INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Enterococci rank among the top four pathogens causing nosocomial infections over the past decade (11, 40). Infections caused by enterococci have always been among the most difficult to treat, given the enterococci's extensive array of intrinsic resistance characteristics, their tolerance to the bactericidal activity of all antimicrobial agents, and their frequent acquisition of novel resistance determinants (27). Ampicillin and vancomycin, however, have until recently remained almost universally active against enterococci and are the cornerstones of effective treatment of enterococcal infections.

Since 1990, a pronounced shift in the susceptibility of enterococci to ampicillin and vancomycin has been observed. The overall prevalence of ampicillin-resistant Enterococcus faecium, which had been increasing slowly over the prior decade, has increased substantially since 1990 (15). The emergence of vancomycin resistance in enterococci has been even more dramatic. Virtually nonexistent prior to 1989, by 1993 vancomycin-resistant enterococci (VRE), the vast majority of which were E. faecium strains, represented 7.9% of all enterococci collected by the Centers for Disease Control and Prevention and 13.6% of those isolates collected from intensive care units (4). More than 95% of vancomycin-resistant E. faecium strains in the United States now also express high levels of resistance to ampicillin. In many cases, VRE strains causing serious infections in hospitalized patients are resistant to all clinically available antibiotics.

Resistance to ampicillin (MIC = 32 to 64 µg/ml) in E. faecium results from increased production of a low-affinity penicillin-binding protein (PBP), PBP5, that is thought to be intrinsic to all E. faecium strains and can assume the functions of all of the other PBPs in cell wall synthesis (48). In Enterococcus hirae, which is closely related to E. faecium, increased production of PBP5 has been attributed to a deletion in psr, a presumed repressor of pbp5 expression (22). Similar regulatory mutations in an E. faecium psr analogue are thought to be important in increased expression of PBP5 in E. faecium, although direct evidence for this is lacking (48). Mutations in the structural pbp5 gene resulting in a decrease in PBP5 penicillin-binding affinity have been found in E. faecium strains with high-level ampicillin resistance (MICs of 128 to >512 µg/ml) (48).

Vancomycin inhibits cell wall (peptidoglycan) synthesis by binding to the terminal D-alanyl-D-alanine of the pentapeptide precursors, preventing the polymerization and cross-linking that are important for peptidoglycan structural stability. Glycopeptide (vancomycin) resistance in enterococci results from the acquisition of resistance operons, expression of which results in the synthesis of precursors terminating in D-alanine-D-lactic acid that bind glycopeptides with low affinity and in the destruction of normal pentapeptide precursors (2). Two such operons, vanA and vanB, have been described to date.

The close association between ampicillin and vancomycin resistance phenotypes in VRE is not explained by synergistic or duplicated mechanisms of resistance. In fact, early studies indicated that expression of vancomycin resistance in E. faecium resulted in hypersusceptibility to penicillin (41). Investigators hypothesize that this synergy results from the inability of PBP5 to process lactated peptidoglycan precursors, necessitating the use of other PBPs that are highly susceptible to ampicillin. Some highly ampicillin-resistant VRE, however, are resistant to the penicillin-vancomycin synergism, suggesting that mutations within pbp5 may result in enzymes both resistant to inhibition by ampicillin and able to process lactated peptidoglycan precursors (48).

The exchange of antimicrobial resistance genes between enterococci is facilitated by conjugative plasmids and transposons. Among the most interesting of enterococcal transposons are the conjugative transposons, which are capable of transferring between enterococcal chromosomes and exhibit a very broad host range (7). Conjugative transposons resemble lambdoid bacteriophages in their mechanism of transposition, employing a nonreplicating circular form as a transposition intermediate. The two most extensively studied conjugative transposons are Tn916 and Tn1545 (8, 9). Although these two transposons differ in size (18 versus 25.4 kb, respectively) and resistance determinants (tetracycline/minocycline versus tetracycline/minocycline, erythromycin, and kanamycin, respectively), the genes encoding their transposition functions are identical, as are their termini.

Ampicillin resistance mediated by PBP5 has never been shown to be mobile or transferable in E. faecium. A gene encoding a related low-affinity PBP, PBP3r, has been localized to a plasmid in E. hirae (32). The vanB operon is most often localized to the bacterial chromosome and transfers between enterococci at low frequency, if at all. The appearance of the vanB operon in a wide variety of clonally distinct, high-level ampicillin-resistant E. faecium strains in several regions suggests the existence of interenterococcal transfer of this operon and linkage to the ampicillin resistance determinant. In this paper, we report the discovery of a 27-kb vanB-containing Tn916-like transposon that is itself integrated within a larger transferable element that also contains a mutated pbp5 gene encoding high-level resistance to ampicillin.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Strains and plasmids. Relevant bacterial strains, cloning vectors, and plasmids are listed in Table 1. More than 400 VRE strains from northeast Ohio were collected as part of a study of the molecular epidemiology of VRE spread in the region. C68 is a clinical E. faecium strain isolated from the feces of a patient hospitalized in northeast Ohio. It represents the predominant clone in the region, with a genotype that is found in more than 50% of area isolates (data not shown). GE-1 is a plasmid-free, rifampin- and fusidic acid-resistant E. faecium strain used as a recipient in conjugation experiments. CV133 and CV142 are two transconjugants that resulted from independent matings between C68 and GE-1. Enterococcus faecalis JH2-7 is a plasmid-free recipient strain used in mating experiments (19). E. faecalis OGIXRF is an OG1 derivative used as a recipient in mating experiments (18). It was selected by sequential inoculation of fusidic acid (25 µg/ml) and rifampin (100 µg/ml) plates with ca. 10⁸ CFU of an overnight culture of OG1X (streptomycin resistant). Single colonies were harvested and purified, and their resistance phenotype was confirmed by replating on selective plates containing both rifampin and fusidic acid at the concentrations stated above.

Conjugation experiments. Conjugation experiments for transfer of chromosomal elements were carried out by mixing 50 µl of overnight cultures of donor and recipient strains (grown in nonselective brain heart infusion [BHI] broth) in a sterile test tube and then spreading the mixture across a BHI agar plate. Plates were incubated at 37°C overnight. The following day, the confluent growth on the plate was removed with a platinum loop and suspended in 500 µl of sterile saline. Aliquots (150 µl) of this suspension were then plated onto selective plates containing vancomycin (10 µg/ml), fusidic acid (25 µg/ml), and rifampin (100 µg/ml). The plates were incubated for 5 days at 37°C and examined each morning for the appearance of colonies. Colonies were restreaked onto identical plates and tested for associated antimicrobial resistance by streaking onto BHI agar plates containing fusidic acid (25 µg/ml), rifampin (100 µg/ml) and erythromycin (10 µg/ml), gentamicin (500 µg/ml), streptomycin (2,000 µg/ml), or tetracycline (10 µg/ml). MICs of vancomycin, teicoplanin, and ampicillin were determined in BHI broth according to standard techniques (30). Conjugation experiments designed to detect transposition of Tn5382 were performed by introducing broad-host-range plasmid pAM1 (kindly provided by Barbara Murray) into CV133 by conjugation with selection on BHI plates containing erythromycin (10 µg/ml) and vancomycin (10 µg/ml). CV133(pAM1) was then mated with JH2-2 (ciprofloxacin resistant) (kindly provided by Roland Leclerq) by standard filter mating techniques as described by Christie et al. (5). Selection for transconjugants occurred on BHI plates containing either erythromycin (10 µg/ml) and ciprofloxacin (20 µg/ml) or erythromycin (10 µg/ml), vancomycin (10 µg/ml), and ciprofloxacin (20 µg/ml).

Hybridization experiments. Genomic DNA was extracted as described previously (38) with the following modifications. After the lysozyme-RNase-proteinase K step (which was shortened to 2 h), the resulting suspension was mixed with a CTAB (hexadecyltrimethyl ammonium bromide)-NaCl solution and heated at 68°C for 20 min. This mixture was then extracted once with phenol-chloroform-isoamyl alcohol (25:24:1) and once with chloroform-isoamyl alcohol. DNA was precipitated with 100% isopropanol, washed with 70% ethanol, and resuspended in TE (Tris [50 mM], EDTA [10 mM], pH 7.0) buffer. Genomic DNA was digested with restriction enzymes for 1 to 2 h at 37°C according to the specifications of the manufacturer (Promega, Madison, Wis.). The digested DNA was separated on 0.7 to 1% agarose gels overnight. Separated DNA was denatured, neutralized, transferred to nylon filters by using a negative pressure transfer apparatus (Pharmacia LKB, Uppsala, Sweden), and baked at 80°C for 1 to 2 h to fix the DNA to the filter. Filters were prehybridized and hybridized with digoxigenin-labeled probes overnight at 68°C and washed under conditions of high stringency according to the specifications of the manufacturer (Boehringer-Mannheim, Indianapolis, Ind.).

PFGE. Genotypic characterization of VRE isolates was accomplished by separating SmaI-digested genomic DNA by pulsed-field gel electrophoresis (PFGE). DNA was extracted by using the GenePath Group 1 Reagent Kit (Bio-Rad, Hercules, Calif.) with the following modifications: (i) proteinase K (Sigma Chemical Company, St. Louis, Mo.) (25 mg/ml) in sterile water was used, and (ii) after the initial proteinase K incubation, plugs were incubated for 16 to 20 hours overnight in 20 µl of proteinase K and 500 µl of 0.5 M EDTA (pH 7.6) plus 0.5% Sarkosyl and restriction enzyme SmaI (25 U per plug) with the appropriate restriction enzyme buffer (Gibco-BRL, Gaithersburg, Md.). Digested DNA was separated on 1% agarose gels for 20 h with the settings described by Murray et al. (28). Southern transfer of DNA from PFGE gels was accomplished as described by Maniatis et al. (25). Transfer proceeded for 48 h. Hybridization and washing steps were as described above.

Cloning of genomic DNA fragments. Once fragments of interest were identified by hybridization, they were removed from agarose gels and purified by a glass bead preparation (Geneclean; Bio 101, La Jolla, Calif.). These fragments were then ligated to like-digested pACYC184 or pUC18 and transformed into Escherichia coli DH5 (16) or E. coli DH10B (Gibco-BRL, Gaithersburg, Md.) by electroporation (Bio-Rad). Transformed preparations were inoculated onto plates containing antimicrobials selective for the cloning vectors, and colonies with the appropriate inserts were identified by colony hybridization techniques as previously described (3). These colonies were purified and subcloned as necessary for further sequencing.

PCR amplification. Amplification of genomic DNA was performed on a 9600 thermal cycler with Taq DNA polymerase according to standard protocols as recommended by the manufacturer (Perkin-Elmer Cetus, Roche Molecular Systems, Branchburg, N.J.). In most cases, an overnight culture of either C68 or CV133 was used as a template. For detection of the joint region of Tn5382, extracted genomic DNA was used as a template. Primers used for joint amplification of Tn5382 were TnOUT 9-3 (5'-TCCGAAAGTAAATTGGTAGTA-3') and TnVAN OUT (5'-CGATCCCGCAAGGCCAGAAATG-3'). Variations were introduced into each individual protocol depending on the expected size of the amplification product and the specific primers used. Ten microliters of a total 50-µl PCR reaction mixture was loaded on a 0.7 to 1.2% agarose gel for analysis.

DNA sequence analysis. DNA sequencing was performed from cloned DNA on double-stranded templates with the A.L.F. automated sequencing kit and fluorescein- or Cy5 indodicarbocyanine dye-labeled forward and reverse primers. DNA was purified by using the Wizard miniprep system (Promega). Sequence was determined with the ALFExpress automated sequencer (Pharmacia LKB). Sequences were compared for homology with sequences entered into the Genbank, EMBL, DDBJ, and PDB databases by using the Blastn and Blastx local alignment search tools (1) and further analyzed by using MacDNAsis version 2.0 (Hitachi, Ltd.) and DNAStar (Madison, Wis.) sequence analysis programs.

Nucleotide sequence accession numbers. The sequences of the region extending from the terminal portion of vanX_B to beyond the end of Tn5382 and of the left end of Tn5382 are entered in GenBank. The accession numbers are AF063010 and AF063900, respectively.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

A large (>25-kb) Tn916-like element containing the vanB operon. Using an internal PCR amplification product of the vanB gene as a probe, we determined restriction maps of regions flanking the vanB operon for more than 20 VanB-type VRE strains collected from northeast Ohio hospitals in 1995 and 1996. These strains were all determined to be clonally distinct by PFGE and IS6770 hybridization (29, 46). Seven groups of common vanB region restriction maps were identified. The largest group (designated group 2) included 11 strains (data not shown). One group 2 strain (E. faecium C68) was chosen for more detailed study. We ligated an 8.5-kb BglII/HindIII vanB chromosomal restriction fragment from C68 to plasmid pACYC184 and transformed the ligation mixture into E. coli DH10B by electroporation. Colonies containing the appropriate insert were identified by hybridization with the vanB probe. One recombinant plasmid (pCWR370) was chosen for further study. A 4.2-kb EcoRV/HindIII subfragment of pCWR370 containing the terminal portion of the vanX_B gene and approximately 4 kb of flanking sequence was cloned into vector pUC18. This recombinant plasmid was designated pCWR374. We subcloned and sequenced this fragment by using pUC18 forward and reverse primers as well as custom synthetic primers based on sequences internal to the insert. We identified several open reading frames (ORFs) with significant homologies to ORFs present within enterococcal conjugative transposon Tn916 (Fig. 1) (20, 42), with translated products exhibiting similarity to the Tn916 excisase and integrase and to the deduced amino acid sequences from orf7 and orf8, whose functions have not been determined. We have designated these ORFs xis-VB, int-VB, orf7-VB, and orf8-VB. The closest relationship observed was between the deduced proteins encoded by xis-VB and xis-Tn from Tn916. Xis-VB is 59 amino acids in length, with 85% identity over amino acids 5 through 54 of Xis-Tn. Int-VB is 397 amino acids in length (compared to an Int-Tn length of 406 amino acids). The similarity between Int-VB and Int-Tn was also significant, with 68% identity over the length of the ORF. This similarity was especially striking on the carboxy-terminal end, where there was 85% identity over 53 amino acids, including the arginine, histidine, and tyrosine residues conserved in this family of integrases (33). Similarities were less striking between ORF7-VB (143 amino acids in length; 35% identical to ORF7 from Tn916 over 129 amino acids) and ORF8-VB (77 amino acids in length; 33% identical to ORF8 from Tn916 over 64 amino acids). An ORF corresponding to traA (Fig. 1), which has been shown to be essential for Tn916-mediated conjugation, could not be identified; however, a 60-bp segment of DNA upstream of xis-VB on the complementary strand (the expected position of traA) exhibited 60% identity with amino acids 33 to 52 of the traA deduced amino acid sequence. This segment fell within an ORF without an identifiable start codon. However, further analysis indicated that a stop codon upstream of the region of homology occurred in an ORF whose start codon overlapped the start of xis-VB on the opposite strand. This relationship is analogous to the relationship of xis-Tn and traA in Tn916. The reading frame of this truncated ORF was the same as that of the sequence homologous to traA, suggesting that it represented at least a vestige of a traA-like ORF. All ORFs are in the same relative orientation to the vanB operon as are their homologues in Tn916 to the tet(M) gene (Fig. 1). No homologies were found between the vanX_B-flanking sequence and orf6, -9, and -10 of Tn916, which are thought to be related to the tet(M) determinant based on the size of tet(M)-related transcripts (42, 44). The GC content of these ORFs is ca. 49%, consistent with the previously reported GC content of the vanB operon and distinctly different from the 38% GC content of Tn916 (14).

View larger version (12K): [in this window] [in a new window]   FIG. 1.   Schematic representation of the right (integrase-encoding) end of Tn5382 and the left end of Tn916 (42). The positions of the vanXB (Tn5382) and tet(M) (Tn916) genes are shown. The approximate positions of orf6, -9, and -10 from Tn916 are indicated. The number of base pairs cited in this region represents the number of bases between the termination codon of vanXB and the start codon of orf7-VB in the upper diagram and the number of bases between the stop codon of tet(M) and orf7 in the lower diagram. Investigators have postulated that orf6, -9, and -10 are transcriptionally related to the tet(M) gene (42, 43). The positions of the individual ORFs and their relative sizes are indicated. Directions of transcription of the ORFs are indicated by the arrows above the ORFs. Amino acid identities between the ORFs of the two transposons are listed between them. The relative GC contents of Tn5382 and the flanking sequences are indicated at the top. Sequence alignments were established by using the Blastn and Blastx basic alignment search tool (1).

View larger version (34K): [in this window] [in a new window]   FIG. 2.   Sequences of the termini of Tn5382 (line A) and comparison with the ends of Tn916 (line B). Identical nucleotides are indicated by vertical lines between Tn5382 and Tn916 (line B). Identical nucleotides are indicated by vertical lines between Tn5382 and Tn916. The boxed sequences represent the 11-bp imperfect inverted repeats of Tn5382. Arrows indicate the direct repeats within the ends of Tn916. Boldface underlining represents the regions of Tn916 protected by the integrase enzyme in DNase protection assays.

View larger version (16K): [in this window] [in a new window]   FIG. 3.   Position of Tn5382 relative to the pbp5 structural gene in E. faecium C68. The stop codon of pbp5 (sequence in uppercase) is indicated by the double underline. The target sequence (and presumed target duplication after insertion) is boxed.

View larger version (24K): [in this window] [in a new window]   FIG. 4.   Relationships between the putative target sequences within E. faecium C68 and E. faecium D366 and the circularized form of Tn5382 prior to insertion. Both target sites exhibit significant homology with the ends of the transposon, suggesting that they may represent hot spots for Tn5382 insertion. The joint region in this diagram is represented by a series of Ns, since we have no knowledge of the flanking regions prior to this episode of circularization.

Circularization of Tn5382. Conjugative transposons such as Tn916 transpose by a conservative mechanism that involves the formation of a nonreplicative circular intermediate. The existence of such circular intermediates of Tn916 and related transposons has been confirmed by using amplification strategies with primers designed to direct polymerization outward from the ends of the transposon (Fig. 5A) (36, 38). Amplification products are obtained only if the element excises and forms a circular intermediate. The amplified area contains the point at which the two ends of the transposon are joined, commonly referred to as the joint. Genomic DNA preparations from C68, CV133, D366, and E. faecalis V583 (kindly provided by Dan Sahm) were used as templates for the amplification experiments. We were able to amplify a fragment of the predicted size by using genomic DNA from C68 as a template (Fig. 5B). Using this labeled joint fragment as a probe, we identified similar amplification products from CV133 and D366 but not from V583 (data not shown). PCR products were cloned into vector PCRII (Invitrogen). Sequence analysis of three cloned PCR products confirm that they represent a joint amplification product. The joint region in two instances was 5 bp in length (TTTGT), whereas the joint in the other was 6 bp (TTTGTA). We have previously observed differences in the number of nucleotides comprising the joint region of Tn916 (36). These data confirm that Tn5382 excises in enterococci, compelling evidence that it is a functional transposon with similarities to Tn916 in its mechanism of transposition.

View larger version (17K): [in this window] [in a new window]   FIG. 5.   (A) Schematic representation of strategy to amplify the putative joint region of Tn5382, based on previous strategies for amplification of similar regions of classic conjugative transposons. P1 and P2 represent the two outward-directed primers used to synthesize the amplification product. (B) Joint PCR product generated from genomic DNA isolated from E. faecium C68. Lane 1, X174 digested with HaeIII (size standard); lane 2, PCR product from E. faecium C68. See the text for details of primers and joint sequences.

Transposition experiments. Transfer of pAM1 from CV133(pAM1) to JH2-2 (ciprofloxacin resistant) at rates of 10⁷ to 10⁶/donor CFU was readily demonstrable. No transconjugants expressing resistance to vancomycin were detectable. Given a donor concentration of roughly 10⁹ CFU in these experiments, transposition into the plasmid would have to occur at a rate of 10² to 10³ per plasmid transfer event to be detectable in these experiments.

Prevalence of Tn5382. To estimate the prevalence of Tn5382, we designed primers (9-3 650 [5'-GTTCTTATTCCGCAGGTGGTGATT-3'] and 9-3 362 [5'-ACGCCATGCTATTTACTTCCGGC-3') to amplify a small product internal to the nonintegrase (left) end of the element. PCR amplification studies were performed directly from overnight cultures of VanB-type E. faecium strains from diverse geographic regions. This end of Tn5382 was present in vanB VRE from Honolulu, Hawaii (one of two strains studied), Los Angeles, Calif. (one of two), Paris (one of one), Pittsburgh, Pa. (one of one), Providence, R.I. (one of one), and Scranton, Pa. (one of one), as well as in 10 of 14 northeast Ohio clones tested. Amplification products were not observed from E. faecalis V583 (St. Louis, Mo.), three strains from Chicago, Ill., and two strains from Houston, Tex.

Structural mutations within the C68 pbp5 gene. Preliminary analysis of the C68 pbp5 gene indicates that the pbp5 ORF is intact and is preceded ca. 80 bp upstream by an ORF similar to previously described psr (negative regulator) genes. Several mutations implicated in high levels of ampicillin resistance have been identified within the pbp5 ORF, including mutations corresponding to an NK change at amino acid 485 and an AT change at amino acid 499 found in highly resistant E. faecium H80721 (48). An extra serine at position 466, first identified in H80721, is also present in C68. Since H80721 was isolated in Pittsburgh, Pa., not far from Cleveland, Ohio, these findings suggest that the two strains may have a common lineage.

Cotransfer of the Tn5382 and the pbp5 gene. To determine whether Tn5382 is conjugative, we performed mating experiments (38) between E. faecium C68 and recipient strains E. faecium GE-1, E. faecalis JH2-7 (19), and E. faecalis OG1XRF. Vancomycin resistance was transferable from C68 to GE-1 in matings on solid media at low frequency (10⁹ to 10⁸/recipient CFU). No transfer to either E. faecalis recipient was observed in several mating experiments. Two E. faecium transconjugants (designated CV133 and CV142) were further characterized by PFGE of SmaI-digested genomic DNA (Fig. 6). In CV133, transfer was associated with the loss of a recipient SmaI band and the appearance of a new band approximately 130 kb larger than the absent band. In CV142, a missing band was not obvious, but a new band was apparent. If insertion occurred within the collection of similar-sized bands observed at the uppermost portion of the recipient digest (Fig. 6), the calculated size of the transferred DNA would be 160 kb. These findings suggested that a 130- to 160-kb transferable element had integrated into two distinct loci within the E. faecium GE-1 chromosome in association with transfer of vancomycin resistance.

View larger version (113K): [in this window] [in a new window]   FIG. 6.   PFGE of SmaI-digested genomic DNA from C68 (lane A), CV133 (lane C), CV142 (lane F), and GE-1 (lane I). Lanes B, D, G, and J, hybridizations of Southern transfers of lanes A, C, F, and I, respectively, with a 2.1-kb EcoRV fragment internal to the vanHB-vanB-vanXB operon. Lanes E, H, and K, hybridization of the gel with a PCR amplification product of a region internal to the pbp5 gene. Blots were not stripped and reprobed; duplicate lanes from the same gel were blotted. The additional gel lanes are not shown to conserve space. Lane L, Megabase II size standard (Bethesda Research Laboratories), with the corresponding band sizes (in kilobases) at the right. Hybridizations were performed and filters were washed under high-stringency conditions.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

The discovery of a genetic linkage between a Tn916-like vanB-carrying putative transposon and a high-level ampicillin resistance determinant within a transferable element is important in several respects. On a basic scientific level, it provides a genetic explanation for the virtually universal observation that vancomycin-resistant E. faecium strains also express high levels of resistance to ampicillin. It also expands the family of Tn916-like transposons, unique elements that are important in the dissemination of antimicrobial resistance in a broad range of bacteria. On a clinical level, the association of the two determinants on a transferable element has important implications for selection of these strains. Specifically, antimicrobial pressure that selects for high levels of resistance to ampicillin will also select for resistance to vancomycin. The vancomycin-ampicillin resistance linkage could therefore explain recent observations that increased use of extended-spectrum cephalosporins (which have no activity in the presence of mutated PBP5) is associated with an increased prevalence of VRE colonization within hospitals (34).

Since we have not demonstrated transposition in a recombination-deficient background, Tn5382 does not formally meet the definition of a transposon. It does, however, demonstrate numerous characteristics that are emblematic of transposons. The structural similarity with Tn916 is striking. In addition, we have identified and characterized two separate insertions of the element, with highly suggestive evidence for the presence of a target duplication in both locations. Finally, our ability to amplify a joint region from Tn5382 confirms that it is capable of excision to form a circular intermediate. Excision is thought to be the rate-limiting step for Tn916 transposition. The fact that we have not been able to demonstrate transposition of Tn5382 in vitro is most likely due to the lack of appropriate tools for selection of transposition events in E. faecium. We were not able to detect transposition into pAM1, but the transfer frequency of this plasmid is relatively low. The most commonly employed plasmids for detection of transposition in enterococci are the pheromone-responsive plasmids, but the host range of these plasmids is generally restricted to E. faecalis. A single pheromone-responsive plasmid, pHKK703, has been described for E. faecium (17). Attempts to introduce this plasmid into CV133 were unsuccessful.

There are several possible explanations for the fact that we were not able to demonstrate transfer of vancomycin resistance independent of penicillin resistance in these experiments. It is possible that Tn5382 is not a conjugative transposon. Alternatively, if Tn5382 is a conjugative transposon, its conjugative genes may not function in enterococci. It is also possible that this particular transposon has a mutated traA equivalent, resulting in loss of transposition function. Interruption of the traA gene of Tn916 eliminates conjugative transposition. Finally, it is possible that independent transfer of Tn5382 falls below the limit of detection with these experiments. Further experiments to examine the conjugative potential of Tn5382 are planned.

The discovery of the vanB operon within a mobile element is not altogether surprising. The genetic heterogeneity of VanB-type strains in clinical outbreaks suggests that the operon is incorporated within structures that are capable of conjugal transfer between enterococcal strains (26). In addition, the GC content of the vanB operon (12) is distinct enough from that of enterococcal genes that some mechanism for its arrival in enterococci must be presumed. It is plausible that conjugal transfer of Tn5382 is the mechanism by which the vanB resistance operons enter into the enterococcus from their species of origin. Equally plausible is a scenario in which a broad-host-range plasmid entered into the Tn5382-harboring species and then returned to E. faecium after incorporating Tn5382. We have reported evidence for a similar mechanism of resistance transfer between staphylococci and enterococci (3).

Although Tn5382 is significantly different than Tn916, its structure is highly analogous. The identical positioning of the resistance determinants relative to the transposition genes in the two transposons is especially noteworthy. In both cases, the orf7 and orf8 homologues begin downstream of ORFs known (in the case of the vanX_B gene) or postulated (in the cases of orf6, orf9, and orf10 of Tn916) to be related to the resistance determinant. It has been suggested that conjugative transposons represent elements highly developed for movement between different genera and that the associated resistance genes represent relatively recent arrivals (31). The data reported in this paper are consistent with this hypothesis and raise the possibility that this location within conjugative transposons represents a hot spot for insertion of resistance determinants, perhaps similar to the well-described integrons found on plasmids of gram-negative species. It will be of interest to determine the nucleotide sequence of the region immediately upstream of the vanR gene (at the 5' end of the vanB operon), to determine whether a novel tet(M) gene is located in that position and to compare the sequence with upstream sequences from the analogous region in Tn916.

The mechanism of excision and integration of conjugative transposons resembles that of lambdoid bacteriophages. Unlike lambdoid bacteriophages, however, the circular intermediates of conjugative transposons are joined at the ends by mismatched strands representing sequence from flanking regions on either side of the previous insertion. Recent data suggest that the heteroduplexed joint region of conjugative transposons is repaired to form a homoduplex, with preference for retention of a specific strand (24). On subsequent insertion, transposon-flanking regions represent the target sequence on one side and one of the mismatched strands (from the circular form) on the other. Our ability to amplify joints from the circularized forms of Tn5382 indicates that it excises and forms circular intermediates, much like Tn916. Unlike for Tn916, however, we have identified what appear to be duplications of the target sequence flanking two Tn5382 insertions. In one case, the apparent target duplication is 6 bp; in the other, it is 5 bp. Tn5276, a 70-kb nisin-sucrose conjugative transposon identified in Lactococcus lactis, is also reported to generate 6-bp duplications of the target sequence (35). Without knowing the sequences flanking the prior insertion, however, we cannot draw firm conclusions regarding the generation of target duplications by Tn5382.

To our knowledge, this paper represents the first report of transferable pbp5 in E. faecium. The close genetic linkage and cotransfer of the vanB operon and the high-level resistance pbp5 gene are of considerable importance. The almost universal association, in the United States, between vancomycin resistance and ampicillin resistance in E. faecium has to date been unexplained. Genetic linkage of the two resistance determinants on a transferable element neatly explains the association. The observation that the restriction map flanking the vanB gene in E. faecium C68 is identical to that observed for 10 other clonally distinct VRE from northeast Ohio (data not shown) suggests that transfer of the larger (130- to 160-kb) element between enterococci is the primary mechanism by which interenterococcal dissemination of VanB-type vancomycin resistance and high-level ampicillin resistance occurs in this region. Since VanB-type strains represent ca. 80% of VRE in northeast Ohio (9a), this dissemination has had a profound impact on the ability to effectively treat enterococcal infections. Further studies are planned to determine the prevalence of this larger element as well as its internal structure and mechanism(s) of transfer.