Society has become imbued with the Superbug label to define strains of antibiotic resistant bacteria that cause hospital-associated outbreaks of infection (Foster, 2004; Abbott, 2005; Brazier, 2008; Guo et al., 2011). This term, denoting the sudden emergence and spread of new antibiotic resistant strains, could also be applied to an historic global pandemic caused by a penicillin-resistant S. aureus clone known as phage type PT80/81, which emerged in Australia, Great Britain, and North America in the early 1950's (Rountree and Beard, 1958; Williams et al., 1959; Wormald, 1961; Tanimoto, 1962). The initial outbreaks occurred in hospitals, especially among newborns and nursing mothers, but quickly spread to the wider community, causing unusually severe invasive skin infections, and fatal sepsis or necrotizing pneumonia in young and healthy individuals (Hassall and Rountree, 1959). Although the pandemic dissipated after 10 years (∼1953–1963), concomitant with the introduction of methicillin, genetically related contemporary strains are prominent in both the community and health-care settings. These consist of clinical methicillin susceptible S. aureus (MSSA), the epidemic EMRSA-16 lineage of hospital associated MRSA (HA-MRSA) which has the Type II Staphylococcal cassette chromosome SCCmec element, and the hyper-virulent Southwest Pacific (SWP) clone of community associated MRSA (CA-MRSA) which, like other unrelated CA-MRSA, has Type IV SCCmec. All of these strains belong to clonal complex CC30 as determined by multi locus sequence typing (MLST) analysis (Robinson et al., 2005).

To better understand the evolutionary development of CC30, we recently employed comparative genome sequencing to evaluate nine CC30 strains (DeLeo et al., 2011), including the reference genome of MRSA 252, representing the EMRSA-16 clone of HA-MRSA (Holden et al., 2004; Lindsay and Holden, 2004). Phylogenetic analyses based on a contiguous 1.4 Mb region of each genome, or with concatenated nucleotide segments, supported the existence of three major branches that evolved from a common ancestor. Clade I consists of the historic PT80/81 pandemic. This clonal type is typically ST30spa43, as determined by MLST and staphylococcal Protein A (spa) gene typing, and possesses the Panton Valentin Leukotoxin (PVL), that is also characteristic of contemporary CA-MRSA, including the SWP clone, which is ST30spa19 and comprises Clade 2. Although temporally separated by nearly 50 years, Clades 1 and 2 share a number of common traits, which in addition to PVL, include abundant production of α-hemolysin Hla, elevated transcription of RNAIII encoded by the accessory gene regulator agr locus, and a hypervirulent trait in murine infection models (DeLeo et al., 2011).

Clade 3 is comprised of the EMRSA-16 clone of HA-MRSA, which is typically ST36spa16, and contemporary clinical methicillin susceptible S. aureus, which are often ST30spa33. Although Clade 3 strains exhibited attenuated virulence in murine infection models relative to Clades 1 and 2, these are still associated with a high burden of disease. EMRSA-16, which is known in the United States as USA200, has become one of the most successful HA-MRSA clones (Cox et al., 1995; Enright et al., 2000; Johnson et al., 2001; McDougal et al., 2003; Seybold et al., 2006; Fowler et al., 2007). Others defined an association of CC30 MSSA, frequently ST30spa33, with bacteremia, infective endocarditis, and osteomyelitis (Cassat et al., 2005; Fowler et al., 2007; Nienaber et al., 2011). Staphylococcal toxic shock syndrome, which emerged in the late 1970's (Altemeier et al., 1981), is also associated with CC30, and the tst gene encoding toxic shock syndrome toxin has a strong clonal association with CC30 nasal carriage and bacteremia isolates (Holtfreter et al., 2007). MSSA that resemble the EMRSA-16 clone were also commonly associated with asymptomatic nasal carriage in the United States (Kuehnert et al., 2006), and other studies concur that CC30 is a major clonal complex associated with nasal carriage (Feil et al., 2003; Melles et al., 2004; Kuehnert et al., 2006; Fowler et al., 2007; Ko et al., 2008; Melles et al., 2008). Therefore, although Clade 3 exhibits attenuated virulence in murine infection models, we proposed that the high burden of disease associated with these hospital associated strains could be due to the high incidence of colonization, affording more opportunity to cause infection.

Several observations support the contention that Clade 3 evolved to favor enhanced colonization, at the expense of attenuated virulence. Foremost, the genome of MRSA 252 has the highest content of pseudogenes compared to other S. aureus genomes (Holden et al., 2004), and gene decay is a major force in niche-adaptation of microbial pathogens (Moran and Plague, 2004). Most notable among the pseudogenes was a CAG to TAG transition at Gln₁₁₃ of hla encoding α-hemolysin (Hla), which is a major lethal virulence factor of CA-MRSA (Bubeck Wardenburg et al., 2007). This mutation, which creates a premature stop codon, is broadly disseminated in Clade 3, including HA-MRSA and clinical MSSA (DeLeo et al., 2011). Clade 3 strains also possessed a single nucleotide polymorphism (SNP) in agrC of the accessory gene regulator agr locus, causing a Gly₅₅ >Arg change in the AgrC sensor protein, leading to attenuated transcription of the RNAIII product that is needed to produce secreted virulence factors (DeLeo et al., 2011). Consequently, attenuated transcription of agr, and inability to produce Hla contribute to the attenuated virulence of Clade 3.

Other defining traits of genomes that are in transition toward niche adaptation include acquisition of mobile genetic elements, and amplification of insertion sequence (IS) elements (Moran and Plague, 2004). However, a common limitation of conducting multiple genome comparisons by mapping short sequence reads from multiple strains on to a known reference genome is that it may not detect large insertions or deletions that differentiate one or more strains from the reference genome. Accordingly, although tst encoding toxic shock syndrome toxin is associated with different S. aureus pathogenicity island (SaPI) structures, and has a strong clonal association with CC30, tst was not present in the MRSA 252 reference genome, and we failed to identify the relevant SaPI through multiple genome comparisons. Herein, we present a detailed analysis of the hypothesis that Clade 3 strains have evolved in favor of niche adaptation, by conducting in silico comparisons of 15 additional CC30 genomes that are available in the public domain. Our analysis of pseudogenes, SaPI and IS elements, and gene deletion events, support the hypothesis that Clade 3 is following an evolutionary blueprint towards host- and niche-adaptation.

Gene gain, gene loss, and gene change are major forces in bacterial genome dynamics (Moran and Plague, 2004; Pallen and Wren, 2007), and we have evaluated these processes in S. aureus clonal complex CC30. It was previously established that MRSA 252, representing the EMRSA-16 lineage, has the highest content of pseudogenes compared to other S. aureus genomes (Holden et al., 2004; Lindsay and Holden, 2004). A bovine adapted strain ET3-1 was a close second, and several of the accumulated pseudogenes would eliminate production of a number of cell surface proteins and iron acquisition pathways (Herron-Olson et al., 2007). Importantly, the most robust examples of gene decay come from recently emerged pathogens that have changed lifestyle, usually to live in a simpler host-associated niche (Moran and Plague, 2004). As summarized in Table 2, defining traits of CC30 Clade 3 include (1), a premature stop codon in hla encoding α-Hemolysin; (2), a SNP that causes an Gly₅₅>Arg substitution in the AgrC membrane sensor protein, leading to attenuated transcription of the regulatory RNAIII that is needed to produce secreted virulence factors; (3), isdH and trpD pseudogenes; (4), acquisition of SaPI4, and in most isolates also SaPI2, which possesses the tst gene encoding toxic shock syndrome toxin, and (5), an increase in copy number of ISSau2 relative to Clades 1 and 2. Our analyses have revealed several features that were not previously identified through comparative genome sequencing (DeLeo et al., 2011). These include the presence of SaPII and a functional TA genomic addiction module in Clades 1 and 2, the identification of SaPI2 in Clade 3, and the occurrence of unique ISSau2 insertions in each of the three major Clades. Cumulatively, these findings are concordant with niche adaptation in CC30 Clade 3.

It has long been known that 80% of TSST producing S. aureus strains are auxotrophic for tryptophan (Chu et al., 1985; Leung et al., 1993), which we now attribute to a TG deletion in trpD. It is believed that amino acid auxotrophy contributes to niche adaptation, because auxotrophic bacteria are restricted to a niche where the appropriate amino acid can be obtained. Lactococcus lactis recovered from dairy products are auxotrophic for histidine due to frame-shift mutations in hisC, hisG, and hisH, while strains from non-dairy sources are prototrophic (Delorme et al., 1993). S. aureus ET3-1, which is a predominant clonal type associated with bovine mastitis, also has a frame-shift mutation in hisC (SAB2553), which does not occur in other S. aureus genomes. The trpD gene, which has a TG insertion in CC30 Clade 3, is part of the trpEGDCFBA transcriptional unit, of which trpBA encode the subunits of tryptophan synthase, and trpEGDCF, encode enzymes necessary for synthesis of indole precursor. Chlamydia trachomatis uniformly lack the genes needed to produce indole, but strains that cause ocular vs. genital infections can be differentiated on the basis of the latter being able to produce a functional tryptophan synthase, and it is postulated that tryptophan can be produced by condensation of serine with exogenous indole produced by microflora in the female genital tract (Fehlner-Gardiner et al., 2002; McClarty et al., 2007). Consequently, tryptophan auxotrophy may contribute to tropism of TSST producing CC30 strains for the vaginal mucosa.

Another important factor in evolution of niche adapted strains is an increase in copy number of IS elements, leading to genome deletions and inversions through recombination between adjacent IS elements. An interesting example relevant to our analysis is a reduction in the numbers of operons encoding 16S-23S rRNA in microbial endosymbionts of insect cells (Andersson and Andersson, 1999; Itoh et al., 2002). This was attributed to IS integration within operons encoding 16S-23S rRNA, followed by recombination to generate deletions (Dale et al., 2003). It is, therefore, striking that all three 16S-23S-5S rRNA loci in the CC30 genome are targeted by ISSau2, with an insertion at rRNA-3 being unique to Clade 1, while insertions at rRNA-1 and rRNA-2 are unique to HA-MRSA in Clade 3 (Figure 1 and Table 2). Our analysis suggests that the insertions adjacent to rRNA operons is due to the propensity of ISSau2 to target inverted repeats, which likely comprise rho-independent transcription terminators. ISSau2 is a member of the IS30 family, and two unusual members of the IS30 family in Mycoplasma fermentans and M. bovis, which are obligate intracellular parasites, also target rho-independent transcription terminators, which remain intact and are partially duplicated on transposition (Calcutt et al., 1999; Lysnyansky et al., 2009). Conversely, our data suggest that ISSau2 either disrupts or weakens stem loop structures, as illustrated in Figures 3D–G.

Depending on the orientation of ISSau2 with respect to the adjacent gene, this could have important consequences with respect to control of transposition. It is widely accepted that transposition must be maintained at a low level, a commonly cited reason being that excessive transposition is detrimental to the stability of the host genome (Doolittle et al., 1984). Therefore, endogenous transposase promoters are generally weak, and often partially located in the inverted flanking repeats, such that strong promoters can only be created by juxtaposition of inverted repeats due to formation of head-to tail dimers, or circular copies of the IS as noted for the IS30 family (Dalrymple, 1987). IS elements also have mechanisms to attenuate their activation by impinging transcription, following insertion into active host genes. Impinging transcription across the inverted flanking repeats can either sequester translation initiation signals, or disrupt complex formation between the transposase and inverted repeats. These considerations may help to explain the insertion of ISSau2 adjacent to highly transcribed genes, including all three rRNA operons in the CC30 genome, and adjacent to rplQ encoding the 50S ribosomal subunit protein L17 (Figures 1 and 3). In these situations, ISSau2 is oriented in the antisense orientation with respect to the adjacent gene, such that impinging transcription would also generate antisense RNA to the transposase genes.

In an example that is unique to HA-MRSA in Clade 3 (Figure 1 and Table 2), ISSau2 is inserted in the sense orientation adjacent to sbcDC (Figure 3F), disrupting the predicted transcriptional terminator. The sbcDC genes encode a protein complex that recognizes and cleaves hairpin structures in DNA, has a major role in promoting genome stability and repair of breaks in double stranded DNA, and is induced by the SOS stress response in S. aureus (Connelly et al., 1998; Mascarenhas et al., 2006; Chen et al., 2007; Eykelenboom et al., 2008; Darmon et al., 2010). It is noteworthy in this respect that loss of DNA recombinational repair occurs in the initial stages of genome degeneration, as bacteria undergo a transition from an autonomous free-living state to permanent intracellular existence (Dale et al., 2003). This leads to active expansion of IS elements, which in turn promotes deletion or inversion of genome segments via IS-mediated recombination. Therefore, based on established evolutionary trends, the ISSau2 insertion adjacent to sbcDC may represent an early stage in the pathway toward genome destabilization.

Whether ISSau2 influences expression of adjacent genes is unknown. However, an insertion that is unique to strain MN8, from a case of menstrual Staphylococcal toxic shock, is adjacent to the saeRS two-component sensory signal transducer. SaeRS is a major regulator of virulence in S. aureus (Geiger et al., 2008; Voyich et al., 2009; Nygaard et al., 2010), and although there is as yet no evidence that ISSau2 influences expression of saeRS, we find that relative to ST30spa33 strains that lack this insertion, MN8 exhibits more abundant production of secreted proteins, including the HglA, HglB, and HglC components of γ-hemolysin, as well as TSST and SAR0385 gene product encoded on SaPI2 and SaPI4, respectively. This is consistent with a recent finding that, with the exception of strain MN8, other ST30spa33 strains and EMRSA-16 exhibited strongly attenuated transcription of the RNAIII effector component of the agr global regulator, due to a common SNP in agrC (DeLeo et al., 2011). Although this SNP is also present in strain MN8, transcription of RNAIII was not influenced to the same extent as other strains, suggesting that there was a compensatory mechanism in this strain. Additional work is warranted to determine if this is related to the ISSau2 insertion adjacent to saeRS.

The absence of the yefM-yoeB TA module (toxM) in Clade 3 is also consistent with established evolutionary pathways toward niche adaptation. Most free-living bacteria have multiple genomic TA loci, which are thought to help cope with nutritional stress by inducing a reversible state of dormancy during periods of nutrient depletion (Pedersen et al., 2002; Gerdes et al., 2005), although this has been disputed in E. coli (Tsilibaris et al., 2007). However, obligate intracellular pathogens and symbionts experience a less variable environment, and do not have TA modules (Pandey and Gerdes, 2005). Only a few free-living bacteria lack TA modules, the most notable being Lactococcus lactis, which is niche-adapted in its association with dairy products. Intriguingly, outside of CC30, the only other S. aureus genomes that lack this TA module thus far are restricted to ST398 porcine adapted S. aureus, ovine adapted strains 011 and 046, and as yet undefined strains A9635 and 21200. Although we cannot exclude the possibility that this TA module represents a gene acquisition in Clades 1 and 2, rather than a gene deletion in Clade 3, the broad distribution of this element in other S. aureus genomes supports the contention that this element is a component of the core genome that is lost in evolutionary development of some strains. However, more work is needed to confirm this hypothesis.

In conclusion, our findings support the notion that Clade 3 is following an evolutionary blueprint toward niche-adaptation, while Clade 2 strains consisting of ST30spa19 CA-MRSA retain the feral nature of the historic PT80/81 Clade 1. It is important to note that CA-MRSA are defined by Type IV SCCmec, which in CC30 is also associated with ST30spa43 and ST30spa16 strains in Clade 3 (Table 2). Therefore, strains of CA-MRSA, which are typically associated with hyper-virulence, are emerging with the niche-adapted trait. Given that CA-MRSA must evolve from MSSA through acquisition of Type IV SCCmec, this suggests that the niche-adapted trait is widely disseminated in the human population, and this is supported by several key observations. First, MSSA that resemble the EMRSA-16 clone of HA-MRSA (Clade 3) were the most common clonal type associated with asymptomatic nasal carriage in the United States (Kuehnert et al., 2006), and several other studies concur that CC30 is a major clonal complex associated with nasal carriage (Feil et al., 2003; Melles et al., 2004, 2008; Kuehnert et al., 2006; Fowler et al., 2007; Ko et al., 2008). Second, S. aureus infections are usually caused by the same strain that is associated with nasal carriage, and in our analysis of a panel of 172 CC30 clinical isolates, the occurrence of the hla pseudogene, the agrC SNP, and tst encoded by SaPI2, which are key markers of Clade 3 (Table 2), was 70.9%, 72.1%, and 75.6%, respectively, (DeLeo et al., 2011). Third, in a study that assessed nasal carriage isolates from 107 healthy blood donors, 27% were CC30, and 62% of these CC30 strains possessed tst (Holtfreter et al., 2007), which is a marker of Clade 3. Moreover, in the same study, tst was present in 90% of CC30 bacteremia isolates. In this context, although our studies indicate that the niche adapted trait is associated with attenuated virulence in murine infection models (Holtfreter et al., 2007), a benefit to Clade 3 in having premature stop codons in Hla and IsdH (Table 2) is that both proteins are considered as potential vaccine antigens (Clarke et al., 2006; Wardenburg and Schneewind, 2008; Kennedy et al., 2010; Ster et al., 2010), and Clade 3 strains would be immune to this vaccine strategy.