|Home | About | Journals | Submit | Contact Us | Français|
Enterococcus faecalis, a common cause of endocarditis and known for its capacity to transfer antibiotic resistance to other pathogens, has recently emerged as an important, multidrug-resistant nosocomial pathogen. However, knowledge of its lineages and the potential of particular clones of this species to disseminate and cause disease is limited. Using a nine-gene multilocus sequence typing (MLST) scheme, we identified an evolving and widespread clonal complex of E. faecalis that has caused outbreaks and life-threatening infections. Moreover, this unusual clonal complex was found to contain isolates of unexpected relatedness, including the first known U.S. vancomycin-resistant enterococcus (E. faecalis strain V583), the first known penicillinase-producing (Bla+) E. faecalis isolate, and the previously described widespread clone of penicillinase producers, a trait found in <0.1% of E. faecalis isolates. All members of this clonal cluster (designated as BVE for Bla+ Vanr endocarditis) were found to contain a previously described putative pathogenicity island (PAI). Further analysis of this PAI demonstrated its dissemination worldwide, albeit with considerable variability, confirmed its association with clinical isolates, and found a common insertion site in different clonal lineages. PAI deletions, MLST, and the uncommon resistances were used to predict the evolution of the BVE clonal cluster. The finding of a virulent and highly successful clonal complex of E. faecalis with different members resistant to the primary therapies of choice, ampicillin and vancomycin, has important implications for the evolution of virulence and successful lineages and for public health monitoring and control.
Enterococcus faecalis, a natural inhabitant of the gastrointestinal tract and a known cause of infective endocarditis since ca. 1900 (14), has more recently emerged as a significant nosocomial pathogen (12). Interest in enterococci derives in part because of their prominence in multidrug-resistant nosocomial infections (15), which are difficult to control or treat, their propensity for incorporation of mobile elements (28) and their ability to transfer these resistance phenotypes to other pathogens, including the apparent transfer of vancomycin resistance from E. faecalis to methicillin-resistant Staphylococcus aureus in humans (39). The present understanding about the clonal relationships of E. faecalis isolates is limited to sporadic outbreak studies, and further knowledge about its population structure is important for understanding what makes this organism successful.
Molecular typing has shown that discrete lineages of pathogenic bacteria can arise periodically and then spread locally or globally in the presence of strong selective pressure (26). Thus, the identification of E. faecalis clones that are successful in achieving prolonged, widespread outbreaks and the unraveling of their genetic background may shed light on the question of how this opportunist adapts to clinical settings and behaves as a pathogen, causing a range of infections such as intraabdominal, genitourinary, endovascular, or meningeal infections, among others. Among typing methods for examining relatedness of bacterial genetic backgrounds, multilocus sequence typing (MLST), which is objective and less prone to human error, has gained recognition as one of the best approaches and has been used to identify pathogenic lineages of several species, including Neisseria meningitidis, Streptococcus pneumoniae, S. aureus, and Enterococcus faecium (4, 7, 10, 26), among others. Recently, we derived a four-gene MLST system using one housekeeping gene (pyrC) and three antigen-encoding genes (ace, efaA, and salA), chosen for their likely greater variation, which successfully differentiated E. faecalis at the subspecies level (23). When we subsequently used this MLST scheme to examine additional selected isolates from our 30-year collection, the results suggested an unexpected relationship among clinically important isolates disseminated in several states of the United States. We decided to further investigate these isolates using additional housekeeping genes plus the antigen-encoding genes; this combined use of different types of genes has the potential advantage of revealing both the long-term evolutionary history of the chromosome and a short-term differentiation resulting from the more variable antigen-encoding genes. After identifying a circulating E. faecalis lineage that had acquired resistances to the primary and secondary drugs of choice, ampicillin and vancomycin, we further explored the virulence gene profiles and predicted the evolution of this clonal complex based on acquired resistance genes and variations observed in the previously reported pathogenicity island (PAI) (34).
(A part of this work was presented at the 43rd Interscience Conference on Antimicrobial Agents and Chemotherapy, 2003, Chicago, Ill.)
Twenty-one E. faecalis isolates that were recovered from a broad geographic region, including previously defined β-lactamase-producing (Bla+) isolates (5, 11, 13, 17, 20, 23, 27, 33, 38), were chosen for this study based on preliminary results suggesting an unexpected relatedness of some and because of the general interest to the field of others (e.g., V583, MMH594, OG1RF, and JH2-2) (5, 8, 11, 21, 23, 28, 32, 34, 38). Relevant background and characteristics are detailed in Table Table1.1. β-Lactamase production was reconfirmed using nitrocefin disks. To assess the widespread nature of PAI (34), a total of 454 E. faecalis strains isolated over 30 years from diverse locations (United States, Thailand, China, Argentina, Chile, Spain, Canada, Belgium, United Kingdom, France, and Lebanon), including nosocomial clinical isolates, nosocomial- and community-derived fecal isolates, and animal isolates, were included. The “other clinical” group includes isolates from blood, bile, bone, catheters, cervix, cerebrospinal fluid, placenta, peritoneal fluid, sputum, and several types of wounds, among others. In this extensive collection, most of the E. faecalis strains were typed previously by pulsed-field gel electrophoresis (PFGE), and isolates with identical patterns were excluded for the analysis of PAI presence in distinct clones.
E. faecalis isolates freshly streaked from freezer vials onto brain heart infusion agar (Difco Laboratories, Detroit, Mich.) were cultured in brain heart infusion broth. Genomic DNA was extracted by the hexadecyltrimethyl ammonium bromide method as described previously (41). The ef numbers used in this study are from the V583 genome annotation (28). Internal fragments of three antigen-encoding genes (ace, encoding a collagen and laminin adhesin; efaA, encoding an endocarditis antigen; and salA, encoding a cell wall-associated antigen) and six housekeeping genes (pyrC, ef1718 coding for dihydroorotase; gki, ef2788 coding for glucokinase; gdh, ef1004 coding for glucose-6-phosphate 1-dehydrogenase; aroE, ef1561 coding for shikimate 5-dehydrogenase; xpt, ef2365 coding for xanthine phosphoribosyltransferase; and yqiL, ef1364 coding for acetyl coenzyme A acetyltransferase) were amplified using the optimized buffer B (1× buffer: 60 mM Tris-HCl [pH 8.5], 15 mM ammonium sulfate, and 2 mM MgCl2) obtained from Invitrogen (Carlsbad, Calif.). PCR was performed in volumes of 50 μl, with an initial denaturation at 94°C for 2 min followed by 30 cycles of 94°C for 30 s, 55°C for 30 s, and 72°C for 30 s to 1 min (depending on size of the amplicon) and a final extension of 72°C for 7 min. The PCR primers used for amplification and sequencing of all nine genes are listed in Table Table2.2. PCR amplicons purified using the Wizard PCR DNA Cleanup system (Promega Corporation, Madison, Wis.) were sequenced using an Applied Biosystems Prism 377 automated DNA sequencer using the Taq dye-deoxy terminator method (PE Applied Biosystems, Foster City, Calif.). Sequences were assembled using the SeqMan program of DNASTAR software (Lasergene, Madison, Wis.).
Sequence alignments for the nine gene fragments were done by the Jotun Hein method (6) using the MegAlign program of DNASTAR software. To identify nucleotide variation, sequences from the different isolates were compared to the corresponding sequences in the well-studied E. faecalis strain OG1RF (21). For each locus, every sequence with ≥1 nucleotide change was classified as a distinct allele (no weight was given to the degree of sequence divergence between alleles) and each isolate was defined by its allelic profile (a series of numbers corresponding to the alleles at the nine loci). In keeping with other studies, isolates with the same nine allelic profiles were assigned the same sequence type (ST), and isolates that shared alleles at ≥7 loci (single or double locus variants [SLVs or DLVs]) were called a clone. For linked STs differing by SLVs, the term clonal complex was used, with the implication that these are descendants of a common ancestor. Clonality was assessed using BURST, a clustering algorithm designed for use with MLST data sets of bacterial pathogens (9).
PFGE was performed with some modifications of a previously described method (19). Agarose plugs containing genomic DNA were digested with SmaI (Invitrogen), and electrophoresis was carried out using clamped homogeneous electric field (CHEF-DRII; Bio-Rad Laboratories, Richmond, Calif.), with ramped pulse times beginning with 5 s and ending with 45 s, at 200 V for 26 h. The PFGE patterns were interpreted using the criteria suggested by Tenover et al. (37), with closely and possibly related patterns being designated as belonging to a single clone. PFGE pattern names that were presented in earlier publications are used here. Southern and colony lysate hybridizations were performed under high-stringency conditions (36) with probes labeled using the RadPrime DNA labeling system (Invitrogen). Probe details and primers used for amplification are listed in Tables Tables33 and and44.
When we first applied our established four-gene MLST scheme (23) to additional diverse isolates from our collection, we noticed the apparent, but unexpected and previously unrecognized, relatedness of clinically important E. faecalis isolates of diverse origins and resistance profiles. To further investigate the suggested relatedness, we used five additional housekeeping genes (gki, gdh, aroE, xpt, and yqiL), representing altogether nine genes (loci) spread around the chromosome (28) (Fig. (Fig.11 and Table Table2).2). A total of 5,287 bases, including some results with the four genes from our previous study (23), was sequenced from each of 21 selected isolates (Table (Table1).1). Overall, we found 132 point mutations, two deletions, and 60 alleles (Fig. (Fig.2).2). The collagen adhesin gene ace, known to be expressed during serious human E. faecalis infections (24), showed greatest variability with 54 point mutations and one 69-bp in-frame deletion. Using the definitions described in Materials and Methods, 14 STs consisting of six single isolate STs, two clones (ST-2 and ST-12 are SLVs and ST-9 and ST-13 are DLVs), and one clonal complex (ST-6, ST-7, ST-14, and ST-15) were identified among the 21 isolates (Table (Table11).
The nine-gene MLST confirmed the relatedness of the previously studied (23) penicillinase-producing (β-lactamase, Bla+) isolates of ST-6 and its SLV, ST-7 (Fig. (Fig.3A3A and Table Table1).1). The earliest isolate of this clone, HH-22, the first known Bla+ isolate of E. faecalis, is a multidrug-resistant urine isolate recovered in Texas in 1981 (18); the blaZ gene of this strain is located on a plasmid that also harbors a gene for high-level gentamicin resistance (HL-Gmr) (16). Other members of this clone (which represents the majority of known Bla+ isolates) were subsequently found in five states of North America (20), including a large, prolonged outbreak in a Virginia hospital (33, 40) and a hospital outbreak that included five bloodstream infections in North Carolina (13). Surprisingly, the first known U.S. vancomycin-resistant enterococcus, E. faecalis strain V583 (32) (from Missouri), which has recently been sequenced (28), was also found to be a member of this clonal cluster (ST-14), differing from ST-6 by a single nucleotide in efaA. A Bla−, vancomycin-susceptible, HL-Gmr E. faecalis strain, MMH594, representative of an outbreak (1984 to 1987) with increased risk of death and isolated prior to V583 in a city from a nearby state (8), also belongs to ST-14. Among several endocarditis isolates tested, one (vancomycin susceptible, Bla−, and HL-Gmr) belongs to ST-15, differing from ST-14 by a single nucleotide (pyrC); this isolate was recovered 6 years after V583 from another Missouri city.
This group of isolates, clustered as shown in Fig. Fig.3A3A by BURST analysis (9), was named the “BVE” (Bla+-Vanr-endocarditis) clonal complex; all members have a maximum difference of one allele from at least one other member. Because of the overrepresentation of Bla+ ST-6 isolates among isolates tested, the central circle in Fig. Fig.3A3A denotes the predominant type and not the ancestral type (see below for predicted ancestral type). Scenarios to explain these results include the possible existence of an ancestral lineage that spread and evolved slowly over many years, until its disease-causing and resistance acquisition potentials were recognized, or that the group recently evolved into a more favorable form and then spread rapidly, achieving distribution of a subpopulation in various locations. This clonal cluster is notable because it has not only demonstrated pathogenic potential (causing serious infections in outbreaks [8, 13, 33, 40] as well as endocarditis) but also has acquired two uncommon to rare (for E. faecalis) resistances. Vancomycin resistance, seen predominantly in the species E. faecium, is an uncommon property of E. faecalis, found in ≤2% of isolates, while Bla producers are even more rare; of note, however, a vancomycin-resistant E. faecalis was the donor of the vanA genes in at least one of the recent descriptions of vancomycin-resistant methicillin-resistant S. aureus (39). The BVE clone does not appear to be predominant among E. faecalis isolates in general and was found in only 2 of over 70 other independent isolates examined, analyzed by two or more methods of PFGE, multilocus enzyme electrophoresis (38), and MLST.
Analysis of PFGE fingerprints of this BVE clonal complex also showed related PFGE patterns, differing by only a few fragments, which would categorize them as closely or possibly related (Fig. (Fig.3B)3B) using criteria for analysis of potential nosocomial outbreaks (37). Hybridization of PFGE Southern blots of chromosomal digestion fragments with probes for gdh, aroE, yqiL, gki, and xpt showed hybridization to, for the majority of fragments, the same-sized band in isolates of the BVE clonal complex, with sizes as expected from the V583 genome. One of the exceptions (a difference in size of the xpt hybridizing band of V583) is known to be due to insertion of the vancomycin resistance (vanB) element in this strain (28). Although PFGE and hybridizations confirmed the relatedness of all the MLST-defined BVE clonal complex isolates, the degrees of difference inferred by the two techniques did not always strictly overlap; this is not surprising, since PFGE and MLST have different molecular bases.
Among two other new relationships identified, one clone named HV1 (Houston Vanr #1) represents the unrecognized persistence of a clone identified earlier (1, 11, 23) in a single hospital. Another unexpected observation was that two other Bla+ isolates (ST-9 and its DLV, ST-13, differing only in antigenic genes), representative of small nosocomial outbreaks in Connecticut (27) and in Argentina (17), respectively, belong to the same clone, named ACB (Argentina-Connecticut-Bla+); this clone is unrelated to the BVE clonal complex, with variations at eight of nine loci. Among other Bla+ E. faecalis isolates known to date, one isolate from Lebanon (20, 23, 38) and an outbreak strain (20, 30) from Boston (not included in this study) are not related to the BVE or ACB groups (20). This implies that the staphylococcal blaZ gene has spread only a few times into E. faecalis and suggests that certain backgrounds may be particularly receptive to blaZ acquisition.
To further assess the genetic content of the isolates in this study, we generated a profile based upon hybridization to 14 chromosomally encoded potential virulence genes (Fig. (Fig.4A4A and Table Table3).3). Many, including six recently described immunoglobulin (Ig)-like fold-containing putative microbial surface adhesins (35), were present in all 21 isolates. However, four genes were variably present and the differences corresponded to different STs; the variably present genes (Fig. (Fig.4A)4A) were fsrB (encoding part of the Fsr two-component system of E. faecalis, which regulates the virulence genes gelE and sprE ); ef1824 (encoding a predicted adhesin  with a characteristic Ig-like fold ); and hylA and hylB (ef3023 and ef0818, respectively), each encoding a putative hyaluronidase, an enzyme implicated in pathogenesis in other organisms. Although gelE was found in all 21 isolates, the lack of gelatinase production by isolates in ST-4, ST-8, and ST-11 is related to a previously described 23.9-kb deletion (22) which includes fsrB. Based on the observation that ace B repeats (24) are separated by recer sequences, which may promote recombination and thus variable repeat numbers (2), we also tested the number of B repeats of the E. faecalis-specific collagen adhesin Ace (Fig. (Fig.4A).4A). While neither the gene profile nor ace B repeats were alone sufficiently discriminatory, the combination of these two profiles successfully distinguished the lineages from one another, reflective of the MLST and PFGE types (Fig. (Fig.4A).4A). Notably, the BVE clonal complex, including the two additional members recognized by multilocus enzyme electrophoresis or PFGE (both HL-Gmr) (38) but not tested by MLST, contained all of the potential virulence genes.
Shankar et al. (34) recently proposed an ~150-kb region as an E. faecalis PAI (ef0479 to ef0628 of V583) and indicated that there were only subtle differences in this region in two of the E. faecalis isolates described above, V583 (28, 32) and MMH594 (8). However, our recognition that these two isolates are actually members of the same ST suggests that the highly similar nature of the putative PAI of these isolates is a function of their close evolutionary relationship. To further assess whether a similar PAI was present in the other BVE clonal complex isolates and in unrelated strains, we tested the 21 isolates described above, which belong to 14 STs within nine different clonal lineages, with probes representing different genes dispersed over the entire PAI region (Fig. (Fig.4C).4C). Colony lysate hybridization results with the 18 individually labeled intragenic probes (Table (Table4)4) showed that all 18 genes (which are all present in MMH594, in which the PAI was first identified ), are also present in two unrelated STs (ST-4 and ST-11, both represented by HL-Gmr isolates from Thailand isolated in 1980) (Fig. (Fig.4C).4C). The remaining isolates representing 10 STs contain an incomplete PAI with deletions in different regions, except the ST-1 isolate (OGIRF ), which contains none of the 18 genes (Fig. (Fig.4C).4C). Hybridization of BVE clonal complex members to xylA and gls24-like gene probes confirmed the presence of these PAI genes within the same-sized PFGE fragments of the BVE clonal complex. However, within this complex, there were isolate-specific PAI deletions localized to three regions, one including ef0530 and ef0534, a second in the middle (ef0571), and the third including ef0604 and ef0609 (Fig. (Fig.4C).4C). This finding of PAI variability is not unexpected, considering the frequent occurrence of IS-like elements and integrase and recombinase genes in the PAI region of MMH594 and V583 (28, 34), including the previously described 17-kb deletion in PAI of V583 versus MMH594, both ST-14 isolates (34). Although there are many differences within the PAI of isolates of different lineages, possibly due to deletions, the finding of PAI-associated genes in eight of nine lineages containing isolates from around the world corroborates the earlier proposal that PAI is disseminated among different strains.
To investigate a possible common insertion site of apparently transferable PAI (25) in isolates from different STs, PCR was performed using one primer located outside the PAI (PAIout) and the second located within the PAI region (ef0481forward or ef0482forward [Table [Table4]);4]); two isolates lacking both ef0481 and ef0482 were not tested. Products of 3.8 kb (with ef0481forward primer) and 4.2 kb (with ef0482forward primer) (sizes were as anticipated from the V583 genome sequence) were obtained with DNA from 16 of 19 isolates (Fig. (Fig.4B),4B), suggesting the same PAI insertion site in different E. faecalis clones. Using the same sets of primers, ~1.7-kb and ~1-kb larger PCR products were obtained with DNA from the HV1 clone and from the ST-8 isolate, respectively, indicating further small insertions in this region. Thus, these results predict that at least seven of the nine lineages have a common PAI insertion site.
We also tested an additional 454 geographically and temporally diverse isolates for the presence of three selected PAI-associated genes, one close to each end and one in the middle. Hybridization results indicate that all three PAI-associated genes (esp, xylA, and gls24-like) are distributed worldwide and are enriched in infection-derived isolates (P < 0.0025 compared to community-derived isolates from human stools or animals), extending a previous report using 80 isolates of unknown clonal relatedness (34). Among the 341 clinical isolates, 17.6% were found to contain all three genes, 41.4% contained combinations of esp plus xylA or xylA plus gls24-like, and 27% of isolates lacked all three genes. The variability of the PAI region is consistent with the many deletions identified above for the well-characterized lineages (Fig. (Fig.4D).4D). The frequent finding of two or more PAI genes, together with the results for a common insertion site, suggests that these PAI genes were acquired as a unit with subsequent deletions. The less frequent occurrence of PAI-associated genes in nosocomial stool isolates compared to nosocomial clinical isolates (49.9% versus 31% for esp; 62.5% versus 39.7% for xylA) is likely because fecal isolates of hospitalized patients include a mixture of both nosocomially derived and community-derived organisms. The uncommon occurrence of even one of these three PAI-associated genes in non-human-derived isolates (9.1%) plus the high frequency of occurrence of at least one of the three genes in clinical isolates (73%) support the hypothesis that this genomic region may be helpful during some stage of human infection. The results also suggested that some deletions may be favored, or may be a clonal marker, in specific clinical settings, as exemplified by the very frequent absence of the gls24-like region in endocarditis isolates.
In a further analysis of individual isolates of the unusual BVE clonal complex, we used the PAI region variability, together with locus variations and the presence of antibiotic resistances, to predict the evolutionary pathway of this distinctive lineage (Fig. (Fig.5).5). The most complete PAI region (like that present in two Thailand strains) was found in MMH594, which lacks blaZ and vanB, and so this isolate or some predecessor was positioned in an ancestral position. A large PAI region (including cylM, the 17-kb region previously noted as deleted from V583 , araC [ef0530], and ef0534) was missing from all Bla+ (ST-6 and ST-7) isolates of this clonal complex (Fig. (Fig.4C4C and and5,5, ΔB), and long-range PCR with primers outside this deletion (ef0521reverse and ef0539forward [Table [Table4])4]) yielded the same-sized PCR fragments (~10 kb) with all Bla+ isolates of the BVE clonal complex, confirming the same deletion in these isolates. An additional region (ef0604 to ef0628) (Fig. (Fig.5,5, ΔE) of the PAI was absent in a single ST-7 isolate, HH-22, suggesting that this isolate, although it was the first Bla+ enterococcus identified, was not the ancestor of the later, more widespread ST-6 isolates. The endocarditis isolate of the BVE clonal complex showed two independent PAI deletions, one in the middle (ef0571) and the other including ef0604 and ef0609 (Fig. (Fig.4C4C and and5,5, ΔC and ΔD). In our scheme, the acquisition of HL-Gmr is inferred to have occurred after PAI acquisition, although the reverse could also be true.
In summary, we have identified and characterized a unique E. faecalis clonal complex which can cause outbreaks and life-threatening infections and has acquired HL-Gmr as well as, at different times, β-lactamase and vancomycin resistance, two unusual resistances for this species. These three resistances eliminate the activity of the cell wall-active agents most commonly used for E. faecalis infections, ampicillin and vancomycin, and of gentamicin, the aminoglycoside most often used for synergism when treating enterococcal endocarditis. Heightened awareness and the ensuing study of this unusual clonal complex may lead to improved understanding of its incidence, pathogenicity, clinical associations, and evolving patterns of antimicrobial resistances and thus may provide valuable information for control of spread and human disease caused by E. faecalis.
We acknowledge the many physicians and researchers around the world for providing isolates for our 30-year strain collection. We thank Kavindra V. Singh for his help and Karen Jacques-Palaz for her technical assistance.
This work is supported by NIH grant R37 AI47923 from the Division of Microbiology and Infectious Diseases to B. E. Murray.