ABSTRACT
Enterococcus faecalis isolates are resistant to clindamycin (CLI) and quinupristin-dalfopristin (Q-D), and this is thought to be a species characteristic. Disruption of a gene (abc-23, now designated lsa, for “lincosamide and streptogramin A resistance”) of E. faecalis was associated with a ≥40-fold decrease in MICs of Q-D (to 0.75 μg/ml), CLI (to 0.12 to 0.5 μg/ml), and dalfopristin (DAL) (to 4 to 8 μg/ml) for the wild-type E. faecalis parental strain (Q-D MIC, 32 μg/ml; CLI MIC, 32 to 48 μg/ml; DAL MIC, 512 μg/ml). Complementation of the disruption mutant with lsa on a shuttle plasmid resulted in restoration of the MICs of CLI, Q-D, and DAL to wild-type levels. Under high-stringency conditions, lsa was found in 180 of 180 isolates of E. faecalis but in none of 189 other enterococci. Among 19 erm(B)-lacking Enterococcus faecium strains, 9 (47%) were highly susceptible to CLI (MIC, 0.06 to 0.25 μg/ml) and had DAL MICs of 4 to 16 μg/ml; for the remaining erm(B)-lacking E. faecium strains, the CLI and DAL MICs were 4 to >256 and 2 to >128 μg/ml, respectively. In contrast, none of 32 erm(B)-lacking E. faecalis strains were susceptible (CLI MIC range, 16 to 32 μg/ml; DAL MIC range, ≥32 μg/ml). When lsa was introduced into an E. faecium strain initially susceptible to CLI, the MICs of CLI and DAL increased ≥60-fold and that of Q-D increased 6-fold (to 3 to 6 μg/ml). Introduction of lsa into two DAL-resistant (MICs, >128 μg/ml), Q-D-susceptible (MICs, 0.5 and 1.5 μg/ml) E. faecium strains (CLI MICs, 12 and >256 μg/ml) resulted in an increase in the Q-D MICs from 3- to 10-fold (to 8 and >32 μg/ml), respectively. Although efflux was not studied, the similarity (41 to 64%) of the predicted Lsa protein to ABC proteins such as Vga(A), Vga(B), and Msr(A) of Staphylococcus aureus and YjcA of Lactococcus lactis and the presence of Walker A and B ATP-binding motifs suggest that this resistance may be related to efflux of these antibiotics. In conclusion, lsa appears to be an intrinsic gene of E. faecalis that explains the characteristic resistance of this species to CLI and Q-D.
Over the past few years, enterococci have emerged as important bacterial pathogens in nosocomial infections (12, 25-27, 40). These organisms have acquired and/or intrinsic resistance to many different antibiotics (18, 19), which poses a serious problem for the treatment of patients infected with these organisms. Studies have shown that Enterococcus faecalis, unlike E. faecium, is usually resistant to quinupristin-dalfopristin (Q-D), with MICs of 4 to ≥32 μg/ml (3, 10, 12, 35), and that both species are typically resistant to clindamycin (CLI) (12). Acquired resistance to Q-D in E. faecium has also been described, and contributing mechanisms include drug inactivation by enzymes, structural or conformational alterations in ribosomal target binding sites, and efflux of antibiotic out of cells (3, 10). In E. faecalis, however, the mechanism of resistance to Q-D has not been well studied.
We recently investigated the presence of putative transporters in E. faecalis, identified 34 possible transporter homologs, and made disruption mutants of 31 of these (8). Among these mutants were ones with increased susceptibility to novobiocin, pentamidine, daunorubicin, and norfloxacin and one, whose disrupted gene was originally designated abc-23, with reduced susceptibility to Q-D and CLI; the MICs of ∼20 other compounds were similar to those for wild-type OG1RF (8). In the present study, we have further studied this gene and its effect by comparing it with known ABC transporters, by complementing the disruption mutant and introducing the gene into E. faecium strains on a shuttle vector, and by determining its distribution among Enterococcus spp. Based on these results, we have renamed this gene lsa in recognition of its apparent role in the intrinsic resistance of E. faecalis to lincosamides (CLI) and streptogramin A (dalfopristin [DAL]).
MATERIALS AND METHODS
Bacterial strains and MIC studies.The bacteria used in this study were obtained from the collection of our laboratory, which was compiled over the past 20 years. The recipient strains and plasmids used in the study are listed in Table 1. These include E. faecalis strain OG1RF (29) and E. faecium isolates SE34 (TX1330; recovered from feces of a healthy community volunteer [7]), TX2466 (a clinical isolate [23]), and D344-S (36); the E. faecium strains were chosen because of their differing susceptibilities to CLI (MICs of 8 to 16, <0.25, and >256 μg/ml, respectively). A total of 492 isolates of enterococci, including 257 of E. faecalis, 216 of E. faecium, 6 of E. hirae, 5 of E. durans, 2 of E. casseliflavus, 2 of E. mundtii, 1 of E. gallinarum, 2 of E. solitarius, and 1 of E. raffinosus (12, 44), were used for lsa probing and/or susceptibility testing. MICs were determined by agar dilution (30, 31) or by the E test (PDM Epsilometer test; AB BIODISK North America, Inc., Piscataway, N.J.) in accordance with the manufacturer's instructions. Erythromycin (ERY), CLI, kanamycin (KAN), and chloramphenicol (CHL) were purchased from Sigma Chemical Co., St. Louis, Mo., and quinupristin, DAL, and Q-D were provided by Aventis Pharma S.A., Vitry-sur-Seine Cedex, France.
Bacterial strains and plasmids used in this study
DNA extraction, PCR, sequencing, and cloning.DNA extraction (46) and PCRs were performed with the PCR Optimizer kit (Invitrogen, San Diego, Calif.); PCR products were analyzed by automated DNA sequencing at the Microbiology and Molecular Genetics core facility, University of Texas Medical School, Houston, Tex. Sequence analysis was done by using the BLAST network service of the National Center for Biotechnology Information. The Genetics Computer Group software package (Genetics Computer Group, Madison, Wis.) was used to compare similarities among other sequences. ClustalW, at the Baylor College of Medicine website, was used, and the GeneDoc software was used for editing and shading of sequences. Cloning was done with standard methods (42).
Disruption mutation in lsa (abc-23) of E. faecalis.The disruption mutation in strain OG1RF was created previously (8). Briefly, the disruption mutant was created by using a PCR-amplified ∼700-bp intragenic DNA fragment from E. faecalis strain V583 inserted into previously described pBluescript derivative pTEX4577 containing aph(3′)-IIIa (13, 43), resulting in pTEX5332, which was electroporated into competent cells of OG1RF and selected with KAN at 2,000 μg/ml. The resulting mutant was previously shown by PCR to have the targeted insertion of the plasmid (8). In the present study, the insertion was also confirmed by hybridizing EcoRI digests of genomic DNAs of wild-type E. faecalis OG1RF and the lsa disruption mutant with an intragenic lsa DNA probe under high-stringency conditions. In addition, the susceptibility of this mutant to quinupristin, DAL, ampicillin, tetracycline, and ERY was determined by the E test. Recombinant colonies of TX5332 (lsa disruption mutant) were further analyzed by pulsed-field gel electrophoresis (28) of SmaI-digested genomic DNA by comparison with wild-type OG1RF to confirm the host background.
Complementation of TX5332 (lsa disruption mutant).An ∼2-kb fragment (the 1,497-bp lsa ORF, ∼300 bp upstream and ∼200 bp downstream obtained from the V583 E. faecalis TIGR database) was PCR amplified from wild-type E. faecalis V583 and was first cloned into the pCR2.1 vector of the TA cloning kit. This PCR fragment was excised from vector pCR2.1 by digestion with restriction enzymes XbaI and BamHI and then recloned into shuttle vector pWM401 (47), resulting in pTEX5333. Plasmid pTEX5333 DNA was electroporated into competent cells of TX5332 (lsa disruption mutant), and selection was made on Todd-Hewitt agar (Becton Dickinson, Cockeysville, Md.) supplemented with 0.25 M sucrose, KAN at 2,000 μg/ml and CHL at 8 μg/ml. The resulting colonies were restreaked on KAN-CHL plates and analyzed for resistance to CLI, Q-D, and DAL by the E-test or agar dilution method.
Determination of the stability of the components of TX5333.Growth curves comparing wild-type OG1RF, TX5332 (the lsa disruption mutant), and TX5333 (the complemented lsa disruption mutant) were determined with and without antibiotics selective for the chromosomal disruption and/or the shuttle plasmid. Observations were made by measuring optical density at 600 nm hourly, and CFU determinations at 24 h were made on brain heart infusion (BHI) agar (Difco Laboratories, Detroit, Mich.) with and without antibiotics. BHI broth was used to grow wild-type OG1RF; BHI and BHI-KAN were used to grow TX5332; and BHI, BHI-KAN, and BHI-KAN-CHL were used to grow TX5333.
Effect of lsa on E. faecium antibiotic resistance .Electrocompetent cells of E. faecium strains TX1330, TX2466, and D344-S were prepared as previously described (14, 20). Following electroporation of pTEX5333, selection was made on Todd-Hewitt agar supplemented with 0.25 M sucrose and CHL at 8 μg/ml. The resulting colonies were restreaked onto BHI agar-CHL plates and tested for susceptibility to CLI, Q-D, DAL, and quinupristin by the E-test or agar dilution method and also by pulsed-field gel electrophoresis (28) of SmaI-digested genomic DNA and compared to each parental E. faecium strain.
Distribution of lsa among Enterococcus spp.Three hundred sixty-nine enterococcal isolates were tested for the presence of lsa by colony lysate hybridization under high-stringency conditions with an lsa intragenic DNA probe as previously described (43). The erm(B) DNA gene probe was PCR amplified as previously described (44) and used for hybridization under high-stringency conditions. The DNA gene probes and hybridization conditions for efaAfs and aac(6′)-Iifm were the same as those used in a previous study (12).
RESULTS AND DISCUSSION
Determination of stability of components of the E. faecalis lsa disruption mutant by complementation and susceptibility testing.Testing of the complemented lsa disruption mutant (TX5333) showed that when TX5333 was grown in BHI broth or in BHI with CHL at 8 μg/ml, there was an ∼3-log reduction in the number of CFU of Kanr colonies per ml versus when this strain was grown in the presence of KAN at 2,000 μg/ml-CHL at 8 μg/ml. Because of the apparent instability of the chromosomal disruption (Kanr), the mutant was subsequently tested for susceptibility in the presence of KAN at 2,000 μg/ml in Mueller-Hinton II agar (Becton Dickinson) for TX5332 to maintain the chromosomal disruption and in Mueller-Hinton II agar with KAN at 2,000 μg/ml and CHL at 8 μg/ml to maintain the chromosomal disruption and shuttle plasmid.
The MICs determined for wild-type E. faecalis OG1RF, TX5332, and TX5333 are presented in Table 2; the MICs of ampicillin, tetracycline (data not shown), norfloxacin, ciprofloxacin, ethidium bromide, and other compounds previously tested (8) showed no difference among these strains. The lsa mutant was tested on multiple occasions and showed a >40-fold decrease in the MIC of Q-D (0.75 μg/ml) and a ≥64-fold decrease in the MICs of CLI (0.12 to 0.5 μg/ml) and DAL (4 to 8 μg/ml) versus the wild-type E. faecalis parental strain (Q-D MIC, 32 μg/ml; CLI MIC, 32 to 48 μg/ml; DAL MIC, 512 μg/ml). This indicates that lsa or some downstream function is necessary for resistance to CLI and DAL in E. faecalis. Complementation of the disruption mutant with lsa (and ∼300 bp of the upstream sequence and ∼200 bp of the downstream sequence) on a shuttle plasmid resulted in restoration of the MICs of CLI (from 0.12 to 0.5 μg/ml to 32 to 48 μg/ml), Q-D (from 0.75 μg/ml to 32 μg/ml), and DAL (from 4 to 8 μg/ml to 512 μg/ml) (Table 2). This confirms the importance of lsa, as opposed to a possibly cotranscribed downstream gene. The MIC of quinupristin was 16 μg/ml for OG1RF and the two derivatives, and the ERY MIC was 1 μg/ml. The increase in the MICs of CLI (a lincosamide) and DAL (streptogramin A) and the lack of a change in the MICs of macrolides (ERY) or quinupristin (streptogramin B) correspond to the LSA phenotype (5), and the data presented here clearly indicate the involvement of lsa in this phenotype in E. faecalis.
MICs for E. faecalis OG1RF, TX5332 (lsa disruption mutant), and TX5333 (complemented mutant)
Effect of lsa on the antibiotic resistance of E. faecium.To determine if the 2-kb lsa region could function in a heterologous host, we selected strains of E. faecium to serve as recipients of the lsa gene by evaluating 68 E. faecium strains for susceptibility to CLI and for the presence of erm(B), since the latter influences CLI susceptibility (Table 3). While most of the isolates were erm(B)+, an interesting observation was the bimodal distribution of CLI MICs among the erm(B)-lacking E. faecium isolates. Of 19 erm(B)-lacking E. faecium isolates, 10 showed high CLI MICs (MIC, 4 to >256 μg/ml) and 9 showed low CLI MICs (MIC, 0.06 to 0.25 μg/ml). In contrast, none of the 32 erm(B)-lacking E. faecalis isolates showed low CLI MICs (MIC, 16 to 32 μg/ml). We chose three E. faecium strains (Tables 1 and 4) with different CLI and DAL susceptibilities [one of which is erm(B)+] and introduced the lsa gene on a shuttle plasmid. For the most susceptible strain (TX2466 [CLI MIC, 0.19 μg/ml; DAL MIC, 2 μg/ml; Q-D MIC, 0.5 μg/ml]), there was a marked increase in the MICs of CLI (12 μg/ml) and DAL (>128 μg/ml) and an increase to 3 to 6 μg/ml in the Q-D MIC. The ranges of the MICs of Q-D and the other agents were derived by testing on three or more different occasions. For the highly DAL-resistant (MIC, >128 μg/ml) and moderately CLI-resistant (MIC, 12 to 24 μg/ml) recipient TX1330, the MICs of DAL and CLI changed very little, if at all, after lsa was introduced but the Q-D MIC increased from 0.5 μg/ml to 6 to 8 μg/ml. The most pronounced increase in the Q-D MIC (from 3 to ≥32 μg/ml) was seen in strain D344 erm(B)+, which was initially highly DAL resistant (MIC, >128 μg/ml) and CLI resistant (MIC, >256 μg/ml). These data show the interspecies function of lsa and show a marked increase in the MICs of CLI and DAL and a moderate increase in the MIC of Q-D when it is introduced into E. faecium strains susceptible to these antibiotics. Bozdogan and Leclercq (5) also noted the influence of an LSA phenotype in E. faecium on Q-D MICs, where introduction of the sat(A) or vgb gene into a Q-D-susceptible E. faecium strain with the LSA phenotype conferred resistance to Q-D while, in contrast, introduction of these genes into another E. faecium strain susceptible to lincosamide, streptogramin A, and streptogramin B resulted in a 1- or 2-dilution increase in the MIC of Q-D (5).
Susceptibility of E. faecalis and E. faecium strains to CLI and streptogramins
MICs for E. faecium strains with and without lsa
Distribution of lsa among Enterococcus spp.Under high-stringency conditions, hybridization of colony lysates of 369 enterococci showed that lsa was present in all 180 E. faecalis isolates but not in 189 other enterococcal isolates (data not shown). Although most of our isolates were of human origin and animal isolates may differ, these results suggest that lsa is species specific for E. faecalis and may be an intrinsic gene of this species.
Characterization of lsa.The 2-kb region used for complementation, consisting of the 1,497-bp ORF of lsa, ∼300 bp upstream, and ∼200 bp downstream, was analyzed. The predicted Lsa protein (498 amino acids [aa]) showed similarities to known or postulated ABC proteins of other gram-positive bacteria [64% similarity to YjcA (513 aa) of Lactococcus lactis (4), 42% similarity to MsrC (493 aa) of E. faecium (44), and ca. 41% similarity to Vga(A) (523 aa) (1), Vga(B) (2), and Msr(A) (38) of Staphylococcus aureus (Fig. 1 )]. ABC transporters usually contain four single or joined components that are arranged into two homologous halves, each containing an ATP-binding domain and a membrane-spanning domain composed of several (usually six) putative α-helical transmembrane segments (9, 16, 17, 41). In the case of Msr(A), the two ATP-binding regions are fused into a single protein with internally homologous domains while in other instances, the ATP-binding regions are monomeric and likely form dimers in vivo (9, 22, 24, 41). ABC-type ATPase characteristic features, including a putative ABC signature sequence and the Walker A and B motifs, as reported in the literature for Msr(A), Vga(A), and Vga(B) (1, 2, 37), were identified in the corresponding regions of Lsa (Fig. 1). Hydropathy analysis of Lsa with the TMAP and Tmpred website programs revealed no transmembrane helix or a single strong transmembrane helix, respectively. This is similar to Msr(A), which contains no hydrophobic stretches that might be potential membrane-spanning domains, and it remains unclear for Msr(A) whether it utilizes hydrophobic proteins encoded by the genes smpA, smpB, and smpC mapping on the staphylococcal chromosome (37). Genes encoding other ABC proteins that contain two ATP-binding domains but no hydrophobic domain in gram-positive organisms include lmrC, a lincomycin resistance gene from Streptomyces lincolnensis (33); oleB, an oleandomycin resistance gene from S. antibioticus (32); srmB, a spiramycin resistance gene from S. ambofaciens (15); and a tylosin resistance gene from S. fradiae (39). A 45-aa putative peptide was also identified preceding the lsa start codon; the presence of this sequence seems to be important for the expression of drug resistance, as attempts to complement the disruption mutant with only cloned lsa failed to restore resistance to CLI, Q-D, and DAL. The presence of leader peptide sequences for msr(A), erm(A), erm(B), and erm(C) has been reported or postulated to be involved in posttranscriptional regulation of the expression of these resistance genes (6, 11, 21, 24, 34, 38, 45).
Multiple-sequence alignment of Lsa, Msr(A), Msr(C), Vga(A), Vga(B), and YjcA. ClustalW, at the Baylor College of Medicine website, was used, and the GeneDoc software was used for editing and shading. Shown are the two ATP-binding domains, consisting of Walker A and B motifs (underlined) and an SGG sequence (underlined). The bottom line shows the consensus sequence. Regions in which the six proteins are 100% identical are marked in solid black, dark gray shows regions >80% identical, and light gray shows regions with >60% identical amino acids.
In conclusion, we have shown the importance of the lsa gene of E. faecalis for the intrinsic LSA phenotype (CLI and DAL resistance) and Q-D resistance of this species. The apparent species specificity of lsa also suggests that it may be useful for the identification of E. faecalis isolates. We did not study efflux, but Lsa showed sequence similarities to known and postulated ABC transporters, including Msr(A), Vga(A), and Vga(B), suggesting that the protection of E. faecalis against CLI and DAL may be related to ATP-energized efflux of these antibiotics. We have also shown that lsa is functional in isolates of E. faecium and increases the MICs of CLI, DAL, and/or Q-D to various degrees, depending on the initial host's level of susceptibility to these agents.
ACKNOWLEDGMENTS
This work was supported, in part, by USPHS grant AI47923 to Barbara E. Murray from the Division of Microbiology and Infectious Diseases of the National Institutes of Health.
FOOTNOTES
- Received 24 October 2001.
- Returned for modification 29 January 2002.
- Accepted 21 March 2002.
- Copyright © 2002 American Society for Microbiology