Pseudomonas Aeruginosa: Resistance to the Max

Pseudomonas aeruginosa is intrinsically resistant to a variety of antimicrobials and can develop resistance during anti-pseudomonal chemotherapy both of which compromise treatment of infections caused by this organism. Resistance to multiple classes of antimicrobials (multidrug resistance) in particular is increasingly common in P. aeruginosa, with a number of reports of pan-resistant isolates treatable with a single agent, colistin. Acquired resistance in this organism is multifactorial and attributable to chromosomal mutations and the acquisition of resistance genes via horizontal gene transfer. Mutational changes impacting resistance include upregulation of multidrug efflux systems to promote antimicrobial expulsion, derepression of ampC, AmpC alterations that expand the enzyme's substrate specificity (i.e., extended-spectrum AmpC), alterations to outer membrane permeability to limit antimicrobial entry and alterations to antimicrobial targets. Acquired mechanisms contributing to resistance in P. aeruginosa include β-lactamases, notably the extended-spectrum β-lactamases and the carbapenemases that hydrolyze most β-lactams, aminoglycoside-modifying enzymes, and 16S rRNA methylases that provide high-level pan-aminoglycoside resistance. The organism's propensity to grow in vivo as antimicrobial-tolerant biofilms and the occurrence of hypermutator strains that yield antimicrobial resistant mutants at higher frequency also compromise anti-pseudomonal chemotherapy. With limited therapeutic options and increasing resistance will the untreatable P. aeruginosa infection soon be upon us?

Extended-spectrum -lactamases. More commonly reported in the Enterobacteriaceae, though present also in P. aeruginosa, ESBLs typically hydrolyze and, so, provide resistance to broad-spectrum cephalosporins (e.g., the third generation oxyiminocephalosporins cefotaxime and ceftazidime) and aztreonam, in addition to penicillins and narrow-spectrum cephalosporins (reviewed in Paterson and Bonomo, 2005;Bush, 2008). Classical ESBLS have evolved from restricted-spectrum class A TEM and SHV b-lactamases although a variety of non-TEM, non-SHV class A ESBLS have been described (e.g., CTX-M, PER, VEB, GES, BEL; Poole, 2004b;Paterson and Bonomo, 2005) and class D ESBLs derived from narrow-spectrum OXA b-lactamases are also well-known (Paterson and Bonomo, 2005;Poirel et al., 2010b).
Carbapenemases. Carbapenems (e.g., meropenem, imipenem) are an important class of anti-pseudomonal -lactam owing to their stability to most -lactamases (see El Gamal and Oh, 2010 for a recent review of carbapenems) and are of particular use in treating infections associated with ESBL-and AmpC-producers. -lactamases capable of hydrolyzing carbapenems are known (Poole, 2004b). MexAB-OprM accommodates the broadest range of -lactams (amongst these pumps) and is most frequently linked to -lactam resistance in clinical isolates (Drissi et al., 2008;Tomas et al., 2010). The MexXY-OprM efflux system has also been linked to -lactam resistance in clinical isolates of P. aeruginosa (as one of several contributors; Maniati et al., 2007;Vettoretti et al., 2009). While MexAB-OprM, MexCD-OprJ, and MexXY-OprM have all been shown to accommodate carbapenems (except imipenem; Okamoto et al., 2002) MexAB-OprM is by far the better exporter of these agents and the pump has been shown to contribute to reduced susceptibility to meropenem in clinical isolates (Pai et al., 2001;Pournaras et al., 2005). Still, efflux appears to be a minor contributor to carbapenem resistance in this organism, typically operating in conjunction with other mechanisms (Quale et al., 2006;Dotsch et al., 2009;Hammami et al., 2009;Wang et al., 2010). MexAB-OprM has also been implicated in resistance to the penicillin ticarcillin (Boutoille et al., 2004;Cavallo et al., 2007;Hocquet et al., 2007) and its expression linked statistically to aztreonam resistance (Quale et al., 2006). MexXY production, too, has been noted in ticarcillin-resistant P. aeruginosa  although a contribution to resistance was not proven and this efflux system is more commonly associated with resistance to the fourth generation cephalosporin cefepime in clinical isolates (Hocquet et al., 2006;Pena et al., 2009). Indeed, cefepime commonly selects for MexXY-derepressed mutants in vitro (Queenan et al., 2010). MexXY-OprM was also responsible for reduced susceptibility to ceftobiprole in a clinical study of this the novel broad-spectrum cephalosporin (Baum et al., 2009) and mutants expressing mexXY are readily selected by this -lactam in vitro (Queenan et al., 2010). Although MexCD-OprJ accommodates cefepime (Masuda et al., 2000) it has rarely been linked to resistance to this agent in clinical isolates (Jeannot et al., 2008).

Resistance to fluoRoquinolones
Fluoroquinolones (FQs), particularly ciprofloxacin, are commonly used in the treatment of P. aeruginosa infections. Resistance to these agents, particularly high-level resistance, is predominantly mediated by mutations in the DNA gyrase and topoismerase IV enzymes that are the targets of the FQs, though efflux is a significant (reviewed in Queenan and Bush, 2007;Walsh, 2010) and include class A and class D carbapenemases (the latter also referred to as carbapenem-hydrolyzing class D -lactamases, CHDLs; Poirel et al., 2010b) and class B metallo--lactamases (MBLs; reviewed in Walsh et al., 2005), though there are no hitherto reports of CHDLs in P. aeruginosa. Class A -lactamases with activity against carbapenems are uncommon and can be divided into five groups (GES, IMI, KPC, NMC-A, and SME; reviewed in Walther-Rasmussen and Hoiby, 2007) of which only GES and KPC enzymes have been described to date in P. aeruginosa (Zhao and Hu, 2010). KPC enzymes show activity against most -lactams including oxyiminocephalosporins, monobactams, and carbapenems and while they occur as yet rarely in P. aeruginosa (only KPC-2 and KPC-5 have been reported in this organism) the number of reports of KPC-producing P. aeruginosa is increasing (Villegas et al., 2007;Akpaka et al., 2009;Wolter et al., 2009a;Poirel et al., 2010c). Interestingly, KPC-2 is more active against carbapenems than is KPC-5 while the latter shows better activity against ceftazidime (Wolter et al., 2009b). Of note, too, the presence of KPC enzymes in carbapenem-resistant isolates is often coupled with loss of the OprD outer membrane porin (Villegas et al., 2007;Wolter et al., 2009a) that is the primary route of entry of these agents into P. aeruginosa (Trias and Nikaido, 1990). While all GES enzymes are ESBLs three of these also show reasonable activity against carbapenems (GES-2, -4, and -5), with GES-2 and -5 having been reported in P. aeruginosa (Walther-Rasmussen and Hoiby, 2007;Viedma et al., 2009;Wang et al., 2010).
Class B MBLs are by far the major determinants of -lactamasemediated resistance to carbapenems and the major cause of highlevel resistance to these agents. Acquired MBLs include the VIM and IMP enzymes, of which there are numerous variants of the original VIM-1 and IMP-1 MBLs, as well as the SPM-1, GIM-1, NDM-1, AIM-1, and SIM-1 enzymes (Gupta, 2008;Walsh, 2010). The VIM and IMP enzymes are by far the most common MBLs found in carbapenem-resistant bacteria (Walsh et al., 2005), including carbapenem-resistant P. aeruginosa (Gupta, 2008). The predominance of VIM vs. IMP in P. aeruginosa appears to be geographical, with IMP-type MBLs predominating in Asia where it was first discovered and VIM-type enzymes predominating in Europe though both enzymes are now disseminated globally, with VIM-2 in particular well established on five continents (Gupta, 2008;Walsh, 2010;Zhao and Hu, 2010). There are single reports, only, of the GIM-1 (found in five isolates from Germany; Castanheira et al., 2004) and the AIM-1 (Gupta, 2008) MBLs in P. aeruginosa. SPM-1 is the predominant MBL in Brazil (Sader et al., 2005;Picao et al., 2009a) and while previously found only in Brazilian clinical isolates it has now been reported in Europe (Salabi et al., 2010). efflux Five families of efflux systems that export and provide resistance to antimicrobials in bacteria have been described (Li and Nikaido, 2009) although members of the Resistance Nodulation Division (RND) family appear to be the most significant contributors to antimicrobial resistance in P. aeruginosa (Poole, 2004a(Poole, , 2007. There are 12 RND-type efflux systems present in P. aeruginosa of which three, MexAB-OprM, MexCD-OprJ, and MexXY-OprM have been shown to accommodate and provide resistance to -lactams Hyperexpression of this efflux system (and reduction in OprD production) is also seen in lab isolates disrupted in the mexS gene encoding a putative oxidoreductase (a.k.a qrh; Köhler et al., 1999) of unknown function (Sobel et al., 2005). Expression of mexXY is controlled by a single known regulator, the MexZ repressor (Matsuo et al., 2004), and mexZ mutations have been reported in lab-selected FQ-resistant isolates hyperexpressing mexXY . mexXY-hyperexpressing FQ-resistant isolates lacking mutations in mexZ have also been described although the mutation(s) responsible were not identified . Despite its ability to accommodate FQs, however, MexXY-OprM has seldom been linked to FQ resistance in clinical isolates (Wolter et al., 2004).

Resistance to aminoglycosides
A number of aminoglycosides are commonly used in the treatment of P. aeruginosa infections (e.g., tobramycin, gentamicin, amikacin; Gilbert et al., 2003;Bartlett, 2004), particularly pulmonary infections in patients with cystic fibrosis (CF) where amikacin and, in particular, tobramycin are routinely employed (Canton et al., 2005;Taccetti et al., 2008). Their use is, however, linked to resistance development, with acquired aminoglycoside-modifying enzymes (AMEs) and rRNA methylases, and endogenous efflux mechanisms typically responsible (Poole, 2005).
The MexXY-OprM system is encoded by the mexXY operon that is under the control of the MexZ repressor (Matsuo et al., 2004) and the oprM gene of the mexAB-oprM multidrug efflux operon. Mutations in mexZ are common in pan-aminoglycosideresistant CF isolates of P. aeruginosa expressing mexXY (Poole, 2005;Hocquet et al., 2006;Henrichfreise et al., 2007;Islam et al., 2009;Feliziani et al., 2010) with mexZ, in fact, identified as the most commonly mutated gene in CF isolates (Smith et al., 2006;Feliziani et al., 2010). A number of studies highlight, however, the absence of mutations in mexZ or the mexXY promoter region in mexXY-expressing aminoglycoside-resistant CF isolates (Sobel et al., 2003;Hocquet et al., 2006;Islam et al., 2009), indicating that additional genes/mutations are linked to expression of this efflux locus in P. aeruginosa. A recent report of an in vitro-selected mexXY-expressing aminoglycoside-resistant mutant lacking a mexZ mutation identified a novel gene, parR, as the site of mutation (Muller et al., 2010). parR forms part of a two-gene operon, parRS, encoding a two-component regulatory systems that impacts expression of several antimicrobial resistance determinants in P. aeruginosa (e.g., oprD), including mexXY. Significantly, mutations in parR are present in some clinical isolates that express mexXY but lack mutations in mexZ (Muller et al., 2010).

16s rRna methylases
A more recently discovered aminoglycoside resistance mechanism involves methylation of the 16S rRNA of the A site of the 30S ribosomal subunit, which interferes with aminoglycoside binding and so promotes high-level resistance to clinically relevant aminoglycosides like gentamicin, tobramycin, and amikacin (reviewed in Doi and Arakawa, 2007). A number of different pan-aminoglycoside resistance-promoting 16S rRNA methylases have been described in P. aeruginosa, including RmtA (Yamane et al., 2004;Jin et al., 2009), RmtB , RmtD Lincopan et al., 2010), and ArmA (Gurung et al., 2010;Zhou et al., 2010). RmtD is frequently co-produced with the SPM-1 MLB that predominates in Brazil Lincopan et al., 2010) and co-carriage of ArmA and the IMP-1 MBL has also been reported in P. aeruginosa isolates from Korea (Gurung et al., 2010).

Aminoglycoside nucleotidyltransferases
The most prevalent nucleotidyltransferase in P. aeruginosa is the ANT(2′)-I enzyme which inactivates gentamicin and tobramycin but not amikacin and is, thus, found in gentamicin-and tobramycin-resistant clinical isolates (Poole, 2005). A less common nucleotidyltransferases associated with aminoglycoside resistance in P. aeruginosa is ANT(4′)-II which provides resistance to tobramycin and amikacin (Poole, 2005;Ramirez and Tolmasky, 2010). Two variants of this enzyme, ANT(4′)-IIa (Shaw et al., 1993) and -IIb (Sabtcheva et al., 2003) have been described in amikacinresistant clinical isolates and there is a report of an ant(4′)-I gene in P. aeruginosa although its contribution to resistance was not established (Jin et al., 2009). While there are a number of reports of the ANT(3′) nucleotidyltransferase in P. aeruginosa (Ramirez and Tolmasky, 2010) this enzyme is active against streptomycin and none of the clinically used anti-pseudomonal aminoglycosides.

Aminoglycoside phosphoryltransferases
Aminoglycoside phosphoryltransferases found in P. aeruginosa are almost invariably 3′ enzymes that act on the 3-OH of target aminoglycosides and generally provide resistance to aminoglycosides not typically used to treat P. aeruginosa infections (kanamycin, neomycin, and streptomycin; Poole, 2005). APH(3′)-II predominates in clinical isolates resistant to kanamycin (and neomycin; Miller et al., 1994;Poole, 2005) and, indeed, a chromosomal aphA-encoded APH(3′)-II type enzyme, APH(3′)-IIb (Hachler et al., 1996) is likely responsible for the general insensitivity of P. aeruginosa to kanamycin. APH enzymes that provide resistance to other aminoglycosides have also been described in P. aeruginosa and include APH(3′)-VI (amikacin; Kettner et al., 1995;Kim et al., 2008;Jin et al., 2009), APH(3′)-IIb-like (amikacin, weakly; Riccio et al., 2001), and APH(2") (gentamicin and tobramycin; Kettner et al., 1995). efflux Aminoglycoside resistance independent of inactivating enzymes has been known for some time in P. aeruginosa (Bryan et al., 1976). Characterized by resistance to all aminoglycosides and often P. aeruginosa infections in CF (Mulcahy et al., 2010). The idea of a sub-population of biofilm cells displaying different patterns of antimicrobial susceptibility is supported by a recent study showing that only the mobile cells responsible for forming the "cap" component of the typical P. aeruginosa biofilm mushroom structures exhibited tolerance to colistin, as a result of colistin triggering PmrAB-dependent expression of the arn LPS modification locus (Haagensen et al., 2007;Pamp et al., 2008). While the details of persister formation and the mechanism(s) responsible for persister resistance remain unknown, a preliminary screen of a transposon insertion mutant library for mutants showed altered persister formation identified several genes whose disruption either increased or decreased persister formation (De Groote et al., 2009).
Aminoglycosides have been shown to induce biofilm formation by P. aeruginosa, in a process that requires a gene, arr (aminoglycoside response regulator; Hoffman et al., 2005). arr encodes a phosphodiesterase that impacts the levels of bis-(3′,5′)-cyclic-diguanidine monophosphate (c-di-GMP; Hoffman et al., 2005), a second messenger known to influence biofilm formation (Harmsen et al., 2010) and lack of arr compromises biofilm resistance to aminoglycosides (Hoffman et al., 2005). Given that c-di-GMP production is generally correlated with biofilm formation (Harmsen et al., 2010) it is unclear how Arr-promoted turnover of this second messenger would promote biofilm formation. A second gene linked to biofilm-specific resistance to aminoglycosides in some strains only, ndvB, is involved in the synthesis of periplasmic (and intracellular) glucans that bind aminoglycosides (tobramycin), suggestive of a mechanism of resistance whereby aminoglycosides are sequestered and prevented from reaching their targets in the cytosol (Mah et al., 2003). These glucans, which have recently been purified and identified as highly glycerol-phosphorylated -(1 → 3) glucans, actually form part of the biofilm matrix where they do, indeed, bind aminoglycosides (Sadovskaya et al., 2010). A tripartite ABC-family efflux system that is preferentially expressed in biofilm vs. planktonic cells, PA1875-PA1876-PA1877, has also been linked to biofilm-specific aminoglycoside résistance (Zhang and Mah, 2008). Efflux (mediated by MexCD-OprJ) has also been linked to biofilm-specific resistance to azithromycin in P. aeruginosa (Gillis et al., 2005;Mulet et al., 2009).
reports of resistance to both polymyxin B (Landman et al., 2005;Abraham and Kwon, 2009;Barrow and Kwon, 2009) and colistin Matthaiou et al., 2008;Samonis et al., 2010) in clinical isolates. While in many cases the mechanism(s) of clinical polymyxin resistance are unknown, substitution of LPS lipid A with aminoarabinose has been shown to contribute to polymyxin resistance in P. aeruginosa in vitro (Moskowitz et al., 2004) and in CF isolates (Ernst et al., 1999). This modification is carried out by the products of the arnBCADTEF-ugd locus (a.k.a. pmrHFIJKLM-ugd and PA3552-59) that is regulated both by PhoPQ (Macfarlane et al., 2000) and a second two-component regulatory system, PmrAB (McPhee et al., 2003;Moskowitz et al., 2004), with mutations in phoQ and pmrB shown to promote ArnBCADTEF-dependent polymyxin B resistance in clinical isolates (Abraham and Kwon, 2009;Barrow and Kwon, 2009). A third two-component system, ParRS, also controls arnBCADTEF-ugd expression (Fernandez et al., 2010), with a mutation in parR linked to ArnBCADTEF-mediated polymyxin resistance in a lab isolate (Muller et al., 2010). parR (and parS) mutations have been noted in clinical isolates, although there was no indication that the arn locus was upregulated, and the polymyxin resistance of these isolates was minimal (Muller et al., 2010).

biofilm Resistance
Biofilms, surface-attached three-dimensional structures in which bacteria are imbedded in a matrix comprised of polysaccharide, protein, and DNA, are increasingly recognized as the preferred mode of bacterial growth in nature and infectious disease (Lopez et al., 2010). This is true of P. aeruginosa (Harmsen et al., 2010), particularly in the case of pulmonary infections in patients with CF (Wagner and Iglewski, 2008;Davies and Bilton, 2009). An important consequence of P. aeruginosa biofilm growth and one that is particularly relevant in a clinical context is marked resistance to antimicrobial agents (Davies and Bilton, 2009;Hoiby et al., 2010). Antimicrobial resistance of P. aeruginosa biofilms appears to be complex, multifactorial, and in many instances not well understood (Drenkard, 2003;Hoiby et al., 2010). Some studies indicate that P. aeruginosa within biofilms are metabolically less active and grow more slowly than cells at the biofilm periphery (owing to limited access to nutrients and oxygen; Werner et al., 2004), which may contribute to increasing biofilm tolerance to antimicrobials since antimicrobials often target metabolically active cells (Pamp et al., 2008). Certainly, the suggestion that biofilm-grown P. aeruginosa from CF patients are anaerobic (Hassett et al., 2009) is likely to be significant in the context of antimicrobial resistance since many agents are inactive or less active under anaerobiosis (Schobert and Tielen, 2010). Oxygen limitation has, in fact, been shown to contribute significantly to the antimicrobial resistance of in vitro-grown P. aeruginosa biofilms (Borriello et al., 2004).
One explanation for biofilms being generally refractory to antimicrobial chemotherapy is the presence, in biofilms, of a highly resistant sub-population of cells called persisters (Lewis, 2008). Intriguingly, "late" isolates of P. aeruginosa in CF (those recovered later in infection) produce increased levels of drug-tolerant persister cells, which may be the primary "mechanism" for surviving chemotherapy and, so, may explain the general recalcitrance of Alterations in two-component ity (e.g., the siderophore-monobactam hybrid, BAL30072, the anti-pseudomonal cephalosporin CXA-101, and the MBL inhibitor ME1071) are, unfortunately, negatively impacted by known resistance mechanisms (Page and Heim, 2009). While the lack of classical antimicrobial options has prompted research into novel anti-pseudomonal strategies/agents, including a humaneered anti-P. aeruginosa Fab antibody fragment, KB001, cationic antimicrobial peptides, efflux pump inhibitors, modulators of virulence (Page and Heim, 2009;Veesenmeyer et al., 2009), and phage therapy (Wright et al., 2009), only KB001 is in later stage clinical trials (Page and Heim, 2009). Clearly, more therapeutic options are needed. Given the resistance armamentarium available to P. aeruginosa and the observation that drug use begets resistance, more also needs to be done in the areas of antimicrobial stewardship, resistance surveillance, and infection control (Kerr and Snelling, 2009). With limited (and shrinking) options, and an environment where anti-infectives, generally, are not being developed and fewer and fewer resources are being devoted to this therapeutic area by the major pharmaceutical companies (Boucher et al., 2009) prudent management of available agents and more robust resistance monitoring and infection control practices are essential. While these will likely not prevent the rise of untreatable pan-resistant P. aeruginosa, hopefully their numbers and impact can be limited.

acknowledgments
Work on antimicrobial resistance in the Poole lab is supported by operating grants from the Canadian Cystic Fibrosis Foundation and the Canadian Institutes of Health Research.
A second DNA repair system less commonly linked to the mutator phenotype in P. aeruginosa is the DNA oxidative repair (GO) system charged with repairing and preventing incorporation into DNA of an oxidatively damaged form of guanosine (8-oxo-2′deoxyguanosine, 8-oxodG; Oliver and Mena, 2010). In vitro studies have shown that knockouts in the GO genes mutT and mutY yield increased mutation rates concomitant with increased oxidative damage of DNA (Mandsberg et al., 2009), with mutT (Mandsberg et al., 2009;Morero and Argarana, 2009) and mutY (Mandsberg et al., 2009) strains also showing higher rates of antimicrobial resistance. Given that the characteristically chronically inflamed CF lung is an environment rich in reactive oxygen species (ROS) that can damage DNA, the potential for ROS-promoted hypermutability owing to defects in the GO system is certainly real. Although uncommon, mutator strains with lesions in mutT and mutY have been recovered from CF patients (Mandsberg et al., 2009).

concluding RemaRks
Rates of infection and resistance are increasing in P. aeruginosa (Talbot et al., 2006;Kerr and Snelling, 2009), and with reports of colistin-only sensitive P. aeruginosa and the presence of colistinresistance in this organism the untreatable P. aeruginosa infection may be imminent. Compounding the increasing lack of effective anti-pseudomonal agents is the paucity of new drugs being developed that are active against P. aeruginosa and, indeed, the absence of any late-stage agents effective against pan-resistant P. aeruginosa (Talbot et al., 2006;Boucher et al., 2009;Page and Heim, 2009). The few novel agents with anti-pseudomonad activ-