Pathogenicity of Enterococci

ABSTRACT

Enterococci are unusually well adapted for survival and persistence in a variety of adverse environments, including on inanimate surfaces in the hospital environment and at sites of infection. This intrinsic ruggedness undoubtedly played a role in providing opportunities for enterococci to interact with other overtly drug-resistant microbes and acquire additional resistances on mobile elements. The rapid rise of antimicrobial resistance among hospital-adapted enterococci has rendered hospital-acquired infections a leading therapeutic challenge. With about a quarter of a genome of additional DNA conveyed by mobile elements, there are undoubtedly many more properties that have been acquired that help enterococci persist and spread in the hospital setting and cause diseases that have yet to be defined. Much remains to be learned about these ancient and rugged microbes, particularly in the area of pathogenic mechanisms involved with human diseases.

INTRODUCTION

Enterococci are hardy, Gram-positive cocci that are common residents of the gastrointestinal tracts of nearly all land animals, including humans. While they are a core member of the microbiome, they are also capable of causing a variety of severe infections, most often among antibiotic-treated hospitalized patients with perturbed intestinal microbiota. Here, we present an overview of the pathogenicity of enterococci and discuss the most prominent features of this hospital-associated pathogen.

The origin of the term “enterococcus” dates to the end of the 19th century when Thiercelin described a saprophytic Gram-positive coccus of intestinal origin capable of causing infection (1, 2). In the same year, a very detailed picture of enterococcal pathogenesis was reported by MacCallum and Hastings (3), who isolated and characterized an organism (now believed to be a cytolytic strain of Enterococcus faecalis) from a lethal case of acute endocarditis. Termed Micrococcus zymogenes by these authors, bacteria recovered from the blood and cardiac tissue of the patient and inoculated into pure culture were able to recapitulate elements of endocarditis in animals, satisfying Koch’s postulates and establishing Enterococcus as a bona fide pathogen of humans and animals. From these early descriptions, the paradigm of enterococci as a commensal opportunist was established. More than a century later, enterococci are widely recognized as leading hospital pathogens (4, 5).

Infections occurring mainly among hospitalized patients and caused by microbes of the genus Enterococcus (most notably, E. faecalis and Enterococcus faecium) include urinary tract infections, bacteremia, intra-abdominal infections, and endocarditis (6). Enterococci are now the third most common hospital-acquired pathogen, causing 14% of hospital-acquired infections in the United States between 2011 and 2014, an increase from 11% in 2007 (4, 7). In addition to hospital-acquired infections, enterococci are responsible for 5 to 20% of community-acquired endocarditis (8).

This review focuses on the mechanisms by which enterococci cause human disease, with particular attention given to advances in the field since the last edition of this text (9). Additional perspectives and information may be found in the complete volume on enterococcal biology (10).

BIOLOGY AND CHARACTERISTICS

Enterococci can be easily isolated from a wide range of hosts, including invertebrates, insects, and mammals (Fig. 1). This host range highlights the ability of enterococci to survive a diversity of innate host defenses in the gut (11). In humans, enterococci are one of the earliest colonizers of the infant gut, and as a result are core members of the intestinal microbiome (12, 13). They are also common among commensal microbiota of domesticated and wild animals and can cause infections in these animals as well (14–16). Due to their widespread occurrence among humans and animals, they are readily isolated from the environments inhabited by those hosts, often associated with plants, soil, and water (17, 18). Because of their numerous auxotrophies, enterococci persist in the environment but do not likely proliferate to high numbers there.

FIGURE 1.

Animal hosts that have been associated with enterococcal colonization. A simplified tree of life with blue shading indicating animals from which enterococci have been isolated. Corresponding geologic periods are indicated on the left. Reproduced with permission from reference 11.

Enterococci were formerly classified as group D streptococci (19) but were given genus status in 1984 (20) based on nucleic acid hybridization studies that showed a more distant relationship to the streptococci. The genus Enterococcus contains over 50 species and appears to have branched from its last common ancestor approximately 425 million years ago (21). Members of this genus are endowed with intrinsic properties that confer the ability to survive host defenses and compete in the intestinal tract and then persist and spread in the natural environment or hospital, leading to colonization of new hosts. Among these traits are the ability to grow over wide temperature and pH ranges, survive desiccation, and grow in the presence of 6.5% NaCl and 40% bile salts (17).

ANTIBIOTIC RESISTANCE

The challenge of disease caused by enterococci is compounded by their limited susceptibility to antibiotics, due to both intrinsic and acquired antibiotic resistances. Enterococci are intrinsically resistant to cephalosporins, aminoglycosides, lincosamides, and streptogramins (6, 22). Intrinsic resistance of enterococci undoubtedly positions them well to acquire additional resistances on mobile genetic elements. In the antibiotic-treated patient, enterococci accumulate to high numbers (23) and coexist in intimate association with other overtly antibiotic-resistant microbes, the precise environment to enhance the probability of contact between enterococci and other species harboring new resistances on mobile elements, leading to element acquisition.

Because enterococci are not killed by many β-lactams, new approaches to treatment were sought. The discovery that aminoglycosides were synergistic with β-lactams and could achieve cidality (24) resulted in combination therapy becoming the standard of care. However, only a few years later, aminoglycoside resistance in enterococci was reported (25) and found to be due to the acquisition of a plasmid-borne resistance factor. High-level resistance to aminoglycosides is conferred by aminoglycoside-modifying enzymes, especially a bifunctional enzyme (26), and mobile elements conferring this trait have spread throughout the enterococci (27). Further compromising combination therapy, enterococci that expressed a β-lactamase have been identified (28), but these remain surprisingly rare.

Vancomycin was introduced in the late 1950s (29) and because of its limited spectrum and intravenous dosing received limited application in hospitalized patients with β-lactam allergies and infections due to organisms with other antibiotic resistances (30). Vancomycin was found to be effective against aminoglycoside-resistant enterococci (31) and became a leading alternative therapy in the late 1970s and early 1980s (32). However, vancomycin resistance emerged in the mid-1980s—first in Europe (33, 34) and then in the United States (35). Enterococci resistant to vancomycin are now globally disseminated, with as many as 80% of E. faecium isolates being resistant to vancomycin in some hospitals (4, 36).

The rapid rise of vancomycin resistance in enterococci at the beginning of the 21st century raised alarms, because it represented the loss of an important bactericidal, last-line therapy. Daptomycin was shown to be effective against vancomycin-resistant enterococci (VRE) (37), but its value was found to be limited by the rate of daptomycin resistance development during the course of treatment (38–40). Resistant isolates had mutations in several genes that affect cell membrane composition and charge, including the cardiolipin synthase cls, the glycerophosphoryl-diester-phosphodiesterase gdpD, and the stress-sensing response component liaF (41, 42). Confounding a precise understanding of its mechanism of action, Staphylococcus aureus was observed to alter different pathways in achieving daptomycin resistance (43). Furthermore, resistance-conferring mutations derived in enterococci in vitro only partially overlap with those that arise in vivo, potentially because of differences in lipid incorporation from the environment into the enterococcal membrane (44). Lipidomic studies of daptomycin-resistant strains showed differences in lipid profiles compared to sensitive strains, particularly in phosphatidylglycerols, cardiolipins, and glycolipids (45).

Linezolid, the first oxazolidinone antibiotic, was approved by the FDA in 2000, and its introduction into clinical use was quickly followed by the emergence of linezolid-resistant VRE strains in the United States in 2001 (46) and in the United Kingdom in 2002 (47). Sequencing of the resistant isolates revealed a G2576U mutation in the 23S ribosomal RNA subunit (48). This mutation was also seen in resistant organisms isolated from a gnotobiotic mouse model of infection (49), where its occurrence was found to depend on the dose of antibiotic that reached the intestine. A linezolid-resistant isolate from Thailand subsequently was shown to have acquired the cfr methyltransferase on a plasmid, leading to methylation of position A2503 in the 23S rRNA and causing high-level resistance (50).

A third antibiotic that has found utility in treating VRE is the glycylcycline tigecycline. It has been used successfully, alone and in combination with other antibiotics, to treat VRE infection, but resistance has also emerged. A study of 73 tigecycline-nonsusceptible E. faecium and E. faecalis isolates collected from 2007 to 2015 found mutations in various efflux pumps that were identified and associated with tigecycline resistance (51). Serially passaged E. faecium in vitro in the presence of tigecycline selected for mutations in ribosomal protein subunit rpsJ, but this mutation has not yet been functionally validated (51, 52). Nonetheless, enterococci appear able to evolve resistance to antibiotics of all classes that have so far been introduced to practice.

ENVIRONMENTAL PERSISTENCE

As implied by the clonality of infection, epidemiological studies highlight person-to-person transmission of endemic strains within the hospital (53–55), implying that enterococci persist for extended periods in the hospital environment. Transmission occurs through contact with health care personnel and inanimate objects such as bedrails, nursing station keyboards, hospital drapery, and ear-probe thermometers (56). Enterococci are unusually resistant to common antiseptics and disinfectants, as well UV radiation, starvation, and desiccation (21, 57, 58). The ability to survive nutrient-poor environments as well as desiccation has led some to speculate that enterococci may enter a viable but nonculturable state as an adaptation to poor growth conditions (59–61), but details of such a mechanism have yet to be elucidated.

Studies have examined the environmental ruggedness mainly of E. faecalis (62–70), which develops tolerance to otherwise lethal levels of bile salts and detergents, such as sodium dodecyl sulfate, if first subcultured in the presence of sublethal levels (62). This phenomenon likely contributes to the ability of enterococci to survive the cleaning regimens employed in most hospitals as part of infection control programs. Regulatory systems govern the adaptive response to environmental insults (71, 72). In a study by Hancock and Perego (71), inactivation of a response regulator, designated RR06, led to an increased sensitivity to heat stress (growth at 46°C), as well as detergent stress (0.003% sodium dodecyl sulfate). Le Breton et al. (72) showed that another response regulator, RR10, negatively regulates the heat shock proteins DnaK and GroEL. Inactivation of the rr10 gene resulted in acid sensitivity but also enhanced survival at 50°C.

To identify the genetic contributors to this intrinsic ruggedness, a spectrum of enterococcal species, including clinical isolates as well as species never reported to be associated with human infection, were compared for resistances to chemical compounds as well as environmental stresses (21). All enterococci were found to be intrinsically much more resistant to most insults than other related microbes, indicating that many of the underlying traits were acquired as the genus branched from its ancestors. The two enterococcal species that are most pathogenic to humans, E. faecalis and E. faecium, showed the greatest resistance to desiccation and starvation (21). These two species also show high levels of resistance to the common hospital disinfectants chloroxylenol and chlorhexidine. The molecular mechanisms that contribute to these phenotypes were narrowed to a set of 126 genes that distinguished enterococci from ancestors (21). Additionally, the two-component system ChtRS was recently identified as in important regulator of chlorhexidine resistance in E. faecium (73).

PATHOGENIC MECHANISMS

For enterococci to cause disease, several barriers must first be overcome. An initial barrier is the ability to overcome colonization resistance provided by competing microbes, and host defenses such as gastric acid and bile, and colonize the intestinal tract. From this reservoir the bacteria can amplify in number and spread to sites vulnerable to infection. A basic prediction from such a model is that the probability of infection should be a function of the intestinal burden of bacteria in the gut reservoir—the more bacteria, the greater the probability of contamination of a potential infection site in numbers large enough to overcome host defenses. Indeed, colonization of the gastrointestinal tract has been shown to be directly associated with risk of infection (23, 74). Infection occurs when enterococci overwhelm host defenses, when they replicate at rates that exceed clearance, and when pathologic changes result through direct toxin activity, or infection occurs indirectly by bystander damage from the inflammatory response (75, 76).

HUMAN COLONIZATION

Enterococci are core but numerically minor members of the human microbiome (77). The occurrence of two species, E. faecalis and E. faecium, in similar abundance, suggests that neither evolved in humans but, rather, likely entered the primate lineage early and from organisms consumed from lower in the food chain. Both E. faecalis and E. faecium are found in relative abundance in human feces (10⁵ to 10⁷ organisms per gram) (78) but typically represent less than 1% of the total microbial population (17). “Colonization resistance” (79) is a term applied to the challenge an organism faces in colonizing a stable, established community. It is a main mechanism of innate defense of the gastrointestinal tract. Colonization and proliferation of hospital-adapted lineages of enterococci are usually associated with antibiotic-induced disruption of the community structure (80). Therefore, effectively managing the human microbiome in health and disease represents a theoretically promising strategy for preventing hospital infection.

A primary barrier to invading the gut consortium and colonization by orally acquired microbes is gastric pH, which is inhospitable for most microorganisms, including enterococci. Compromise of this barrier enhances the oral acquisition of enterococci from a contaminated environment. In the intensive care unit setting in particular, patients are placed on H₂-receptor antagonists as prophylaxis for treating stress ulcers; the consequence of this action is a pH increase from pH 2 to pH 3.5 to 5.3, depending on the antagonist employed (81). In a study by Başaran et al. (82), H₂-receptor antagonists were shown to promote colonization of the small bowel, as well as translocation of enteric bacteria from the intestinal tract to extraintestinal tissues.

Several studies have examined the ability of enterococci to tolerate acidic pH (64, 83). Flahaut et al. (64) demonstrated that exposure of E. faecalis to a sublethal pH level (pH 4.8) for 15 to 30 minutes protected the organism from a normally lethal challenge at pH 3.2, suggesting an adaptive response to changing acid stress. Suzuki and colleagues (83) showed that an E. faecalis mutant defective in F₁-F₀ H⁺-ATPase activity was unable to grow at a pH of <6. The H⁺-ATPase is used to regulate the cytoplasmic pH of E. faecalis by proton extrusion, and this enzyme has been shown to be activated at low pH (83). Teng et al. (84) identified a two-component regulatory system, termed EtaRS, which appears to be involved in acid stress response and virulence. Inactivation of the response regulator, EtaR, results in increased acid sensitivity and decreased virulence in a murine peritonitis model.

Colonization resistance is also imposed by the stability of a mature complex community of gut microbes. The use of broad-spectrum antibiotics, many of which possess little or no antienterococcal activity, destabilizes this community by removing susceptible members (85–87), opening opportunities for invasion by new organisms. In a study designed to examine the persistence and density of colonization by vancomycin-resistant E. faecium in a murine model, Donskey et al. (88) showed that suppression of the anaerobic microflora by the use of antimicrobials that target this microbial population resulted in increased colonization density and prolonged VRE persistence. This observation highlights the importance of the anaerobic flora in suppressing enterococcal growth within the intestinal microenvironment. Several other studies have also examined factors that lead to overgrowth of enterococci, especially VRE, in the gastrointestinal tract of hospitalized patients. Treatment with broad-spectrum antibiotics, especially cephalosporins and metronidazole, to which enterococci are intrinsically resistant, leads to clearance of commensals and overgrowth of enterococci (80, 89). In extreme cases, hospitalized patients treated with these antibiotics develop a near monoculture of VRE in their intestine and are at high risk for enterococcal bacteremia and other enterococcal infections (80). Factors that limit the outgrowth of E. faecium in the antibiotic-perturbed gut include reduced expression of the host innate immune factor RegIIIγ, which results indirectly from the loss of Gram-negative members of the gut microflora (90). RegIIIγ production relies on signaling through Toll-like receptor-5 and interleukin-22 (IL-22), and supplementation of Toll-like receptor agonists has been shown to restore RegIIIγ expression and suppress E. faecium growth (90). In mice, the IL-22 signaling pathway and, in particular, the IL-22 receptor has been shown to be required for E. faecalis colonization resistance. Il22ra1^–/– mice were observed to be particularly susceptible to E. faecalis overgrowth and subsequent systemic dissemination (91). This restriction of E. faecalis outgrowth is controlled in part by fucosylated glycans, which are downregulated in Il22ra1^–/– mice. Oral supplementation of fucosylated glycans was shown to promote diversity of the gut microbiota and restore the growth restriction of E. faecalis (91).

In addition to antibiotic and host immune factors, hospital-adapted strains of enterococci are endowed with properties that allow them to colonize microniches that are likely less suitable for colonization by commensal strains. For example, some hospital-associated strains of E. faecalis contain an additional bile salt hydrolase (92). E. faecalis also produces extracellular superoxide (93), leading to DNA damage in colonic epithelial cells (94) and chromosomal instability (95), potentially relating to local inflammation, changes in the nutritional properties of the mucosa, and development of colorectal cancer. Enterococci also have been associated with inflammatory bowel disease, Crohn’s disease and ulcerative colitis. Enterococcal abundance, particularly of E. faecium, is higher in fecal samples from Crohn’s disease patients compared to healthy controls (96, 97). Increased E. faecalis abundance and anti-E. faecalis antibodies have also been observed in ulcerative colitis patients (98). Germ-free IL-10 knockout mice colonized with E. faecalis develop chronic inflammatory bowel disease-like symptoms (99), highlighting the bacteria’s ability to perpetuate gut inflammation. A subset of virulence factors in E. faecalis, including the metalloprotease GelE, enterococcal polysaccharide antigen Epa, and cell surface-associated lipoprotein Lgt, were shown to influence induction of inflammatory bowel disease in these mice (Table 1) (99).

TABLE 1.

Enterococcal virulence factors

Virulence factor	Description	Reference(s)
Cytolysina	Hemolysin/bacteriocin, lyses a broad range of eukaryotic and Gram-positive cells	128, 126, 129, 218
Gelatinasea	Secreted zinc metalloprotease	216, 221, 222
Serine proteasea	Secreted serine protease	216, 221, 222
Hyaluronidaseb	Degrades hyaluronic acid	236
Espa,b	Cell wall anchored protein, enhances biofilm formation and colonization of bladder epithelium	92, 237
Aggregation substancea,b	Cell wall anchored protein, involved in conjugation and adhesion to eukaryotic cells	128, 132, 214, 267
Acea,b	Cell wall anchored protein, collagen binding protein	135, 136, 138, 139
Epaa,b	Enterococcal polysaccharide antigen, anti-phagocytic cell wall polysaccharide	200
Cpsa,b	Capsular polysaccharide, antiphagocytic cell wall polysaccharide	194, 196
LTAa,b	Lipoteichoic acid, enterococcal group antigen, binding substance for conjugation	214, 268
Toxic metabolitesa,b	Reactive oxygen species: extracellular superoxide, hydrogen peroxide	94, 239

^a Virulence factors identified in E. faecalis.

^b Virulence factors identified in E. faecium.

Enterococci must compete with other microbes in the gut for space, binding sites, and nutrients. Enterococci have long been known to be prolific producers of bacteriocins, many of which are encoded by mobile elements (100, 101). Plasmid pPD1, harbored by some E. faecalis strains, encodes a bacteriocin that inhibits other E. faecalis strains, eventually leading under some conditions to strain replacement in vivo (102). Importantly, the bacteriocin-expressing strain was able to reduce the intestinal load of VRE in a cocolonization study, highlighting the potential therapeutic value of such agents.

The apparently hasty accretion of mobile elements by enterococci affects their ability to compete in surprising ways. It was recently shown that commensal E. faecalis is able to induce self-killing of the vancomycin-resistant strain V583, through the production of the pheromone peptide cOB1 (103). This pheromone induces a cascade of events that involves expression of conjugation-associated functions on a pheromone-responsive plasmid resident within V583, which then engage in a lethal cross talk with a gene or genes in another mobile genetic element in the chromosome of V583, inducing death of V583 by an as yet unknown mechanism.

The ability to obtain nutrients within the competitive environment of the gut is an important aspect of enterococcal colonization efficiency (104). Within E. faecium, the two main clades, A and B, differ in the availability of carbohydrate utilization pathways. Strains in clade A1, which includes most infection-associated strains, express more transporters and metabolic pathways for the utilization of carbohydrates derived from the gut mucosa, as opposed to the diet, such as mucins and epithelial cell glycosidic surface decorations (104). Inactivation of a phosphotransferase system related to this ability reduced gut colonization in an antibiotic-treated mouse model (104, 105). Clade B strains do not possess this pathway (104), highlighting ecological differences between pathogenic and commensal strains. Meanwhile, clade B commensal E. faecium strains outcompete clade A strains in the murine intestinal tract in the absence of antibiotics (106).

Antagonisms within the gut community also influence colonization. Among commensals, Barnesiella species have been observed to be antagonistic to enterococcal colonization (107). Colonization resistance to VRE could also be artificially effected with a four-species cocktail containing Bacteroides sartorii, Parabacteroides distasonis, Clostridium bolteae, and Blautia producta (108). To colonize and establish residence within the gut, then, multidrug-resistant strains of enterococci must successfully navigate the hazards of innate host defenses and enter a destabilized consortium lacking overt antagonists.

TRANSLOCATION FROM THE GUT INTO THE BLOODSTREAM

Enterococal overgrowth in the colon increases the chance, by simple numeric probability, of dissemination into the bloodstream and contamination of other body sites in numbers large enough to establish infection (Fig. 2). Mechanisms that contribute to enterococcal translocation are still being discovered. In some cases, enterococci may be phagocytosed by intestinal epithelial cells, dendritic cells, or other tissue-resident leukocytes and transported across the intestinal wall to the underlying lymph system. Failure to kill the phagocytosed organism could then lead to abscess formation in reticuloendothelial organs and systemic spread (109). Gentry-Weeks et al. (110) showed that E. faecalis possesses the ability to survive for up to 72 hours in peritoneal macrophages. Wells et al. (111, 112) induced intestinal E. faecalis overgrowth by antibiotic treatment and observed organisms adhering to epithelial surfaces of the ileum, cecum, and colon. They also showed that enterococci possess the ability to translocate from the intestinal lumen into the mesenteric lymph nodes, liver, and spleen (111, 112).

FIGURE 2.

Routes of dissemination of enterococci from the intestinal reservoir of a hospitalized patient. Bacteria from the intestine can seed infections throughout the body and contaminate surfaces, leading to patient-to-patient spread. Reproduced with permission from reference 122.

Bacteria also diffuse through the gut barrier at low levels, and enterococci may gain access to the bloodstream that way. Inert particles the size of bacteria, and without any special properties, penetrate the gut mucosal barrier (113); special invasive properties do not appear to be essential for at least some level of translocation into the bloodstream. In a healthy, immune-competent host, innate defenses are well equipped to eliminate such invaders. So long as the microbes leaking into the bloodstream and occurring in the lymphatics are easily killed, such as the predominating anaerobes in the healthy gut community, there is little ill effect. However, because enterococci persist in phagocytic cells, their extraintestinal accumulation may more readily lead to infection, especially in an immunocompromised or disease-stressed patient.

A variety of enterococcal factors have been examined for their role in colonization and translocation out of the gastrointestinal tract. Olmsted et al. (114) found a plasmid-encoded surface protein, aggregation substance (AS), to promote internalization of E. faecalis by cultured intestinal epithelial cells (HT-29). To further dissect the functional domains of AS involved in internalization by HT-29 cells, Waters et al. (115, 116) examined the various domains of AS to determine their roles in this process. Surprisingly, the aggregation domain, and not the N-terminal RGD motif of AS, was found to be important for internalization by HT-29 cells (116). An additional finding from this study was the apparent requirement for a cofactor (likely lipoteichoic acid), because expression of AS in the binding substance mutant INY3000 failed to efficiently internalize. In a subsequent study it was shown that the amino-terminal aggregation domain of the protein could bind efficiently to purified lipoteichoic acid (LTA) as well as LTA purified from INY3000, suggesting that another cell component other than LTA was responsible for the cooperative internalization of E. faecalis by HT-29 cells (115).

BACTEREMIA

Enterococcal bloodstream infections (BSIs) are associated with a high level of mortality. In a recent large survey study of Canadian hospitals (117), the incidence of enterococcal BSI of hospitalized patients was 6.9 per 100,000, with most cases due to E. faecalis (4.5 per 100,000). The incidence of enterococcal BSI increased between 2008 and 2014 (9 to 14 per 100,000) in Switzerland (118), suggesting that enterococcal BSIs are on the rise. Overall, mortality from nosocomial enterococcal BSI is quite high, ranging from 25 to 50% (119, 120).

Most cases of enterococcal BSI are thought to result from translocation of enterococci from the gut into the bloodstream. Other routes of infection include along intravenous lines, endocarditis, urinary tract infections, and abscesses (109, 121, 122). The risk factors for mortality associated with enterococcal bacteremia include severity of illness (based on APACHE II scores), patient age, and use of third-generation cephalosporins or metronidazole (123). Huycke et al. (54) found that patients infected with hemolytic, gentamicin-resistant E. faecalis strains had a 5-fold increased risk for death within three weeks compared to patients with nonhemolytic, gentamicin-susceptible strains. Moreover, mode of treatment was not associated with outcome, discounting the direct contribution of aminoglycoside resistance. In another study, Caballero-Granado et al. (124) analyzed the clinical outcome, including mortality, for bacteremia caused by Enterococcus spp. with and without high-level gentamicin resistance. Mortality associated with high-level gentamicin resistance (29%) was not significantly different from gentamicin-susceptible strains (28%). Taken together, these two studies suggest that high-level aminoglycoside resistance itself does not explain clinical outcome and that the presence of other factors, such as the E. faecalis cytolysin (hemolysin), appears to contribute to enhanced lethality. In support of the latter finding, the E. faecalis cytolysin has been shown to result in acute toxicity in a number of animal models. Cytolysin significantly lowers the 50% lethal dose of the infecting strain for mice (125–127). As will be discussed later, the cytolysin also contributes to the acute toxicity of lupine endocarditis and endophthalmitis models (128, 129) and contributes to nematode killing in Caenorhabditis elegans (130).

URINARY TRACT INFECTION

In a nosocomial population surveyed between 2011 and 2014, enterococci were the most commonly isolated Gram-positive bacteria from catheter-associated urinary tract infections, with over 20,000 cases reported to the CDC National Healthcare Safety Network between 2011 and 2014 (4). Of these cases, just over 50% were caused by E. faecalis, followed by “other Enterococcus species” (∼30%) and E. faecium (∼20%). Most worrisome was the fact that ∼85% of E. faecium isolates were vancomycin resistant, with the fraction of resistant isolates increasing each year (4).

A few studies have been aimed at understanding the interaction of enterococci with uroepithelial tissue (131, 132). Kreft et al. (132) showed a potential role for the plasmid-encoded AS in the adhesion of enterococci to renal epithelial cells. E. faecalis harboring the pheromone-responsive plasmid pAD1, or various isogenic derivatives, were better able to bind to the LLC-PK cultured pig renal tubular cell line than plasmid-free bacteria. Their findings also showed that a synthetic peptide containing the fibronectin motif, Arg-Gly-Asp-Ser, could inhibit binding. This structural motif mediates the interaction between fibronectin and eukaryotic surface receptors of the integrin family (133). Guzmàn and coworkers (131) analyzed the ability of strains of E. faecalis isolated from either urinary tract infections or endocarditis to adhere to urinary tract epithelial cells and to the Girardi heart cell line. Urinary tract infection isolates adhered to urinary tract epithelial cells in vitro, whereas strains from endocarditis adhered efficiently to the Girardi heart cell line, suggesting that environmental adaptations can facilitate interactions with host tissues. Shankar et al. (134) showed that the E. faecalis Esp surface protein colocalizes in the bladder, but not the kidneys, in an ascending urinary tract infection model, suggesting that the protein has tissue specificity for bladder epithelium. Tomita and Ike (135) showed that highly adherent enterococcal urinary tract infection strains recognize the extracellular matrix proteins fibronectin, laminin, and collagen types I, II, IV, and V. One of these strains, AS14, was subjected to transposon mutagenesis with Tn916, and mutants with altered collagen IV and laminin binding were identified. Out of 14 single transposon insertion mutants, 13 mapped to the Ace protein (136–140). The level of adherence of mutants to collagen IV and laminin was 1 to 2 orders of magnitude lower than the wild-type strain, but the level of adherence to fibronectin remained the same as that of the wild-type strain. The collagen binding proteins Ace (E. faecalis) and Acm (E. faecium) were originally identified because of their homology to the amino terminal end of Cna, the collagen binding protein from S. aureus (138, 139). The ability of most E. faecalis Ace-positive strains to adhere to collagen and laminin was highly dependent on growth temperature, with binding occurring at 46°C, but not at 37°C (136). No plausible explanation existed for this unusual binding phenotype until observations by Tomita and Ike (135) showed that gelatinase-positive E. faecalis strains adhered poorly to collagen and laminin due to the ability of the protease to cleave these proteins. Shiono and Ike (141) showed that some E. faecalis clinical isolates adhere efficiently to human bladder carcinoma cells, as well as to human bladder epithelial cells, and that these interactions could be inhibited by pretreating the bacteria with fibronectin or trypsin.

More recently, several studies have identified a pilus in E. faecalis (140, 142–145) and have shown that Ebp pili are important for adhesion to the urinary epithelium and urinary catheters through the binding of fibrinogen released into the bladder during catheterization (146–148). In a mouse model of catheter-associated urinary tract infection, E. faecalis infection could be blocked by the administration of antibodies raised against the tip subunit of the Ebp pili, EbpA (149). It is hoped that these antibodies could potentially be used to prevent infection in catheterized patients without the use of antibiotics.

ENDOCARDITIS

Of the infections caused by enterococci, infective endocarditis (IE) is one of the most therapeutically challenging (8, 150). Enterococci account for between 10 and 20% of all cases of IE (151, 152), with E. faecalis causing a majority of these. In IE, bacterial colonization of the endocardium leads to vegetations within the tissue and a biofilm-like matrix surrounding the bacteria, which is difficult to penetrate either by the immune system or by antibiotics (153–155).

Several virulence factors have been identified as playing a major role in the pathogenesis of enterococcal IE, with adhesion to host tissues by cell surface adhesins among the best characterized. In E. faecalis, AS represents a class of large, pheromone-inducible surface proteins involved in mediating cell-cell adhesion as the first step in conjugal plasmid transfer (156). These proteins also appear to be key players in both attachment to heart tissue and the development of biofilm during IE. Three highly related AS proteins have been well studied in this regard, including Asa1, Asc10, and Asp (155, 157, 158). All three proteins appear to aid in cell-cell adhesion and contribute to the exopolymeric matrix that accumulates within the biofilm.

The EfaA antigen is another adhesin that was originally identified as being expressed in serum but not broth culture (159), and database homology searches revealed extensive sequence similarity with several streptococcal adhesins. Shepard and Gilmore (160) showed that of all the suspected E. faecalis virulence factors, EfaA resulted in the most dramatic increase in expression when serum-grown and broth-grown E. faecalis were compared. It is now known that EfaA forms part of a putative ABC-type transporter specific for manganese, along with EfaB and EfaC. Because manganese is tightly sequestered within the mammalian host, it is not unexpected that E. faecalis grown in serum would require the induction of a high-affinity manganese transporter, EfaABC. The induction of the efaCBA operon occurs at relatively low manganese levels, because under high manganese levels the product of the efaR gene, a manganese-responsive transcriptional regulator, represses transcription from the efaCBA operon (159). More recently, EfaR inactivation has also been shown to impair the ability of E. faecalis to form biofilms, survive inside macrophages, and tolerate oxidative stress (161).

Pili encoded by the Ebp gene cluster in E. faecalis have also been shown to be important for both adherence and biofilm formation in IE (162). High titers of antibodies against the EbpABC proteins in patients with enterococcal IE have been demonstrated, indicating that they are expressed at high levels within the host during infection (162). Other important virulence factors in E. faecalis IE include the metalloprotease GelE (163), collagen-binding adhesin Ace (164), fibronectin-binding protein EfbA (165), cytolysin (128), and phosphotransferase system permease BepA (Table 1) (166). A review of the many virulence factors that play a role in E. faecalis endocarditis was recently published (167).

In E. faecium, less is known regarding specific virulence factors and the roles they play in IE. However, the enterococcal surface protein Esp and the collagen-binding protein Acm have been shown to be important for the pathogenesis of E. faecium IE in a rat model of infection (168, 169). In infected animals, E. faecium strains that lack either Esp or Acm were recovered at lower numbers than their wild-type counterparts. Antibodies against both proteins have also been detected in patients diagnosed with E. faecium IE, and pretreatment of bacteria with anti-Acm antibodies derived from an infected patient was shown to reduce the ability of the bacteria to bind to collagen (162, 168). It seems that targeting the adhesive and biofilm development properties of enterococci, in addition to treatment with antibiotics, may be a promising method of treating enterococcal IE in the clinic.

SURGICAL SITE INFECTIONS

Enterococci are common causes of surgical site infection at all anatomical sites (170), including the eye (129, 171). The eye affords a unique opportunity to visualize the evolution of such an infection using readily available equipment. Small numbers of organisms injected into the vitreous of a rabbit (129) or mouse (172), can be directly studied using an ophthalmoscope or slit lamp, and declining organ function can be measured quantitatively by electroretinography (173). Enterococci rapidly proliferate at this site, with a 10-μl injection of enterococci suspended in phosphate-buffered saline remaining as a light-refracting bead. After a few hours, the small bead acquires a coat of fibrin. About 12 hours later, the microvessels in the retina begin to dilate and subsequently turn white as they become coated with adherent neutrophils. After about 24 hours, neutrophils can be seen streaming from vessels in the optic nerve head to the infection nidus. These measurement parameters allow various steps in the pathogenesis of infection to be dissected and the roles of various traits expressed by the bacterium to be defined. In this model, the enterococcal cytolysin makes a profound contribution to organ destruction, which is not salvaged by otherwise effective antimicrobial or anti-inflammatory therapy (173). In contrast, identical infections initiated with isogenic cytolysin knockout strains resolved completely with antimicrobial and anti-inflammatory therapy, with little loss of organ function. Using this model, GelE and SprE proteases, both controlled by the Fsr quorum-regulated system, were observed to make measurable contributions to virulence in the rabbit model of endophthalmitis (174).

ROOT CANAL FAILURE

Due in part to its resistance to many disinfectants and antiseptics, E. faecalis is an important cause of root canal failure (175, 176). This is particularly true in cases of chronic and persistent periradicular lesions leading to root canal failure, where E. faecalis was isolated in 24 to 77% of cases (176). Genetic analysis of E. faecalis strains isolated from infected root canals suggests that these strains, while not clonal, are highly related and may share additional traits that lead to efficient colonization of the oral cavity (177, 178). Calcium hydroxide, a common disinfectant used within the root canal, has been shown to be ineffective at killing E. faecalis on its own (179), as have other disinfectants such as EDTA and citric acid (180, 181). It appears that the bacteria can resist starvation within the root canal for long periods of time and are then able to be resuscitated by serum that enters the canal from the alveolar bone and periodontal ligament (59). Provided with an adequate food supply, E. faecalis may utilize proteases such as SprE and GelE, and the collagen binding protein Ace, to bind to dentin and form biofilms (182). Once the infection is established, the most efficient treatments include prolonged washing of the canal with sodium hypochlorite or chlorhexidine solution (176). Using chlorhexidine or sodium hypochlorite washes during the initial procedure are thought to be the best options for preventing E. faecalis infection of the root canal and preventing failure.

IMMUNE EVASION

For enterococci to maintain an infection, they must successfully evade both specific and nonspecific host defense mechanisms. Other Gram-positive pathogens possess attributes that allow them to survive in the host in spite of powerful nonspecific host defenses mediated primarily by professional phagocytes, i.e., neutrophils, monocytes, and macrophages. These factors include antiphagocytic polysaccharide capsules, antiphagocytic surface proteins, such as the group A streptococcal M protein, and various secreted toxins with direct phagocytic cell toxicity. The most important cell type for clearing bacteria during the early innate immune response to enterococcal infection is neutrophils, with complement and antibody-mediated killing also playing a role (183). Neutropenic mice are less able to clear E. faecium from their bloodstream compared to their wild-type counterparts and show higher inflammatory cytokine levels after intraperitoneal infection (184).

Numerous studies have characterized the host response to enterococcal infection (183, 185–187). E. faecalis has been found to be able to actively cleave the complement protein C3, mediated by the metalloprotease GelE, leading to reduced phagocytosis and higher rates of bacterial survival (188). In a case report, Bottone and colleagues (189) isolated three highly mucoid encapsulated strains of E. faecalis from patients with urinary tract infections. These mucoid isolates appeared to persist longer in the mouse following intraperitoneal injection compared to nonmucoid controls, but crude mortality was not changed, suggesting a role in preventing or delaying phagocytic clearance (190). Once phagocytosed, E. faecalis is able to resist intracellular killing in both polymorphonuclear leukocytes and macrophages (110). In polymorphonuclear leukocytes, opsonized bacteria are more readily killed than unopsonized cells, and AS appears to limit opsonization, leading to survival within phagosomes (191). Recent studies examining E. faecalis survival in macrophages showed that phagosomes containing E. faecalis did not acidify or fuse with autophagosomes, leading to greater bacterial survival (192). E. faecalis is also able to survive the oxidative burst within macrophages, which is in part mediated by the production of catalase (112, 193). Survival within macrophages may also allow for bacterial translocation across the intestinal epithelium, leading to dissemination of E. faecalis throughout the body (112).

Several studies have examined the diversity of cell wall polysaccharides produced by E. faecalis (194–197). Based on genetic and serologic data, E. faecalis capsular polysaccharides fall into four serogroups (197, 198). Antibodies to E. faecalis capsular polysaccharides do, however, cross-protect across the species boundary (196), because antibodies to E. faecalis capsular antigens are reactive with some E. faecium capsular antigens and mediate opsonophagocytosis. The genetic basis for E. faecalis capsular biosynthesis is encoded by the cpsC-K operon (194, 199). In addition to capsular antigens, Murray and coworkers identified another cell wall polysaccharide, called the enterococcal polysaccharide antigen, whose synthesis is encoded by the epa locus (200–202). This pathway appears to be widespread and conserved within E. faecalis, and it has been designated the E. faecalis group antigen (194), analogous to the streptococcal group antigens. Antibodies to the Epa polysaccharide have also been shown to promote opsonophagocytic clearance by neutrophils (200), but this prospect remains controversial (196). A recent study showed that inactivation of a predicted wall teichoic acid biosynthetic gene in E. faecalis caused differences in wall teichoic acid and Epa composition and led to increased susceptibility to complement-mediated killing by neutrophils (203). Finally, enterococcal cell surface-associated polysaccharides have been found to impact gastrointestinal tract colonization and infection in a variety of settings (99, 195, 204–207). The extent to which enterococcal surface-associated polysaccharides could be used as potential vaccines awaits further study.

TISSUE DAMAGE

Following adhesion to host cell surfaces and evasion of the host immune response, the last step in the pathogenesis of enterococcal infection is the production of pathologic changes in the host. Such changes can be induced by the host inflammatory cascade or by direct damage as a result of secreted toxins or proteases, as well as the production of toxic metabolic byproducts. Each of these mechanisms has been observed in studies of E. faecalis pathogenesis.

Indirect Tissue Damage

Enterococcal LTA, also known as the group D streptococcal antigen, has been implicated in a variety of biological processes (204, 208–211). Some properties ascribed to LTA include modulation of the host immune response and mediating the adherence of enterococci to host cells. Bhakdi et al. (212) found LTA from enterococci to be as inflammatory as lipopolysaccharide of Gram-negative bacteria, although more recent findings dispute this claim (213). Strains of E. faecalis defective in enterococcal binding substance (EBS) (which is at least partially derived from LTA) and also defective in the protein adhesin AS did not induce clinical signs of illness when injected intraventricularly into rabbits (214). While EBS⁺AS^– or EBS^–AS⁺ strains caused signs of illness and pericardiac inflammation, all rabbits injected with the EBS⁺AS⁺ strain developed illness and died. Surprisingly little inflammation was observed in rabbits injected with the EBS⁺AS⁺ strain, despite the lethality observed. The authors state that such observations are consistent with the presence of a superantigen. The presence of LTA (EBS) and AS together may mediate effects on the host immune response that differ from those seen when either component acts alone.

Direct Tissue Damage

The enterococcal cytolysin and two proteases, a zinc metalloprotease (gelatinase) and a serine protease, are secreted factors well suited to contribute to disease severity (126, 128, 129, 215, 216). The enterococcal cytolysin is well established in several independent labs and animal models (126, 128, 129). The presence of the cytolysin has also been shown to promote the appearance of E. faecalis in the bloodstream (217). The genetics and biology of the enterococcal cytolysin have been reviewed elsewhere (218, 219). The toxin is genetically and structurally related to the lantibiotic family of bacteriocins but diverges from this family in that it possesses both bacteriocin activity and toxicity toward a variety of mammalian cell types. The cytolysin operon is typically encoded on large pheromone-responsive plasmids but has also been shown to reside within an E. faecalis pathogenicity island (134). The operon consists of six genes, designated cylL_L, cylL_S, cylM, cylB, cylA, and cylI, with each gene product playing an essential role in toxin synthesis, modification, secretion, activation, and immunity. Transcription of the operon is repressed by the products of a divergent operon encoding CylR1 and CylR2, and repression is alleviated by the accumulation of fully processed CylL_S, which serves as a quorum sensing molecule to prime the synthesis of more toxin components as cell density increases (220).

The most direct and quantitative evidence for pathologic damage attributable to the cytolysin was obtained using a rabbit model of endophthalmitis (129). This model was selected because a robust infection can be established with as few as 10 organisms due to the natural aberrations in the intraocular immune response. This limited response provides the offending bacterium an opportunity to adapt to in vivo growth conditions and environmental cues. Moreover, highly sensitive and quantitative measurements of the evolution of disease can be made. A role for the cytolysin in tissue pathology was unambiguously demonstrated by both a reduction in β-wave response and complete destruction of retinal architecture 24 hours postinfection (129).

Because of its broad protease specificity, the enterococcal gelatinase may also play a measurable role in systemic disease (127), as well as in a model of dental caries using germ-free rats (215). In the germ-free rat model, Gold et al. (215) showed that a proteolytic (Gel+) strain exhibited cariogenic activity, whereas three nonproteolytic strains exhibited little cariogenicity. Dupont et al. (127) showed a reduced 50% lethal dose for mice injected with gelatinase-producing (Gel+) strains, and these findings were subsequently confirmed using isogenic strains defective in protease production (216). On the same transcript as the gelE gene resides the sprE gene, which encodes a serine protease. Both proteases have been shown to play important roles in C. elegans nematode killing (221), lupine endophthalmitis (222), and murine peritonitis (216). As discussed previously, gelatinase has been shown to be important in E. faecalis biofilm formation (223, 224) and also appears to play a role in the degradation of the host extracellular matrix proteins collagen and laminin (135).

Both gelatinase and the SprE serine protease are regulated at the transcriptional level by a quorum sensing signal transduction system termed fsr (225). The fsr locus encodes a response regulator, FsrA, which is thought to be activated by phosphorylation at a conserved aspartyl residue by phosphotransfer from a histidine kinase, FsrC. The kinase is thought to undergo auto-phosphorylation upon sensing the accumulation of a peptide lactone quorum molecule (226), which is encoded at the carboxy-terminus of FsrB. Most if not all of the phenotypic effect of inactivating the fsr locus resides in the lack of protease production. While initially the only genes thought to be transcriptionally controlled by activated FsrA were fsrB/C and gelE/sprE, various studies have suggested that additional genes are controlled by the fsr locus (221, 222, 227). More recently, a third locus under the control of FsrA was identified (228); this locus encodes the antimicrobial peptide enterocin O16 (229).

There is some indication that clinical isolates of E. faecalis may be enriched for the proteolytic trait, because greater than 50% of isolates from both endocarditis and other clinical sources exhibited gelatinase activity, whereas only 27% of community fecal isolates possessed this trait (230). Studies that have explored the presence of fsr genes and gelE in epidemiology have found widely differing results (231–234). In addition to differences in sample size and setting, this could be because a large percentage of gelatinase-negative clinical isolates contain a 23.9-kb chromosomal deletion that results in the deletion of both fsrA and fsrB and yields a null protease phenotype (235). Because this represents a defined deletion, it is tempting to speculate why E. faecalis would delete these important virulence factors. Analogous to the description of the E. faecalis pathogenicity island, which is known to be modulated through the insertion and deletion of genetic information (92), the organism may simply choose to modulate its virulence for fear of overtly offending its host.

While E. faecium appears to lack both proteolytic and cytolytic activities, it has been shown to encode a hyaluronidase (236), which appears to be enriched in nonstool isolates. Hyaluronidase activity has yet to be demonstrated in these isolates, but it is conceivable that this trait is emerging within the E. faecium population. An E. faecium pathogenicity island has also been described, which contains an ortholog of the E. faecalis esp gene (237, 238). Apart from the esp gene and an araC-like transcriptional regulator, none of the remaining open reading frames resemble those present in the E. faecalis pathogenicity island, and the nature of this element remains largely unknown.

TOXIC METABOLITES

In addition to secreted proteins, E. faecalis and E. faecium have also been shown to produce toxic oxygen metabolites that can cause cell or organ damage (94, 239). Using a worm model, Moy et al. (239) demonstrated that E. faecium produces hydrogen peroxide at levels that cause cellular damage. E. faecium transposon insertion mutants were identified which altered C. elegans killing activity and displayed altered levels of hydrogen peroxide production. Mutation of an NADH oxidase-encoding gene eliminated nearly all NADH oxidase activity and reduced hydrogen peroxide production and killing, whereas mutation of a gene encoding an NADH peroxidase resulted in enhanced levels of hydrogen peroxide and more rapid nematode killing. Depending on the culture conditions, E. faecium can produce hydrogen peroxide at varying levels, and this production appears to correlate inversely with nematode survival.

According to one study, the vast majority (87/91) of E. faecalis strains tested produce superoxide (O₂^–), whereas E. faecium isolates (5/13) did so less frequently (240). The study authors speculated that membrane-damaging effects of oxygen radicals might potentiate cellular damage to nearby intestinal epithelial cells. In proof, it was shown that the production of extracellular superoxide and hydrogen peroxide damages colonic epithelial cell DNA (94). The answer to why E. faecalis produces extracellular oxygen radicals stems from the fact that this oxygen radical is a byproduct of an incomplete respiratory chain. E. faecalis is capable of reconstituting a cytochrome complex in the presence of exogenous heme (241), enabling it to effectively respire. The fact that enterococci generate a byproduct that damages its host appears to be an unintended consequence of incomplete respiration.

COMPARATIVE GENOMICS OF E. FAECALIS AND E. FAECIUM: EMERGENCE OF PATHOGENIC, RESISTANT LINEAGES

With the development and widespread use of next-generation sequencing and comparative analysis of bacterial genomes, a great deal has been learned about the genetic makeup of pathogenic E. faecalis and E. faecium strains. The first enterococcal genome to be completely sequenced was a vancomycin-resistant E. faecalis strain isolated from an HIV-infected patient (35, 242). The genome of this isolate, termed V583, made apparent the importance of mobile genetic elements in the development of E. faecalis into a hospital-adapted pathogen (35, 242). When compared to a commensal-like E. faecalis strain, OG1RF (243), which harbors no known functional mobile elements, V583 was found to possess a genome more than 25% larger as the result of harboring three plasmids, a pathogenicity island (92, 243), and several integrated prophages (242).

First identified in the multiple drug-resistant strain MMH594 (92), the pathogenicity island has been shown to be widely distributed among hospital-adapted, pathogenic strains of E. faecalis. It encodes the virulence factors cytolysin and AS, ESP, and a variety of other elements believed to play a role in pathogenesis (92, 244). Overall, the trend of increased mobile genetic element content among multidrug-resistant, hospital-adapted strains holds true within E. faecalis and is strongly correlated with the absence of a functional clustered regularly interspaced short palindromic repeat-Cas system (245).

The population structures of E. faecalis and E. faecium have been widely studied, initially by multilocus sequence type and related techniques (246–249) and more recently by comparing sequences of entire genomes of many isolates of both species (104, 250, 251) (Fig. 3). These studies revealed a pronounced clade structure within the E. faecium species. The two main branches, clades A and B, are so divergent from one another that they approach the limits for designation as separate species (Fig. 3A) (104, 248). Most hospital-adapted E. faecium isolates belong to clade A and closely resemble strains isolated from agricultural and companion animals. Non-hospital associated human fecal isolates of E. faecium, on the other hand, mainly belong to clade B and appear to have diverged from clade A strains several thousand years ago (104, 248, 252–254).

FIGURE 3.

Population structure of E. faecium and E. faecalis based on comparisons of whole genomes. (A) RAxML tree of 1,344 single copy core genes found in 73 E. faecium genomes, showing the clade structure of the E. faecium population. Reproduced with permission from reference 104. (B) Phylogeny of 515 E. faecalis isolates based on 1,293 conserved core genes found in 99% of isolates. The L1, L2, and L3 lineages are highlighted with red, purple, and turquoise lines, respectively. Reproduced with permission from reference 250.

In contrast to the pronounced clade structure of E. faecium, the population structure of E. faecalis is less discrete (Fig. 3B) (250, 251). A study of E. faecalis mainly derived from Great Britain revealed three dominant lineages, named L1, L2, and L3 (Fig. 3B) (250). However, these lineages are fairly closely related compared to the A and B clades of E. faecium. This finding supports previous work comparing a smaller set of E. faecalis genomes (253).

ANIMAL MODELS USED TO STUDY PATHOGENIC MECHANISMS

Several animal models have been developed to study different aspects of enterococcal pathogenesis. Using both invertebrate and vertebrate host organisms, many important discoveries have been made in identifying the roles of specific virulence factors in enterococcal disease (Table 1). Model organisms have also been employed in the discovery of novel antibacterial compounds. Here, we summarize the major advances in the establishment of relevant models of enterococcal disease; a more complete examination of models of enterococcal infection has been presented recently in two excellent reviews (255, 256).

Invertebrates

Invertebrate models of enterococcal infection have been very useful due to the ability to use large numbers of organisms in each experiment and because they are known to be natural hosts of enterococci in the environment. C. elegans, Drosophila melanogaster, and Galleria mellonella have all been used in different infection contexts, largely to examine the effects of virulence factors on lethality of the whole organism (130, 257, 258). It has been observed that there is good correlation between virulence factors in invertebrate hosts and in mammalian infection models (130, 221). Invertebrates, particularly C. elegans, have also proven to be very fruitful for high-throughput screening for novel antienterococcal compounds, because they allow for identification of bactericidal compounds with low host toxicity, as well as host immunomodulatory molecules that impact the course of infection (259).

Vertebrates

Several infection models have been established in small mammals for many of the most important enterococcal diseases. Two of the earliest infection models were developed for enterococcal IE, first in rabbits by MacCallum and Hastings in the first description of enterococcal disease (3) and later modified by Garrison and Freedman (260). Santoro and Levison developed a rat model of infection in the 1970s (261). Both the rabbit and rat IE models involve the insertion of a catheter into the heart valve followed by administration of bacteria. The introduction of a foreign body allows for initial adherence of enterococcal cells, resulting in a robust infection of the cardiac tissue. The determination of many important virulence factors for endocarditis have been shown using these animal models (128, 157, 255).

Intestinal colonization by E. faecalis and E. faecium has been studied in antibiotic-treated mice to model antibiotic treatment in hospitalized human patients. Generally, mice are treated with high concentrations of broad-spectrum antibiotics, particularly spectinomycin or metronidazole, to allow enterococcal invasion into the microbiota. Kommineni and colleagues obtained stable, long-term colonization of an introduced E. faecalis strain into an intact mouse microbiota through long-term administration of the bacteria in drinking water (102). Recently, studies examining the localization of E. faecalis within the intestinal tract have been performed in germ-free mice, allowing for direct visualization of bacteria in contact with intestinal tissue (262). How closely this localization resembles that of enterococci in a fully colonized animal has yet to be determined.

Mice have also been used with great success to study bacteremia and peritonitis caused by E. faecalis and E. faecium (126, 263). Urinary tract infection models, both catheter-associated and endogenous, have been established in female mice and were instrumental in identifying virulence factors, particularly adhesins, in enterococcal pathogenesis (134, 147, 264, 265). Finally, to study postoperative infections, Stevens and colleagues established a highly tractable rabbit endophthalmitis model in 1992 (266). This model involves the direct placement of small numbers of bacteria into the vitreous, directly behind the lens, where the evolution of infection can be easily observed by direct ophthalmoscopy as well as slit lamp biomicroscopy. This allows the influence of a number of virulence traits on the host-microbe dynamic to be assessed in real time (129).

From one of the first descriptions of enterococcal infection in 1899 (3) to the understanding of the role of virulence factors in disease and identification of novel antimicrobials, animal models have proven to be extremely important in the study of enterococcal pathogenesis, and their further development will continue to provide insight into diseases caused by enterococci.

CONCLUSIONS AND PERSPECTIVES

Enterococci are unusually well adapted for survival and persistence in a variety of adverse environments, including inanimate surfaces in the hospital environment and at sites of infection. This intrinsic ruggedness undoubtedly plays a role in providing opportunities for enterococci to interact with other overtly drug-resistant microbes and acquire additional resistances on mobile elements. The rapid rise of antimicrobial resistance among hospital-adapted enterococci has rendered hospital-acquired infections a leading therapeutic challenge. Having added about a quarter of a genome of additional DNA conveyed by mobile elements, there are undoubtedly many more properties that have been acquired that help enterococci persist and spread in the hospital setting and cause diseases that have yet to be defined. Much remains to be learned about these ancient and rugged microbes.