to S. aureus was immediately recognized and became a major global health concern.

The first clinical isolate of S. aureus with decreased susceptibility to vancomycin [minimal inhibitory concentration (MIC) = 8 μg/ml] was isolated in Japan in 1997 (5). Additional strains have been isolated in several countries, including the United States, and are designated as vancomycin-intermediate S. aureus (VISA). Analysis of these isolates revealed a common structural change, a significant thickening of the cell wall (6). This alteration increases the number of D-ala-D-ala targets in the outer layers of the cell wall that may then “trap” the large vancomycin molecules, preventing inhibitory concentrations of the agent from reaching precursor targets before they are incorporated into the cell wall structure (7, 8). Although concern for the emergence of high-level resistance was renewed, vancomycin MICs among the VISA strains remained relatively low (≤16 μg/ml). However, in June 2002 a high-level vancomycin-resistant S. aureus (VRSA, MIC = 1024 μg/ml vancomycin) was isolated from a dialysis patient in Michigan (9). This report provides a genetic analysis of the VRSA clinical isolate.

Identification of the VRSA isolate as S. aureus was confirmed by DNA sequence analysis of rDNA, gyrA, and gyrB (10). Contamination with enterococci was ruled out, because we were unable to amplify genes encoding enterococcal ligases (11). The SmaI pulsed-field gel type of the VRSA was identified as USA100 (the New York/Japan lineage) (1, 12, 13), the most common staphylococcal pulsed-field type found in U.S. hospitals.

An MIC of 1024 μg/ml of vancomycin was determined by broth microdilution (10). The VRSA isolate was also resistant to aminoglycosides, β-lactams, fluoroquinolones, macrolides, rifampin, and tetracycline, but was susceptible to linezolid, quinupristin/dalfopristin, and trimethoprim-sulfamethoxazole. Unlike laboratory mutants selected for vancomycin resistance, such as VM50 (14), the VRSA clinical isolate retained its MRSA phenotype (oxacillin MIC >128 μg/ml).

Vancomycin resistance in the VRSA isolate was identified as vanA-mediated by polymerase chain reaction amplification (15). The prototype VanA-encoding element is a 10.8-kb transposon, Tn1546, frequently found on plasmids in vancomycin-resistant enterococci (VRE). Therefore, we isolated plasmids from the VRSA, a co-isolate of VRE, and a vancomycin-susceptible MRSA isolated from the same patient (10). The VRSA harbored a single 57.9-kb plasmid, pLW1043. The MRSA contained a single plasmid, pAM829, of ∼47 kb. Two plasmids (45 kb and 95 kb) were found in the VRE (16). Comparisons of HindIII restriction fragments from these plasmids (Fig. 1) indicated that the VRSA plasmid (lane 4) and the MRSA plasmid (lane 3) were very similar, although the VRSA plasmid was 11 kb larger. (Two of the bands in the VRSA plasmid are doublets.) Restriction fragments from the two VRE plasmids did not suggest any similarity to the staphylococcal plasmids. Southern hybridization localized vanA to a 7.1-kb fragment of the VRSA and VRE plasmids (Fig. 1B) not present in the MRSA plasmid. These results are consistent with the reported sequence of Tn1546 (17), in which an internal 7.1-kb HindIII fragment contains the complete coding region of vanA. The antibiogram of the MRSA and VRSA differed only in resistance to the glycopeptides, vancomycin and teicoplanin, for which the MICs increased by a factor of 1000 and 64, respectively, in the VRSA. In addition, the VRSA and MRSA were the same SmaI pulsed-field gel types, USA 100. When genomic DNA was digested with EagI, an additional pulsed-field gel fragment was detected in the VRSA that corresponded in size with the plasmid and hybridized with the vanA probe (fig. S1). These data suggest that the MRSA plasmid acquired Tn1546 to produce a vancomycin-resistant staphylococcal isolate.

Fig. 1.

Restriction analysis of pLW1043 and localization of vanA by Southern hybridization. (A) Isolated plasmids were digested with HindIII, and the resulting fragments were separated on an agarose gel and stained with ethidium bromide. Lane 1, molecular size marker; lane 2, VRE plasmids; lane 3, pAM829 (MRSA); lane 4, pLW1043 (VRSA). (B) DNA fragments from the gel were transferred to nitrocellulose and hybridized with a probe for vanA. Lane numbers in (B) correspond to those in (A).

Filter mating studies identified both the VRE plasmid containing Tn1546 and the VRSA plasmid as conjugative plasmids. Vancomycin resistance was transferred in vitro from the VRE isolate to E. faecalis JH2-2 and from the VRSA isolate to the S. aureus COL strain (10). Filter matings between the patient's VRE and MRSA isolates were not successful, possibly because the only selection available for an MRSA-to-VRSA transconjugant was vancomycin. The lack of transconjugant selection on vancomycin has been previously described (18). Noble et al. attribute this lack to the induction time required for vancomycin resistance to be expressed. Also, one would expect the frequency of transfer to be relatively low because the enterococcal plasmid was not maintained in the MRSA. Therefore, two sequential genetic events, conjugative transfer of the plasmid followed by excision and integration of the transposon into the resident MRSA plasmid, must occur. Alternatively, transfer of the transposon could have occurred by transduction.

Vancomycin resistance has been transferred from E. faecalis to S. aureus when selection for transconjugants was possible with a different antibiotic. These transfers rarely occurred in vitro, but were documented at a relatively high frequency in vivo (2 × 101 to 6 × 103 per 106 recipients on mouse skin) when transconjugants were selected on agar containing rifampin plus erythromycin (18). However, plasmids did not appear to play a role in the transfer, and the mechanism was not determined.

The complete sequence of the VRSA plasmid, pLW1043, was determined and submitted to the GenBank database (accession number AE017171). The 57,889–base pair (bp) plasmid (Fig. 2A) encoded 57 proteins, including six copies of the IS257 transposase. There were also two truncated remnants of the IS256 transposase. The overall G+C content was 30.8%, which is consistent with the relatively low (33%) G+C content of S. aureus. pLW1043 appears to be a composite plasmid, sharing homologous regions with the pSK1 family of multidrug resistance plasmids (particularly pSK4) and the pSK41/pGO1 family of multiresistance conjugative plasmids of staphylococci (19). Other than the incorporation of Tn1546, the resistance elements on pLW1043 are most similar to those reported for pSK4, including Tn4003 (dfrA, trimethoprim resistance), Tn552 (blaZ β-lactamase, resistance to penicillins), and Tn4001 (aacA-aphD, gentamicin/kanamycin/tobramycin resistance). Also present is qacC, encoding an energy-dependent mechanism of active efflux that mediates resistance to quaternary ammonium compounds (disinfectants). This differs from pSK4, which carries qacA, a larger version of the Qac multidrug efflux system (19).

Fig. 2.

(A) Circular representation of pLW1043. The outer circle shows predicted coding regions, color coded by role categories: gold, DNA metabolism; olive, regulatory functions and signal transduction; light green, cell envelope; red, cellular processes; light blue, biosynthesis of cofactors, prosthetic groups, and carriers; blue-green, mobile and extrachromosomal element functions; gray, unknown function; dark blue, conserved hypothetical proteins; black, hypothetical proteins. The oriT and EcoRI restriction sites are shown on the rim of the outer circle. The inner circle shows predicted resistance elements, insertion sequences, and transposons color coded as: green, vancomycin resistance (Tn1546, vanA); magenta, β-lactamase resistance (blaZ/RI/I); dark blue, quaternary ammonium compound resistance (qacC); orange, gentamicin resistance (aacA/aphD); red, trimethoprim resistance (dfrA/thyA); salmon, IS/Tn elements; yellow, conjugative transfer genes (tra). Color blocks are scaled to the size of the coding regions. The apparent break in the vancomycin resistance element represents the distance between vanX and vanY (448 bp) [see (B)]. (B) Insertion site and genetic elements of Tn1546. Nine genes are encoded on the Tn1546 transposon: ORF1 and ORF2 are transposase and resolvase enzymes, respectively, required for mobilization of the transposon; vanR and vanS encode a two-component, inducible regulatory system; the vanH dehydrogenase produces d-lactate from pyruvate; vanA encodes the d-ala-d-lac ligase; vanX encodes a D,D-dipeptidase that destroys vancomycin-susceptible d-ala-d-ala dipeptides in the cell wall precursor pool, and vanY encodes a D,D-carboxypeptidase, which removes the carboxy terminal d-ala from the cell wall precursor pentapeptides; the exact function of vanZ is unclear, but it is known to be important for teicoplanin resistance. Arrowheads indicate the direction of transcription.

The 14-kb transfer region of the VRSA plasmid (Fig. 2A, tra), including artA, traA-traM, topB, and trsO, shares 99% homology with analogous regions of pSK41(20) and pGO1 (21). The origin of transfer, oriT, is located 11 kb upstream from the tra region in pSK41 and 7.4 kb upstream from tra in pLW1043. The difference is due to the presence of pUB110 on pSK41. This element, which confers resistance to bleomycin (ble) and resistance to kanamycin, neomycin, paromomycin, and tobramycin (aad) (22), was not a component of pLW1043. However, the position of oriT relative to nes was the same for pLW1043 and pSK41/pGO1 (109 bp downstream from the gene for the nicking enzyme). The putative oriT nick region of pLW1043 (5′-TAAGTGCGCCCT-3′) was identical to those of pSK41 and pGO1.

A Tn552-like element, which mediates resistance to penicillins, was also present on pLW1043. This element, which encodes a β-lactamase, blaZ, and its regulators, blaR1 and blaI, is not found on pSK41 or pGO1. When Tn552 does insert in a pSK41 family plasmid (e.g., pUW3626), the insertion site is located preferentially within an inverted repeat near the gene for a resolvase, res (20). The insertion site in pLW1043 is consistent with this report, as the blaZ genes were found 1 kb upstream from res.

Tn1546, encoding the vancomycin resistance gene cluster, integrated into the region between aac-aphD and blaZ (Fig. 2B). The transposon was complete, without insertions or deletions, with 100% nucleotide identity with the prototype Tn1546 (GenBank M97297). Also conserved were the 38-bp imperfect terminal inverted repeats, a characteristic of the Tn3 family of transposable elements. The G+C content of the nine individual genes of Tn1546 varies from 29% to 45%, with an overall average of 39%. The very high level of vancomycin resistance indicates that Tn1546 genes were expressed as efficiently in this staphylococcal isolate as they are in enterococci.

In summary, the molecular basis of vancomycin resistance in a clinical isolate of VRSA was identified as an in vivo, interspecies transfer of Tn1546 to a multiresistance conjugative plasmid in an MRSA, most likely from a co-isolate of vancomycin-resistant E. faecalis. The VRSA plasmid was transferable to other strains of S. aureus, reinforcing concerns of potential widespread resistance to one of the few classes of agents still active against multidrug-resistant S. aureus.

Supporting Online Materials

Materials and Methods

Fig. S1


References and Notes

View Abstract

Navigate This Article