Most enterococci have naturally occurring or inherent resistance to various drugs, including cephalosporins and the semisynthetic penicillinase-resistant penicillins (e.g., oxacillin) and clinically achievable concentrations of clindamycin and aminoglycosides. Compared with streptococci, most enterococci are relatively resistant to penicillin, ampicillin, and the ureidopenicillins, with MICs of 1 µg/ml to 8 µg/ml for most Enterococcus faecalis and even higher for most E. faecium. Many enterococci are also tolerant to the killing effects of cell-wall active agents, including ampicillin and vancomycin; recent data suggest that this property may not be inherent, but rather acquired after exposure to antibiotics (4). Inherent in vivo resistance of E. faecalis to trimethoprim-sulfamethoxazole may explain the lack of efficacy in animal models. In vitro, trimethoprim-sulfamethoxazole readily inhibits most enterococci at low concentrations, but this activity is lessened by exogenous folates (5). Moreover, bactericidal activity against E. faecalis seems unreliable and very method dependent (6). In animal models, this combination has not shown good activity and is not generally accepted as an effective antienterococcal therapy, especially for systemic infections (7,8).

In addition to natural resistance to many agents, enterococci have also developed plasmid- and transposon-mediated resistance to tetracycline (as well as minocycline and doxycycline), erythromycin (plus the newer compounds azithromycin and clarithromycin), chloramphenicol, high levels of trimethoprim, and high levels of clindamycin.

Figure 1

Figure 1. Enterococcus faecalis pheromone-responsive conjugative system.
Pheromone A released from the potential recipient cell (right) interacts with plasmid A in the potential donor cell (left) to induce synthesis of aggregation substance. Attachment of...

The propensity of E. faecalis to acquire multiple antibiotic-resistance traits may result from a variety of distinctly different mechanisms for conjugation, i.e., bacterial mating. The best studied system of conjugation involves oligopeptides called pheromones and pheromone-responsive plasmids (9;Figure 1). Briefly, strains of E. faecalis typically secrete into the culture medium a number of different small peptide sex pheromones specific for different types of plasmids. When a cell containing a pheromone-responsive plasmid (the potential donor cell) comes into contact with its corresponding pheromone, transcription of a gene on the plasmid is turned on, resulting in the synthesis of a sticky substance (called aggregation substance) on its surface. When the donor cell bumps into another E. faecalis, aggregation substance, which contains two Arg-Gly-Asp motifs, sticks to the binding substance on the surface of most E. faecalis cells, causing them to clump together. In the test tube, clumps of cells actually fall to the bottom of the tube, resulting in a visible aggregate. By a process not yet well understood, the pheromone-responsive plasmid can then transfer from the donor bacterium to the other (recipient) bacterium. Once the recipient cell has acquired this particular plasmid, the synthesis of the corresponding sex pheromone is shut off to prevent self-clumping. This system of conjugation, which occurs primarily in E. faecalis, is highly efficient and results in transfer of plasmids in both filter and broth matings.

Another system of conjugation, also not well understood, involves broad host-range plasmids that can transfer among species of enterococci and other gram-positive organisms such as streptococci and staphylococci (10). The transfer frequency is generally much lower than with the pheromone system and is much more efficient with filter than with broth matings. Since staphylococci, streptococci, and enterococci share a number of resistance genes, these broad host-range plasmids may be a mechanism by which some of these resistance genes have spread among different genera.

A third type of conjugation, which involves conjugative transposons, may also explain the spread of resistance genes to many different species (11). As opposed to ordinary transposons, which can jump within a cell from one DNA location to another, conjugative transposons also encode the ability to bring about conjugation between different bacterial cells. Since plasmids typically require rather complex machinery for replication (often depending on successful interactions with host proteins) and must face additional problems of surface exclusion and incompatibility, conjugative transposons (which do not replicate, but instead insert into the chromosome or into a plasmid of the new host) appear to be an even more efficient and far-reaching way of disseminating a resistance gene. This may explain why the tetM gene of the conjugative transposon Tn916 has spread beyond the gram-positive species into gram-negative organisms, including gonococci, meningococci, and Haemophilus ducreyi, as well as into mycoplasma and ureaplasma, among others (12,13). Other resistance genes, including those encoding resistance to erythromycin and kanamycin, are also found on conjugative transposons; these frequently contain or are related to Tn916. Such transposons may have evolved from a Tn916 ancestor; their emergence suggests the possibility of further dissemination of resistance among gram-positive organisms. Particularly ominous are reports of the vanB gene cluster within large conjugative chromosomal elements that appear similar, at least in function, to conjugative transposons (14).

Although some acquired resistance of enterococci is not clinically important because the agents involved are not commonly used, other resistance greatly affects enterococcal therapy; high-level resistance (HLR) to aminoglycosides is an example. This resistance is added onto the normal low-level resistance of enterococci to aminoglycosides and typically results in MICs of > 2,000 µg/ml. This degree of resistance predicts, without exception, resistance to synergism between cell-wall active agents and the aminoglycoside to which the organism is highly resistant (1). High-level aminoglycoside resistance is most often due to aminoglycoside-modifying enzymes; HLR to streptomycin can also be ribosomal, that is, due to a mutation that results in ribosomes resistant to streptomycin inhibition. HLR to kanamycin (without gentamicin) is a fairly common trait and is due to the production of a 3'-phosphotransferase, APH(3')-III. This enzyme is important because it also eliminates synergism between cell-wall active agents and amikacin (through phosphorylation of the 3'-hydroxyl group), although it does not necessarily confer HLR to amikacin. HLR to gentamicin results from the bifunctional protein (AAC(6')-I/APH(2")-I), encoded by a single gene with two active sites, one with 6'-acetyltransferase activity and the other, 2"-phosphotransferase activity (15). The combination of these activities results in HLR or resistance to synergism for all commercially available aminoglycosides except streptomycin, which is not modified by this enzyme. However, HLR to streptomycin (due to either ribosomal resistance or a streptomycin adenylyltransferase) is also common and can coexist with the gene(s) for HLR to other aminoglycosides. Spectinomycin is also not modified by the bifunctional enzyme, but this agent, which is not a true aminoglycoside, is not generally bactericidal against enterococci and does not appear to show synergism with cell-wall active agents.

Strains of enterococci from patients with endocarditis and other serious infections for whom combination therapy is desired should be screened for HLR to streptomycin and gentamicin. HLR screening for tobramycin is not generally performed or advisable It could, in principle, be used for E. faecalis, but E. faecium isolates have a chromosomally encoded, naturally occurring gene for a 6'-acetyltransferase that eliminates synergism with tobramycin, although it does not cause HLR (MICs are typically 128-500 µg/ml); a probe for this gene has been used to confirm the identification of E. faecium isolates to species. In addition, HLR of an E. faecium isolate (without HLR to gentamicin) to tobramycin, due to an adenylyltransferase, was recently described (16). Therefore, the use of tobramycin for possible synergism in serious enterococcal infections would need to be preceded by screening for HLR to tobramycin (a test that is not commonly available), as well as for identification to species, neither of which is practical.

More recently, veterinary and human isolates resistant to moderate levels of gentamicin (256 µg/ml) were found to have a new gentamicin-modifying enzyme encoded by a gene designated aph(2")-Ic (17). This gene conferred resistance to synergism between gentamicin and cell-wall active agents and may be less easily detected than strains producing the bifunctional enzyme.

The first known penicillinase-producing isolate of enterococcus was an isolate of E. faecalis recovered from a patient in Houston, Texas, in 1981 (18). Although rare, these isolates have been reported from the United States (Texas, Florida, North Carolina, Delaware, Pennsylvania, New York, Massachusetts, and Connecticut), Lebanon, Canada, and Argentina (19). Like other enterococci, beta-lactamase­producing strains have been found as colonizers, as in the large "outbreak" of colonization in a Boston children's hospital, but they have also been associated with true infections, as was demonstrated by cases at a Virginia Veterans Administration hospital, by isolates from Argentina, and in other reports (20,21). The enterococcal penicillinase gene, identical to the gene encoding staphylococcal type A penicillinase, almost always occurs in strains with HLR to gentamicin and is often found on a transferable plasmid that also contains aph(2")-Ia/aac(6')-Ie . The relatively low levels of beta-lactamase produced by enterococci result in a marked inoculum effect with these strains so that at low and even moderate inocula (10³-10⁵ CFU/ml), penicillinase-producing enterococci usually appear no more resistant than other enterococci, while at high inocula (> 10⁷ CFU/ml), these organisms are usually highly resistant to penicillin, ampicillin, and ureidopenicillins. The activity of the penicillinase is reversed by the beta-lactamase inhibitors clavulanate, sulbactam, and tazobactam; in animal models of endocarditis, beta-lactamase inhibitors have been shown to markedly enhance the therapeutic efficacy of ampicillin or penicillin. In the clinical laboratory, penicillinase-producing enterococci are generally not detected by routine laboratory susceptibility testing, such as MICs or disk diffusion. For this reason, if a penicillin is to be used for therapy, enterococcal isolates from patients with endocarditis or other serious infections should be tested for penicillinase production by using a specific beta-lactamase test such as the chromogenic cephalosporin nitrocefin.

Nonpenicillinase-producing, penicillin-resistant enterococci have been reported for decades and usually are E. faecium. Until recently, MICs of penicillin typically ranged from 8 µg/ml to 64 µg/ml, with an occasional isolate having higher levels of resistance. However, increasingly, strains with much higher levels of penicillin resistance have been reported (22). Whether a large number of strains have converted from low-level to high-level resistance or a more limited number of strains have been disseminated is unclear. The mechanisms involved in this resistance are overproduction of a low-affinity penicillin-binding protein (a cell-wall synthesis enzyme) and a further decrease in the affinity of one of these enzymes for penicillin (23). As a possible explanation for why many vancomycin-resistant E. faecium also have very high levels of resistance to ampicillin, Rice and colleagues (24) showed that transfer of vancomycin resistance from one strain to another was linked to transfer of ampicillin resistance.

Most surprising in recent years has been the emergence among enterococci of acquired resistance to vancomycin. Vancomycin had been in clinical use since the 1950s, although it was not heavily used until the late 1970s and particularly the 1980s. Because multiple genes are involved in generating vancomycin resistance, the development of resistance was neither easy nor recent. Three phenotypes of vancomycin resistance (types A, B, and C) are now well described; a fourth, type D, has been recently reported (25). VanA-type strains are typically highly resistant to vancomycin and moderately to highly resistant to teicoplanin. This phenotype is often plasmid or transposon mediated and is inducible (i.e., exposure of bacteria to vancomycin results in the induction of the synthesis of several proteins that together confer resistance) (26).

In vancomycin-susceptible enterococci, D-alanyl-D-alanine (formed by an endogenous D-alanine-D-alanine ligase) is added to a tripeptide precursor to form a pentapeptide precursor. The D-Ala-D-Ala terminus is the target of vancomycin; once vancomycin has bound, the use of this pentapeptide precursor for further cell-wall synthesis is prevented. In the VanA phenotype, one of the proteins whose synthesis is induced by exposure of bacterial cells to vancomycin is called VanA; VanA is a ligase and resembles the D-alanine-D-alanine ligase from Escherichia coli and other organisms, including vancomycin-susceptible enterococci (27). VanA generates D-Ala-D-X, where X is usually lactate; the formation of D-lactate is due to the presence of VanH, a dehydrogenase encoded by vanH. The depsipeptide moiety, D-Ala-D-Lac, is then added to a tripeptide precursor, resulting in a depsipentapeptide precursor. Vancomycin does not bind to the D-Ala-D-Lac terminus, so this depsipentapeptide can be used in the remaining steps of cell-wall synthesis. However, when the normal pentapeptide precursor ending in D-Ala-D-Ala is also present, cells are not fully vancomycin resistant, despite the presence of D-Ala-D-Lac containing precursors. This apparent problem is taken care of in large part by vanX, which encodes a dipeptidase, VanX, that cleaves D-Ala-D-Ala, preventing its addition to the tripeptide precursor. Should any D-Ala-D-Ala escape cleavage and result in a normal pentapeptide precursor, vanY encodes an ancillary or back-up function. That is, it codes for a carboxypeptidase, VanY, which cleaves D-alanine and D-lactate from D-Ala-D-Ala and D-Ala-D-Lac termini, respectively, resulting in tetrapeptide precursors, to which vancomycin does not bind. The other genes involved in the VanA resistance complex include vanR and vanS, whose encoded proteins are involved in somehow sensing the presence of extracellular vancomycin or its effect and signaling intracellularly to activate transcription of vanH, vanA, and vanX (27). A final gene in the vanA cluster is vanZ, which encodes VanZ, the role of which is not known.

VanB, encoded by vanB in the vanB gene cluster, is also a ligase that stimulates the formation of D-Ala-D-Lac. The VanB phenotype is typically associated with moderate to high levels of vancomycin resistance but is without resistance to teicoplanin. This is explained by the observation that vancomycin, but not teicoplanin, can induce the synthesis of VanB and of VanH_B and VanX_B. However, because mutants resistant to teicoplanin can readily be selected from VanB strains on teicoplanin-containing agar, clinical resistance would likely occur among VanB strains if teicoplanin were widely used. Most of the proteins encoded by the vanA gene cluster have homologues encoded by the vanB gene cluster, except for VanZ. The vanB gene cluster has an additional gene, vanW, of unknown function.

The VanC phenotype (low-level resistance to vancomycin, susceptible to teicoplanin) is an inherent (naturally occurring) property of E. gallinarum and E. casseliflavus. This property is not transferable and is related to the presence of species-specific genes vanC-1 and vanC-2, respectively (28); a third possible species, E. flavescens and its gene vanC-3, are so closely related to E. casseliflavus and vanC-2 that different names are probably not warranted (29). These species appear to have two ligases; the cell-wall pentapeptide, at least in E. gallinarum, ends in a mix of D-Ala-D-Ala and D-Ala-D-Ser (29,30). The genes vanC-1 and vanC-2 apparently lead to the formation of D-Ala-D-Ser containing cell-wall precursors, while D-Ala-D-Ala ligases, also present in these organisms, result in D-Ala-D-Ala. The presence of both D-Ala-D-Ala and D-Ala-D-Ser precursors may explain why many isolates of these species test susceptible to vancomycin and why even those isolates with decreased susceptibility display only low-level resistance.

VanD-type glycopeptide resistance has been recently described in an E. faecium isolate from the United States (25). The organism was constitutively resistant to vancomycin (MIC > 64 µg/ml) and to low levels (4 µg/ml) of teicoplanin. Following polymerase chain reaction amplification with primers that amplify many D-Ala-D-Ala ligases, a 605-bp fragment was identified whose deduced amino acid sequence showed 69% identity to VanA and VanB and 43% identify to VanC.