Population biology of Gram-positive pathogens: high-risk clones for dissemination of antibiotic resistance

Abstract

Infections caused by multi-resistant Gram positive bacteria represent a major health burden in the community as well as in hospitalized patients. Staphylococcus aureus, Enterococcus faecalis and Enterococcus faecium are well-known pathogens of hospitalized patients, frequently linked with resistance against multiple antibiotics, compromising effective therapy. Streptococcus pneumoniae and Streptococcus pyogenes are important pathogens in the community and S. aureus has recently emerged as an important community-acquired pathogen.

Population genetic studies reveal that recombination prevails as a driving force of genetic diversity in E. faecium, E. faecalis, S. pneumoniae, and S. pyogenes and thus, these species are weakly clonal. Although recombination has a relatively modest role driving the genetic variation of the core genome of S. aureus, the horizontal acquistion of resistance and virulence genes plays a key role in the emergence of new clinically relevant clones in this species. In this review we discuss the population genetics of E. faecium, E. faecalis, S. pneumoniae, S. pyogenes, and S. aureus. Knowledge of the population structure of these pathogens is not only highly relevant for (molecular) epidemiological research but also for identifying the genetic variation that underlies changes in clinical behaviour, to improve our understanding of the pathogenic behaviour of particular clones and to identify novel targets for vaccines or immunotherapy.

Introduction

Enterococcus faecium, Enterococcus faecalis, Streptococcus pneunmoniae, Streptococcus pyogenes, and Staphylococcus aureus are low GC Gram-positive bacteria belonging to the phylum Firmicutes. All of these species are human commensals, and as such are part of the normal human microflora having a benign relationship with their host. However, they are also human pathogens capable of infecting a broad range of body sites. The occasional virulence of these species requires a susceptible host, but also reflects a history of selection for specific bacterial variants or clones with enhanced pathogenic potential that are particularly difficult to manage when these clones acquire antibiotic resistance. Despite the fact that enterococci, streptococci and staphylococci all represent important opportunistic pathogens in which resistance has evolved against multiple antibiotics, the epidemiology of resistant clones in these pathogens differ substantially. While antibiotic resistance in S. pneumoniae and S. pyogenes has evolved in clones causing community-acquired infections, multi-resistant S. aureus and enterococci clones are primarily causing infections in hospitalized patients, although community-acquired infections with resistant S. aureus have emerged the last decade. In contrast to S. pneumoniae and S. pyogenes, which are strict human pathogens, S. aureus and enterococci are capable of colonizing and causing infections in a wide range of animals. Consequently, multi-resistant S. aureus and enterococci are also found in non-human reservoirs (e.g. livestock) in which the selective pressure imposed by antibiotic use favours the selection of resistant clones. This review aims to give an overview of the evolution and population dynamics of specific multi-resistant clones in these Gram-positive species, in the context of the differing ecological and epidemiological characteristics of these opportunistic pathogens.

The genus Enterococcus

Enterococci are lactic acid bacteria belonging to the ileal microbiota (Booijink et al., 2010). They represent typical examples of opportunistic pathogens that have long been regarded as relatively harmless commensals colonizing the gastro-intestinal tract of humans and animals. Although considered normal colonizers of the digestive tract, they have also been recognized for more than 100 years as etiological agents of hospital-acquired infections in debilitated patients (Murray, 1990). The perspective on enterococci started to change in the 1980s when these organisms were found to have gained high-level resistance to ampicillin and their growing importance was cemented by the rapid emergence of vancomycin-resistance in the 1990s in hospitals in the US (Rice, 2006). These events catapulted enterococci in the Premier League of multi-drug resistant Gram-positive pathogens, together with pathogens like methicillin-resistance S. aureus and penicillin-resistant S. pneumoniae (Woodford & Livermore, 2009; Nordmann et al., 2007).

The digestive tract, the main habitat of Enterococci in humans

Of all Enterococcus species, E. faecalis and E. faecium are the species most frequently found as colonizers of the human gastro-intestinal tract, and are similarly most commonly responsible for infections in hospitalized patients. Undoubtedly, antibiotic resistance has played a pivotal role in the emergence of E. faecalis and E. faecium as nosocomial pathogens. Intrinsic resistance to certain classes of beta-lactam antibiotics (e.g. cephalosporins), low-level resistance to aminoglycosides, lincosamides, streptogramins (in case of E. faecalis) and monobactams provided enterococci with a selective advantage in the hospital environment (Klare et al., 2003). Level of resistance increased over the last 20–30 years due to the ability of E. faecalis and E. faecium to acquire foreign DNA resulting in the rapid accumulation of antibiotic resistance genes (Hegstad et al., 2010). Tetracycline and erythromycin resistance genes were reported to be located on mobile genetic elements as early as the 1970s, signifying the rapid dissemination of resistance traits among enterococci (Courvalin et al., 1974; Clewell et al., 1974). Today, enterococci have acquired resistance against a broad range of antibiotics covering most antimicrobial classes (Klare et al., 2003; Hegstad et al., 2010). Most notably, from a clinical point of view, is the acquired resistance to aminoglycosides, penicillines, especially ampicillin, and vancomycin.

Vancomycin resistance a serious threat

The emergence of vancomycin-resistance by enterococci has become the paradigm of the post-antibiotic era. Acquisition of plasmid-encoded high-level vancomycin resistance, first isolated in 1986 in Europe (Leclercq et al., 1988; Uttley et al., 1988), definitively marked enterococci as one of the most notorious antibiotic resistant bacteria. During the 1990s vancomycin-resistant enterococci (VRE) have disseminated rapidly thoughout hospitals in the US, first in intensive care units (ICU) and subsequently in essentially all hospital wards (National Nosocomial Infections Surveillance (NNIS) System Report, data summary from January 1992 through June 2004, issued October 2004, 2004). At that time VRE prevalence rates in hospitals in Europe were still low, while VRE carriage rates in the community, and in meat products via animal reservoirs were high (Bonten et al., 2001). The latter was most probably due to the Europe-wide use of avoparcin, a glycopeptide antibiotic conferring cross-resistance to vancomycin, as an antimicrobial-growth promoter from the early 1970s until 1997, when it was banned in Europe (Goossens, 1998).

Although vancomycin-resistance in enterococci can be mediated by six vancomycin-resistance determinants, the major vancomycin-resistance phenotypes are VanA and VanB (Courvalin, 2006; Boyd et al., 2008). Virtually all VRE recovered from nosocomial infections are also ampicillin resistant. In fact, the emergence of high-level ampicillin resistance, specifically in US hospitals in the early 1980s, preceded the epidemic rise of vancomycin resistance with increasing numbers of high-level ampicillin-resistant enterococci being isolated in US hospitals in the early 1980s (Grayson et al., 1991; Iwen et al., 1997). Acquired high-level resistance to beta-lactam antibiotics in clinical isolates of E. faecium is conferred by mutations of the low-affinity penicillin-binding protein, PBP5, which leads to a lower affinity of this PBP for beta-lactam antibiotics, or by overproduction of PBP5 (Rybkine et al., 1998; Klare et al., 2003). The association between ampicillin and vancomycin resistance phenotypes can sometimes be explained by genetic linkage of high-level ampicillin-resistance and vancomycin-resistance genes (Carias et al., 1998). More frequently this phenotypic linkage probably reflects sequential and independent acquisition of resistance genes resulting in the selective dominance of a small subset of hospital -adapted clones. The progressive increase in high-level ampicillin resistance and vancomycin resistance coincided with the increase of nosocomial infections caused by E. faecium relative to E. faecalis. While up to the 1980s/early 1990s more than 90% of all enterococcal infections were caused by E. faecalis and only 5–10% by E. faecium (Iwen et al., 1997; Murdoch et al., 2002; Treitman et al., 2005) this ratio has gradually changed in favour of E. faecium. Today, between 38–75% of enterococcal infections are caused by E. faecium (Hidron et al., 2008; Markogiannakis et al., 2009; Chiang et al., 2007; Mikulska et al., 2009). The observed ecological shift in enterococcal infections is probably due to the fact that resistance to ampicillin and vancomycin is much more prevalent in E. faecium than in E. faecalis isolates recovered from clinical infections. While to date approximately 90% of E. faecium from healthcare-associated infections (HAI) in the US are resistant to ampicillin and 80% to vancomycin these resistance percentages are significantly lower in E. faecalis (~4% resistance to ampicillin and 7% resistance to vancomycin) (Hidron et al., 2008). Also, globally, the epidemiology of VRE is dominated by E. faecium with vancomycin-resistant E. faecium being isolated from all continents (Tenover & McDonald, 2005; Werner et al., 2008; Camargo et al., 2006; Dahl et al., 2007; Zheng et al., 2007; Lee et al., 2004; Hoshuyama et al., 2008; von Gottberg et al., 2000; Padiglione et al., 2003).

Molecular epidemiology of E. faecium

Molecular epidemiological studies of VRE outbreaks in the US using pulsed-field gel electrophoresis (PFGE) not only reported intra- and inter-hospital dissemination of single clones (Handwerger et al., 1993; Boyce et al., 1994; Moreno et al., 1995; Dunne & Wang, 1997) but the presence of polyclonal VRE in single hospitals suggested simultaneous spread of multiple VRE clones (Bonten et al., 1996; Mato et al., 1996). However methods like PFGE, that index rapidly evolving variation, may give misleading results for epidemiological typing in rapidly recombining and evolving bacterial genera such as Enterococcus (Paulsen et al., 2003; Willems et al., 2005; Ruiz-Garbajosa et al., 2006). The observation that isolates belonging to the same clone may differ in up to 7 PFGE bands (Morrison et al., 1999) illustrates that PFGE is too discriminatory to study the long term and global epidemiology of enterococci.

The first insights into the genetic relatedness of a large sample of E. faecium isolates came from a study in 2000 when 255 vancomycin-resistant E. faecium from different ecological niches and geographic locations were genotyped by amplified fragment length polymorphism (AFLP) (Willems et al., 2000). Since AFLP combines the analysis of small conserved and variable DNA regions distributed over the whole genome, this technique will identify clusters of related isolates that would not be detected by PFGE. In this study, 84% of isolates recovered from hospitalized patients from different parts of the world clustered together in a single genogroup C, distinct from 75% of isolates from non-hospitalized persons and 100% of the isolates from pigs (genogroup A), 92% of poultry isolates (genogroup B) and 70% of veal calf isolates (genogroup D). The existence of a specific hospital subpopulation in E. faecium was also demonstrated by comparative genomic hybridisation (CGH) of 97 E. faecium isolates from diverse ecological and geographic background and representing both vancomycin-resistant and –susceptible isolates to an E. faecium mixed whole genome array. Clustering of CGH data based on the presence or absence of a hybridisation signal grouped 75% of the globally dispersed hospital outbreak and clinical isolates into a single hospital clade. Furthermore, this clade contained only 7% of the community (human, animal and environmental) isolates (Leavis et al., 2007). This finding indicated that the genetic content of hospital isolates differed from non-hospital isolates.

Genetic evolution of hospital-associated E. faecium

The question is whether relatedness-based on shared DNA fragments (AFLP) or genes/genetic elements (CGH) also means that hospital isolates share a common evolutionary history. Whole genome sequencing of 7 E. faecium strains, representing four clinical isolates from 4 hospitals in three different countries on two continents, one carrier isolate from a hospitalized patient and two carrier isolates from two non-hospitalized persons, provided a detailed analysis on diversity and evolutionary relatedness of E. faecium isolates (van Schaik et al., 2010). The observation that a neighbour-joining tree based on shared gene content and a maximum likelihood tree based on concatenated alignment of 649 orthologous proteins showed a similar topology, with the community isolates clearly distinct from the hospital isolates, is consistent with the hypothesis that hospital isolates not only share DNA content but are also descended from a comparatively recent common ancestor.

More in depth analysis of the evolutionary relatedness of E. faecium genotypes on a population level came from multilocus sequence typing (MLST) data (Homan et al., 2002). The first E. faecium population-wide study using MLST characterized a global collection of human (hospital and community-acquired) and non-human (animals and the environment) E. faecium and defined 175 sequence types (STs). STs were grouped with the eBURST algorithm. eBURST, which divides an MLST data set of any size into groups of related isolates and clonal complexes, predicts the founding (ancestral) genotype of each clonal complex, and computes the bootstrap support for the assignment (Feil et al., 2004). This clustering indicated that the majority of the globally representative hospital isolates were genotypically and evolutionary closely related and belonged to a single clonal-complex (CC), CC17 (Willems et al., 2005). However, the E. faecium population structure based on all STs (n=554) available in the MLST database (http://efaecium.mlst.net/ accessed on July 19^th 2010) inferred by eBURST resulted in one major large straggly CC, which includes the previously designated CC17, 18 minor CCs and 110 singletons, with sixty-nine percent of the E. faecium STs in the database clustering in the major CC (Fig. 1). Recently, Turner and co-workers showed that eBURST links, i.e. eBURST inference on recent evolutionary descent, in populations in which >25% of STs belong to a single large straggly group, may not be accurate (Turner et al., 2007). Consistent with an inaccurate eBURST-based clustering is the observation that major hospital-associated subgroup founders (ST17, ST18, and ST78) that are linked together using eBURST belong to different genetic lineages in a neighbour-net tree or ClonalFrame (Didelot & Falush, 2007) based phylogentic tree constructed from a concatenation of the seven MLST housekeeping genes (Willems, 2010) (Fig. 2)(Willems & van Schaik, 2009). Also preliminary analysis of the population genetics of E. faecium using a recent Bayesian modeling approach (BAPS software) (Tang et al., 2009) demonstrated that the major hospital clones ST78 and its single locus variants ST192 and ST203 group in a different BAPS group than ST17 and ST18 (Jukka Corander, Personal communication, 2010). This demonstrates that hospital-derived E. faecium strains have not recently evolved from a single common ancestor and consequently the designation of CC17 as a hospital-selected clonal complex may well be erroneous. Instead, clinical and outbreak-associated isolates, collectively referred to as hospital-derived isolates, form a polyclonal E. faecium subpopulation harbouring evolutionary distinct clones. Despite the apparent lack of a common evolutionary history, hospital-derived clones are, based on AFLP, MLST, and comparative genomics (see above), genetically and evolutionary distinct from non-hospital isolates. This is also illustrated by the fact that the seven major ampicillin-resistant hospital-derived clones represented by ST16, ST17, ST18, ST78, ST117, ST192 and ST203, accounting for the majority (56%) of all 910 hospital-derived isolates, are only found sporadically among non-hospital isolates (41/513) (http://efaecium.mlst.net/). Furthermore, these clones have spread globally among hospitalized patients and can therefore be considered as highly successful and high-risk clones for dissemination of antibiotic resistance in the hospital environment thus contributing most to the hospital-burden of E. faecium (Werner et al., 2008; Deplano et al., 2007; Panesso et al., 2010; Camargo et al., 2006; Zheng et al., 2007; Lester et al., 2008; Klare et al., 2005; Damani et al., 2010; Bonora et al., 2007; Ko et al., 2005; Top et al., 2008; Khan et al., 2008; Koh et al., 2006; Valdezate et al., 2009; Billstrom et al., 2010; Hsieh et al., 2010; Ergani-Ozcan et al., 2008; Caplin et al., 2007; Galloway-Pena et al., 2009; Freitas et al., 2010; Willems et al., 2005; Libisch et al., 2008; Coque et al., 2005; Homan et al., 2002).

Figure 1.

Population snapshot based on all unique STs of the entire E. faecium MLST database accessed on 19/7/2010 and visualized using eBURST. The shaded area indicates the previously designated CC17. ST16, ST17, ST18, ST64, ST78, ST132, ST148, ST192, ST203, ST252, ST280, and ST384, representing multidrug resistant hospital isolates and belonging to the previously designated CC17 are indicated.

Figure 2.

Phylogenetic analysis based on nucleotide sequences of the E. faecium MLST genes of 73 STs representing major subgroup founders (>3 SLV links) and singletons with no double locus variant links to CCs. The phylogenetic tree was based on all trees sampled after the burn-in periode obtained using ClonalFrame after 100,000 iterations, including 50,000 burn-in. ST17, ST18, ST78, ST192 and ST203, representing multidrug resistant hospital isolates are indicated.

There is one other reservoir known where these typical hospital clones seem to reside. Damborg and co-workers showed in two recent publications that ampicillin-resistant E. faecium clones that are frequently associated with infections in hospitalized patients were present in 48 of 63 dogs (Damborg et al., 2008; Damborg et al., 2009). It remains to be investigated whether this has resulted from anthropo-zoonotic transfer or whether dogs form a risk for zoonotic transfer of these clones to hospitalized patients.

Acquisition of adaptive elements by hospital-association E. faecium

The success of these hospital clones most probably is due to the cumulative acquisition of adaptive elements like genes or genetic elements encoding antibiotic resistance determinants (e.g. high-level ampicillin-and quinolone-resistance) (Leavis et al., 2003), and also genes linked to virulence. Well characterised virulence genes in E. faecium include (Leavis et al., 2006; Willems et al., 2005; Coque et al., 2002; Galloway-Pena et al., 2009; Hegstad et al., 2010)esp and other genes on the esp pathogenicity island, hyl and several genes encoding surface proteins (Willems et al., 2001; Coque et al., 2002; Rice et al., 2003; Leavis et al., 2004; Vankerckhoven et al., 2004; Coque et al., 2005; Klare et al., 2005; Camargo et al., 2006; Galloway-Pena et al., 2009; Leavis et al., 2007; Hendrickx et al., 2007). Character evolution analysis based on CGH data identified IS elements, specifically IS16, to be associated with hospital-derived isolates. It has been argued that the acquisition of IS elements might facilitate the subsequent acquisition of virulence determinants like the esp gene (Leavis et al., 2007), and the availability of multiple whole genome sequences in the near future will greatly help to explore this possibility. The process of cumulative acquisition of adaptive elements, combined with the progressive increase in fitness, has been termed genetic capitalism (Baquero, 2004).

What biological characteristics of E. faecium allowed for the rapid evolutionary development of hospital-selected E. faecium clones that started to dominate enterococcal hospital epidemiology from the 1980s is not known. It is interesting to note however, that these hospital clones are enriched in particular insertion sequence (IS) elements (Leavis et al., 2007). Transposable elements (TEs), like IS elements, may dramatically increase the rate of molecular evolution thereby significantly increasing genome restructuring and adaptability to changing requirements, as in hospital environments (Oliver & Greene, 2009). This may have promoted genetic diversification in specific subpopulations and the evolutionary development of the successful E. faecium clones mentioned above.

Shared hospital and community clones in E. faecalis

As in E. faecium, the genome of E. faecalis is also infested by TEs. The E. faecalis V583 genome has 38 IS elements and more than a quarter of the genome consists of probable mobile or foreign DNA (Paulsen et al., 2003; Manson et al., 2010). The extreme high abundance of TEs in E. faecalis and the advanced pheromone system in E. faecalis that enables not only spread of plasmid but also transfer of large genomic regions including virulence and resistance genes (Manson et al., 2010), may have increased genome plasticity and species adapatability to changing conditions in hospital environment (Oliver & Greene, 2009). Comparison of gene tree topologies of individual MLST genes indicate recombination rates in E. faecalis that are even higher than in E. faecium (Ruiz-Garbajosa et al., 2006; Willems et al., 2005). Recent multi-genome analysis of E. faecalis and E. faecium showed that both have an open pan-genome indicating that both organisms can efficiently acquire and integrate foreign DNA in their gene pool (Nelson et al., 2010; van Schaik et al., 2010).

It is interesting to note that high prevalent E. faecium hospital clones are virtually absent in the community (Werner et al., 2011), while this is not the case in E. faecalis. The seven most prevalent E. faecalis clones among clinical and outbreak-associated isolates (n=355), based on MLST (Ruiz-Garbajosa et al., 2006), ST6, ST9, ST16, ST21, ST28, ST40, and ST87, account for only 37% of the hospital-derived isolates (http://efaecalis.mlst.net/), while this is 56% for the seven most prevalent hospital-derived E. faecium clones (see above). These clones are also found frequently in the community, including farm animals and food products (in 17% of 351 isolates (http://efaecalis.mlst.net/) (Ruiz-Garbajosa et al., 2006), while in E. faecium major hospital-derived clones are more rarely found in the community (8% of 513 non-hospital isolates; see also above (McBride et al., 2007; Larsen et al., 2010; Freitas et al., 2009; Lopez et al., 2010; Burgos et al., 2009). This indicates that in E. faecalis clones are more often shared between hospitalized patients and other reservoirs. In line with this result is the finding based on CGH, and confirmed by PCR, that genes involved in the virulence of E. faecalis, such as gelatinase, cytolysin, the enterococcal surface protein Esp and aggregation substance, can also be frequently found in nonclinical E. faecalis strains, such as isolates from food or healthy babies (Semedo et al., 2003; Solheim et al., 2009; van Schaik & Willems, 2010). This suggests that there is no clear distinction between pathogenic and non-pathogenic clones in E. faecalis. Nevertheless, some E. faecalis clones, like ST6, ST9, ST28 and ST40, represented by at least 10 isolates in the database (http://efaecalis.mlst.net/) are clearly enriched (>70% of isolates) among hospitalized patients and are distributed world wide (McBride et al., 2007; Nallapareddy et al., 2005; Ruiz-Garbajosa et al., 2006; Freitas et al., 2009; Solheim et al., 2009; Damani et al., 2010; Lopez et al., 2010; Lester et al., 2010; Kawalec et al., 2007; Sun et al., 2009; McBride et al., 2009).

Streptococcus pneumoniae

The pneumococcus (S. pneumoniae) is the leading cause worldwide of community-acquired pneumonia. Other disease manifestations associated with pneumococcal infection include common, mild self-limiting infections such as acute otitis media, but extend to rare cases of invasive disease associated with high mortality, such as meningitis (Durbin, 2004). Its clinical burden is concentrated among the very old and very young, and it is estimated to be responsible for more than 800,000 deaths per annum in children under 5 years of age (O’Brien et al., 2009). Despite this high toll, the vast majority of pneumococci are found in asymptomatic nasopharyngeal carriage, the prevalence of which varies by age and region (Crook et al., 2004). The carriage state is responsible for transmission, and is the stage of pneumococcal life history at which interventions such as antibiotics and vaccines exert their selective pressure.

Pneumococci with reduced susceptibility to penicillin were first noted in a clinical setting in the 1960s (Kislak et al., 1965), but the risk posed by increasing resistance was not heeded until much later. By the mid 1980s penicillin resistant pneumococci had been reported worldwide (Appelbaum, 1987), and by the 1990s were responsible for >50% cases of invasive disease in some samples (Jacobs, 2003; Fenoll et al., 1998). More recently erythromycin resistance has emerged and posed an increasing threat (Jacobs, 2004; Leclercq & Courvalin, 2002). The implementation of a conjugate vaccine effective against a subset of pneumococcal serotypes has had considerable benefit (Kyaw et al., 2006), but increasing resistance has been documented in serotypes not targeted by the vaccine (Farrell et al., 2007; Gertz et al., 2010).

The pneumococcus is both antigenically and genetically diverse. It can be divided by serology into more than 90 serotypes that reflect the structure of a complex polysaccharide capsule (Aanensen et al., 2007). As well as being an important virulence factor, capsule is the target antigen for conjugate vaccines. Historically, serotyping was an important epidemiological tool, and it has been known for some time that some serotypes are more likely than others to be found in cases of invasive disease. Similarly, it was noted early in studies of pneumococcal penicillin resistance that high level resistance, and the vast majority of intermediate level resistance, was concentrated in a few serotypes (Klugman, 1990). To a large degree these serotypes were the same as those that were particularly associated with carriage and disease in young children, and were later the targets for conjugate vaccination: serogroups or types such as 19, 23, 6, 9 and 14.

While serotypes remain an important part of pneumococcal biology, modern epidemiology focuses on identifying the individual clones or lineages that make up each serotype. This is particularly important in the pneumococcus, because each serotype may typically be made up of a number of clones, which are not closely related and are not equivalent in terms of antibiotic resistance.

It is known that pneumococci can receive homologous DNA from other pneumococci and indeed other oral streptococci, and incorporate it into their own genome. It has been shown conclusively that this process has led to the acquisition of resistance to penicillin mediated by penicillin binding proteins (pbps) originating in other oral streptococcal species (Dowson et al., 1993), which have then been acquired by the pneumococcus through homologous recombination. Mosaic pbp2x, which has developed in several different lineages and species, has been studied in some detail, demonstrates the importance of horizontal transfer in the evolutionary history of this locus (Chi et al., 2007), and indicates diverse oral streptococcal species as possible ancestors of the resistant pbp2x locus. Such studies should however be interpreted with some caution, because our knowledge of the population structures of related oral streptococci is so very poor, these organisms appear to be extremely diverse (Bishop et al., 2009) and as a result it is hard to pinpoint with confidence the specific source of genetic variation. Fluoroquinolone resistance may also be acquired horizontally (Balsalobre et al., 2003). This is of particular interest because recombination can transfer not only mutations causing resistance, but also additional compensatory mutations ameliorating any cost associated with that resistance (Balsalobre & de la Campa, 2008; Rozen et al., 2007). In other cases, the basis of resistance is carried on a transposon or similar genetic elements, which are also capable of horizontal transmission. Multiple resistances may be combined in single transposable elements, such as erm(B) (erythromycin) and tet(M) (tetracycline) on Tn3872 (resulting from the insertion of Tn917 into orf9 of Tn916) and Tn1545 (Varaldo et al., 2009; Cochetti et al., 2008).

The Pneumococcal Molecular Epidemiology Network

To identify important lineages/clones the customary arsenal of molecular epidemiology has been directed at pneumococcal resistance. The pneumococcal molecular epidemiology network (PMEN) combines PFGE and MLST to identify the important internationally and intercontinentally distributed clones, define their resistance phenotypes and genotypes, how they may be identified, and produce a standardized nomenclature (McGee et al., 2001). PMEN also examines the molecular causes of resistance, and records the alleles of penicillin binding proteins and, determines the presence of the mef and erm loci, which produce erythromycin resistance. Through the work of PMEN we may speak with confidence, for example, of the Spain 23F-1 clone. This is a particular lineage, originally noted as having spread from Spain to the US, with a 23F capsule exhibiting high level resistance to penicillin and with a distinctive PBP profile (Munoz et al., 1991). It has since spread to every continent, and been recorded with several other serotypes in addition to the presumably ancestral 23F. As it was the first PMEN clone it is identified as such by the suffix -1. At the time of writing PMEN recognizes 27 clones associated with at least one resistance determinant (http://www.sph.emory.edu/PMEN/).

We can explore the relationships between the PMEN clones and other pneumococci using eBURST (Feil et al., 2004). Figure 3 shows the largest clonal complex presently to be found in the MLST database, which descends from ST 156 (itself the Spain^9V-3 PMEN clone). Analyses of this nature must be undertaken with caution due to known biases in the submission of strains to the database, and the potential for recombination to obscure the history of a group of strains (Turner et al., 2007). Nevertheless, those clones that are closely linked by eBURST indisputably share a recent common ancestor, even if their precise relationships are difficult to discern. With this in mind it is interesting to note that no fewer than 9 of the 27 recognized PMEN clones associated with resistance are to be found in this single clonal complex. Moreover, five of these clones, all originally noted with a 6B capsule, are evident in a relatively small cluster towards the right of the diagram.

Figure 3.

The largest clonal complex in the S. pneumoniae MLST database (CC156), accessed on 6/8/2010 and visualized using eBURST. Resistant PMEN clones are highlighted in dark pink, and their ST indicated. A cluster of 6B clones is evident at right. The pbp profiles and erm/mef data are drawn from the PMEN website at http://www.sph.emory.edu/PMEN/.

Given that these five are clearly closely related, we should ask whether they are genuinely five clones with separate origins, or one clone, which has subsequently diversified. The close relationship argues in favour of the latter interpretation, yet the pbp alleles also recorded by PMEN are distinct, suggesting that they are the product of independent events (Fig. 3).

This is also evident considering another clonal complex. CC15 contains three PMEN clones (Fig. 4) all of them associated with serotype 14. Again, it appears that distinct resistance profiles have emerged from this clonal complex on three occasions. In reality, these must be considered minimal estimates, as this analysis is based on the published profiles of the PMEN clones, which may not represent all the resistance phenotypes to be found in that clone in nature. A recent project using whole genome sequencing to probe the evolution of the PMEN 1 clone has produced evidence that this lineage has acquired macrolide resistance, due to both mef and erm elements, on multiple occasions (Bentley SD, Personal communication, 2010).

Figure 4.

The 3rd largest clonal complex in the S. pneumoniae MLST database (CC15), accessed on 6/8/2010 and visualized using eBURST. Resistant PMEN clones are highlighted in dark pink, and their ST indicated. All three PMEN clones in this CC are serotype 14. The pbp profiles and erm/mef data are drawn from the PMEN website at http://www.sph.emory.edu/PMEN/.

Figure 5 shows the location of the resistant PMEN clones in the entire MLST database. As noted, they are concentrated in a few lineages. Given that resistance loci seem so ‘easy’ to obtain, it is hard to explain why this should happen in some lineages and not others, unless the cost of acquiring resistance is not constant across the pneumococcal population. This possibility is discussed further below.

Figure 5.

Population snapshot of the entire S. pneumoniae MLST database accessed on 6/8/2010 and visualized using eBURST. Resistant PMEN clones are highlighted in dark pink. CC 156 and CC 15 are indicated

Pneumococcal vaccination: old clones, new serotype

The development of a conjugate vaccine against seven pneumococcal serotypes has been a watershed in the control of pneumococcal disease. The vaccine is highly effective against invasive disease of vaccine serotype (Black et al., 2000), and with a 50% efficacy against carriage of those serotypes (Ghaffar et al., 2004). In communities with high coverage, the targeted serotypes have dwindled to the point where they are at a tiny fraction of their prior prevalence, in carriage and disease. Because these serotypes were previously associated with the majority of antibiotic resistance, their decline has had concomitant benefits in this regard (Kyaw et al., 2006).

If we assume that vaccine type strains are sufficiently disadvantaged that they have no long term future in vaccinated communities, there are two means by which any resistance loci associated with them might persist through recombination: either resistant vaccine type strains might generate escape variants through serotype switching to gain a new non-vaccine serotype, or the resistance loci themselves might be transferred into a new non-vaccine type background. Several cases of the former have been identified, and are discussed below. In the US resistant non-vaccine type (NVT) strains have been increasingly common following vaccination (Farrell et al., 2007; Gertz et al., 2010; Hanage et al., 2007; Beall et al., 2011)

Following PCV-7 introduction in the US in 2000, the Centers for Disease Control Active Bacterial Core Surveillance has undertaken surveillance of the pneumococcal population causing invasive disease in order to examine the impact of the vaccine. The results form a remarkable body of work, which can only be briefly described here for reasons of space. We will discuss three PMEN clones that were common causes of invasive disease prior to vaccination, and were also associated with antibiotic resistance: Spain^23F-1, Spain^9V-3 and Taiwan^19F-14 (McGee et al., 2001). Using MLST these and related strains may be assigned to ‘clonal complexes’ (CCs). The respective CCs for these clones are CC 81, CC 156 and CC 236.

All three clones have persisted into the post-vaccine population with a new serotype not targeted by the vaccine, and in each the predominant serotype is 19A. The most successful and clinically significant is ST 320 (Beall et al., 2006; Beall et al., 2011; Brueggemann et al., 2007; Pai et al., 2005b; Pelton et al., 2007; Techasaensiri et al., 2010), which is a close relative of ST 236, Taiwan^19F-14. In addition to an important role in invasive pneumococcal disease (IPD), ST 320 is increasingly common in carriage (unpublished observations). It maintains resistance to multiple antibiotic classes. As noted above, pneumococci are a frequent cause of otitis media, and a variant of the Spain^9V-3 clone (ST 2722) has been documented which exhibits resistance to all FDA approved antimicrobials for treatment of acute otitis media (AOM) (Pichichero & Casey, 2007a). Alongside the 19A variant, 11A and 15C capsules have been found among isolates closely related to this clone (Xu et al., 2009b), illustrating that serotype switching has already produced escape variants which will not be covered by the next generation of conjugate vaccines. While 19A variants of the Spain^23F-1 clone (ST 81) have been found, and indeed considerably predate the introduction of PCV-7 (Coffey et al., 1998), they do not seem common. Despite this, preliminary results of a whole genome analysis suggest that the 23F to 19A switch has occurred at least three times (Croucher et al., 2011) again underlining the capacity of resistant pneumococci to evade vaccine pressure.

In at least one case in which a successful non-vaccine type version of a PMEN clone has emerged (a 23A variant of the Columbia^23F-26 clone (Pai et al., 2005a) it is clear that the vaccine escape clone was already present before vaccination. In most other cases it is unknown whether this was the case. It may be important to note that while ST 320 is a close relative of ST 236, it was not particularly common in the US with a 19F capsule prior to vaccine introduction (Gertz et al., 2003). As such it is likely to represent an import from elsewhere.

Alongside the cases of serotype switching from well-known PMEN clones, antimicrobial resistance has been increasing among NVT strains that were not previously associated with vaccine serotypes. Among these, the 19A ST 199, the 35B ST 558 and the 6C ST 1379, are particularly noteworthy (Gertz et al., 2010; Hanage et al., 2007; Richter et al., 2009).

At the time of writing, conjugate vaccines against 7 to 13 serotypes are being implemented around the world. As a result the pneumococcal population, and pneumococcal population biology, are in a state of flux. PCV-7 was introduced into the US in 2000. In the decade since then many formerly prominent vaccine type clones have almost disappeared in the face of vaccine pressure. Others have persisted through serotype switching, and yet others have emerged from obscurity. Given differences in vaccine availability, vaccination schedules, and the pneumococcal population in different parts of the world, it would be foolish to make hard and fast predictions as to which clones will be important in the future.

The survival of lineages such as ST 320, and ST 2722 through serotype switching should teach us a sobering lesson in the potential of this pathogen to respond to diverse selective pressures. When considering which clones are at high risk for dissemination of resistance, we should at least consider the possibility that some clones are a higher risk than others, not as a result of resistance itself, but as a result of their ability to acquire it.

Why resistance in S. pneumoniae?

A survey of the literature can make it seem that resistant pneumococcal isolates are overwhelmingly common, but this reflects our clinical focus. No matter how interested we might be in resistant clones, and for good reasons, the majority of pneumococci remain susceptible to all classes of antibiotics. Moreover, as shown in figures 3–5, the PMEN clones and resistance to multiple antibiotics are concentrated in a few lineages. Unlike S. aureus, pneumococcal disease is more commonly acquired in the community than in health care settings, and so it is hard to suggest that multiple resistance represents adaptation to a specific niche, or if it is, the niche in question is not so well defined.

One recent proposal is that the probability of recombination is not constant throughout the pneumococcal population (Hanage et al., 2009). If some pneumococci are more likely to undergo recombination than others, then they are more likely to acquire resistance determinants. This can be seen as an idea similar to the proposal that bursts of hypermutation can allow bacteria to scale local fitness peaks. In both cases, it is thought that maintaining a high rate of recombination or mutation is detrimental in the long term, but for a short period it may be adaptive. There is some evidence for this in the observation that strains with MLST loci that are identified as having been imported from a distinct and divergent genetic background (ie through recombination), are significantly more likely to be resistant to antibiotics (Hanage et al., 2009).

Streptococcus pyogenes

S. pyogenes is a human-specific pathogen that causes disease throughout the world, ranging in severity from mild superficial infections at the epithelium of the throat or skin (pharyngitis, impetigo), to life threatening invasive disease (egs., bacteremia, toxic shock syndrome, necrotizing fasciitis). There are an estimated 730 million cases of pharyngitis and impetigo per year caused by S. pyogenes (Carapetis et al., 2005). In addition, S. pyogenes can colonize the throat in the absence of disease; carriage rates can exceed 30% in school-aged children (Kaplan, 1980). Included among the important ecological features of S. pyogenes is its narrow range of biological hosts (i.e., humans), high prevalence of colonization and disease throughout the world, and direct person-to-person transmission.

Resistance development in S. pyogenes

S. pyogenes stands apart from most other gram positive cocci pathogens in that it has not acquired natural resistance to the β-lactam class of antibiotics. Penicillin treatment failure is sometimes attributable to co-colonization with a β-lactamase-producing bacterium of another species (eg., S. aureus) (Quie et al., 1966; Brook & Gober, 2008). “Tolerance” of S. pyogenes to penicillin has been sporadically observed; however, the mechanistic underpinnings have yet to be explained (Pichichero & Casey, 2007b). Why S. pyogenes has (luckily) failed to evolve resistance to this drug remains a mystery (Horn et al., 1998). The lack of β-lactamase acquisition by S. pyogenes is difficult to explain by genetic barriers alone because they share with staphylococci other highly homologous resistance genes carried by mobile genetic elements. The failure of S. pyogenes to evolve genes encoding penicillin-binding proteins having a lower binding affinity for the drug might be explained by reduced fitness (eg., impaired peptidoglycan biosynthesis) that exacts a large biological cost. β-lactams remain the antibiotics of choice for the treatment of most infections due to S. pyogenes.

Macrolides are the drug of choice for patients with β-lactam allergies or a prior treatment failure. Clindamycin (a lincosamide) is preferred for patients with life-threatening soft tissue infections such as toxic shock syndrome or necrotizing fasciitis because it has good tissue penetration and quickly halts synthesis of the exotoxin that is responsible for much of the disease pathology. Clinically significant macrolide resistance in S. pyogenes emerged during the mid-1970s and has gradually become widespread (Seppala et al., 1997; Gerber, 1996; Perez-Trallero et al., 1999; Yan et al., 2000; Cha et al., 2001; Espinosa de los Monteros et al., 2001; Syrogiannopoulos et al., 2001; Cresti et al., 2002; Martin et al., 2002; Alos et al., 2003). In local surveys, resistance rates sometimes exceed 30% and can climb even higher. Importantly, there is a strong correlation between macrolide consumption and levels of resistance among S. pyogenes (Fujita et al., 1994; Seppala et al., 1997; Freeman & Shulman, 2002; Garcia-Rey et al., 2002; Albrich et al., 2004; Gagliotti et al., 2006; Hsueh et al., 2006).

Resistance to macrolides in S. pyogenes is mostly attributable to the erm(A), erm(B) and mef(A) genes; less common determinants are other forms of erm and mef genes and ribosomal mutations. The erm gene products lead to ribosome modification and the MLS phenotype that includes resistance to lincosamides, whereas the mef gene confers macrolide efflux and the M phenotype, resulting in resistance to macrolides but susceptibility to lincosamide and streptogramin B antibiotics (Sutcliffe et al., 1996). Mobile genetic elements harboring one of the macrolide resistance genes, sometimes coupled with a gene for tetracycline resistance, have been well characterized (Banks et al., 2004; Beres & Musser, 2007; Varaldo et al., 2009; Brenciani et al., 2010). In common with the above sections dealing with enterococci and pneumococci, many of the genetic elements are present in other streptococcal species, indicative of a larger network for interspecific movement of resistance genes.

Assigning S. pyogenes clones based on sequenced-based emm typing and MLST

The S. pyogenes population displays a very high level of genetic diversity (Bessen, 2009). The hair-like surface fibrils, known as M proteins, are targets of a protective host immune response; organisms undergo immune escape via diversifying selection of emm genes. M proteins provide the basis for classical serotyping of S. pyogenes. More recently; this scheme was replaced with emm sequence-based typing (www.cdc.gov/ncidod/biotech/strep/strepindex.htm) (Beall et al., 1996) and today, emm typing is almost universally used for molecular analysis of S. pyogenes isolates obtained via population-based epidemiological surveillance, with > 200 emm types identified.

In a large multicenter study of pediatric cases of pharyngitis in the USA over 3 years (2000–2003), ~3,000 S. pyogenes isolates were obtained and of these, 4.1% were macrolide resistant; 70% of resistant isolates harbored the mef(A) gene (Tanz et al., 2004). The most prevalent emm type among the macrolide resistant isolates was emm12, followed by emm75 and emm4, which together accounted for more than half of the resistant isolates; the emm75 isolates all harbor erm(A), whereas emm12 and emm4 isolates are associated with > 1 resistance gene type, indicative of distinct clones. Two parallel but independent studies show dominance of emm75, emm12, emm4 and/or emm58 genotypes among macrolide resistant S. pyogenes isolates, with overall resistance rates of ~6% (Richter et al., 2005; Green et al., 2006). Other dominant emm types associated with macrolide resistance in local surveys include emm1, emm2, emm6, emm22, emm77, and emm89 (Jasir et al., 2001; Dicuonzo et al., 2002; Martin et al., 2002; Katz et al., 2003; Zampaloni et al., 2003; Creti et al., 2007; Chan et al., 2009; Michos et al., 2009; Meisal et al., 2010).

For many strains, emm type does not correlate with clone (Enright et al., 2001; Bessen et al., 2008). Clones of S. pyogenes are more precisely defined by a combination of emm type and sequence type (ST) based on alleles at multiple housekeeping loci. To date, MLST of S. pyogenes reveals 485 STs, organized into 73 clonal complexes with 207 singleton STs, as predicted by eBURST (http://spyogenes.mlst.net/).

For macrolide-resistant S. pyogenes, the combination of emm type, ST and resistance gene type provides a good working definition of clone. The MLST database, which is publicly available, contains genotyping data for > 400 macrolide resistant isolates, deposited by numerous investigators (Perez-Trallero et al., 2004; Reinert et al., 2004; Szczypa et al., 2004; Littauer et al., 2006; Montes et al., 2006; Robinson et al., 2006; Silva-Costa et al., 2006; Perez-Trallero et al., 2007; Strakova et al., 2007; Silva-Costa et al., 2008). According to the MLST database (http://spyogenes.mlst.net/), 100 unique combinations of ST, emm type and resistance gene type [erm(A), erm(B) and/or mef] can be identified. Based on analysis of single locus variants (SLVs), as many as 73 genotypes may represent independent acquisitions of a resistance gene, whereas 27 of the clones may have arisen from another resistant clone via diversification of housekeeping alleles (Robinson et al., 2006). Thus, the population of macrolide-resistant S. pyogenes has extensive diversity, whereby new resistant clones have emerged on a wide variety of genetic backgrounds.

Phylogeography of antibiotic resistant S. pyogenes clones

The geographic location from which a macrolide resistant clone was isolated can provide the basis for estimating the relative prevalence of individual clones, whereby a wide geographic distribution for a clone is an indicator of its transmission success. Of the fully genotyped macrolide-resistant S. pyogenes in the current MLST database (http://spyogenes.mlst.net/), isolates were recovered from Europe (70.5%), North America (12.8%), Asia (11%), South America (3.2%), Oceania (1.4%), and Africa (1.1%); overall, 35 countries are represented. Only one clone [emm94/ST89/erm(A)] was recovered from all six major continents (Table 1). Other widely dispersed clones, each recovered from five continents, are emm4/ST39/mef(A) (17 countries), emm12/ST36/mef(A) (16 countries) and emm58/ST176/erm(A) (10 countries). The emm4/ST39/mef(A) clone is the predicted founder of a clonal complex having five additional STs, whereby all clones harbor emm4 and mef(A); thus, the emm4/ST39/mef(A) clone has diversified at multiple housekeeping loci, perhaps signifying that it is among the oldest and most successful of macrolide resistant S. pyogenes clones. All three resistance gene types are represented among widely dispersed clones of macrolide resistant S. pyogenes, and two emm-ST genotypes are associated with > 1 resistance gene type (emm12-ST36 and emm22-ST46) (Table 1). The oldest clone in the MLST collection is emm12/ST36/erm(B), recovered in Canada in 1976. Within more localized outbreaks, there is often a single highly prevalent strain, such as emm11/ST43/erm(B) in Gipuzkoa, Spain (Perez-Trallero et al., 2007) and emm6/ST382/mef(A) in Pittsburgh, USA (Martin et al., 2002; Banks et al., 2003).

Table 1.

The 12 most widely dispersed clones of macrolide resistant S. pyogenes.

emm type	Resistance gene	ST	No. of countries	No. of continents	SLVs sharing the same emm type and resistance gene as the founder clone
94	erm(A)	89	12	6	None
4	mef(A)	39	17	5	ST38,373,421,422,423
12	mef(A)	36	16	5	None
58	erm(A)	176	10	5	None
12	erm(B)	36	9	3	None
22	erm(B)	46	8	3	None
73	erm(A)	331	5	3	None
1	erm(B)	28	5	3	None
75	mef(A)	49	5	3	ST388,548
22	mef(A)	46	3	3	ST389
28	erm(B)	52	11	2	ST244
77	erm(A)	63	10	2	ST369

Nine of the 10 emm types (all except for emm1) belonging to the widely dispersed, macrolide-resistant strains (Table 1) are typically associated with the sof gene, which lies ~10 kb upstream from the emm locus and is present in ~50% of strains. Sof encodes a multifunctional surface protein that can bind human fibronectin (note - the sof gene may not encode a fully functional product in emm12 strains). A strong correlation between the sof gene and macrolide resistance in S. pyogenes was previously noted (Dicuonzo et al., 2002). Perhaps related to the strong association with sof is the finding that erm-positive strains bear significant positive correlations with other fibronectin-binding proteins and the ability to invade epithelial cells, and negative correlations with biofilm formation (Facinelli et al., 2001; Baldassarri et al., 2007; Hotomi et al., 2009), in at least some studies.

Fluoroquinolone resistance has recently emerged among S. pyogenes, due to mutations in the parC and/or gyrA locus. A highly prevalent, non-susceptible emm6/ST382 clone was identified in numerous studies in many parts of the world, although most isolates appear to have low-level resistance (Alonso et al., 2005; Orscheln et al., 2005; Powis et al., 2005; Yan et al., 2008; Malhotra-Kumar et al., 2009; Smeesters et al., 2009; Montes et al., 2010; Pires et al., 2010). Other emm types associated with fluoroquinolone resistance include emm1, emm5, emm28, emm75 and emm89. The close relative of S. pyogenes - Streptococcus dysgalactiae subspecies equisimilis - may be a source for some parC and gyrA mutations (Pinho et al., 2010).

It is unusual for tetracycline to be administered for the treatment of a S. pyogenes infection, however, the widespread use of this antibiotic in general creates strong selection pressures for the emergence of a resistant clone. Indeed, in one study on a genetically diverse set of S. pyogenes strains, tetracycline resistance is estimated to have been acquired in > 80 independent genetic events (Ayer et al., 2007). Because of its minimal clinical impact, relatively little strain typing has been done on tetracycline-resistant S. pyogenes, and it remains largely unknown as to which tetracycline-resistant clones are among the most prevalent. The majority of tetracycline resistance in S. pyogenes is conferred by the tet(M) or tet(O) genes; these genes have been found within the same mobile genetic element as erm or mef genes (Varaldo et al., 2009; Brenciani et al., 2010). Because resistance to tetracycline is often more prevalent among S. pyogenes as compared to macrolide resistance, it follows that tetracycline usage for other diseases may facilitate the acquisition of macrolide resistance via co-inheritance of resistance genes (Nielsen et al., 2004).

Since macrolide consumption in the community is a strong predictor of the prevalence of macrolide resistant strains, the resistant clones that have come to dominate may have emerged simply by being in the right place at the right time. A meta-analysis of >38,000 S. pyogenes global isolates reveals that the most common emm types (in decreasing order) are emm1, emm12, emm28, emm3 and emm4 (http://www.cdc.gov/ncidod/biotech/strep/strepindex.htm) (Steer et al., 2009); with the exception of emm3, the highly prevalent emm types are also associated with macrolide resistance (Table 1). However, the opposite does not always hold true. The most widely dispersed macrolide-resistant clone is emm94; this emm type is relatively rare throughout the world when all GAS isolates - susceptible and resistant - are considered. Thus, the finding on emm94 clones is consistent with the idea that, at least in this one instance, macrolide resistance coincides with a large leap in transmission success.

Staphylococcus aureus

S. aureus asymptomatically colonises the skin and anterior nares (nostrils) of approximately one third of the human population at any given point in time (Peacock et al., 2001; Nulens et al., 2005; van Belkum et al., 2006). Although primarily a commensal species, asympomatic nasal carriage is known to be associated with infection (Deleo et al., 2010), which leads to a range of conditions from boils to life-threatening endocarditis. S. aureus is traditionally considered a nosocomial pathogen, as serious infections are more common within health-care settings, such as hospitals and nursing homes, than in the community. However, S. aureus is not limited to colonizing humans, and can cause infection in livestock and companion animals including cows, sheep, goats, buffalo, pigs and chickens (Aires-de-Sousa et al., 2007; Vanderhaeghen et al., 2010; van Belkum et al., 2008; Lowder et al., 2009). Although the true ecological range in non-human hosts is not known, there is increasing awareness of the possible risks of zoonotic transmission from livestock (Catry et al., 2010).

Disease management within health-care settings and, increasingly, within the community is substantially hampered by the rapid spread of resistance to β-lactam antibiotics. The first penicillinase producing S. aureus strains isolated from hospitalized patients were observed in 1942, just two years after penicillin was first used to treat wounded soldiers, and reported in 1944 (Kirby, 1944). The frequency of pencillin resistance rose rapidly after the World War II (Barber & Rozwasowska-Dowzenko, 1948); by 1947, 25% of strains from hospitalized patients were resistant, and this figure rose to >80% by the late 1950s (Jessen et al., 1969). For reasons that remain unclear, this frequency then levelled off, and has remained broadly consistent at 80% up to the present day. Resistance to penicillin in the community followed a similar trajectory but with an approximate 20 year lag. By the late 1950s to early 1960s, ~ 25% of isolates recovered from sporadic community-acquired infection were resistant to penicillin, by the mid 1970s this figure had risen to 70–85% (Chambers, 2001).

MRSA, emergence of a “superbug”

Methicillin was first introduced in 1959, but the first methicillin resistant S. aureus (MRSA) isolates were reported only two years later (Barber, 1961; Jevosn & Parker, 1964). Resistance to methicillin is conferred via the horizontal acquisition of a large (20–60Kb) chromosomal cassette (SCCmec), which is thought to have been introduced from naturally resistance commensal Staphylococcal species on multiple occasions (Enright et al., 2002). Different “types” of SCCmec element have been described, and the origin and evolution of these elements has been extensively discussed elsewhere (Oliveira & de Lencastre, 2002; Chongtrakool et al., 2006; Hanssen & Ericson Sollid, 2006; Deurenberg & Stobberingh, 2008; Deurenberg & Stobberingh, 2009). The increase of MRSA appears to be generally following a parallel trend to that observed previously for penicillinase producing strains; resistant strains initially gained a foothold within health-care settings, then over time resistance has also subsequently become an increasing problem in the community (Chambers, 2001; Deleo et al., 2010; Chambers & Deleo, 2009; Deleo et al., 2010).

The problems associated with methicillin resistance vary considerably across Europe and other parts of the world (Tiemersma et al., 2004). Data from the European Antimicrobial Resistance Surveillance System (EARSS; http://www.rivm.nl/earss/) for 2008 show the prevalence of MRSA from all cases of invasive disease ranging from <1% (Scandinavian countries and The Netherlands), to >50% (Portugal and Malta). Although the reasons behind this regional variation are not precisely known, differences in antibiotic usage and in the effectiveness of local infection control strategies in health care settings are likely to be important.

S. aureus has also evolved resistance to the glycopeptides vancomycin and teicoplanin, which are among the most important drugs for treating MRSA infections. These resistant strains emerged in the 1980s by two distinct mechanisms. First, chromosomal mutatations have been identified which confer a moderate increase to glycopeptides by increasing the thickness of the cell wall. (VISA/GISA phenotypes; (Howden et al., 2010)). More marked gycopetide resistance is noted in the VRSA strains, first reported in 2002 (Sievert et al., 2008). These strains have evolved though the acquisition of a van gene, which results in the synthesis of a modified peptidoglycan precursor. The prevalence of both VISA/GISA and VRSA strains have remained very low, probably due to the fitness costs associated with these genetic changes.

The newest drugs to treat MRSA are linezolid, daptomycin and tigecycline, and these remain broadly effective against VISA and VRSA strains. Although very rare, mutations in the 23S rRNA gene have been identified which reduce susceptibility to linezolid, and increasingly so as more gene copies are mutated (Besier et al., 2008). Perhaps more worringly, linezolid resistance can also be acquired horizontally through the acquisition of the cfr gene which encodes a protein which modifies (methylates) the target ribosomal gene (Long et al., 2006; Locke et al., 2010). Cfr-positive strains have recently caused nosocomial outbreaks in Spain (Sanchez Garcia et al., 2010; Morales et al., 2010). Further, this modification confers cross-resistance to phenicols, streptogramin A, lincosamides, oxazolidinones and pleuromultilins, which raises the problem of co-selection in both medical and veterinary settings where anti-ribosomal agenets may be used. It is noteworthy in this context that the plasmid harbouring the cfr gene was first noted in a strain of animal origin (Kehrenberg & Schwarz, 2006). There is also evidence for a link between reduced daptomycin susceptibility and the VISA phenotype (Wootton et al., 2006), possible related to changes in the thickness of the cell wall (Cui et al., 2006), although the potential clinical impact of this remains equivocal (Nannini et al., 2010).

Typing schemes used for S. aureus

Several typing methods have been used for epidemiological surveillance of this species, including pulse-field gel electrophoresis (PFGE), MLST (Enright et al., 2000; Cookson et al., 2007) and MLVA (multiple loci VNTR analysis, where VNTR stands for variable number of tandem repeats) (Melles et al., 2009; Schouls et al., 2009). VNTR loci are hyper-variable microsatellites, the most notable example in S. aureus being the spa gene which has been used extensively for typing studies in this species (Mellmann et al., 2008; Basset et al., 2009). Typing techniques based on variation within the different SCCmec elements is also widely used for differentiating different MRSA clones (Oliveira & de Lencastre, 2002; Chongtrakool et al., 2006). Whilst these methods present a range of utility for more detailed evolutionary analyses, they (more or less) consistently delimit the S. aureus population in to the same discrete clusters, or clonal complexes. There are exceptions however, such as agr groups (Robinson et al., 2005), and SCCmec typing. The latter approach is based on a highly mobile accessory element so does not always accurately reflect the genetic background (Nubel et al., 2008). Ultimately, a pluralistic approach of different approaches (e.g. MLST + spa typing + SCCmec typing), along with as much appropriate metadata as possible, provides the optimal utility (van Belkum et al., 2007).

The low rate of homologous recombination in S. aureus (Feil et al., 2003) is thought to explain in part how (if not why) the discrete clusters have emerged and been maintained. The consistent delineation of the same clonal complexes, even from gene content data generated using microarrays (Lindsay et al., 2006), underline that they are real biological entities, representing different fitness peaks, and as such are both of evolutionary interest and of epidemiological utility (Feil et al., 2003; Turner & Feil, 2007). These clusters can be visualised using UPGMA dendrograms or eBURST (Feil et al., 2004) (eburst.mlst.net). Extensive typing studies have revealed that hospital-acquired (HA-) MRSA isolates are particularly clonal, in fact since the 1960s only 5 clonal complexes (CCs 5, 8, 22, 45 and 30 can account for the vast majority of nosocomial infection world-wide (Crisostomo et al., 2001; Aires de Sousa et al., 2005; Gomes et al., 2006; Conceicao et al., 2007). These five clonal complexes are visualized using eBURST in Figure 6.

Figure 6.

The five major HA-MRSA containing clonal complexes as visualised by MLST/eBURST (CCs 45, 5, 8, 30, 22 shown in red). Names given to MRSA clones corresponding to highlighted STs are given in parentheses. Each circle is an ST. Linked STs differ at only one locus out of 7. The size of the circle reflects the frequency of the ST. Blue circles represent clonal founders, yellow circles are sub-founders. Full details concerning eBURST can be found at eburst.mlst.net.

Hospital and community acquired MRSA: Clinical, epidemiological and genotypic comparisons

Whereas HA-MRSA isolates have evolved from a small number of globally disseminated clonal groups, community-acquired (CA-) methicillin sensitive S. aureus (MSSA) and CA-MRSA isolates are more diverse than HA-MRSA. A recent impressive study by the European Staphylococcal Reference laboratory Working Group, led by Hajo Grundmann, characterized 2,890 MSSA and MRSA isolates from 26 European countries by spa typing (Grundmann et al., 2010). These data revealed fundamental differences in the geographic structure and diversity of MRSA and MSSA. The MSSA spa types were uniformly diverse and homogenously mixed throughout Europe, such that a large proportion of the overall variation was recovered on a local scale. Such a pattern is indicative of an endemic pathogen which has become established in a population over a long period of time. In contrast substantial geographic structure was apparent in the spa data for the MRSA isolates, such that specific types tend to be predominantly observed in particular countries, and only a small fraction of the overall diversity is observed on a local scale. This structuring was particularly apparent for the HA-MRSA isolates, and illustrates how these clones have spread through recent and rapid epidemic expansions through public health networks on a local scale.

In addition to being distinct from MSSA, HA-MRSA have historically been genetically and epidemiologically distinct from CA-MRSA, although these distinctions are now being eroded. The earliest HA-MRSA clones, such as those corresponding to ST239 (UK EMRSA-1, -4, -7, -11, the Brazilian, Portuguese and Viennese clone)are multiply resistant due to the acquisition of the large type III SCCmec element, and the rarity with which they are observed outside of health-care settings points to their adaptation to a hospital environment. In contrast, CA-MRSA isolates are more likely to be susceptible to non-β-lactam antibiotics as they harbour smaller SCCmec elements (e.g. type IV) (Feng et al., 2008; Deurenberg & Stobberingh, 2009). However, these smaller cassettes may impose a smaller fitness burden, both in vitro and in vivo (Okuma et al., 2002; Lee et al., 2007; Diep et al., 2008). Thus, to use an analogy, if HA-MRSA is burdened by a suit of armour, CA-MRSA can be thought of as wearing a bullet-proof vest, as a means of optimizing the trade-off between protection and competitiveness. For an excellent and comprehensive review of the distinct “waves” of MRSA infection, see (Chambers & Deleo, 2009).

Whilst ST239 remains the most common HA-MRSA clone globally, it has long since been replaced by other clones in Western Europe. Currently disseminated HA-MRSA clones in Europe include ST22 (EMRSA15), ST36 (EMRSA16), ST225 and ST228. These are less multiply resistant than the earliest HA-MRSA, showing resistance mostly to β-lactams, fluoroquinolones, and Macrolide-Lincosamide-Streptogramin (MLS), and have more commonly acquired SCCmec elements other than the large stereotypical HA-MRSA types I-III. The resistance profiles of these currently circulating HA-MRSA clones are broadly similar to the most common CA-MRSA clones, such as the epidemic USA300 (ST8), and the clones making up the more hetergenous population of CA-MRSA currently circulating in Europe. The most common European CA-MRSA clone is currently ST80-IV (the European clone) (Otter & French, 2010) which has been reported throughout central Europe (Vandenesch et al., 2003; Denis et al., 2005; Witte et al., 2005) Scandanavia (Hanssen et al., 2005; Fang et al., 2008) and Greece (Katopodis et al., 2010). This clone is characteristically resistant to fusidic acid, tetracycline, kanamycin and variably resistant to ciprofloxacin. It is PVL-positive and possibly originated from the Middle East or North Africa (Goering et al., 2009; Otter & French, 2010). Other CA-MRSA clones of note circulating in Europe include the PVL-positive ST152 clone (Perez-Roth et al., 2010) which is circulating throughout central Europe and the Balkans. The high frequency of ST152 in asymptomatic carriage in Mali (Ruimy et al., 2008) and from clinical isolates in Nigeria (Okon et al., 2009) points to sub-Saharan Africa as the possible origin of this clone, a scenario which points to the acquisition of an SCCmec element into an already PVL-positive background (Perez-Roth et al., 2010).

Typically there are predisposing risk factors for HA-MRSA infection. Although risk factors such as socioeconomic status and drug usage are clearly implicated in CA-MRSA (Bratu et al., 2006), these infections commonly cause serious skin and soft tissue infections in otherwise healthy individuals (Young et al., 2004). In the United States, CA-MRSA isolates recovered from sporadic cases of disease before 2001 corresponded to CC1 (USA400). This group has subsequently been largely replaced by USA300 (CC8), which is currently the leading cause of CA bacterial infections in this country (Kennedy et al., 2008; Miller & Diep, 2008). USA300 isolates have acquired elements which have been suggested to play a key role in virulence, notably the prophage encoded toxin, Panton-Valentine Leukocidin (PVL) and the arginine catabolic mobile element (ACME) (Diep et al., 2006; Diep et al., 2008). However, it has recently been reported that the USA300 lineage (and its progenitor lineage USA500) acquired increased virulence through changes in expression of core virulence genes, a finding which challenges the paradigm that virulence primarily results from the acquisition of new genes (Li et al., 2009)). USA300 has also acquired increased resistance, and has begun to spread outside of the USA (Tenover & Goering, 2009). In regions where cases of CA-MRSA are high, such as Taiwan and the USA, an increasing number of hospital-acquired infections are being caused by typical CA-MRSA clones such as USA300 (Klevens et al., 2006; Huang et al., 2007), a clear sign that the epidemiological distinction between HA-MRSA and CA-MRSA is beginning to break down (Deurenberg & Stobberingh, 2008).

Livestock-associated MRSA, a new threat?

In addition to HA-MRSA and CA-MRSA, recent attention has begun to focus on livestock-associated (LA-) MRSA. Of particular note is ST398 which is a common LA-MRSA associated with pigs. First reported in 2005, the distribution of ST398 has hitherto largely been restricted to Holland, Belgium, Northwest Germany and Denmark, although this clone has started to spread to other parts of the world (Golding et al., 2010). In addition to SCCmec, LA-MRSA ST398 have also typically acquired resistance to tetracycline, an antibiotic intensively used in pig farming. Nasal colonization of ST398 has been recorded in farmers and veterinarians, who come into close contact with pigs (Wulf & Voss, 2008), although the public health risk associated with this clone is unclear as a recent report has suggested it is poorly transmissible within the hospital environment (Wassenberg et al., 2010). Interestingly, a high frequency of asymptomatic paediatric carriage of MSSA ST398 has been noted in China (Fan et al., 2009), raising the possibility that a susceptible predecessor of this clone was imported from this country.

Next-generation sequencing illuminates the microevolution and transmission of MRSA clones

The current typing schemes have been successful in revealing the changing frequencies of the clusters over time and space (over years and decades, and on both national and international scales). However, these datasets still tend to lack the discrimination required for tracking strain transmission patterns on local scales (within and between hospitals) over short time scales (weeks and months). This is because a very small number of genotypes tend to predominate at a given location at any given point in time, a result of sequential waves of infection (de Lencastre et al., 2007). Thus more powerful techniques are required to resolve sufficient variation within clonal complexes. Nubel et al addressed this using a mutation discovery procedure to identify SNPs in ~45.5Kb (1.6% of the genome) in 135 S. aureus CC5 isolates from 22 countries (Nubel et al., 2008). These data pointed to multiple independent acquisitions of the SCCmec element, even within a single clonal group. The authors argued against the prevailing model of the rapid global dissemination of a few MRSA strains, in favour of a model of frequent emergence of “home-grown” locally restricted MRSA clones, some of which happen to all belong to the same globally disseminated clone. Such a model is consistent with the strong geographic structuring of MRSA across Europe (Grundmann et al., 2010)). A more recent paper using similar methodology focused on a single sub-cluster within the broader CC5 group, called ST225 (Nubel et al., 2010). The universal presence of a defining deletion within SCCmec showed that this cluster corresponds to a single acquisition of this element, and the authors were able to posit that this sub-group was introduced into central Europe from the USA approximately a decade ago, and has subsequently spread rapidly between hospitals.

The first clone to be characterized on a genome-wide basis using next-generation sequencing was ST239 (Harris et al., 2010). All ST239 isolates are resistant to methicillin (MRSA) and possess the large type III SCCmec cassette which confers multiple resistance. ST239 is responsible for ~90 % of hospital-acquired MRSA infection throughout most of mainland Asia (from the Middle-East to China), and much of South America (Diekema et al., 2001; Chongtrakool et al., 2006; Xu et al., 2009a). ST239 was also the predominant MRSA clone in Western Europe during the 1980s and 1990s, although it has subsequently been largely replaced by other strains (Conceicao et al., 2007) it is still circulating in Eastern Europe. Harris et al characterized 62 ST239 isolates, 42 of which were globally representative, whilst the remaining 20 were recovered from patients in a single hospital in northeast Thailand over a 7 month period (Harris et al., 2010). The study was thus designed to address both the global diversity of this clone and the utility of next-generation sequencing for very localised epidemiology.

These data point to a European origin of ST239 in the mid 1960s, and a single introduction of ST239 into South America followed by dramatic clonal spread. The data also confirmed that a wave of infection caused by ST239 in Portugal in the late 1990s was sparked by a transmission event from South America. More surprisingly, the data pointed to a South East Asian origin of the sequenced reference strain (TW20) which recovered from an outbreak in a London hospital (Edgeworth et al., 2007; Holden et al., 2010). Most notably, however, the study supports the possibility that this approach could inform on transmission chains even at the scale of a single hospital, which has clear implications for infection control (Harris et al., 2010). These data also point to an unusually high transmissibility of this clone within and between hospitals. When combined with multiple resistance and high virulence (Amaral et al., 2005; Edgeworth et al., 2007), it is highly likely that ST239 currently represents a greater public health burden globally than any other MRSA clone (Smyth et al., 2010).

By synthesizing the data from the recent studies discussed, the dynamics of HA-MRSA clones are clearly characterized by occasional international or intercontinental transmission combined with subsequent rapid localized clonal expansion. However, the relative importance of the de novo emergence of new HA-MRSA variants over transmission into a locale from elsewhere remains contentious, and may even vary between clonal lineages. Whereas the data of Nubel point to rapid and recurrent acquisitions of SCCmec variants within a single clonal group (Nubel et al., 2008), the data of Harris et al point to a stable association between a type III SCCmec type and ST239, a globally disseminated global lineage, over at least four decades (Harris et al., 2010). Although sub-variants are detected within the type III SCCmec, this observation probably reflects the high transmissibility, virulence and resistance particular to the ecology and epidemiology of ST239.

Concluding remarks

MLST schemes developed and applied to diverse strain collection from various ecological sources and geographic locations has not only improved world-wide tracking of clinically relevant (i.e particularly virulent or resistant) circulating enterococcal, pneumococcal, group A streptococcal and staphylococcal clones but has also provided a more detailed insight into the population structure of these pathogens. MLST revealed extensive genetic variation within these low GC Gram-positive bacteria with profound differences in the mechanisms driving genetic variation. In S. aureus genetic variation is mainly driven by mutation which means that the core genome of S. aureus clonal lineages is relatively stable with gene flow structured according to clonal complexes and with rare recombination events between clonal complexes. This implies that S. aureus clonal complexes defined by MLST probably represent ancient lineages in which, over time virulent and or resistant variants have arisen through changes in the accessory genome resulting in a wide range of resistance and virulence potential (Turner & Feil, 2007). One might even argue that these ancient clonal lineages correspond to ‘biological species’ which may explain the existence of host-specific clonal complexes like CC151 and CC133, predeominantly associated with cows and small ruminents (sheep and goat) (Guinane et al., 2010). In contrast, high rates of recombination drive genetic diversity in the strict human pathogens S. pyogenes and S. pneumoniae. In these species each ST represent lineage of recent age and therefore more accurately reflect virulence and or resistance potential. In both species hyper-virulent or resistant clones can be distinguished (e.g. S. pneumoniae PMEN1 or S. pyogenes M1T1), that have evolved rapidly the last 40 years through recombination. Natural competence explains high rates of recombination in S. pneumoniae while the mechanisms underlying genetic change in S. pyogenes involve transduction and conjugation. The biological characteristics of enterococci resemble that of S. aureus as well as that of S. pneumoniae and S. pyogenes. Like S. aureus, enterococci exhibit a broad host range but like S. pneumoniae and S. pyogenes, diversification has largely been driven by recombination rather than mutation. Genetic exchange in enterococci mainly involves mobilization and conjugation of plasmids that can be accompanied by chromosome-chromosome transfer of resistance and virulence but also of MLST genes (Manson et al., 2010). High recombination rates in enterococci leading to a rapidly changing core genome, combined with the presence of host-specific STs, at least in E. faecium, indicates that STs in enterococci represent relatively recent lineages and that these genetic lineages are more likely to reflect recent changes in the accessory genome, such as acquisition of adaptive elements required for succesful maintenance in the hospitalized patients in the typical hospital-derived clones.

A systems biology approach integrating transcriptome, proteome, metabolome analysis and functional studies including animal infection models and human patient studies will increasingly give us insights in the population structure of E. faecium, E. faecalis, S. pneumoniae, S. pyogenes, and S. aureus, will reveal the extent of lateral gene transfer and identify genetic lineages that are particularly enriched in antibiotic resistant isolates that are clinically relevant. In addition and boosted by the rapid expansion in sequencing capacity and reduction in costs through the development of novel next-generation sequencing techniques, an increasing number of staphylococcal, streptococcal and enterococcal whole genome sequences will become available providing the opportunity to obtain insights into population genomics of these species and molecular correlates for disease specificity. This may open up avenues for novel intervention approaches.