Emergence of CC17 Enterococcus faecium: from commensal to hospital-adapted pathogen


  • Editor: Willem van Leeuwen

Correspondence: Janetta Top, Department of Medical Microbiology, University Medical Center Utrecht (UMCU), G04.614, PO box 85500, 3508 GA Utrecht, The Netherlands. Tel.: +31 88 7557627; fax: +31 30 2541770; e-mail: j.top@umcutrecht.nl


For many years, Enterococcus faecium was considered to be a commensal of the digestive tract, which only sporadically caused opportunistic infections in severely ill patients. Over the last two decades, vancomycin-resistant E. faecium (VREF) has emerged worldwide as an important cause of nosocomial infections, especially in immunocompromised patients. The global Vancomycin-resistant enterococci (VRE) epidemic was preceded by the emergence of ampicillin-resistant E. faecium (AREfm) in the United States in the early 1980s, followed by the rapid emergence of VRE in the 1990s. A similar increase of VRE may occur in countries with still low levels of VRE in hospitals (such as The Netherlands), but increasing incidence of AREfm infections. Molecular epidemiological studies of both human- and animal-derived E. faecium isolates using multilocus sequence typing revealed the existence of host-specific genogroups, including a specific genetic lineage designated CC17, associated with hospital-related isolates. These strains were characterized by ampicillin and quinolone resistance. In addition, the majority of these CC17 isolates contain over hundred hospital-clade-specific genes, including mobile elements, phage genes and plasmid sequences, hypothetical and membrane proteins and antibiotic and regulatory genes and a putative pathogenicity island including the esp gene.


Enterococci are widespread in nature and are commonly found in alimentary tracts of humans and other animals as well as in soil, water and food. In human adults, enterococci account for 1% of the intestinal microflora (Sghir et al., 2000). Enterococci are facultative anaerobic, catalase-negative Gram-positive cocci that occur singly, in pairs or as short chains. The optimum growth temperature is at 35 °C, but the growth temperature can range from 10 to 45 °C (Facklam & Collins, 1989). Furthermore, enterococci are able to survive extreme temperature and chemical disinfectants like chlorine, glutaraldehyde and alcohol, which is an important characteristic for their survival and spread in hospitals (Freeman et al., 1994; Kearns et al., 1995; Bradley & Fraise, 1996). All enterococci grow in broth containing 6.5% NaCl and hydrolyze esculin in the presence of 40% bile salts (bile–esculin medium) (Facklam & Collins, 1989). Among the enterococcal species, Enterococcus faecalis and Enterococcus faecium are the most commonly encountered species in human feces (Noble, 1978; Murray, 1990). Although for years enterococci were for years considered to be harmless inhabitants of the gut flora, they are now among the leading causes of nosocomial infections of humans. Originally, the majority of clinical infections like bacteremia, endocarditis, urinary tract and surgical wound infections were caused by E. faecalis (80–90%), while E. faecium was found much less frequently (isolated in almost 10% of the infections) (Ruoff et al., 1990; Gordon et al., 1992; Jones et al., 1995; Patterson et al., 1995; Low et al., 2001). However, the ratio E. faecalis to E. faecium infections changed in favor of E. faecium in the United States in the late 1990s (Oppenheim, 1998; Mundy et al., 2000; Treitman et al., 2005), while in Europe the first reports on increased infections due to E. faecium were published in the mid-1990s (Torell et al., 1999; Fortun et al., 2002; Simonsen et al., 2003; Thouverez & Talon, 2004; Klare et al., 2005). Other enterococcal species that occasionally cause infections in humans are Enterococcus durans, Enterococcus avium, Enterococcus casseliflavus, Enterococcus hirae, Enterococcus gallinarum, Enterococcus raffinosus and Enterococcus muntdii (Murray, 1990; Ruoff et al., 1990; Gordon et al., 1992).

Antimicrobial resistance in enterococci

Antimicrobial resistance in enterococci can be divided into two classes: intrinsic resistance and acquired resistance (Table 1). Intrinsic resistance is due to either lack of target sites for the antimicrobial drug or insufficient penetration of the drug to the intracellular target site. For example, enterococci do not possess penicillin-binding proteins (PBPs), which bind cephalosporins with high affinity (Georgopapadakou & Liu, 1980; Neu, 1987), and as a result of poor permeability of the enterococcal cell wall, aminoglycosides are unable to reach their target site, although the combination of aminoglycosides and cell-wall synthesis inhibitors, like β-lactam antimicrobials and glycopeptides, has a synergistic killing effect on enterococci (Moellering Jr & Weinberg, 1971). Enterococcus faecium strains contain the gene aac(6)-Ii encoding the enzyme AAC(6′)Ii, which eliminates the synergism between cell-wall inhibitors and aminoglycosides like kanamycin, netilmycin and tobramycin. Gentamicin, however, is not affected by this enzyme (Costa et al., 1993). More important in the emergence of resistance is the ability of enterococci to acquire resistance through either chromosomal mutations or genetic exchange of mobile elements like transposons or plasmids (Witte et al., 1999). For example, mutations in the DNA gyrase or topoisomerase genes reduce the affinity of quinolones for these genes (Tankovic et al., 1996). In E. faecalis and E. faecium, many different transposons and plasmids have been identified conferring resistance to a wide variety of antimicrobial drugs, including erythromycin, gentamicin, kanamycin, streptomycin, tetracycline and vancomycin (Weaver et al., 2002). These resistance genes are present in combinations on large composite elements or as single genes. It has been hypothesized that E. faecium plays a central role in the acquisition, conservation and transfer of antimicrobial resistance genes among bacteria (Witte, 2004).

Table 1.   Antimicrobial susceptibility of enterococci
 AntibioticSpeciesMechanism of resistance
Intrinsic resistanceβ-LactamsAll enterococciLow affinity penicillin-binding proteins (PBP)
Penicillins (low level)  
Carbepenems (moderate level)  
Cephalosporins (high level)  
Aminoglycosides (low level)All enterococciInefficient uptake
Aminoglycosides (moderate level)E. faeciumProduction of chromosomal AAC(6′)Ii enzyme
Lincosamides and streptogramins AE. faecalis, E. avium, E. gallinarum, E. casseliflavusPutative efflux
Glycopeptides (low level)E. gallinarum, E. casseliflavusProduction of d-Ala-d-Ser ending peptidoglycan precursors
Acquired resistanceAmpicillin (high level)E. faecium, E. hiraeOverproduction or alterations of PBP5
 E. faecalisβ-lactamase (rare)
Aminoglycosides (high level)E. faecalis, E. faecium, E. gallinarum, E. casseliflavusAminoglycosides modifying enzymes e.g. AAC(6′)-APH (2″)
MacrolidesMost enterococciRibosomal methylation
ChloramphenicolE. faecium, E. faecalisCAT encoding enzymes
TetracyclineE. faecium, E. faecalisModification of ribosome protein
QuinolonesE. faecium, E. faecalisAlterations in DNA gyrase and Topoisomerase IV
Glycopeptides (high level)E. faecium, E. faecalisPrecursor modification
OxazolidinonesE. faeciumMutation in 23S rRNA gene

Ampicillin resistance

Enterococci are intrinsically resistant to β-lactam antimicrobials due to the low affinity of their PBP to β-lactam agents (Table 1) (Williamson et al., 1985; Fontana et al., 1992; Kak & Chow, 2002). They possess at least five and sometimes more than nine different PBPs (Williamson et al., 1986). The level of intrinsic resistance differs among the β-lactam antimicrobials. Generally, penicillins (e.g. ampicillin) have the highest activity, carbapenems slightly lower and cephalosporins have the lowest activity (Murray, 1990; Kak & Chow, 2002). Except for a few β-lactamase-producing E. faecalis isolates identified in the US (Murray et al., 1986, 1988; Murray, 1990), high-level ampicillin resistance is mainly found in E. faecium isolates derived from clinical specimens. High-level ampicillin resistance in E. faecium is due to either alterations by mutations in PBP5 resulting in even lower affinity for ampicillin (Al Obeid et al., 1990; Ligozzi et al., 1996; Zorzi et al., 1996; Rybkine et al., 1998) or by overproduction of PBP5 (Klare et al., 1992; Fontana et al., 1994; Zorzi et al., 1996). In 2000, a novel mechanism of β-lactam resistance has been described in a laboratory mutant of E. faecium not involving PBPs (Mainardi et al., 2000). In this strain, cross-linking during cell-wall elongation occurred by an ld-transpeptidation, which bypasses the usual β-lactam-susceptible dd-transpeptidation (Mainardi et al., 2000, 2002). So far, no clinical isolates with this type of resistance have been reported.

Glycopeptide resistance

Vancomycin, as well as teicoplanin, belong to the group of glycopeptide antimicrobials. These antimicrobials bind with high affinity to the d-alanyl-d-alanine (d-Ala-d-Ala) C-terminus of peptidoglycan pentapeptide precursors and block the addition of pentapeptide precursors by transglycosylation to the nascent peptidoglycan chain, thereby preventing subsequent cross-linking catalyzed by transpeptidation (Barna & Williams, 1984; Reynolds, 1989).

At present, six types of vancomycin resistance have been described in enterococci [Table 2, adapted from Courvalin (2006)]. Of the six phenotypes, the VanA and VanB types of glycopeptide resistance are reported most frequently (Coque et al., 1996; Sahm et al., 1999; Rice, 2006). Sequencing and functional analysis of the genes encoded by vanA and vanB gene clusters revealed that glycopeptide resistance is due to enzymes that encode for (1) synthesis of low-affinity precursors, in which the C-terminal d-Ala residue is replaced by d-lactate (d-Lac) or d-serine (d-Ser), thus modifying the vancomycin-binding target, and (2) for elimination of the high-affinity precursors that are normally produced by the host, thus removing the vancomycin-binding target (Brisson-Noel et al., 1990; Arthur et al., 1993, 1996; Reynolds & Courvalin, 2005).

Table 2.   Level and type of resistance to vancomycin in enterococci*
High levelVariable levelModerate levelLow levelLow level
  • *

    Adapted from Courvalin (2006)

Type MIC (mg L−1)VanAVanB1/B2/B3VanDVanEVanGVanC1/C2/C3
Mobile elementTn1546Tn1547UnknownUnknownUnknownNot applicable
SpeciesE. faeciumE. faeciumE. faeciumE. faecalisE. faecalisE. gallinarum
E. faecalisE. faecalisE. faecalis  E. casseliflavus
E. gallinarumS. bovis   E. flavescens
E. casseliflavus     
E. avium     
E. durans     
E. mundtii     
E. raffinosus     
S. aureus     
ExpressionInducibleInducibleConstitutiveInducibleInducibleConstitutive, Inducible
LocationPlasmid, chromosomePlasmid, chromosomechromosomechromosomechromosomechromosome
Modified targetd-Ala-d-Lacd-Ala-d-Lacd-Ala-d-Lacd-Ala-d-Serd-Ala-d-Serd-Ala-d-Ser

The first clinical isolates of vancomycin-resistant enterococci (VRE, both E. faecium and E. faecalis) were detected in Europe in 1986 (Leclercq et al., 1988; Uttley et al., 1988). Since then, VRE have spread rapidly all over the world. Especially in the US, VRE prevalence rates increased from 0% in 1989 to 28.5% in 2003 (Guru, 1993, 2004). Consequently, in the early 1990s, VRE were already the second most common nosocomial pathogen in the US (Frieden et al., 1993) and became endemic in many hospitals (Murray, 2000). In Europe, VRE prevalence rates in hospitals have been increasing since the year 2000 (Schouten et al., 2000; EARSS, 2004).

In 1988, French researchers discovered that glycopeptide resistance was plasmid-mediated (Leclercq et al., 1988). A few years later, the same group identified that vancomycin resistance was located on a small mobile genetic element designated transposon Tn1546, encoding the VanA phenotype (Arthur et al., 1993). Tn1546 belongs to the Tn3 family of transposons, which do not encode conjugative functions, and therefore dissemination of the transposon can only occur after integration into transferable elements such as plasmids and conjugative transposons (Arthur et al., 1993; Weaver et al., 2002). Furthermore, the same year, a second phenotype, VanB, was identified on a different mobile element, designated transposon Tn1547 (Quintiliani Jr et al., 1993). Based on nucleotide sequence differences in the vanB ligase gene, the vanB gene cluster can be subdivided into three subtypes, vanB1-3 (Patel et al., 1998; Dahl et al., 1999). The vanB2 gene cluster, the most frequently reported subtype worldwide, has been linked to the conjugative transposons Tn1549/Tn5382 (Carias et al., 1998; Dahl et al., 2000). A detailed overview of composite elements encoding antibiotic resistance in E. faecium and E. faecalis, including the vanA and vanB gene clusters, has been published in the book ‘Drug Resistance of Enterococci’ (Werner et al., 2006). Owing to these conjugative transposons and plasmids, dissemination of vancomycin resistance is not only the result of clonal expansion of resistant strains but also of horizontal gene transfer between strains and even species. Already at that time the potential transfer of these easily movable resistance genes to more pathogenic Gram-positive bacteria like methicillin-resistant Staphylococcus aureus (MRSA) was feared. At that time, vancomycin was the critically important antimicrobial to treat patients with MRSA infections. The first high-level vancomycin-resistant S. aureus (VRSA) was identified in Michigan (US) in 2002 (Chang et al., 2003). Up till now, five additional VRSA isolates have been identified (Appelbaum, 2006). In one case, a vancomycin-resistant E. faecalis was a likely source of the vanA gene cluster (Weigel et al., 2003), while VRSA may have acquired the vanA gene cluster from an E. faecium isolate in another case (Weigel et al., 2007).

Recently, a novel mechanism of in vitro resistance to glycopeptide antimicrobials in E. faecium has been described (Cremniter et al., 2006). The E. faecium laboratory mutant strain with the novel resistance mechanism to β-lactam antibiotics (see last paragraph of ‘Ampicillin resistance’) also appeared to be resistant to glycopeptides, while the activation of the ld-transpeptidation pathway can also release inhibition of transglycosylation, leading to cross-resistance to glycopeptides and β-lactams (Cremniter et al., 2006).

Clinical epidemiology of ampicillin and VRE

In the US, the first reports on an increase of infections and outbreaks due to ampicillin-resistant E. faecium (AREfm) were published in the early 1980s (Coudron et al., 1984; Murray, 1990; Grayson et al., 1991; Jones et al., 1995). In several European countries, a similar increase of AREfm has been observed, but with a 10-year delay (Lavery et al., 1997; Dornbusch et al., 1998; Fontana et al., 1998; Suppola et al., 1999; Torell et al., 1999; Harthug et al., 2002; Top et al., 2007).

Although the first clinical VRE were detected in Europe, a remarkable difference exists in the epidemiology of VRE between Europe and the United States. In the United States, colonization of hospitalized patients with VRE rapidly increased in the 1990s, up to the current endemic levels in many hospitals. In parallel, nosocomial VRE infection rates increased as well, while colonization in healthy people appeared to be absent. In Europe, the prevalence rates in hospitals have remained much lower and only started to increase since the year 2000 (Schouten et al., 2000; EARSS, 2004). In The Netherlands, outbreaks due to vancomycin-resistant E. faecium (VREF) have been reported in three different hospitals. In all cases, intervention measurements were successful in controlling the outbreak (van der Steen et al., 2000; Timmers et al., 2002; Mascini et al., 2006). It has been suggested that the rapid increase of VRE in the United States was due to a 5–10-fold higher use of vancomycin in the United States compared with five European countries, including France, Italy, Germany, United Kingdom and The Netherlands, which have, in total, a similar number of inhabitants (Kirst et al., 1998; Bonten et al., 2001).

In contrast to the United States, where VRE is restricted to hospitals, a large community reservoir of VRE among healthy people and farm animals exists in Europe, which is most probably linked to massive use of avoparcin in animal husbandry (Van Belkum et al., 1996; Van der Auwera et al., 1996; van den Bogaard et al., 1997a, b, 2002; van den Braak et al., 1998; Stobberingh et al., 1999). Avoparcin is a glycopeptide antimicrobial, like vancomycin, and has been used as a growth promoter in the agricultural industry since the 1970s in most European countries. Because the presence of a large community reservoir of VRE was thought to pose a threat for VRE transmission into hospitals either by enterococcal strains harboring the vancomycin resistance genes or by horizontal transfer of plasmids containing Tn1546 from animal strains to human strains, the European Union banned the use of avoparcin in April 1997. Since then, the prevalence rates of VRE colonization among farm animals and human volunteers have decreased (Klare et al., 1999; van den Bogaard et al., 2000; Aarestrup et al., 2001), although several countries report persistence of VRE, especially in poultry, 3–8 years after the ban (Borgen et al., 2000a, b, 2001; Heuer et al., 2002; Manson et al., 2004; Johnsen et al., 2005; Sorum et al., 2006).

Methods to study the genetic relatedness of E. faecium

Molecular typing methods are essential to determine, in detail, the epidemiology of E. faecium and its resistance traits, and to identify outbreaks in hospitals. Furthermore, the recognized presence of E. faecium in different ecological niches created an additional need to determine its population structure and genetic evolution.

The first molecular typing methods used for enterococci were based on the analysis of plasmid profiles, including plasmid composition and restriction endonuclease analysis of specific plasmids (Luginbuhl et al., 1987; Zervos et al., 1987). In the late 1980s, a new typing method was developed based on analysis of chromosomal DNA restriction endonuclease profiles by pulsed-field gel electrophoresis (PFGE) (Chu et al., 1986), which was soon adapted for enterococci (Murray et al., 1990; Goering & Winters, 1992). Until recently, many laboratories considered PFGE to be the ‘Gold Standard’ typing method. However, in the literature, there is inconsistency about the stability of DNA banding patterns of enterococci (Bonten et al., 1998; Morrison et al., 1999; Werner et al., 2003b) and therefore PFGE is suitable for the purpose of tracing transmission of strains that are associated in time and location, but not for long-term epidemiological studies. For several bacterial species, an international database, designated PulseNet, was developed with standardized protocols, aimed at early recognition of outbreaks of foodborne diseases (Swaminathan et al., 2001, 2006), but for enterococci interlaboratory data exchange is still problematic as there is a lack of standardized conditions for electrophoresis and criteria for interpreting PFGE banding patterns (Tenover et al., 1995; Kim et al., 1999; Morrison et al., 1999; Facklam et al., 2002).

To study the genetic relatedness between epidemiologically nonrelated VREF, amplified-fragment length polymorphism analysis (AFLP) was developed, which allows analysis of polymorphisms among small restriction fragments (Willems et al., 2000). With this technique, particular E. faecium genogroups appeared to be associated with particular hosts, like pigs, calves, poultry and humans. Most importantly, though, there were genetic differences between VREF isolated from feces of nonhospitalized persons without infection (genogroup A) and isolates from hospitalized patients of fecal origin or from infected body sites like blood (genogroup C). Other studies confirmed the existence of these genogroups among vancomycin-susceptible E. faecium (VSEF) isolates originating from different sources (Borgen et al., 2002; Bruinsma et al., 2002; Jureen et al., 2003, 2004; Coque et al., 2005). Furthermore, AFLP exhibited a discriminatory power comparable to PFGE and discriminated outbreak-related isolates from other isolates (Jureen et al., 2004).

Although AFLP appeared to be a robust and fast typing method generating reproducible data within a given laboratory, this method was less suitable for data exchange between different laboratories and for studying the global epidemiology and the evolution of E. faecium. For this, a typing method is required, that generates unambiguous data suitable for the development of web-based databases. In 1998, multi locus sequence typing (MLST) was proposed for Neisseria meningitides with the afore mentioned properties (Maiden et al., 1998). MLST is based on identifying alleles from DNA sequences of internal fragments of housekeeping genes, resulting in a numeric allelic profile. Each profile is assigned a sequence type (ST). In addition, an Internet site with the possibility for data exchange was developed (http://www.mlst.net), which, currently, together with http://www.pubMLST.org, contains MLST schemes of 38 different bacterial species, including E. faecium (Homan et al., 2002) and E. faecalis (Ruiz-Garbajosa et al., 2006). MLST of 123 isolates, including VREF and VSEF originating from human (nonhospitalized, clinical and hospital outbreak) and animal sources from various countries, confirmed the genogroups as determined by AFLP, including the hospital-related genogroup C (Homan et al., 2002). MLST typing of the hospital-related isolates revealed that the outbreak isolates clustered in a sub-population designated lineage C1, which was subsequently confirmed in many studies performed world wide (Werner et al., 2003a; Bonora et al., 2004; Coque et al., 2005; Ko et al., 2005; Lee et al., 2006).

In a more recent study, MLST analysis of 411 E. faecium isolates, including VREF and VSEF from different human and animal sources and different continents, explored the evolutionary origin of epidemic isolates by discerning patterns of microevolution within the C1 lineage and determined the population structure of E. faecium (Willems et al., 2005). ST17 is the predicted founder of the C1 lineage and therefore this genetic subpopulation is renamed in clonal complex-17 (CC17), which is characterized by ampicillin and quinolone resistance.

The recognition of CC17 as a hospital-adapted E. faecium subpopulation, responsible for the majority of the hospital burden, created a need for rapid typing of E. faecium, in order to better target infection control measures in hospitals. With this purpose, a multiple-locus variable-number tandem repeat (VNTR) analysis (MLVA) typing scheme has been developed (Top et al., 2004). MLVA is based on variation in six, over the genome dispersed, VNTR loci of E. faecium. For each VNTR locus, the number of repeats is determined by PCR using primers based on the conserved flanking regions of the tandem repeats. PCR products are separated on agarose gels and the band size determines the number of repeats. The six numbers together result in the MLVA profile and each profile is assigned an MLVA type (MT). An internet site has been developed (http://www.mlva.umcutrecht.nl) to serve as a database and also for the submission of MLVA profiles to assign MTs. The MLVA scheme was validated by typing 291 E. faecium isolates both with MLVA and with MLST. The discriminatory power of both methods was comparable, but more importantly certain MLVA profiles were predictive of whether isolates will belong to the MLST-based subpopulation CC17 with a sensitivity of 97% and a specificity of 90% (Top et al., 2004).

Molecular characterization of DNA heterogeneity in the vanA gene cluster of Tn1546 in isolates from humans and animals revealed high degrees of DNA polymorphisms due to point mutations, deletions and insertions of different insertion sequences e.g. IS1216 V and IS1251, which can be used to study the epidemiology of Tn1546 (Willems et al., 1999; Schouten et al., 2001). Identical Tn1546 variants among VREF recovered from farm animals and humans were identified, which could be a result of either colonization of animal-derived VREF in humans or transfer of Tn1546 from animal VREF to human enterococcal isolates (Jensen et al., 1998; van den Braak et al., 1998; Woodford et al., 1998; Descheemaeker et al., 1999). Combining molecular typing data of isolates and the vanA transposon has improved our insights into the transmission of vancomycin resistance between different ecological niches (Jensen et al., 2003).

Virulence determinants in E. faecium

In contrast to E. faecalis, little is known about the virulence of E. faecium (Gilmore et al., 2002). Many clinical isolates of E. faecium are resistant to phagocytosis by neutrophils (Arduino et al., 1994), which might be considered to be a pathogenic property.

Other putative virulence factors are the secreted antigen SagA (Teng et al., 2003) and a surface-exposed antigen designated Acm (Nallapareddy et al., 2003). Both antigens are able to bind to human extracellular matrix proteins. In contrast to the specific collagen-binding adhesin Acm, SagA has broad-spectrum binding to fibrinogen, collagen type I, collagen type IV, fibronectin and laminin. Although the exact role of both antigens in the pathogenesis of E. faecium infections is not well understood, adherence to extracellular matrix proteins might be the first step in colonization of the host.

In Caenorhabditis elegans, E. faecium produces hydrogen peroxide at levels that cause cellular damage (Moy et al., 2004). Additional studies are necessary to investigate the relevance of hydrogen peroxide production by E. faecium in the human host.

Gelatinase is an extracellular zinc mettaloprotease, which contributes to E. faecalis virulence in some animal models and is regulated in a cell-density-dependent manner by the fsr operon (Qin et al., 2000, 2001). Recently, dissemination of gelatinase was also described in E. faecium (Lopes et al., 2006).

The identification of the E. faecium subpopulation CC17 raised the question of whether this population contained specific traits, which contribute to increased abilities in spread and/or infections among hospitalized patients. Screening of human and animal isolates for the presence of the esp gene, which has been associated with increased virulence and biofilm formation in E. faecalis (Shankar et al., 1999, 2001; Toledo-Arana et al., 2001; Tendolkar et al., 2004, 2005), revealed that in E. faecium the esp gene is restricted to hospital-derived isolates belonging to the CC17 (Baldassarri et al., 2001; Bonten et al., 2001; Willems et al., 2001; Eaton & Gasson, 2001, 2002; Coque et al., 2002; Hammerum & Jensen, 2002). Interestingly, in E. faecalis this gene is contained on a pathogenicity island (Shankar et al., 2002) and was identified among clinical and animal-derived isolates (Shankar et al., 1999; Hammerum & Jensen, 2002; Dupre et al., 2003; Creti et al., 2004). Analysis of the up- and downstream regions of the E. faecium esp revealed that as in E. faecalis, the esp gene in E. faecium is contained on a putative pathogenicity island, which is different from the E. faecalis pathogenicity island (Leavis et al., 2004). In E. faecium, Esp expression on the surface was shown to be growth condition dependent and Esp is involved in initial adherence and biofilm formation (van Wamel et al., 2007). Recently, the role of Esp in biofilm formation was confirmed by a biofilm deficient esp insertion-deletion mutant (Heikens et al., 2007), which was constructed using a temperature-sensitive vector (Nallapareddy et al., 2006).

In 2003, another putative virulence gene, hyaluronidase (hylEfm), with homology to the same gene in Streptococcus pyogenes and Streptococcus pneumoniae, was described to be enriched among clinical E. faecium isolates (Rice et al., 2003). Although the presumed function of hyaluronidase in E. faecium is still unknown, in S. pneumoniae it is suggested that hyaluronidase may contribute to the invasion of the nasopharynx.

Concluding remarks

The rapid increase of infections and outbreaks with VRE in the United States was preceded by an increase of infections and outbreaks with AREfm. In Europe, a similar increase can be observed but with a 10-year delay. Recent molecular epidemiological studies revealed that this worldwide emergence is due to a specific E. faecium subpopulation CC17, which is characterized by ampicillin and quinolone resistance. Furthermore, the majority of the CC17 isolates contain a putative pathogenicity island, including the esp gene, and another putative virulence gene hylEfm was found to be enriched among clinical isolates. All these findings suggest that E. faecium ST17, the predicted founder of CC17, has been able to adapt to the hospital environment in a multi step process, in which ST17 isolates first acquired ampicillin resistance, and by cumulative acquisition of adaptive elements, like resistance genes and putative virulence genes, ST17 gained selective advantage. This was followed by an increase in frequency facilitating further adaptive possibilities and genetic diversification, resulting in a complex of closely-related genotypes (CC17) (Willems et al., 2005). This process has been called genetic capitalism (Baquero et al., 2003). Interestingly, mixed whole-genome microarray analysis based on comparative genome hybridization of 97 E. faecium strains isolated from different epidemiological niches worldwide revealed a hospital clade, that largely overlapped with the MLST-based CC17 subpopulation (Leavis et al., 2007). Furthermore, the whole-genome analysis revealed over hundred hospital-clade specific genes, including mobile elements like insertion sequence (IS) elements, phage genes and plasmid sequences, hypothetical and membrane proteins and antibiotic resistance and regulatory genes (Leavis et al., 2007).

With the development of new molecular tools, construction of isogenic mutants of e.g. virulence genes in E. faecium will help to investigate the role of these genes in the pathogenesis of E. faecium. This knowledge may lead to the development of novel intervention strategies to prevent infections and spread of CC17 E. faecium.