Antibiotic Resistance and the MRSA Problem

Alexander Fleming’s observation that the growth of S. aureus can be inhibited by a contaminating mold led to the groundbreaking discovery of penicillin and initiated the antibiotic era. In the past 75 years beta-lactams have been the most important class of antibiotics for the treatment of S. aureus infections. However, even before penicillin was marketed, some S. aureus strains were in possession of a powerful resistance mechanism in the form of beta-lactamases. Carried by plasmids (5), the beta-lactamase gene spread rapidly and soon obviated penicillin as the first line of treatment for S. aureus infections. As a consequence, in the following years, several hundred antibiotics of four structural subclasses, penicillins, cephalosporins, monobactams, and the more powerful carbapenems, were developed and marketed. The ability of S. aureus to become resistant to almost all of these through the acquisition of a single genetic element, the staphylococcal cassette chromosome mec (SCCmec), giving rise to the notorious methicillin-resistant S. aureus (MRSA) strains, is beyond doubt the major resistance concern. The amount of scientific literature covering the epidemiology of MRSA and beta-lactam resistance mechanisms is overwhelming. Nonetheless, important knowledge gaps still exist, as will be highlighted below.

All beta-lactam antibiotics possess the defining four-membered beta-lactam ring that is essential for the biological activity of these compounds. Early work on the mechanism of action of penicillin culminated in the discovery that penicillin inhibits cross-linking of peptidoglycan, the central component of bacterial cell walls (6). In Gram-positive bacteria, the cell wall is composed of glycan chains of N-acetylglucosamine and N-acetylmuramic acid that are heavily cross-linked by pentapeptides (UDP-MurNAc-penta). Each precursor component is synthesized in the cytoplasm and transported to the division septum of the growing cell, where the structural similarity between penicillin and d-alanyl-d-alanine (d-ala-d-ala) residues of newly synthesized UDP-MurNAc-penta, allows penicillin to inhibit crosslinking of peptidoglycan by acting as a substrate analog (6). The enzymes mediating transpeptidation of peptidoglycan strands, the target of penicillin, were hence designated penicillin binding proteins (PBPs). The binding of beta-lactams to PBPs leads to formation of long-lasting acyl-enzyme complexes, thereby essentially blocking the transpeptidase activity of these enzymes. The realization that penicillin reduces peptidoglycan cross-linking led to the classical model explaining penicillin-mediated cell lysis as a consequence of a mechanically weakened cell wall incapable of withstanding the high internal osmotic pressure (7, 8). The killing effect of beta-lactam antibiotics, however, has turned out to be much more complex, and the downstream effects of blocking transpeptidation that lead to cell death remain an unsolved mystery (8). In staphylococci, the killing mechanism has been shown to be highly cell cycle dependent, and it has been proposed that a mis-coordination in the spatial and temporal deposition of peptidoglycan during cell division leads to cell death and that cell lysis and cell death may be two separate events (9, 10). Similar to other bacteria, S. aureus possesses several PBPs that in addition to providing transpeptidase activity seem to fulfill distinct functions in coordinating peptidoglycan synthesis (see below). Because different classes of beta-lactams target individual PBPs with variable affinity, this may explain why different beta-lactams introduce various morphological changes (10). Hence, the molecular events underlying the bactericidal effects of beta-lactam antibiotics may depend on the affinity of the specific compounds for the various PBPs.

Penicillin-resistant S. aureus isolates emerged shortly after penicillin was introduced for clinical treatment of bacterial infections (11), and less than 2 decades later, about 80% had become resistant to penicillin through acquisition of the blaZ gene encoding the enzyme beta-lactamase, which inactivates penicillin by hydrolyzing the critical beta-lactam ring (12). The blaZ gene is part of a transposable element that can either be integrated into the chromosome or be located on large conjugative plasmids that, in addition to blaZ, often harbor resistance to mercury and other heavy metals as well as additional antibiotic resistance genes (13–16). Today as many as 99% of clinical S. aureus strains are resistant to penicillin, underscoring the adaptive power of S. aureus when exposed to a strong selective pressure such as antibiotics (1).

In response to the worldwide spread of penicillin-resistant S. aureus, semisynthetic penicillinase-resistant beta-lactams such as, first, methicillin and, later, oxacillin were developed for clinical use. Methicillin was introduced into clinical practice in 1961, and later the same year, the first clinical MRSA isolates were identified in a UK hospital (1, 17). MRSA strains are resistant to methicillin through horizontal acquisition of the mecA gene encoding PBP2a, an alternative transpeptidase with low affinity for most beta-lactam antibiotics, hence conferring resistance not only to methicillin but to virtually all members of the large class of beta-lactams. Despite the broad resistance provided, and even though methicillin is no longer used clinically, MRSA has remained the preferred designation for mecA-positive isolates.

The mecA gene is present on a family of structurally complex genetic elements designated SCCmec that often carry additional genes conferring resistance to other antibiotics and harmful substances such as heavy metals (18, 19). Several lines of evidence suggest that SCCmec evolution occurred in the Staphylococcus sciuri group, a primitive group of Staphylococcus species that are frequent colonizers of the skin of domestic animals (20–23). PBP2a is very similar to the native PBPs encoded by species belonging to the S. sciuri group, but the evolution of the mecA ancestor gene into a resistance determinant required numerous genetic changes (23). Consistent with this finding, it seems to be the widespread use of penicillin and other first-generation beta-lactams in the years prior to the introduction of methicillin which selected for S. aureus strains carrying the mecA resistance determinant (24). A novel genetic determinant, mecC, encoding a transpeptidase enzyme with only 63% identity to mecA-encoded PBP2a, was recently described (25). Interestingly, mecC, similar to mecA, seems to have evolved in animal-adapted lineages of S. aureus, emphasizing the importance of the animal staphylococci as a reservoir of resistance genes that can potentially contribute to the evolution of antibiotic-resistant human pathogens (26).

SCCmec elements are highly diverse in their structural organization and genetic content, with sizes varying from approximately 20 kb to >60 kb (27). They share a modular structure with two essential gene complexes, the mec complex, consisting of mecA plus its regulators (mecR1 and mecI), and the ccr complex, encoding recombinases of the invertase/resolvase family that mediate site-specific integration of the element (18). The element has a single chromosomal integration site (designated attB) located downstream of the rlmH gene (previously orfX), encoding an rRNA methyltransferase (28). The SCCmec elements additionally carry a highly variable number of insertion sequences, transposons, and plasmids inside the so-called joining regions J1 to J3 (also known as junkyard regions). Support for SCCmec as a mobile genetic element was provided by the finding that different staphylococcal strains harbor identical SCCmec elements (29, 30), but it was only recently that in vitro transfer of SCCmec by transducing phages or conjugative plasmids was finally demonstrated (31–34). Epidemiological data likewise support the idea that acquisition of SCCmec elements by methicillin-susceptible S. aureus (MSSA) strains is a relatively rare event (35, 36), and restriction modification barriers between major MRSA clones are thought to limit transfer (37). Thus, the primary spread of MRSA seems to occur via clonal expansion of a relative low number of epidemic clones.

The structural differences between the mec and ccr gene complexes form the basis for SCCmec typing, a PCR-based typing method that is used worldwide for epidemiological surveillance of MRSA (see 38 for details). To date, 11 SCCmec types have been identified (27). Early hospital-acquired MRSA isolates such as the archaic MRSA strain COL typically harbor large SCCmec elements belonging to type I, II, or III (>50 kb). In contrast, more recent strains typically possess the much smaller SCCmec IV (21 to 24 kb) or V (27 kb). The smaller SCCmec cassettes were first identified in community-associated MRSA (CA-MRSA) that emerged outside of hospitals in the mid-1990s (until then MRSA isolates were largely confined to hospitals) (15). Today, the traditional grouping of MRSA isolates into health care-associated MRSA (HA-MRSA) and CA-MRSA is losing relevance, because CA-MRSA has also become established in health care settings, and it has been suggested that these clones, which are more contagious and more virulent than the classic HA-MRSA, will replace the HA-MRSA clones over time (39).

A notable feature of most MRSA isolates is that resistance to beta-lactams is expressed in a heterogeneous manner (40, 41). For these strains, populations arising from a single cell display widely different resistance levels, with the majority of cells exhibiting a low level of resistance and a minority of cells being highly resistant. While some HA-MRSA isolates exhibit high-level, homogeneous methicillin resistance, CA-MRSA isolates often exhibit low-level, heterogeneous resistance (42). Insight into the molecular mechanisms underlying this phenomenon has come from the identification of mutations that convert strains expressing low, heterogeneous resistance into homogeneous, highly resistant strains. These mutations map to genes associated with cellular stress responses, such as stringent response signaling via ppGpp and the ClpXP protease controlling the Spx stress response (42–45), indicating a close link between bacterial physiology and resistance levels.

Despite the fact that MRSA strains have become resistant to beta-lactams through acquisition of one specific resistance determinant, the mecA gene, clinical MRSA isolates exhibit highly variable levels of resistance: in some MRSA strains resistance is barely above that displayed by susceptible isolates (methicillin MICs <3 μg/ml), while other strains are highly resistant (methicillin MICs up to 1,600 μg/ml) (46). The mechanisms underlying these intriguing differences in resistance level remain poorly understood. In some cases, high resistance levels were attributed to increased expression of PBP2a due to duplication or enhanced transcription of the mecA gene (47). Two functionally similar two-component systems, BlaR1/BlaI and MecR1/MecI, control transcription of mecA in response to beta-lactams, and because genetic mutations are common in the regulatory elements controlling mecA expression, the PBP2a level varies widely between MRSA strains (see 48 for a recent review of mecA regulation). In several cases differences in resistance levels did not correlate to PBP2a expression, suggesting that factors other than PBP2a modulate the strain-specific level of beta-lactam resistance (46, 49, 50). Indeed, genetic screens have identified a number of auxiliary factors (also designated “fem- factors”) essential for methicillin resistance that are critical for PBP2a-mediated resistance to beta-lactam antibiotics (51, 52). Examples include cell division proteins, native PBPs, and enzymes involved in the synthesis of teichoic acids and peptidoglycan precursors. In some cases, the requirements for these auxiliary factors have been explained. For example, the essential glycosyltransferase domain of the native PBP2 is needed to cooperate with the transpeptidase activity of PBP2a in the building of peptidoglycan (53). The fact that beta-lactam resistance depends on auxiliary factors opens up new possibilities for the treatment of MRSA infections, because drugs that inhibit the functions of auxiliary factors work synergistically with beta-lactams to kill MRSA (52, 54, 55).

The discovery that MRSA strains can be resensitized to beta-lactams by interfering with the activity of auxiliary factors opens up a novel paradigm in the treatment of MRSA, because drug-like leads targeted against these factors could be used in combination with beta-lactams to restore the sensitivity of MRSA to beta-lactams (56). Indeed, a combination agent strategy of pairing beta-lactamase inhibitors such as clavulanic acid with beta-lactams has proven highly successful in restoring beta-lactam efficacy against Gram-negative bacteria that have become resistant to beta-lactams through acquisition of beta-lactamase enzymes (57). Additionally, a new generation of cephalosporins, fifth-generation cephalosporins such as ceftaroline, with efficacy against MRSA have been marketed (58). The active site of PBP2a is normally protected from beta-lactam antibiotics because it is present in a narrow, inaccessible groove. To accommodate the native peptidoglycan strand, the active site is made accessible via a conformational change regulated by an allosteric switch (59, 60). Ceftaroline binds to the allosteric site of PBP2a and thereby opens the active site for binding of a second molecule of ceftaroline that inhibits transpeptidase activity (60). Despite the short time that ceftaroline has been in clinical use, MRSA strains with reduced susceptibility to ceftaroline (obtained through mutations in PBP2a) have already been isolated (61, 62). Hence, the battle continues.

Although it was discovered in the late 1950s, it was not until the emergence of MRSA 30 years later that the glycopeptide vancomycin became the choice for treatment of MRSA infections (63). Since then, vancomycin has been one of the preferred drugs for treatment of MRSA infections. The target of vancomycin is the cell wall, where it binds with high affinity to the penultimate d-ala-d-ala residues of newly synthesized UDP-MurNAc-penta, thereby disrupting peptidoglycan assembly (64). Strains of S. aureus susceptible to vancomycin (VSSA) have MICs of <2 μg/ml, whereas intermediately resistant (VISA) strains display MICs of 4 to 16 μg/ml, and for the resistant strains the MIC is ≥16 μg/ml. Vancomycin-resistant S. aureus (VRSA) carries the vanA operon that provides the cell with two important activities, namely, the hydrolysis of the d-ala-d-ala precursors and the synthesis of the d-ala-d-lactate precursor, which cannot bind vancomycin (65). The first case of VRSA was reported in 2002, when during coinfection with Enterococcus faecalis a MRSA strain acquired the vanA operon from a conjugative E. faecalis plasmid and displayed a MIC to vancomycin of >1,000 μg/ml (66, 67). Fortunately, since then, only a handful of VRSA cases have been reported, and usually they involve transfer of the vanA operon from enterococci (68). The low prevalence of VRSA may be due to limited fitness of the vanA-containing enterococcal plasmids in S. aureus (69) or, for the MRSA strains, the incompatibility with methicillin resistance, because the mecA-encoded transpeptidase, PBP2a, is not able to cross-link the modified wall precursor in the VRSA wall (70). The incompatibility between glycopeptide and methicillin resistance has been termed the “seesaw” effect and can be exploited clinically (71).

VISA strains are associated with serious clinical complications, such as prolonged hospitalization, persistent infections, prolonged vancomycin treatment, and/or treatment failure (71–75), although they appear compromised when examined in animal model systems (76). Vancomycin treatment failure has even been reported for strains with marginally decreased susceptibility and MIC break- points of 1 to 2 μg/ml, the latter being the concentration of an antibiotic, which defines whether a bacterium is susceptible or resistant to the antibiotic (77). In contrast to VRSA, VISA’s reduced susceptibility to vancomycin is not due to an acquired antibiotic resistance gene but results from the accumulation of mutations leading to one or more of a number of characteristic phenotypes. These include increased cell wall thickness, cell wall changes leading to anomalous diffusion of vancomycin through the VISA cell wall, decreased negative cell surface charge, decreased autolysis, increased cell wall synthesis, and decreased peptidoglycan cross-linking resulting in high-affinity binding of vancomycin to nonamidated muropeptides (78–89). The mutations associated with VISA were recently summarized in reference 90 and often involve genes associated with (i) the cell wall stress regulon, e.g., the two-component regulatory systems graRS, vraSR, and walRK (yycFG) that stimulate expression of the dlt operon and mprF, leading to reduced negative cell surface charge and consequently less vancomycin binding; (ii) agr encoding the virulence regulatory quorum sensing system; (iii) rpoB encoding the RNA polymerase; and (iv) other transcriptional regulators or clpP encoding the proteolytic component of the Clp protease. There are multiple evolutionary pathways for a VSSA strain to become VISA (91), and although the number of mutations needed to display the VISA phenotype is usually less than 10 (92), it can be challenging, particularly with respect to clinical isolates, to determine the contribution of individual mutations to the VISA phenotype. In one study of the well-known VISA strain Mu50, Katayama and coworkers introduced mutations of six genes associated with decreased vancomycin susceptibility to VSSA strains and tracked the contribution (93). Importantly, all six mutated genes contributed to the VISA phenotype and were directly or indirectly involved in the regulation of cell physiology (93). In other studies, sequential tracking of strains during infection and chemotherapy has revealed a much greater number of genes involved (80, 94–96), stressing the importance of bacterial processes for the development of VISA strains.

Similar to methicillin resistance, the VISA phenotype is commonly preceded by a hetero-VISA phenotype (hVISA), where multiple mutations in hVISA strains lead to the VISA phenotype (97). Mu3 was the first hVISA to be characterized, and similar to hetero-MRSA, the hVISA phenotype is revealed as an uneven killing of a seemingly homogenous cell population. For hVISA the majority of cells have little or no resistance to vancomycin and are killed by 2 μg/ml, whereas a subpopulation survives vancomycin concentrations of >4 μg/ml and thus behaves like VISA strains (89). The hVISA strains are characterized by a thickened cell wall in the absence of mutations associated with VISA strains, and the phenotype can be triggered by exposure to nonglycopeptide antibiotics such as beta-lactams (98, 99). Interestingly, the hVISA phenotype does not develop in cells carrying a mutation in trfA that influences resistance to teicoplanin, another glycopeptide antibiotic (100), and encodes an adaptor of the ClpC ATPase (101) that together with the proteolytic subunit, ClpP, is responsible for the degradation of nonnative proteins (101, 102). Clinically, hVISA strains are associated with persistent infections and attenuated host immune response, and differential gene expression changes seem to underlie their development (103).

Recently, yet another VISA phenotype was discovered, termed slowVISA (sVISA), which is characterized by very slow growth, requiring 72 hours or more for colony formation, relatively high MICs (>8 μg/ml) to vancomycin, an unstable resistance profile, and colony morphology that reverts in the absence of vancomycin (104). Altered expression of the stringent response, as well as mutations in rpoB and rpoC encoding subunits of the RNA polymerase, are associated with the sVISA phenotype (105, 106). Importantly, low concentrations of mupirocin, a well-known inducer of the stringent response, enabled the isolation of stable sVISA strains, and this approach was used to demonstrate the presence of sVISA among clinical isolates (107).

A significant challenge with the hVISA and sVISA strains is that traditional susceptibility testing does not reveal their presence. Both forms are induced by vancomycin exposure, but hVISA also arises in response to beta-lactam antibiotics, while stabilization of sVISA occurs upon induction of the stringent response—neither of which are present during susceptibility testing. A related finding was made by Haaber et al., who observed that upon exposure to the antimicrobial peptide antibiotic colistin, S. aureus elicited reversible and reduced susceptibility to vancomycin in the absence of genetic change (108). Collectively, these findings indicate that VISA derivatives may develop not only through mutations but also through unrelated phenotypic processes (90) and that such phenotypic tolerance could contribute significantly to the clinical failures of vancomycin chemotherapy.

The reduced susceptibility of MRSA strains to vancomycin has called for new treatment options for MRSA infections, leading to the development of the lipoglycopeptides, telavancin, oritavancin, and dalbavancin. They are derivatives of vancomycin that have been modified by lipophilic side chains, where the heptapeptide core inhibits transglycosylation and transpeptidation reactions in cell wall synthesis and the lipophilic side chains prolong half-life and allow interactions with the cell membrane (109). While dalbavancin inhibits the late stages of peptidoglycan synthesis mainly by impairing transglycosylase activity, oritavancin and telavancin anchor in the bacterial membrane by the lipophilic side chain linked to their disaccharide moiety, disrupting membrane integrity and causing lysis (110).

Telavancin is active against both MSSA and MRSA, and it causes rapid, concentration-dependent depolarization of the bacterial plasma membrane, increased permeability, and leakage of cellular ATP and K⁺ (111). It is 16- to 32-fold more active than vancomycin and displays activity against hVISA and VISA but reduced activity against VRSA (112). Dalbavancin is also active against VISA and hVISA, but strains with reduced susceptibility to vancomycin also demonstrate reduced susceptibility to dalbavancin (113, 114). Importantly, oritavancin is active against both VISA and VRSA (115). This is likely related to its ability to bind not only to the d-ala-d-ala peptidoglycan termini similar to vancomycin but also to interact with peptides near the terminal d-ala-d-ala (or d-lactate in VRSA), thus maintaining binding affinity for the modified peptidoglycan peptide termini of vancomycin-resistant organisms (116). In comparison to vancomycin; dalbavancin, oritavancin, and telavancin all have longer half-lives, which allow once daily dosing for telavancin, once weekly dosing for dalbavancin, and potentially one dose per treatment course for oritavancin (109, 117, 118).

In general, resistance to lipoglycopeptides is remarkably rare. The long half-life of these compounds in vivo may contribute to this but may also provide selection for resistance at subinhibitory concentrations. In a study of telavancin, resistance development was assessed by in vitro passages, and only 1 out of 10 passaged MRSA clones showed increasing MIC. This was from 0.25 μg/ml to 2 μg/ml after 43 days, and subsequently, no further increase occurred (119). Resistance to dalbavancin could not be obtained by direct selection of MSSA, MRSA, or VISA strains on plates containing dalbavancin (120), whereas passaging in the presence of subinhibitory concentrations increased the dalbavancin MIC of MRSA strains by 2 or 3 dilutions (to 0.25 and 0.5 μg/ml) (120). Recently, a clinical case of decreased susceptibility to dalbavancin was reported, where the use of a dalbavancin and vancomycin therapy led to development of a vancomycin- and dalbavancin-nonsusceptible phenotype. Importantly, these susceptibility changes were accompanied by a 4-fold increase in daptomycin and telavancin MICs, although the patient did not receive treatment with these agents (121). This observation may be explained by the finding that decreased dalvabancin susceptibility was a result of mutations of yvqF (vraT), which regulates cell wall metabolism and has been implicated in the VISA phenotype (122).

Due to the increasing problems with antibiotic resistance, older antibiotics are being re-evaluated for clinical use to expand treatment opportunities against difficult to treat infections (123, 124). Fosfomycin is a phosphoenolpyruvate analogue that inhibits the enzyme MurA, which catalyzes the first step in peptidoglycan biosynthesis, and prevents the formation of N-acetylmuramic acid, which is an essential precursor for the peptidoglycan cell wall (125). Fosfomycin is minimally used for other than uncomplicated urinary tract infections. However, it generally displays good activity against MRSA and can be orally administered; hence, this antibiotic may be of clinical importance for difficult to treat S. aureus infections (124).

Fosfomycin resistance can arise via enzymatic modification of the agent, reduced uptake, or target-site modifications. Enzymatic modification of the antibiotic by a thiol-S-transferase encoded by the fosB gene is a common resistance mechanism among clinical S. aureus isolates (126, 127). Several other enzymes, such as FosA and FosX, have been identified in other bacterial species that can modify fosfomycin, whereas in S. aureus only the fosB gene has thus far been identified (128). Reduced uptake has been characterized in E. coli by chromosomal mutations in two transporter systems, GlpT and UhpT, which enable fosfomycin uptake (129). Mutations in the genes encoding for GlpT and UhpT have also been identified among clinical fosfomycin-resistant S. aureus isolates, along with target-site modifications of MurA (127). However, the individual roles of such mutations have not been elucidated in naive strain backgrounds (127).

Daptomycin, a cyclic lipopeptide produced by Streptomyces roseosporus, was approved by the FDA in 2003 for treatment of complicated skin and soft tissue infections caused by Gram-positive bacteria, for S. aureus bacteremia, and in 2006 for right-sided infective endocarditis. (See 130 for an extensive review of the journey of daptomycin from drug discovery to the bedside.) Since then, daptomycin has become an important treatment option for deep-seated infections caused by MRSA (131). Daptomycin targets the cytoplasmic membrane, but the mechanistic details underlying the bactericidal effect of this binding remain elusive (reviewed in 130). The activity of daptomycin is completely dependent on calcium ions, and it is generally accepted that the positive charge of calcium-complexed daptomycin facilitates binding and insertion of daptomycin into the cytoplasmic membrane, where it oligomerizes to form pore-like structures (130). Demonstrated downstream effects that directly or indirectly may cause cell death involve membrane depolarization, ion leakage, and delocalization of enzymes synthesizing the cell wall (130). Notably, daptomycin, in contrast to most antibiotics, has bactericidal activity even against stationary S. aureus cells, supporting the view that it targets a process that is essential in nongrowing cells (132).

The development of daptomycin resistance (an official daptomycin resistance breakpoint has not been declared, and therefore the official terminology is “daptomycin nonsusceptibility,” but for ease of presentation we will use the term “daptomycin resistance”) in staphylococci seems to be a relatively rare phenomenon, and no trend toward increased daptomycin resistance was noted in a worldwide surveillance program (2005 to 2012) (133). Nonetheless, S. aureus isolates that have evolved daptomycin resistance during clinical treatment are well described (134–136). Daptomycin resistance is obtained through selection of multiple spontaneous mutations with hotspots in genes affecting the composition or charge of the cell envelope (134, 136). The clinically most significant example of this is mutations increasing the activity of the multipeptide resistance factor (MprF) that catalyzes synthesis of the positively charged lysylphosphatidylglycerol (134–138). The mprF mutations correlate with excess lysylphosphatidylglycerol in the cell membrane but not always with augmented positive surface charge, suggesting that electrostatic repulsion alone does not explain the improved ability to withstand daptomycin (138, 139). Other genetic changes frequently implicated in decreased tolerance to daptomycin map to rpoB and to genes controlling activity of the YycFG (WalRK) two-component system, which responds to perturbations in cell membrane homeostasis (94, 95, 134, 140, 141). Consistent with these genetic changes, the prevailing phenotypic changes reported for daptomycin-resistant S. aureus strains include increased positive charge of the cell membrane and a thickened cell wall (134, 136, 142–144). Additionally, the genetic changes associated with daptomycin resistance often correlate with decreased expression of virulence genes and altered pathogenesis of the isolates (144, 145). An intriguing explanation for this finding was proposed when it was found that mutants lacking the agr quorum sensing system survived daptomycin exposure by releasing membrane phospholipids, which bound and inactivated the antibiotic. Although wild-type bacteria also released phospholipid in response to daptomycin, agr-triggered secretion of small cytolytic toxins, known as phenol soluble modulins, prevented antibiotic inactivation (146).

Linezolid is an oxazolidinone that since its approval in 2000 has been used to treat nosocomial pneumonia and skin and soft tissue infections caused by both MRSA and MSSA (147). Compared to vancomycin, it has favorable pharmacokinetic properties and is applied to difficult to treat infections, including those caused by strains displaying decreased susceptibility to glycopeptides (148). Linezolid targets the ribosome by interacting with the 23S rRNA and blocks protein synthesis by preventing binding or proper placement of aminoacyl-tRNA in the peptidyltransferase center (149, 150). Linezolid also interacts with tRNA, the ribosomal protein L27, and the ribosomal-associated protein LepA (150). The general effect of linezolid on protein expression leads to reduced expression of toxins (151) and particularly reduced amounts of excreted or membrane-associated proteins (152), which likely contributes to the clinical efficacy of the drug.

The primary activity of linezolid is binding to the 23S rRNA, and since there are five to six rRNA (rrn) operons in S. aureus (153), it was suspected that resistance development would be limited (154). However, resistance has been associated with distinct nucleotide substitutions in domain V of the 23S rRNA gene, particularly G2447T, T2500A, and G2576T, where a gradual increase in the number of mutated 23S rRNA gene copies occasionally combined with copy deletions has been associated with increased resistance levels (155–158). In particular, the first mutation appears to be rate limiting in resistance development as demonstrated in one study, where the first G2576T mutation required by far the longest time to occur (157). Subsequent acquisition of the same nucleotide exchange in the other 23S rRNA gene copies suggested that they are not independent mutational events but arise, for example, by homologous recombination or gene conversion (157, 159). Deletions or mutations in the ribosomal proteins L3 and L4 have also been associated with resistance (158, 160) and sometimes in combination with 23S rRNA mutations (147). The fitness of resistant strains obtained by in vitro passage decreases with increasing numbers of mutated 23S rRNA copies and resistance levels (157), but importantly, strains that carry only one mutated copy and are susceptible to linezolid are without apparent fitness defects, and the mutations are stable over extended periods of time (159). In a clinically isolated, linezolid-resistant mutant, the resistance was found to be mediated by a G2576U mutation in all five 23S rRNA copies. The resistance was not accompanied by any growth defect, and passage in antibiotic-free medium did not decrease resistance levels (161). Thus, in successfully resistant clinical strains, compensatory mutations are likely to be present.

Another mechanism by which S. aureus may become resistant to linezolid is through the acquisition of the ribosomal methyltransferase cfr gene (154, 162, 163). It methylates the A2503 in the peptidyl transferase center of the 23s rRNA (164). The cfr gene has been identified in a number of clonal complexes, including the livestock-associated ST398 and ST9 (165), and it is commonly associated with other mobile genetic elements (166). Importantly, the cfr gene is present on plasmids also in other Gram-positive pathogens, including S. epidermidis, from where it can be transferred to clinical MRSA strains (166, 167), highlighting the importance of the resistance gene pool present in other staphylococcal species. Fortunately, tedizolid, a second-generation oxazolidinone, retains activity against strains carrying cfr (168). Tedizolid is 4- to 16-fold more potent toward MRSA compared to linezolid, and this compound causes a dramatic reduction in virulence factor production (169), and resistance is still limited.

Although there are numerous reports of resistance to linezolid (147), generally >98% of S. aureus strains remain susceptible. In some patient groups, however, resistance is significantly more prevalent. For example, in cystic fibrosis patients, 11% of isolates are resistant (170). Surprisingly, linezolid may also stimulate the growth of some staphylococcal strains, as demonstrated for a thymidine-dependent small colony variant of a MRSA strain isolated from the lung of a cystic fibrosis patient (171). In this strain, in the absence of any known linezolid resistance mutations or genes, the presence of linezolid greatly stimulated growth. This finding reflects the complexity and unpredictability of linezolid’s effects on ribosomal functions and stresses the need for further investigations.

Macrolides, lincosamides (including clindamycin [172, 173]), and streptogramins (type A and B) (commonly abbreviated MLS) all inhibit protein synthesis by targeting the peptidyl transferase center of the 50S ribosomal subunit. Macrolides and streptogramin B bind adjacent to the peptidyl transferase center within the exit tunnel and thereby prevent elongation of the peptide chain (174). Binding of lincosamides inhibits peptide-bond formation by preventing correct positioning of the aminoacylated ends of tRNAs in the peptidyl transferase center (174). When macrolides, lincosamides, or streptogramins are used for monotherapy, they are bacteriostatic (175). However, when a type B streptogramin (e.g., quinupristin) is used together with a type A streptogramin (e.g., dalfopristin), the combination, marketed as Synercid, acts synergistically and is bactericidal against staphylococci (176).

Resistance to macrolides, lincosamides, and streptogramin can be mediated via target-site mutation, enzymatic target-site modification, enzymatic inactivation of the antibiotic, and active efflux (177). Target-site modification occurs by acquisition of rRNA methylases encoded by erm genes, including erm(A), erm(B), erm(C), and related genes, which add methyl groups to an adenine residue in the 23S rRNA (177, 178). Also, the 23S rRNA methyltransferase, encoded by the cfr gene, confers resistance to lincosamides and streptogramin A, as well as oxazolidinones, phenicols, and pleuromutilins, but does not confer resistance to macrolides (179).

Ribosomal mutations in the 23S rRNA target site may cause resistance to macrolides, lincosamides, and streptogramins. As is the case for linezolid, the number of rrn genes limits resistance development. However, in cystic fibrosis patients treated with a macrolide, strains have been isolated that contain mutations in five of the six rrn genes, conferring cross-resistance to lincosamides and streptogramins (180). Mutations in the ribosomal proteins L22 (encoded by rplV) and L4 (encoded by rplD) may also contribute to the development of resistance to the MLS antibiotics (180, 181). Enzymatic inactivation of macrolides can be mediated by phosphorylases encoded by mph genes, and type A streptogramin resistance can be mediated by acetyltransferases (177, 182).

Staphylococci can also acquire genes encoding efflux pumps that can extrude one or more of the MLS antibiotics (183). Acquisition of the msr(A) gene encoding for the Msr(A) ABC transporter confers resistance to macrolides and streptogramin B, while lincosamides are not substrates for this pump (184). The plasmid-carried gene vga(A) encodes an ABC transporter with specificity for type A streptogramins, along with the pleuromutilin class of antibiotics (184, 185). The LmrS efflux pump can extrude macrolides and lincosamides, as well as several other classes of antibiotics (186). Finally, overexpression of the chromosomally encoded gene mdeA reduces susceptibility to the streptogramin virginiamycin (187).

Chloramphenicol is a broad-spectrum, bacteriostatic antibiotic that interferes with protein synthesis by binding to the ribosomal 50S subunit. Chloramphenicol binds at the peptidyl-transferase center and thus inhibits peptide bond formation between the tRNAs at the A- and P-sites (174). The wide use of chloramphenicol has been hindered by its toxicity, because systemic administration of chloramphenicol is associated with irreversible aplastic anemia (188). Due to the severity of the adverse effects of chloramphenicol, it is now used primarily for topical applications, such as a treatment for staphylococcal conjunctivitis (189).

Resistance to chloramphenicol is attributed to either enzymatic inactivation of the antibiotic, active efflux, or target-site modification. Inactivation of chloramphenicol occurs through acetylation at the C3 position via various chloramphenicol acetyltransferases, encoded by cat genes (190), which often are located on plasmids (191). Active efflux of chloramphenicol is mediated by the FexA (192) or LmrS efflux pump (186). Finally, chloramphenicol resistance can occur through target-site modification of the 23S rRNA at position A2503 by an rRNA methylase encoded by the gene cfr (164).

The pleuromutilin class of antibiotics was first described in 1951, and the first agent in this class, tiamulin, was approved for veterinary use in 1979 (193). In 2007, the pleuromutilin retapamulin was first approved for human use for treatment of skin infections caused by S. aureus and Streptococcus pyogenes (193, 194). Pleuromutilins inhibit protein synthesis by binding to the ribosomal 50S subunit at the peptidyl transferase center (174).

Pleuromutilin resistance mechanisms include reduced binding to the target and active efflux (194). In vitro selected pleuromutilin resistance can be conferred by spontaneous mutations in the gene rplC, which encodes the ribosomal protein L3. Stepwise mutations in rplC confer higher levels of resistance than single mutations in the gene (195, 196). Reduced binding to the target can also occur by acquisition of the cfr gene (179). Pleuromutilins are also subjected to active efflux by, for example, Vga(A) (185) and related efflux systems (197).

Aminoglycosides, such as gentamicin, interfere with protein synthesis by binding to the A-site of the 16S rRNA of the 30S subunit and promote inaccurate codon-anticodon recognition that ultimately decreases fidelity of translation, leading to the generation of mistranslated proteins (198, 199). The bactericidal activity of aminoglycosides is attributed to the production of mistranslated proteins that may incorporate into the membrane, increasing its permeability and leading to uncontrolled small molecule diffusion and eventually increased uptake of aminoglycosides (198). Aminoglycosides are rarely used as monotherapy; they may be used in combination with other agents for treatment of endocarditis (172).

For S. aureus, resistance to aminoglycosides is conferred by several mechanisms, including enzymatic modification of the aminoglycoside agents, decreased uptake, and target modifications of the 30S ribosomal subunit. Active efflux of aminoglyocsides has not been reported for S. aureus, whereas this mechanism is common among Gram-negative pathogens (200). In contrast, enzymatic modification and inactivation of aminoglycoside agents is common among S. aureus clinical isolates. The aminoglycoside-modifying enzymes can be divided into three subclasses based on the chemical type of modification of the aminoglycosides (200–202).

Aminoglycoside uptake is dependent on the membrane potential, and mutants experiencing depolarization of the membrane display low-level resistance to aminoglycosides (203). On agar plates such mutants often display a small colony phenotype and hence are referred to as small colony variants (203). Aminoglycoside-resistant small colony variants are considered to be of clinical importance, because they have been associated with recurrent and persistent staphylococcal infections such as device-associated infections, bone infections, and airway infections in cystic fibrosis (204).

Plasmid-mediated 16S rRNA methyltransferases can confer a high level of resistance to various aminoglycosides by preventing binding of the aminoglycoside to the active site; however, this mechanism has been reported only among clinical Gram-negative bacteria (205). Nonetheless, an in vitro-generated S. aureus mutant containing the 16S rRNA methyltransferase gene rmtC on a plasmid was capable in conferring resistance to gentamicin and kanamycin (206).

Tetracyclines, including tigecycline, are bacteriostatic antibiotics that inhibit protein synthesis by binding to the ribosomal 30S subunit (207). Tigecycline blocks entry of amino-acyl tRNA molecules into the acceptor site of the ribosome, thus preventing incorporation of amino acids into the elongating peptide chain (174, 208).

The backbone of the tetracycline compound class is characterized by four interlocking six-carbon rings. Due to the emergence of resistance to the marketed tetracyclines, there has been continued development of new analogues that overcome common resistance mechanisms against older versions of tetracyclines (209). In 2005, the FDA approved tigecycline, which belongs to the novel antibiotic class of glycylcyclines. Glycylcyclines are derivatives of the tetracycline class of antibiotics and thus contain the four-ring carbocyclic structure, with a substitution of an N-alkyl-glycylamido group to the D-9 position. This substitution expands the spectrum of activity compared to tetracyclines and overcomes some tetracycline resistance mechanisms (210).

Tetracycline resistance of clinical relevance has been attributed to active efflux or target-site protection (211). Active efflux is mediated via the Tet(K) and Tet(L) efflux pumps of the major facilitator superfamily by exchanging a proton for a tetracycline molecule (211). The efflux pump of Tet(K) can be encoded on plasmid pT181, which may be integrated with SCCmec elements of MRSA CC398 (212). A chromosomally encoded efflux pump, Tet38, confers tetracycline resistance when the repressor protein MgrA is inactivated (213).

Target-site protection is mediated via TetM or TetO determinants, which are often encoded on chromosomally located conjugative transposons. TetO and TetM bind to the elongation factor G (EF-G) binding site on the ribosome, releasing bound tetracycline from the A site in the presence of GTP (214). Strains harboring both TetK and TetM display higher MICs than strains carrying an individual gene (215) and are selected for during exposure to subinhibitory concentrations (216).

Tigecycline susceptibility is minimally affected by the presence of the tetracycline resistance determinants TetM and/or TetK in S. aureus (217). Tigecycline is currently not greatly burdened by resistance in clinical isolates, with only a few reports highlighting this in staphylococci (218). Among 28,278 strains from the United States, more than 99.9% isolated between 2006 and 2012 remained susceptible to tigecycline (219), whereas the susceptibility frequency in Mexico (where prescription control is limited) was down to 91% (220). Mechanisms conferring resistance to tigecycline have been investigated in vitro by selection of resistant mutants. Loss-of-function mutations have been identified in the transcriptional repressor mepR, leading to derepression of the mepA-encoded efflux pump (221, 222). Mutations in mepA itself have also been associated with reduced tigecycline susceptibility (222). Furthermore, mutations have been identified in the gene rpsJ, which encodes the ribosomal protein S10, that are suspected to reduce the access of tigecycline to the binding site (223).

Mupirocin is a topical antibiotic widely used to reduce nasal carriage of MRSA and to treat skin infections. Mupirocin arrests protein synthesis by inhibiting the isoleucyl-tRNA synthetase, which is encoded by the ileS gene. The isoleucyl-tRNA synthetase (IleRS) catalyzes formation of isoleucyl-tRNA from isoleucine and tRNA and thereby prevents incorporation of isoleucine into protein chains (224).

Resistance to mupirocin can arise by chromosomal mutations in the ileS gene. Amino acid substitutions in IleRS reduce the affinity for mupirocin and confer low-level resistance to the antibiotic (224). Resistance to mupirocin can also be conferred by acquisition of the gene mupA, which encodes a mupirocin-insensitive isoleucyl-tRNA synthetase (224). mupA is often located on plasmids that also harbor genes conferring resistance to other antibiotics of greater importance than mupirocin for treatment of S. aureus infections, and hence mupirocin may provide selective pressure for maintenance of nonmupirocin resistance genes (225). Recently, it was demonstrated that mupB confers high-level resistance, but the mechanism of resistance remains incompletely understood (226).

Fusidic acid is a bacteriostatic antibiotic that inhibits protein synthesis by binding to the EF-G on the ribosome, effectively stalling the elongation phase of translation. EF-G is encoded by the chromosomally located gene fusA (227). For systemic application, fusidic acid often is used in combination with other antistaphylococcal antibiotics to reduce the risk of resistance development, because fusidic acid resistance can be readily selected (227, 228). Fusidic acid has been applied as a monotherapy for topical administration, but consequently has experienced a rapid emergence of resistant variants (227).

Fusidic acid resistance can arise due to chromosomal mutations in fusA, leading to reduced fusidic acid binding to EF-G due to altered protein conformation (227, 229) or mutations in rplF (also termed fusE, encoding the ribosome protein L6) (230). rplF mutations can be selected for using fusidic acid or aminoglycosides and may be accompanied with secondary mutations in hemin or menaquinone biosynthesis genes conferring the small colony variant phenotype (230, 231). Acquired fusidic acid resistance genes found in S. aureus include fusB, fusC, and fusD (232). By binding to EF-G on the ribosome, FusB frees stalled ribosome-EF-G-GDP complexes that form in the presence of fusidic acid, which allow the ribosomes to resume translation (233). fusC has been inferred to have a similar mechanism of resistance as fusB, based on sequence similarity (231). The fusD gene is associated with the intrinsic resistance of Staphylococcus saprophyticus to fusidic acid (232). Fusidic acid resistance genes have been identified on SCCmec-like elements termed SCCfar, which lack the mecA gene, and these elements may contribute to the horizontal transfer of fusidic acid resistance genes (234). The fusB gene has also been identified on mobile elements with similarity to staphylococcal pathogenicity islands. In S. aureus the mobile element does not harbor any known virulence determinants, so it is termed the S. aureus resistance island carrying fusB (SaRI_fusB) (235), while in S. epidermidis, the fusB gene is located on the pathogenicity island SePI_fusB-857 (236).

Fluoroquinolones are bactericidal due to the blocking of bacterial DNA replication. They have two molecular targets, where the primary target in S. aureus is topoisomerase IV and the secondary target is DNA gyrase. Topoisomerase IV promotes separation of intertwined daughter chromosomes following replication and is a tetramer complex consisting of two GrlA and two GrlB subunits encoded by grlA and grlB. DNA gyrase is a tetramer complex consisting of two GyrA and two GyrB subunits encoded by gyrA and gyrB, and it is essential for viability because it removes positive supercoils that accumulate during replication (237).

Resistance to different fluoroquinolones is common, with some reporting a prevalence of up to 60% (238, 239). Due to the high prevalence of fluoroquinolone resistance and the risk of resistance developing during treatment, it is primarily recommended that fluoroquinolones be used in combination with other antibiotics (172). Fluoroquinolone resistance mechanisms include mutational changes in the target site of topoisomerase IV and DNA gyrase and increased expression of endogenous efflux pump systems (240). Target-site resistance mutations have primarily been located in the DNA gyrase gene, gyrA, and the topoisomerase IV gene, grlA (241). Combinatorial mutations in topoisomerase IV and DNA gyrase produce higher levels of resistance than a single mutation in either of the targets (242). S. aureus harbors three chromosomally encoded efflux pump systems, namely, NorA, NorB, and NorC, whose overexpression leads to reduced susceptibility to fluoroquinolones. The affinity of these efflux pump systems varies for the individual fluoroquinolone; e.g., NorA primarily extrudes ciprofloxacin and norfloxacin, and its overexpression can increase the MIC approximately 8-fold (243). Fluoroquinolones generally have great activity against S. aureus, so accumulation of multiple mutations altering the target-site and/or conferring overexpression of efflux pump systems is needed to exceed the clinical breakpoint (240).

The selective antibacterial activity of rifampicin is attributed to the inhibition of RNA synthesis, because it binds to prokaryotic RNA polymerases with much greater affinity than to eukaryotic RNA polymerases. The bacterial RNA polymerase is a multisubunit complex, consisting of the subunits α₂, β, β′, and σ. The binding site of rifampicin is located in the β-subunit, which is encoded by the gene rpoB (244).

Rifampicin-resistant mutants are readily selected (228), and resistance-conferring point mutations have been located within the rpoB gene. Alteration of the target decreases affinity for rifampicin and can confer a more than 10,000-fold increase in MIC (245). Rifampicin resistance mutations often confer a fitness cost, but the level of resistance is not correlated with the magnitude of the fitness cost for in vitro selected mutants, and rifampicin-resistant mutants from clinical isolates often carry mutations associated with low fitness costs (245). Available online tools can help to determine chromosomal rpoB point mutations conferring rifampicin resistance from whole-genome sequences, e.g., the Mykrobe Predictor (246). Due to the extensive risk of resistance development, treatment with rifampicin is recommended only for use in combination with a secondary antibiotic (172).

Sulphonamides were the first class of antibiotics to be introduced into the clinic (in 1935). Currently, sulfamethoxazole (SMX) is the most widely used sulphonamide. Trimethoprim (TMP) was introduced in 1962 and has been used in combination with SMX since 1968 in a formulation called co-trimoxazole (247). Both of the agents interfere with bacterial metabolism by inhibiting the synthesis of folic acid, a cofactor in amino acid and nucleotide synthesis. SMX inhibits the enzyme dihydropteroate synthase, which catalyzes the formation of dihydrofolate from para-aminobenzoic acid. TMP inhibits dihydrofolate reductase (dfrB), which catalyzes the formation of tetrahydrofolate from dihydrofolate (247). By inhibiting two enzymes in the same metabolic pathway, it was anticipated that the risk of resistance development would be diminished, yet a high prevalence of resistance to both agents has been reported (247).

Resistance to SMX in clinical isolates arises from chromosomal mutations in the gene encoding the dihydropteroate synthase enzyme, which is expected to reduce the binding affinity of SMX to the target (248). Similarly, TMP resistance arises from chromosomal mutations in the dfrB gene encoding the dihydrofolate reductase enzyme (249). Besides chromosomally encoded TMP resistance, S. aureus can acquire dfr gene variants, dfrA, dfrG, and dfrK, which are less susceptible to inhibition by TMP (250). The two antibiotics are often used in combination to limit resistance development, but in patients exposed to long-term treatment with TMP-SMX, thymidine-dependent TMP-SMX-resistant small colony variants emerge (251). The thymidine-dependent small colony variants acquire mutations in thyA encoding for thymidylate synthase, which is essential for de novo thymidylate biosynthesis. Tetrahydrofolic acid acts as a cofactor for thymidylate synthase, and thyA mutants are therefore dependent on the uptake of thymidine or its metabolite dTMP (252, 253).