INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Organisms of the genus Enterococcus, and in particular Enterococcus faecium, have become a significant cause of nosocomial infections and usually show multiple drug resistance (45). Resistance to the most commonly used antibiotics for gram-positive bacteria provides these organisms with a selective advantage in the hospital environment (40, 43). In Europe, a number of studies have documented the spread of vancomycin-resistant (Van^r) enterococci in sewage, food, animals, and human fecal samples taken from healthy volunteers (1, 5, 8, 17, 37, 38, 49, 54, 56, 57). Paradoxically, in Europe there is a low incidence of Van^r enterococci among human clinical isolates compared with the United States (45, 55, 58). The factors involved in these epidemiological differences remain unknown.

Bacteriocins are peptides or proteins produced by different bacteria that inhibit the growth of strains and species usually related to bacteriocin-producing bacteria (31). The ability to produce bacteriocins has been shown to confer an ecological advantage (48). In the genus Enterococcus, bacteriocin production has been linked to the same genetic determinant as -hemolysin synthesis (4, 7, 13, 27, 30), and its production is a pathogenic marker (30, 39). In E. faecalis, the best-characterized inhibitor substances are the pAD1-encoded bacteriocin-hemolysin (or cytolysin) (50, 52) and the peptide AS-48 (24, 26), both encoded by transferable plasmids (3). Other bacteriocins have been characterized in E. faecalis (bacteriocin 31, encoded by a conjugative plasmid [53]) or in E. faecium (enterocin A [2], enterocin I [22], enterocin P [11], enterocin L50A/L50B [12], and enterocin B [9]). A number of less-characterized bacteriocins in Enterococcus spp. have been reported (enterocin EFS2 [41], enterocin 1146 [47], and enterocin 900 [23], among others). The cotransference of bacteriocin production and the pheromone response, together with antibiotic resistance, have been described for Enterococcus strains (28, 39). The purpose of this study was to determine the relationship of bacteriocin production and vancomycin resistance in Enterococcus isolates of different species and origins.

(Part of this work was presented previously [R. del Campo et al., Abstr. 38th Intersci. Conf. Antimicrob. Agents Chemother., abstr. C95, 1998].)

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial isolates and media. This study included 218 Enterococcus isolates (93 E. faecalis and 125 E. faecium) with different vancomycin susceptibility patterns (162 vancomycin susceptible [Van^s] and 56 Van^r) and from different origins (87 human clinical samples, 78 human fecal samples, 28 sewage samples, and 25 chicken samples) (Table 1). Enterococcal isolates from human clinical samples (blood, urine, wounds, etc.) corresponded to consecutive E. faecalis and E. faecium isolates obtained from different patients from San Millán Hospital, La Rioja, Spain (1996). Enterococcal isolates from human fecal samples were recovered from consecutive samples from in- and out-patients in Hospital Clínico, Zaragoza, Spain (1996). Fecal samples were seeded on M-Enterococcus agar (Biomérieux, La Balme, France), and one colony per plate was studied and retained if it belonged to the species E. faecalis or E. faecium. Identification was carried out by the API-20 Strept System (Biomérieux) and by PCR, using specific primers for E. faecium (16) and E. faecalis (18). Enterococcal isolates from chicken samples corresponded to those recovered from chicken feces or chicken products in the La Rioja area (Spain). All Van^r isolates were characterized as having a vanA genotype by PCRs (59) and were included in this study on the basis of vancomycin resistance. A total of 33 isolates of eight different bacterial genera (Enterococcus, Listeria, Staphylococcus, Lactobacillus, Leuconostoc, Pediococcus, Escherichia, and Bacillus) were used as bacteriocin production indicators (Table 2). These isolates were maintained as frozen stocks at 80°C in skim milk (Difco, Detroit, Mich.) and propagated twice in brain heart infusion agar (Difco), with the exception of the strains of the genera Lactobacillus, Leuconostoc, and Pediococcus, which were grown in Man, Rogosa, & Sharpe (MRS) agar (Biomérieux).

Bacteriocin and -hemolysin assays. For qualitative bacteriocin detection, 50 µl of an overnight culture of the indicator isolate was added to 5 ml of molten soft tryptic soy broth (Difco) supplemented with 0.5% yeast extract and 0.7% agar (Difco), mixed, and poured onto a yeast extract-supplemented tryptic soy agar plate (Difco). A single colony of each Enterococcus isolate to be tested for bacteriocin production was transferred with a sterile toothpick to the agar plate seeded with the indicator. Plates were incubated at 37°C for 48 h and then observed for zones of inhibition around the strains. Isolates were considered bacteriocin producers (BAC⁺) when they showed activity (inhibition zone) against at least 1 of the 33 indicator isolates. This assay does not discriminate between single or multiple bacteriocin production. -hemolysin detection was performed in tryptic soy agar medium containing 5% horse blood (Biomérieux). A clear zone of -hemolysis around the isolate growth was considered a positive reaction.

Susceptibility testing and PCR determinations. The antibiotic resistance phenotype of the enterococcal isolates was determined by agar dilution following the NCCLS standard method (46). For AS-48 bacteriocin and enterocin I detection, PCRs were performed using primers and conditions as described in other studies (22, 36). The pAM401-61 plasmid containing an SphI-BglII fragment of the AS-48 genetic determinant was used as a positive control for AS-48 bacteriocin detection (kindly supplied by M. Martínez-Bueno); E. faecium 6T1a was used as a positive control for enterocin I (22).

Mating experiments. The transferability of bacteriocin production, as well as Van^r and erythromycin resistance (Ery^r) determinants was tested by conjugation using a filter method (19), with E. faecalis strain JH2-2 as recipient (rifampin and fusidic acid resistant, vancomycin and erythromycin susceptible, nonbacteriocin producer [Rif^r, Fus^r, Van^s, Ery^s, BAC]). All donor strains were Rif^s and Fus^s. Vancomycin-resistant transconjugants were first selected onto brain heart infusion agar plates supplemented with rifampin (100 µg/ml), fusidic acid (20 µg/ml), and vancomycin (20 µg/ml); bacteriocin production and Ery^r were then analyzed in the Van^r transconjugants obtained.

Pulsed-field gel electrophoresis (PFGE). Genomic DNA was prepared as previously described (44). A third part of the plug was digested with 10 U of SmaI (Amersham Life Science) for 18 h, and then an additional 10 U was added and the sample was left for another 4 h. Electrophoresis was then carried out (CHEF DR-II; Bio-Rad) in a 1.2% agarose gel with 0.5% Tris-borate-EDTA, and the following settings were applied: 5 to 35 s, 6 V/cm², and 30 h. The gel was stained with ethidium bromide for UV observation. Isolates were classified as indistinguishable, closely related, possibly related, or different according to previously published criteria for bacterial strain typing (51).

Characterization of the bacteriocin produced by E. faecium RC714. To perform a preliminary characterization of the bacteriocin activity from vanA-containing E. faecium RC714, a cell-free, filter-sterilized (0.22-µm-pore-size Millex-GV filter; Millipore SA, Molsheim, France), stationary-phase MRS culture supernatant was tested for stability to heat, pH, and proteolytic enzymes. To test for heat sensitivity, 1-ml samples were heated to 80, 90, and 100°C for 5, 10, and 20 min each. To test for pH sensitivity, 1-ml aliquots of active supernatants were adjusted to different pH values (3, 4, 5, 6, 7, 8, 9, 10, and 11) with 1 M NaOH or 0.6 M HCl. After the different treatments, the remaining bacteriocin activity was then tested by spotting a 25-µl aliquot on a plate seeded with E. faecium AR9 as the indicator strain. Plates were incubated at 37°C for 24 h and then observed for inhibition zones.

Purification of the bacteriocin produced by E. faecium RC714. All the purification steps were carried out at room temperature, and all of the chromatographic equipment and media were purchased from Pharmacia Biotech. Bacteriocin was purified from a 2-liter MRS broth culture of the vanA-containing E. faecium RC714 strain, following the method previously described for the bacteriocin plantaricin S (35). Briefly, the supernantant was ammonium sulfate precipitated, desalted through a PD10 column, consecutively applied to a cation-exchange and a hydrophobic interaction column, and finally subjected to C₂/C₁₈ reverse-phase chromatography. Fractions showing activity after the C₂/C₁₈ reverse-phase column step were pooled and subjected to a second run. Fractions from this second run showing inhibitory activity were stored at 80°C in 30% 2-propanol containing 0.1% trifluoroacetic acid until use.

SDS-PAGE. C₂/C₁₈ reverse-phase column-purified bacteriocin RC714 was analyzed by sodium dodocyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (34) with an 18.5% acrylamide resolving gel. A molecular mass marker (range, 2,512 to 16,946 Da) kit (Pharmacia Biotech) was used for size standards. After electrophoresis, a gel was silver stained (42), and a similar gel was used for detection of antimicrobial activity (6) with E. faecium AR9 as the indicator strain.

N-terminal amino acid sequencing of bacteriocin RC714. Amino acid sequencing was performed by automated Edman degradation with a Beckman LF3000 sequencer-phenylthiohydantoin amino acid analyzer (System Gold) by F. Canals, Institut de Biologia Fonamental "Vicent Villar Palasí", Barcelona University, Barcelona, Spain.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacteriocin production in Enterococcus isolates. One hundred and two out of 218 (46.8%) E. faecalis or E. faecium isolates were found to produce an antibacterial substance active against at least 1 of the 33 indicator isolates, thus being considered BAC⁺. Eighty percent of the E. faecalis isolates were BAC⁺, whereas only 21.6% of the E. faecium isolates were. The proportion of BAC⁺ isolates was significantly higher among isolates from human clinical samples (55 of 87 [63.2%]) than from those of human fecal samples (29 of 78 [37.2%]) (P = 0.00041) (Table 3). The frequencies of BAC⁺ isolates from other origins were as follows: chicken (7 of 25 [28%]) and sewage (11 of 28 [39.3%]). Among the isolates obtained from human clinical samples, Van^r Enterococcus isolates showed a higher proportion of BAC⁺ isolates (9 of 11 [81.8%]) than did Van^s isolates (46 of 76 [60.5%]). This trend to higher bacteriocin production among Van^r isolates from human clinical samples was observed for both E. faecalis and E. faecium isolates (Table 4). However, only 1 out of the 15 (6.7%) vanA-containing Enterococcus isolates from human fecal samples (13 E. faecium and 2 E. faecalis) was BAC⁺ (E. faecalis H1), while 44.4% of human fecal Van^s Enterococcus isolates were BAC⁺ (58.8% in E. faecalis and 39.1% in E. faecium).

Spectrum of activity of bacteriocinogenic isolates. The results of the assay of the inhibitory activity of BAC⁺ enterococcal isolates against the indicators are summarized in Table 5. In the case of the Van^s E. faecalis, Van^s E. faecium, and Van^r E. faecium isolates, more than one isolate was used as the indicator, and the results express the average values. In general, the most frequently inhibited indicators were Listeria monocytogenes (48%), vanA-containing E. faecium (46.6%), Van^s E. faecium (36.6%), and Van^s E. faecalis (32.1%). Pediococcus pentosaceus was the most inhibited of the lactic acid bacteria studied (29.4%). Among vanA-containing Enterococcus indicator strains, E. faecalis and E. hirae were similarly inhibited (20.6%), while vanA-containing E. faecium was inhibited by 46.6% of the BAC⁺ isolates.

Mating experiments. In all 12 vanA-containing Enterococcus isolates tested (nine E. faecalis and three E. faecium), bacteriocin production was cotransferred together with vancomycin and erythromyein resistance to the recipient E. faecalis strain JH2-2, with a mating frequency ranging from 5 × 10² to 6.6 × 10⁸. In 8 of these 12 isolates, high-level kanamycin and streptomycin resistance was also cotransferred. In all cases, vancomycin-resistant transconjugants showed the same spectrum of inhibitory activity against indicator isolates as the donors, suggesting cotransference of the same bacteriocin(s) genetic determinant(s).

Bacteriocin RC714 characterization and purification. The bacteriocin produced by the vanA-containing E. faecium RC714 strain (which was isolated from a human exudate sample) was chosen for further characterization. Bacteriocin RC714 showed inhibitory activity against all E. faecalis (eight Van^s and one vanA- and one vanB-containing isolate) and E. faecium (six Van^s and two vanA-containing isolates) strains tested as indicators, as well as against L. monocytogenes, Listeria innocua, Listeria murrayi, Listeria grayi, Lactobacillus paracasei, Lactobacillus plantarum, Leuconostoc sp., and P. pentosaceus. However, no inhibitory activity was detected against the vanA-containing E. hirae, vanC-1-containing E. gallinarum, S. epidermidis, S. aureus, S. haemolyticus, E. coli, and B. subtilis. E. faecium RC714 did not show -hemolysis, and negative PCR results for the previously reported bacteriocins AS-48 and enterocin I were also obtained. The RC714 strain consistently cotransferred vancomycin and erythromycin resistance and bacteriocin production to the E. faecalis strain JH2-2. This bacteriocin was resistant to heat treatment (100°C for 20 min) and was stable in a wide range of pH values (3 to 11). Bacteriocin RC714 was susceptible to the proteolytic activity of trypsin, -chemotrypsin, papain, alkaline protease, and proteinase K, but it was resistant to lysozyme.

View larger version (15K): [in this window] [in a new window]   FIG. 1.   C2/C18 reverse-phase chromatography of bacteriocin RC714 (second run). Numbers below the graph indicate the fraction (0.15 ml each) exhibiting bacteriocin RC714 activity.

View larger version (31K): [in this window] [in a new window]   FIG. 2.   SDS-PAGE of bacteriocin RC714 and detection of antimicrobial activity. (A) Silver-stained gel. (B) Gel fixed in 20% 2-propanol-10% acetic acid and washed in deionized water as described by Bhunia et al. (6). The gel was then placed on an MRS agar plate and overlaid with MRS soft agar containing E. faecium AR9. Lanes 1 to 5, fraction numbers 13 to 17 of C2/C18 second run-purified bacteriocin RC714 (see Fig. 1); lane 6, purified pIS  peptide (35); lane 7, purified enterocin I (22). Size standards are indicated on the left.

View larger version (11K): [in this window] [in a new window]   FIG. 3.   Comparison of bacteriocin RC714 amino acid sequence with that corresponding to bacteriocin 31 (53). The different amino acids are indicated.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Bacteriocin production has been shown to confer an ecological advantage on the producer strain (48). In this study, a higher proportion of BAC⁺ isolates was detected among E. faecalis (80.6%) than among E. faecium (21.6%) isolates. Similarly, Tomita et al. (53) found that 54% of their E. faecalis isolates were BAC⁺. Both in feces and in invasive isolates, E. faecalis was more commonly found and was at a higher proportion than E. faecium (43). In accordance with our results, this fact may be explained, at least in part, by the ecological advantage of the BAC⁺ E. faecalis strains.

Our data indicate that the proportion of BAC⁺ isolates among human clinical isolates was significantly higher (63.2%) than that among human fecal isolates (37.2%). Among Van^r isolates, bacteriocin production was found in 81.8% of the isolates from human clinical samples and in 6.7% of those from human fecal samples. In addition to the low frequency of bacteriocin production by vanA-containing Enterococcus isolates from human fecal origin, Van^s BAC⁺ enterococcal isolates of fecal origin showed a high inhibitory activity against vanA-containing E. faecium isolates (Table 5). In fact, more than half (60.3%) of all our BAC⁺ human fecal Enterococcus isolates, all of them Van^s, inhibited both clones of vanA-containing E. faecium used as indicators. Only 6.19% of these isolates inhibited the vanA-containing E. faecalis H1 strain used as indicator. Moreover, not only most vanA enterococcal isolates obtained from human clinical samples were BAC⁺, but their bacteriocin activities showed broad interspecific activity and high intraspecific activity in a higher proportion than BAC⁺ isolates from the other origins. Most BAC⁺ and vanA-containing E. faecium and E. faecalis isolates corresponded to different PFGE patterns, thus indicating that the results were not severely biased by the predominance of a particular widespread clone. These observations suggest an ecological advantage of bacteriocinogenic strains for colonization and for invasion, as previously postulated (30, 39).

L. monocytogenes was the Listeria species most inhibited by BAC⁺ enterococcal isolates of different origins. The activities of BAC⁺ enterococcal isolates against lactic acid bacteria indicate that P. pentosaceus is the most susceptible strain to this antimicrobial inhibition. The moderate activities of BAC⁺ isolates against lactic acid bacteria and the absence of activity against Bacillus or staphylococci suggest high bacteriocin specificity, preferentially mediating amensalistic interactions among different enterococcal populations in the intestinal habitat.

The bacteriocins detected in E. faecalis frequently correspond to the bacteriocin/hemolysin encoded by the plasmid pAD1 (32), which usually also confers a sex pheromone response (13, 60). The bacteriocin/hemolysin has been associated with virulence in animal models (10, 29, 33). Nine of our 32 BAC⁺ tested isolates (eight E. faecalis and one E. faecium) (28%) were -hemolytic, and all of them were Van^s and of clinical origin. None of our BAC⁺ Van^r enterococcal isolates showed this -hemolytic activity. Tomita et al. (53) detected a high proportion of BAC⁺ isolates among their E. faecalis isolates (54%), and 68% of these bacteriocin producers showed -hemolytic activity. The occurrence of -hemolysin in clinical isolates of E. faecalis varied from 17 to 60% in different studies (14, 21, 30). Similarly, Coque et al. (15) detected -hemolysin activity among E. faecalis strains of different origins (16 to 37%) but not in non-E. faecalis isolates.

This study has demonstrated the cotransference of vancomycin and erythromycin resistance with bacteriocin production (but not -hemolysin) in 12 vanA-containing enterococcal isolates, corresponding to nine different PFGE patterns of E. faecalis (nine isolates, six clones) and E. faecium (three isolates, three clones). In all cases, the selection for transconjugants was first performed for vancomycin resistance, and bacteriocin production and other antibiotic resistance determinants were then evaluated in Van^r transconjugants. Notably cotransference does not mean association between bacteriocin production and antibiotic resistance, and the eventual presence of more than one bacteriocin in the transconjugants cannot be ruled out. Nevertheless, the frequent cotransference may have ecological consequences. In 1990, Handwerger et al. (28) described the cotransference of vancomycin resistance and a pheromone response as well as -hemolytic activity in E. faecium. Cotransference of high-level aminoglycoside, erythromycin, and chloramphenicol resistance and bacteriocin/hemolysin has also been previously reported (30, 39, 40).

Sequence analysis of the bacteriocin RC714 revealed that our bacteriocin belongs to the class II bacteriocins (small, heat-stable, non-lantionine-containing peptides) (20). A bacteriocin similar to RC714 was previously described in an E. faecalis strain from Japan by Tomita et al. (53) and was named bacteriocin 31. Bacteriocin RC714 has a difference of 5 amino acids from the deduced sequence of 42 amino acids of bacteriocin 31 (Fig. 3) and originated from a clinical vanA-containing E. faecium strain. Also, our bacteriocin showed a wide range of activity against different indicator isolates of different genera (5) and species (10) of gram-positive bacteria. On the basis of the N-terminal amino acid sequences, the bacteriocin RC714 could represent a new bacteriocin (enterocin) different from bacteriocin 31.

The use of glycopeptides in humans and animals has been previously associated with the dissemination of Van^r enterococci (45). The present work suggests that other factors should be taken into account. The very frequent association of bacteriocin production and vancomycin resistance in different enterococcal clones isolated from human clinical samples suggests that the production of amensalistic substances may enhance extra-intestinal colonization. In contrast, bacteriocin production was infrequently found among our vancomycin-resistant enterococcal strains from human fecal samples. That may suggest that these nonbacteriocinogenic vancomycin-resistant fecal isolates may remain at low density in the intestinal tract. Whether these observations help explain the discrepancies in the rates of vancomycin resistance among enterococcal isolates from invasive infections versus fecal isolates in Europe and the United States remains to be evaluated. Such evaluation will require a broader sampling of vancomycin-resistant strains of different origins and/or the use of animal models. This work indicates that a complete understanding of the epidemiology of vancomycin resistance in Enterococcus will probably require the simultaneous consideration of different factors involved in the dissemination and selection of particular strains in different environmental compartments.