Could a Phosphotransferase System Provide the Means to Control Outbreaks of Enterococcus faecium Infection?

Sudha R. Somarajan 2 Barbara E. Murray 0 1 2 0 Department of Microbiology and Molecular Genetics, University of Texas Medical School at Houston 1 Center for the Study of Emerging and Re-emerging Pathogens 2 Division of Infectious Diseases, Department of Internal Medicine EDITORIAL COMMENTARY - Enterococcus faecium; phosphotransferase system; colonization. The epidemiology of enterococcal infections has changed dramatically over the past 100 years [1]. Until relatively recently, the vast majority were caused by Enterococcus faecalis and were acquired in the community; then, during the 1970s 1980s, the percentage of nosocomial infections due to E. faecalis tripled [2, 3], an increase temporally associated with increasing use of third-generation cephalosporins. In the United States, an increase in the proportion of nosocomial infections caused by Enterococcus faecium was observed in the 1980s [4, 5], and by 20062007, a Centers for Disease Control survey of healthcare-associated infections found that approximately 38% of clinical enterococcal isolates identified to the species level were E. faecium [6]. This increase was blamed on the acquisition of vancomycin resistance by E. faecium (which seldom occurred in E. faecalis) and the increasing use of vancomycin in hospitalized patients. However, the epidemiology of vancomycin-resistant E. faecium (VREfm) presented a quandary: at the time, VREfm was already a common cause of US healthcare-associated infections, such organisms were not found in the community [7]. Yet, concurrently, VREfm was rare as a cause of infection in the European Union but were commonly found in feces of food animals in Europe, in products derived from these animals, in feces of individuals handling them, and even in feces of individuals in the general public [8]. These European community-associated VREfm strains were linked to use of a glycopeptide, avoparcin, in animal feeds, a practice that was not legal in the United States and has since been banned in the European Union. What were the reasons for the rapid increase in VREfm infections in US hospitals, and how did they remain very uncommon in the European Union until considerably later, despite the large reservoir of community-based VREfm in Europe [9, 10]? It is now clear that there are many differences between E. faecium isolates recovered from healthcareassociated infections and outbreaks and those that are predominantly communityassociated fecal commensals, such as the early vancomycin-resistant enterococci in the European Union community. First, healthcare-associated isolates usually belong to a small number of prominent clonal clusters, each composed of related multilocus sequence typing (MLST) types distinct from the diverse MLST types typically found, with a few notable exceptions, in community-associated isolates [11, 12]. Other traits noted to be common among healthcare-associated E. faecium but uncommon among community-associated commensals include the presence of esp, IS16, a hyl-like gene (now annotated as family 84 glycosyltransferase), and genes predicting MSCRAMM adhesins [12, 13]. A high minimum inhibitory concentration (MIC) of ampicillin is another characteristic of healthcare-associated E. faecium and is due to a version of PBP5 with markedly reduced affinity for penicillin [11, 14, 15]. Since higher ampicillin MICs predict higher MICs of antipseudomonal penicillins and cephalosporins, which are frequently used in hospitalized patients (some of which reach high gut concentrations), this trait should provide an important selective advantage for gastrointestinal colonization in the hospital setting. E. faecium strains with high-level resistance to ampicillin were first reported in the United States in the 1970s1980s, including several nosocomial outbreaks, prior to the emergence of VREfm [4, 14, 16]; a similar increase in ampicillinresistant E. faecium has more recently preceded the emergence of VREfm in [some European Union nosocomial outbreaks [9]. An important observation that we now know reflects fundamental differences among E. faecium strains was made by Hallgren et al, who found a bimodal distribution of ampicillin MICs among E. faecium isolates (although not among E. faecalis) [17]. In retrospect, we can conclude that, in the United States, it was the ampicillin-resistant E. faecium subgroup found in hospitals that first acquired vancomycin-resistance, while in the European Union, the communityassociated subgroup with greater ampicillin susceptibility was the first to acquire vancomycin resistance. Thus, in the United States, ampicillin-resistant VREfm was likely promoted and perpetuated in hospitals by both vancomycin and -lactam use, while ampicillin-susceptible VREfm strains common in the European Union community, albeit promoted by the use of avoparcin in animals, were much less resistant to -lactams and, thus, not as likely to survive in the hospital milieu. But is the increase in E. faecium as a cause of infection due solely to increased levels of resistance to ampicillin and vancomycin? It is now clear that the differences noted above in the accessory genome (eg, esp, hyl-like, and IS16) and in ampicillin resistance between the healthcare-associated and the communityassociated E. faecium subgroups correlate with significant differences in the sequences of these subgroups shared core genes. This correlation was clearly illustrated by the finding of a difference in approximately 5% of nucleotides through out the entire pbp5 gene (not just those that cause increased ampicillin resistance) [15] and of a difference in 3% 10% of nucleotides between >100 other core genes of strains in the healthcareassociated subgroup versus the communityassociated subgroup [15, 18]. Furthermore, an analysis of synonymous SNPs predicted that the divergence between the healthcare-associated and communityassociated subgroups (now referred to as clades) of E. faecium likely occurred hundreds of millennia ago or longer [18]. These analyses are consistent with a report of differences between clade A (healthcare associated) and clade B (community associated) strains [19]. Interestingly, many E. faecium strains appear to be hybrids, with a mix of core genes from each clade or from different MLST types within the same clade [18, 20]. Given the extensive differences in the core and accessory genomes of the 2 E. faecium clades, how does one begin to tackle the question of which factors are responsible for the increase and predominance of the healthcare-associated clade in nosocomial infections? In this issue of the Journal, Van Schaik et al have approached this complex question from the point of view of the antibiotic-altered gastrointestinal tract, the site where VREfm is first established and from which most, if not all, VREfm infections are thought to originate [21, 22]. The effect of antibiotics on gastrointestinal VREfm is profound, and virtually all patients with VREfm infections are r (...truncated)