|Home | About | Journals | Submit | Contact Us | Français|
Enterococcus faecium has increasingly been reported as a nosocomial pathogen since the early 1990s, presumptively associated with the expansion of a human-associated Enterococcus faecium polyclonal subcluster known as clonal complex 17 (CC17) that has progressively acquired different antibiotic resistance (ampicillin and vancomycin) and virulence (espEfm, hylEfm, and fms) traits. We analyzed the presence and the location of a putative glycoside hydrolase hylEfm gene among E. faecium strains obtained from hospitalized patients (255 patients; outbreak, bacteremic, and/or disseminated isolates from 23 countries and five continents; 1986 to 2009) and from nonclinical origins (isolates obtained from healthy humans [25 isolates], poultry , swine , and the environment ; 1999 to 2007). Clonal relatedness was established by pulsed-field gel electrophoresis (PFGE) and multilocus sequence typing (MLST). Plasmid analysis included determination of content and size (S1-PFGE), transferability (filter mating), screening of Rep initiator proteins (PCR), and location of vanA, vanB, ermB, and hylEfm genes (S1/I-CeuI hybridization). Most E. faecium isolates contained large plasmids (>150 kb) and showed variable contents of van, hylEfm, or espEfm. The hylEfm gene was associated with megaplasmids (170 to 375 kb) of worldwide spread (ST16, ST17, and ST18) or locally predominant (ST192, ST203, ST280, and ST412) ampicillin-resistant CC17 clones collected in the five continents since the early 1990s. All but one hylEfm-positive isolate belonged to the CC17 polyclonal subcluster. The presence of hylEfm megaplasmids among CC17 from Europe, Australia, Asia, and Africa since at least the mid-1990s was documented. This study further demonstrates the pandemic expansion of particular CC17 clones before acquisition of vancomycin resistance and putative virulence traits and describes the presence of megaplasmids in most of the contemporary E. faecium isolates with different origins.
The number of enterococcal nosocomial infections caused by Enterococcus faecium has increasingly been reported since the early 1990s worldwide (18, 47, 55). By using multilocus sequence typing (MLST) and the eBURST algorithm, the majority of contemporary isolates recovered from the hospital environment were initially grouped within a lineage designated clonal complex 17 (CC17), which comprised different sequence types (STs) apparently related and evolved from an ST17 founder (27, 55). However, recent studies using goeBURST and ClonalFrame algorithms indicate that E. faecium CC17 clinical isolates identified to date group in a subcluster composed of different clones that have evolved independently, although they still might be genetically linked based on previous studies (33, 48, 55). Interestingly, half of the STs within this CC17 polyclonal subpopulation have also been identified from samples obtained from healthy humans, swine, poultry, and pets (55).
Although the traits that might have played a role in the recent transition of E. faecium from commensal to nosocomial pathogen have not been identified (55), it seems that the clones able to infect or persistently colonize human hosts have progressively acquired genetic elements which confer selective advantages for different genogroups (18). They include traits coding for resistance to antibiotics (ampicillin and vancomycin) or enhancing colonization and/or virulence such as the fms genes (encoding microbial surface components recognizing adhesive matrix molecules [MSCRAMM]), espEfm genes (encoding a surface protein responsible for biofilm formation), or hylEfm genes (encoding a putative glycoside hydrolase which seems to facilitate intestinal colonization and peritoneal invasion) (2, 18, 39, 40, 55).
HylEfm shows homology to proteins such as glucosaminidases or hyaluronidases, which are considered virulence factors in other Gram-positive human pathogens like Streptococcus pyogenes, Listeria monocytogenes, or Clostridium perfringens (7, 22, 39). HylEfm was initially described as a hyaluronidase but recently was annotated as a putative glycoside hydrolase after in silico analysis comparing this gene to the spy1600 gene from Streptococcus pyogenes (45). Analysis of complete genomes has shown that glycosyl hydrolases facilitate intestinal colonization in many bacterial organisms (15).
The hylEfm gene has been detected in ampicillin-resistant vancomycin-susceptible E. faecium (AREF) CC17 recovered in the United States in 1986 and has been lately identified among AREF CC17 and vancomycin-resistant E. faecium (VREF) strains obtained from hospitals all over the world (10, 18, 39, 54, 55). Other authors have recently demonstrated the location of hylEfm on large plasmids that can also carry genes conferring resistance to glycopeptides in three strains from the United States and Colombia recovered in the mid-1990s and 2005, respectively, and showed that some of these plasmids enhanced gastrointestinal colonization and increased mortality in an experimental peritonitis mouse model (2, 40).
An increase in the number of E. faecium CC17 strains carrying hylEfm in hospitals in different countries has been documented during the last decade, although we lack information about the genetic background associated with this trait (2, 10, 39, 54, 55). The presence and the location of hylEfm among a comprehensive collection of well-characterized ampicillin-susceptible E. faecium (ASEF), AREF, and VREF strains obtained from different countries over the last 2 decades (1986 to 2009) are analyzed in this study.
We studied 455 E. faecium isolates with clinical (n = 255) and nonclinical (n = 200) origins. They included 172 VREF (152 vanA and 20 vanB) strains causing documented hospital outbreaks (only one single isolate per outbreak and phenotype) or disseminated in more than one health care institution in 23 countries in Europe (Denmark, Finland, France, Germany, Greece, Hungary, Italy, Netherlands, Poland, Portugal, Serbia, Spain, and United Kingdom), North and South America (Argentina, Brazil, Canada, Chile, Paraguay, and United States), Asia (Saudi Arabia and Singapore), Africa (Tunisia), and Australia during a 24-year period (1986 to 2009). We also analyzed 52 AREF and 31 ASEF clonally unrelated isolates causing bacteremia in different patients (1995 to 2008) in a teaching hospital (1,200 beds; serving a population of 550,000) in Spain, a country where VREF rates are still under 3% (10; this study). The remaining 200 E. faecium isolates were recovered from farm animals (swine and poultry), healthy humans, and environmental sources (hospital wastewaters and rivers), and they were representative of strain collections obtained in Portugal from 1999 to 2007 (16, 34, 36; this study). Clonal relatedness among all strains containing hylEfm and almost all the isolates lacking hylEfm was established by pulsed-field gel electrophoresis (PFGE) and MLST (10, 21; http://efaecium.mlst.net).
The presence of hylEfm, espEfm, vanA, vanB, and ermB was screened by different multiplex PCR schemes as described elsewhere (14, 29, 49). The chromosomal or plasmid location of these genes was assessed by hybridization of I-CeuI-digested genomic DNA with specific hylEfm, espEfm, vanA, vanB, ermB, and 23S rRNA probes for all strains included in the study (30). Genomic DNAs from E. faecium strain RC714 (vanA and ermB), E. faecalis strain V583 (vanB), and E. faecium strain C68 (hylEfm and espEfm) were used as positive controls and the template for generating specific probes by PCR (8, 12, 37).
Transferability of acquired antibiotic resistance genes frequently associated with CC17 isolates, vancomycin, and erythromycin was addressed by filter mating at a 1:1 donor/recipient ratio, using E. faecium strains GE-1 and BM4105RF as recipients and brain heart infusion (BHI) agar supplemented with 6 mg/liter of vancomycin and 20 mg/liter of erythromycin as selective media (20, 38). The sizes of plasmids containing hylEfm were determined by using a modified protocol of the technique described by Barton et al., followed by hybridization with specific probes in wild-type strains and transconjugants selected with antibiotics (4) (see below). In order to further characterize the enterococcal hylEfm plasmids, screening of replication initiator proteins corresponding to Gram-positive plasmids was assessed by a multiplex PCR method recently described by Jensen et al. (23).
Genomic DNA from isolates, wild types, and transconjugants was extracted using a QIAamp DNA minikit (Qiagen, Hilden, Germany). PCR assays were performed in volumes of 25 μl under the following conditions: 1.5 mM MgCl2, 1× reaction buffer, 0.2 mM each of deoxynucleoside triphosphate, 1 μM each of primer, and 1.5 units of Taq DNA polymerase (GoTaq Flexi DNA polymerase; Promega, Madison, WI), with amplification programs of 10 min at 94°C, followed by 30 cycles of 30 s to 60 s at 94°C, 30 s to 60 s at 45 to 70°C, and 30 s to 2 min at 72°C, and a final extension step of 10 min at 72°C. PCR products were purified with the Wizard SV PCR purification kit (Promega, Madison, WI) and sequenced on an ABI Prism 377 automated sequencer (Applied Biosystems PE, Foster City, CA). DNA Southern transfer and hybridization were performed by standard procedures (43). Labeling and detection were carried out using the Gene Images Alkphos direct labeling system kit by following the manufacturer's instructions (Amersham GB/GE Healthcare Life Sciences UK Limited). A modification of the Barton et al. method was used (4). Briefly, genomic DNA was digested with 14 U of S1 nuclease (Takara Bio Inc., Shiga, Japan) for 15 min at 37°C, and migration of DNA fragments was carried out in a Chef-DR III device (Bio-Rad, La Jolla, CA) using the following settings: switch time of 5 to 25 s for 6 h, followed by 30 to 45 s for 18 h. PFGE was performed as described previously (24) using the following electrophoresis conditions: 5 to 30 s for 22 h, 14°C, and 6 V/cm2 (I-CeuI) and 1 to 20 s for 26 h, 14°C, and 6 V/cm2 (SmaI).
The sequences corresponding to those for hylEfm from representative isolates were assigned the following NCBI accession numbers: HM043740, HM043741, HM043742, HM043743, and HM043744.
The hylEfm gene was detected in 51 isolates (11.2%) which were frequently resistant to vancomycin (n = 34/51; 67%) and contained espEfm (n = 30/51; 59%). These hylEfm-positive strains were collected from clinical samples obtained from hospitalized patients (n = 45) and hospital sewage (n = 6).
The hylEfm gene was detected among clinical VREF strains (n = 28/172, 16%) from 13 countries in Europe (Germany, Greece, Hungary, Netherlands, Poland, Portugal, Serbia, and Spain), North America (Canada and United States), Asia (Saudi Arabia), Africa (Tunisia), and Australia from 1992 to 2009 (Table (Table1).1). They belonged mostly to different STs considered within the CC17 population (n = 27/28). The STs identified comprise predominantly ST16, ST17, and ST18 (n = 19), followed by ST80, ST125, ST192, and ST412 (n = 1 each) and ST173 and ST280 (n = 2 each). It is of interest to note that the vanA VREF ST280 isolate recovered in Portugal in 2003 showed an identical PFGE type to that of another vanB VREF clinical isolate collected in Singapore in 2004. They differed in the presence of hylEfm and the sizes of their plasmids.
The hylEfm gene was also detected in 33% of AREF CC17 isolates causing nosocomial bacteremia (n = 17/52) from 1997 to 2007 (4 ST16 isolates, 9 ST18 isolates, and 1 isolate each from ST17, ST103, ST203, and ST419) and among 6 VREF strains collected from sewage downstream of health care institutions (4 ST18 and 2 nonanalyzed isolates). As with the ST280 isolates, we also identified AREF isolates showing the same PFGE types but with variable content in hylEfm, which corresponded to ST16. The only non-CC17 isolate (ST65) carrying hylEfm caused a hospital outbreak in Greece in 1999 (42). Although in this study all human isolates containing hylEfm caused infections, some of the outbreak clones were also recovered from colonized patients during extended periods of time (data not shown). The remaining 31 clinical ASEF (3 CC17 and 28 non-CC17) and 194 VREF, AREF, and ASEF (mostly non-CC17) isolates obtained from healthy human and animal samples did not contain hylEfm.
HylEfm is a member of family 84 of the glycoside hydrolases, which is conserved among enterococci (http://www.uniprot.org/uniprot/Q8GMZ8). The sequenced hylEfm genes displayed ≥99% identity with those from E. faecium genomes (GenBank accession numbers AF544400, ZP_00604207, ZP_05663180, ZP_05671579, and EFF30024). Analysis of translated HylEfm amino acid sequences showed homology to those of proteins previously described as hyaluronidases which have more recently been classified as members of family 84 of the glycoside hydrolases with β-N-acetylglucosaminidase activity, since it is not clear whether this family does genuinely process hyaluronic acid (45; http://www.cazy.org/fam/GH84.html; http://www.uniprot.org/uniprot/Q8GMZ8; J. A. Laverde-Gomez and G. Werner, unpublished results). Family 84 of the glycoside hydrolases comprises a diverse group of enzymes found in higher eukaryotes and pathogenic or symbiotic bacteria (http://www.cazy.org/Glycoside-Hydrolases.html).
All CC17 isolates studied contained a variable number of plasmids of different sizes, but megaplasmids ranging from 150 to 450 kb were identified in most of them. The hylEfm gene was always located on plasmids ranging from 170 kb to 375 kb (Fig. (Fig.1)1) which did not contain replication proteins of the known Gram-positive plasmids described among Enterococcus, Streptococcus, Staphylococcus, Lactococcus, or Bacillus spp. to date (23). However, other modules from widely spread enterococcal plasmids such as Inc18 or pAD1 (50) were identified in some megaplasmids from AREF and VREF isolates (A. R. Freitas, A. P. Tedim, and T. M. Coque, unpublished data). Under antibiotic selection with vancomycin, hylEfm from four out of 34 VREF isolates producing HylEfm was cotransferred with genetic elements encoding resistance to glycopeptides (3 vanA plasmids of ca. 50 to 60 kb and 1 vanB Tn5382-like chromosomal element). The vanB isolate corresponded to the E. faecium strain C68, which was one of the first hylEfm-positive E. faecium isolates identified (39). The van and hylEfm genes were located on different plasmids in all isolates, while the ermB gene was often detected on hylEfm plasmids which were often transferable (data not shown). The espEfm gene was always chromosomally located.
The global spread of hylEfm associated mainly with megaplasmids of E. faecium CC17 isolates recovered in the last 2 decades is reported. Galloway-Peña et al. have recently suggested that the acquisition of ampicillin resistance, hylEfm, and espEfm by ancestral American human E. faecium isolates carrying different MSCRAMM genes might have contributed to the selection and expansion of the CC17 population in the United States since the early 1980s (18). We describe the variable content in genetic elements containing the van, hylEfm, or espEfm genes for the worldwide spreading STs, such as ST16, ST17, ST18, and also locally predominant ST192 (Germany and Korea), ST203 (Germany and Spain), ST280 (Portugal and also identified in early CC17 from the United States), or ST412 (Greece) AREF clones (this study and references herein; data obtained from the MLST website [http://efaecium.mlst.net]). Since the above-mentioned STs seem to represent different subgroup founders within CC17 (55), our results indicate the occurrence of a global spread of the CC17 polyclonal population before the acquisition of different traits associated with antibiotic resistance or pathogenicity.
The hylEfm gene in three E. faecium strains obtained from the United States and Colombia has recently been identified in large conjugative plasmids (>145 kb) containing antibiotic resistance genes and expressing increased virulence in a mouse peritonitis model (2, 40). In the present study, we document for the first time that the hylEfm gene from clonally unrelated E. faecium strains recovered from hospitals in Europe, Australia, Asia, and Africa has been consistently located on E. faecium plasmids of variable size (175 kb to 375 kb) since at least the mid-1990s, which is almost a decade after the first hylEfm-positive isolates were recovered in the United States (18). Moreover, plasmids larger than 150 kb were present in all E. faecium CC17 and non-CC17 strains with human and animal origins that we have analyzed, with the exception of the first VREF non-CC17 and VREF CC17 strains described in France in 1988 and in the United Kingdom in 1992, respectively.
The existence of large plasmids (>150 kb) among vancomycin-resistant E. faecium strains with human and animal origins recovered from the mid- to late 1990s may be inferred from different recent studies (5, 16, 19, 41, 52). Their presence in most isolates analyzed in this work suggests that they encode traits essential for the adaptation of E. faecium to particular hosts, and thus, they might be defined as megaplasmids in the classical way (31). We should mention that large plasmids had not been recognized among enterococci before the 1990s (51), although the lack of studies analyzing these genetic elements by suitable methods precludes conclusions about their recent acquisition. However, it is of interest to highlight that Teixeira et al. in 1995 documented a great variability in the phenotypic expression of sugar metabolism (mannitol, glycerol, sorbitol, and raffinose) among VREF and VSEF strains from collections of the Center for Diseases Control and Prevention in the United States, although the association of these traits with the plasmid content of the isolates was not analyzed (46). Megaplasmids encoding α-galactosidases isolated from E. faecium clones of different origins have recently been communicated (56); however, the locations of both a α-galactosidase gene(s) and hylEfm on the same plasmid was not addressed. Megaplasmids are common among Lactobacillus, Bifidobacterium, and Bacteroides and confer on these microorganisms the ability to utilize complex carbohydrates from vegetables in the diet, host-derived glycoproteins and glycoconjugates which could enhance their adaptation and persistence in the colon (44).
The origin of these megaplasmids remains undetermined since the plasmid classification based on Rep proteins was designed upon the sequences of small- or medium-size plasmids from known families of Gram-positive bacteria (23). However, their remarkable variation in the size and content of the genes encoding resistance to gentamicin, erythromycin, or vancomycin or modules originated from different plasmids which are widespread among enterococci or streptococci, suggesting that the hylEfm megaplasmids might result from the cumulative effect of lateral transfer events (2; this study). Although we did not identify Tn1546 and hylEfm genes within the same plasmid in this study, the hylEfm megaplasmids often cotransfer with that encoding vancomycin resistance, which is of concern as it could further facilitate the success of the E. faecium CC17 subcluster in hospitals.
In summary, we describe the global distribution of megaplasmids among CC17 and non-CC17 isolates which may contain a glycoside hydrolase, HylEfm. The encoding hylEfm gene seems to have spread in Europe, Africa, Asia, and Australia isolates by at least the mid-1990s. The common presence of large plasmids in human E. faecium CC17 that are able to collect and convey different traits, enhancing antibiotic resistance and/or virulence, might have contributed to the recent emergence and ramping success of E. faecium as a human-adapted, leading nosocomial pathogen. Further characterization of E. faecium large plasmids will contribute to the identification of traits responsible for the epidemicity of CC17 populations and toward a further understanding of the recent changes in the ecology and evolution of this species.
We greatly thank (in alphabetical order) José Campos (ISCIII, Spain), Maciej Chebicki (Singapore General Hospital, Singapore), Keryn Christiansen (Royal Perth Hospital, Australia), Alejandra Corso (Instituto Nacional de Enfermedades Infecciosas, Argentina), Patrice Courvalin (Institut Pasteur, France), Ana Lucia Darini (Universidade São Paulo, Brazil), Rosa del Campo (Hospital Ramón y Cajal, Madrid, Spain), Anette Hammerum (Statens Serum Institut, Copenhagen, Denmark), John Hays (Rotterdam, Netherlands), Herminia de Lencastre (The Rockefeller University, NY), Balasz Libisch (Hungary), Graeme Nimmo (Queensland Hospital Campus, Australia), Annalisa Pantosti (Instituto Superiore di Sanità, Rome, Italy), Susan Richardson and Anne Matlow (The Hospital for Sick Children, Ontario, Canada), Ewa Sadowy (National Institute of Public Health, Poland), Abbassi Mohamed Salah (National Bone Marrow Transplantation Centre, Tunis, Tunisia), Saara Salmenlinna (National Public Health Institute, Helsinki, Finland), Sylvia Valdezate (Centro Nacional de Microbiología, Madrid, Spain), Keith Weaver (University of South Dakota), Rob Willems (University Medical Center Utrecht, Netherlands), and Neil Woodford (Health Protection Agency, London, United Kingdom) for the gifts of the strains. We are grateful to Maria Victoria Francia for continuous helpful discussions about enterococcal plasmid biology.
We thank the Spanish Network for the Study of Plasmids and Extrachromosomal Elements (REDEEX) for encouraging and funding cooperation among Spanish microbiologists working on the biology of mobile genetic elements (Ministry of Science and Innovation from Spain, grant BFU 2008-0079-E/BMC). Ana R. Freitas was supported by fellowships from Fundação para a Ciência e Tecnologia from Portugal (grant SFRH/BD/24604/2005) and the European Union (grant LSHE-2007-037410). Research on enterococci was supported by grants from the European Union (LSHE-2007-037410), from the Ministry of Science and Innovation from Spain (PI 07/1441 and PS09/02381), and from Fundação para a Ciência e Tecnologia of Portugal (POCI/SAL/61385/2004 and POCI/AMB/61814/2004).
Published ahead of print on 12 April 2010.