Journal of Antimicrobial Chemotherapy

JAC Advance Access originally published online on May 24, 2005
Journal of Antimicrobial Chemotherapy 2005 56(1):20-51; doi:10.1093/jac/dki171

This Article Abstract FREE Full Text (PDF) All Versions of this Article: 56/1/20    most recent dki171v1 Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Add to My Personal Archive Download to citation manager Search for citing articles in: ISI Web of Science (64) Request Permissions Disclaimer Google Scholar Articles by Poole, K. Search for Related Content PubMed PubMed Citation Articles by Poole, K. Social Bookmarking          What's this?

© The Author 2005. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved. For Permissions, please e-mail: journals.permissions@oupjournals.org

Review

Efflux-mediated antimicrobial resistance

Keith Poole^*

Department of Microbiology & Immunology, Queen's University, Kingston, ON, Canada K7L 3N6

---

^* Tel: +1-613-533-6677; Fax: +1-613-533-6796; E-mail: poolek{at}post.queensu.ca

	Abstract

Top Abstract Introduction Efflux-mediated resistance to... Efflux-mediated resistance to... Evolution and natural function... Overcoming efflux-mediated... Concluding remarks References

 
Antibiotic resistance continues to plague antimicrobial chemotherapy of infectious disease. And while true biocide resistance is as yet unrealized, in vitro and in vivo episodes of reduced biocide susceptibility are common and the history of antibiotic resistance should not be ignored in the development and use of biocidal agents. Efflux mechanisms of resistance, both drug specific and multidrug, are important determinants of intrinsic and/or acquired resistance to these antimicrobials, with some accommodating both antibiotics and biocides. This latter raises the spectre (as yet generally unrealized) of biocide selection of multiple antibiotic-resistant organisms. Multidrug efflux mechanisms are broadly conserved in bacteria, are almost invariably chromosome-encoded and their expression in many instances results from mutations in regulatory genes. In contrast, drug-specific efflux mechanisms are generally encoded by plasmids and/or other mobile genetic elements (transposons, integrons) that carry additional resistance genes, and so their ready acquisition is compounded by their association with multidrug resistance. While there is some support for the latter efflux systems arising from efflux determinants of self-protection in antibiotic-producing Streptomyces spp. and, thus, intended as drug exporters, increasingly, chromosomal multidrug efflux determinants, at least in Gram-negative bacteria, appear not to be intended as drug exporters but as exporters with, perhaps, a variety of other roles in bacterial cells. Still, given the clinical significance of multidrug (and drug-specific) exporters, efflux must be considered in formulating strategies/approaches to treating drug-resistant infections, both in the development of new agents, for example, less impacted by efflux and in targeting efflux directly with efflux inhibitors.

Keywords: efflux , resistance , antimicrobials , antibiotics , biocides , multidrug

	Introduction

Top Abstract Introduction Efflux-mediated resistance to... Efflux-mediated resistance to... Evolution and natural function... Overcoming efflux-mediated... Concluding remarks References

 
While antimicrobials have proven invaluable in the management of bacterial infectious disease, resistance to these agents actually predates the introduction of the first true antibiotic (penicillin) into clinical usage,¹ and resistance continues to compromise the use of old and new antimicrobials alike.^2–8 The clinical impact of resistance is immense, characterized by increased cost, length of hospital stay and mortality,^9–19 often as a result of inappropriate initial antimicrobial therapy.^19–24 Resistance to antibiotics occurs typically as a result of drug inactivation/modification, target alteration and reduced accumulation owing to decreased permeability and/or increased efflux.^25–27 It may be an innate feature of an organism or, when it is not, occurs as the result of mutation or the acquisition of exogenous resistance genes.^28,29 Specific growth states (e.g. biofilm formation^30–34 and anaerobiosis^35,36) can also negatively impact antimicrobial susceptibility. While biocidal agents generally remain effective at ‘at use’ concentrations, numerous mechanisms of reduced susceptibility have, nonetheless, been reported in bacteria.²⁵ This review provides an overview of efflux determinants of antimicrobial (antibiotic and biocide) resistance, both agent-specific and multidrug, emphasizing recent advances and discussing all efflux mechanisms as determinants of resistance to specific, clinically-relevant antimicrobials. It is hoped that this will provide some insights vis-à-vis the probable clinical significance of drug-specific versus multidrug efflux systems as regards resistance to a given antimicrobial. While the emphasis is on the clinical relevance of efflux mechanisms of resistance, the probable role of Gram-negative multidrug efflux systems in other cellular processes is also addressed. The interested reader is referred to recent reviews of antimicrobial³⁷ and multidrug^37–41 efflux for additional information.

	Efflux-mediated resistance to antibiotics

Top Abstract Introduction Efflux-mediated resistance to... Efflux-mediated resistance to... Evolution and natural function... Overcoming efflux-mediated... Concluding remarks References

 
The last of the resistance mechanisms to be identified, efflux was first described as a mechanism of resistance to tetracycline in Escherichia coli,^42,43 with the plasmid-encoded single component Tet protein export of tetracycline (complexed with Mg²⁺ it turns out) across the cytoplasmic membrane sufficient for resistance. In the intervening years, numerous plasmid- and chromosome-encoded efflux mechanisms, both agent- or class-specific and multidrug have been described in a variety of organisms where they are increasingly appreciated as important determinants of antimicrobial resistance. Bacterial efflux systems capable of accommodating antimicrobials generally fall into five classes, the major facilitator (MF) superfamily, the ATP-binding cassette (ABC) family, the resistance-nodulation-division (RND) family, the small multidrug resistance (SMR) family [a member of the much larger drug/metabolite transporter (DMT) superfamily] and the multidrug and toxic compound extrusion (MATE) family (see Reference ⁴⁴ for an in-depth review of drug efflux families) (Figures 1 and 2). Though not unique to Gram-negative bacteria, RND family transporters are most commonly found in such organisms,⁴⁵ and typically operate as part of a tripartite system that includes a periplasmic membrane fusion protein (MFP) and an outer membrane protein [now called outer membrane factor (OMF)], an organization also seen on occasion with ABC [e.g. the macrolide-specific MacAB-TolC efflux system (Table 1)] and MF [e.g. the VceAB multidrug efflux system of Vibrio cholerae⁴⁶] family exporters (Figure 2). Members of all but the ABC family (whose members hydrolyse ATP to drive drug efflux) function as secondary transporters, catalysing drug–ion (H⁺ or Na⁺) antiport (Figures 1 and 2). Drug efflux systems can be drug-/class-specific as for the original Tet pump and the more recently described Mef exporters of macrolides and various chloramphenicol exporters (reviewed briefly in Reference ⁴⁷) or capable of accommodating a range of chemically-distinct antimicrobials as for the chromosomally-encoded NorA-like MF transporters prevalent in Gram-positive bacteria or RND transporters of Gram-negative bacteria (Tables 1 and 2).

View larger version (16K): [in this window] [in a new window]   Figure 1.. Schematic diagram of representative drug exporting systems in Gram-positive bacteria, highlighting the different families of pumps involved in resistance. FQ, fluoroquinolone; CM, chloramphenicol; TC, tetracycline; ML, macrolides, MD, multidrug. While NorA is, strictly speaking, a multidrug transporter, it exports only FQs (and biocides) as clinically relevant agents and so it is highlighted here as an MF family efflux determinant of FQ resistance.

 

View larger version (48K): [in this window] [in a new window]   Figure 2.. Schematic diagram of representative drug exporting systems in Gram-negative bacteria, highlighting the different families of pumps involved in resistance. (a) Pumps localized solely to the cytoplasmic membrane; (b) Pumps that span the cell envelope. FQ, fluoroquinolone; CM, chloramphenicol; TC, tetracycline; ML, macrolides; BL, ß-lactams. While many of the indicated efflux system are actually multidrug exporters, they are highlighted here for their contribution to resistance to specific agents because their expression in mutants is reportedly selected by those agents. They have been shown to contribute to meaningful resistance to the indicated agent in clinical isolates or the indicated agent is the only clinically-relevant antimicrobial exported by the efflux system.

 

View this table: [in this window] [in a new window]   Table 1.. Efflux-mediated resistance to non-fluoroquinolone antimicrobials

 

View this table: [in this window] [in a new window]   Table 2.. Efflux-mediated resistance to fluoroquinolonesa

 
Chloramphenicol

Owing to a number of adverse affects associated with its use, chloramphenicol is now used sparingly in human medicine and is restricted in veterinary medicine to pets and non-food-producing animals.⁴⁸ The fluorinated analogue, florfenicol, is, however, used in cattle, pigs and salmon, though not human medicine.⁴⁸ Non-enzymic resistance to chloramphenicol has been known for >25 years, being first described in Pseudomonas aeruginosa carrying transposon Tn1696, an element later shown to encode the CmlA chloramphenicol exporter of the MF superfamily⁴⁹ (Table 1). Related MF-family chloramphenicol exporters have been reported in a number of Gram-negative bacteria, usually encoded by genes present on mobile elements (i.e. plasmids, transposons, integrons) that often carry additional resistance genes.^50–58 This may explain the persistence of this (and other) chloramphenicol resistance determinants in, for example, food animals⁵⁹ despite the longstanding prohibition of chloramphenicol use in these animals. Although the nomenclature of chloramphenicol efflux genes has not been standardized, which complicates ready appreciation of relationships between these determinants, several distinct groups of chloramphenicol exporters have been identified in bacteria, with a number of unrelated exporters described in Gram-positive bacteria (i.e. Streptomyces, Corynebacterium and Rhodococcus spp.). The CmlA family remains the largest family of chloramphenicol-specific exporters although an equally large group of transporters, also unique to Gram-negative bacteria, accommodates both chloramphenicol and the fluorinated chloramphenicol analogue, florfenicol (usually designated Flo) (see Reference ⁴⁸ for a review of chloramphenicol and florfenicol exporters). The Flo transporters can be plasmid- or chromosome-encoded, though the flo determinants are typically associated with mobile genetic elements that, as is the case for cmlA determinants, carry additional resistance genes.^60–63 flo is an important determinant of florfenicol resistance in animal isolates of E. coli^64–68 but is also found in human pathogens (e.g. Salmonella enterica^60–62,69 and V. cholerae⁶³). Recently, a unique plasmid- and transposon-encoded chloramphenicol–florfenicol exporter, FexA, was described in Staphylococcus lentus, the first report of a chloramphenicol–florfenicol exporter in Gram-positive bacteria.^70,71 Interestingly, too, fexA expression is inducible, with FexA-mediated resistance to chloramphenicol and florfenicol increasing 2- to 4-fold in the presence of these agents.⁷⁰ This is reminiscent of chloramphenicol induction of the chloramphenicol efflux genes of Tn1696⁷² and Rhodococcus fascians,⁷³ which seems to occur via a posttranscriptional translational attenuation mechanism.

In addition to chloramphenicol/florfenicol-specific exporters, a number of chromosome-encoded, broadly-specific, multidrug transporters of the RND family that are widely distributed amongst Gram-negative bacteria⁷⁴ have been shown to accommodate chloramphenicol.^48,74–77 Moreover, mutant derivatives of P. aeruginosa that hyperexpress the RND family MexEF-OprN multidrug efflux system, which exports and provides resistance to fluoroquinolones and trimethoprim, in addition to chloramphenicol, are readily selected in vitro using chloramphenicol.^78–80 In vitro-selected chloramphenicol-resistant Burkholderia cepacia (cenocepacia) also shows a multidrug resistance phenotype⁸¹ that might be explained by production of this organism's RND family multidrug transporter, CeoAB-OpcM,⁸² inasmuch as CeoAB-OpcM is known to export chloramphenicol and has been implicated in the chloramphenicol resistance of a clinical (cystic fibrosis) isolate of this organism.⁸³ Recently, a plasmid-encoded RND family multidrug transporter responsible for resistance of a swine isolate of E. coli to the growth enhancer olaquindox was described that also provided for substantial resistance to chloramphenicol.⁸⁴ The MF family multidrug transporters MdfA of E. coli and VceAB of V. cholerae⁷⁴ and the ABC type multidrug transporter LmrA of Lactococcus lactis⁸⁵ have been reported to accommodate chloramphenicol as well. Although infrequently tested, RND family exporters also appear to export florfenicol.^75,86

Tetracyclines

Tetracyclines were discovered in the 1940 s and have been used clinically to treat a variety of infections since the 1950 s and are still widely used today.^87,88 Since the original reports of TetA-mediated efflux of and resistance to tetracycline in E. coli, numerous other Tet proteins have been described, in Gram-negative and Gram-positive organisms, with at least 20 different types presently known (reviewed in References ⁴⁷ and ⁸⁹) (Table 1). These MF family exporters are almost invariably encoded by genes present on mobile genetic elements (plasmids, transposons, IS elements and integrons^47,87), often together with additional resistance genes (e.g. in multidrug-resistant S. enterica,^{50,54–56,58,60,62} E. coli⁹⁰ and V. cholerae⁶³), although those found in tetracycline-producing microorganisms (as probable mechanisms of self-defence, see below) are chromosome-encoded.^91–93 Recently, a chromosomal tet efflux determinant, tet38, was reported in Staphylococcus aureus, its overexpression in a mutant strain providing for markedly enhanced tetracycline resistance.⁹⁴ The vast majority of the different Tet proteins have been reported in Gram-negative bacteria (includes the Tet proteins A–E, G, H, I, J, Z and Tet(30) which are found exclusively in Gram-negative bacteria^47,89) with two, TetK and TetL, found predominantly in Gram-positive bacteria.⁸⁹ Novel Tet efflux proteins, Tet(V)⁹⁵ and Tet(39)⁹⁶ have also been reported in Mycobacterium smegmatis and Mycobacterium fortuitum⁹⁵ and Acinetobacter spp.,⁹⁶ respectively. Still, efflux is a very uncommon mechanism of tetracycline resistance in Gram-positive pathogens (e.g. Streptococcus^97–104 and Enterococcus^105,106 spp.) though it is reported in pathogenic staphylococci,^107–109 especially veterinary isolates where it is very common.^110,111 A recent study of tetracycline resistance in Enterococcus faecalis isolated from raw food did, however, note a high frequency of isolates with tetL.¹¹² In Gram-negative pathogens, efflux is the predominant mechanism of tetracycline resistance in several organisms (e.g. Salmonella spp.,^{58,107,113–116} Shigella spp.,^117,118 E. coli,^117,119,120 Acinetobacter spp.¹²¹ and Chlamydia spp.¹²²) and has been noted in Helicobacter pylori,¹²³ but not, for example, in Campylobacter¹²⁴ or Neisseria^125,126 spp. Efflux determinants are also present in tetracycline-resistant fish pathogens of Photobacterium, Vibrio, Pseudomonas, Alteromonas, Citrobacter and Salmonella spp.¹²⁷

The Tet efflux proteins typically export and provide resistance to tetracycline, oxytetracycline and chlortetracycline, with TetB and TetL the only known Tet family exporters of minocycline.^47,87 Unfortunately, however, TetB has the widest host range among Gram-negative pathogens.⁸⁷ Expression of the various tet genes is inducible by tetracyclines, although the induction mechanisms differ for Gram-negative versus Gram-positive tetracycline efflux genes; Gram-negative tet genes are controlled by the tetracycline responsive TetR repressor whereas Gram-positive tet genes are controlled by a ‘translational attenuation’ mechanism.⁴⁷

Many of the RND family multidrug resistance efflux systems of Gram-negative bacteria also accommodate tetracyclines^74–77,128 as do the MF family MdfA,⁷⁴ and the Tap (also known as RV1258c)^129,130 and P55¹³¹ multidrug exporters of E. coli and the mycobacteria, respectively (Table 1). The L. lactis ABC family multidrug transporter LmrA⁸⁵ and the E. coli SMR family multidrug transporter EmrE⁷⁴ also export tetracyclines as does the recently described MF family exporter NorB of S. aureus, though only weakly.⁹⁴ Still, there are few reports of multidrug exporters as primary determinants of tetracycline resistance (i.e. selected by tetracyclines in vitro or in vivo) although tetracycline selection of RND family MexAB-OprM-overproducing multidrug-resistant isolates of P. aeruginosa has been demonstrated in vitro.^78,132,133 Similarly, multidrug-resistant Stenotrophomonas maltophilia, including mutants overproducing the RND family SmeDEF efflux system¹³⁴ can also be selected with tetracycline in vitro.^135,136 Tetracycline has also been shown to positively influence expression of the mexXY genes encoding an RND family multidrug efflux system that contributes to intrinsic resistance to this agent as well as aminoglycosides and macrolides in P. aeruginosa.¹³⁷

Macrolide–lincosamide–streptogramin antibiotics

Antibiotics of the macrolide–lincosamide–streptogramin (MLS) group¹³⁸ are employed widely in the treatment of Gram-positive infections (mainly staphylococci and streptococci) and infections caused by anaerobic microorganisms.^88,139–143 While most Gram-negative bacteria and the mycobacteria are generally resistant to these agents, there are notable exceptions (macrolide susceptibility of mycobacteria¹⁴⁴ and macrolide/lincosamide susceptibility of Bordetella, Campylobacter, Chlamydia, Helicobacter, Legionella, Neisseria, Haemophilus and Moraxella spp.¹⁴⁰). A variety of individual efflux mechanisms of resistance to one or more of these agents has been reported (reviewed in References ⁴⁷ and ¹⁴⁵), typically of the MF and ABC families of drug exporters (Table 1). A common determinant of macrolide-specific (14- and 15-membered macrolides only) resistance/efflux (M resistance phenotype) is the MF family Mef(A) exporter first identified in Streptococcus pyogenes but widely distributed amongst Gram-positive bacteria and also present in Gram-negative organisms.^145,146 A highly related (91% amino acid sequence identity) efflux system, Mef(E), has been described in Streptococcus pneumoniae, and while these determinants are still often referred to separately in the literature, they probably represent a common macrolide-specific efflux system that has been suggested be jointly dubbed Mef(A).¹⁴⁷ Still, in light of differences in genetic context, distribution [mef(E) is disseminated in many more species] and impact on macrolide MICs [mef(A)-containing isolates of S. pneumoniae show higher MICs than do mef(E)-containing isolates], some researchers favour maintaining separate designations for these mef determinants.^147a In the streptococci, Mef(A) is typically encoded by genes carried by mobile elements present in the chromosome,¹⁴⁶ possibly associated with prophages.¹⁴⁸ It represents a significant determinant of macrolide resistance in Streptococcus spp., particularly in S. pneumoniae^149–162 but also in S. pyogenes^152,162,163 and other Streptococcus spp.,^147,163,164 and appears to be common in commensal viridans group streptococci^162,165 (see References ¹⁶⁶ and ¹⁶⁷ for reviews of macrolideresistance, including efflux, in streptococci). While the mef(A) gene has been infrequently reported in Gram-negative bacteria (e.g. Neisseria gonorrhoeae¹⁶⁸), including anaerobes (e.g. Bacteroides spp.¹⁶⁹), a recent study of randomly-selected Gram-negative commensal bacteria of 13 different genera obtained from healthy children revealed that fully 41% carried the mef(A) gene.¹⁷⁰ Recently, too, the mef(A) gene was identified in Neisseria spp. isolated in the 1950 s and 1960 s, representing the oldest of any species to carry this gene.¹⁷¹ A chromosomal mef(A)-like gene (30% identical; 45% similarity at the amino acid level), cme, has recently reported in Clostridium difficile where it contributes to erythromycin resistance.¹⁷²

Inducible resistance to both erythromycin and type B streptogramins (MS_B resistance phenotype) reported in Staphylococcus spp. is also attributable to a putative efflux mechanism encoded by the plasmid-borne msr(A) gene.¹⁷³ This ABC family protein contains the two prototypical ATP-binding domains but lacks any obvious membrane-spanning domains, raising questions about its ability to function as a drug exporter. Still, studies have demonstrated that Msr(A)-containing Staphylococcus spp. exclude/show reduced accumulation of erythromycin that was energy-dependent and abolished by classical ABC inhibitors like arsenate.¹⁷³ One possibility is that Msr(A) associates with another protein that provides the necessary transmembrane domains for drug export.¹⁷³ Msr(A) is implicated in the resistance of clinical isolates of S. aureus,¹⁷⁴ MSSA in particular,¹⁷⁵ but also coagulase-negative staphylococci.¹⁷⁴

Several msr(A)-like elements have been described in Gram-positive bacteria including the plasmid-borne vga(A), a vag(A) variant, vga(A)_v and vga(B) determinants of streptogramin A (e.g. dalfopristin) resistance found on mobile genetic elements in S. aureus,^176–179 and the chromosomal msr(C) determinant prevalent in E. faecalis and associated with modest resistance to macrolides (including 16-membered macrolides) and group B streptogramins.^180–182 A recent study has now confirmed the ability of vag(A) and vga(A)_v to confer low-level lincosamide resistance in S. aureus and Staphylococcus epidermidis, and it has been suggested that the LS_A phenotype occasionally found in staphylococcal isolates may be due to these elements.¹⁸³ This same study also reported a novel finding that vga(B) confers only low-level resistance to group A streptogramins but substantially increases resistance levels to pristinamycin, a mixture of streptogramin A and streptogramin B compounds. Moreover, a chromosomal gene encoding a Vga(A)-like molecule has been reported in Listeria monocytogenes, Lmo0919, and shown, when cloned into a plasmid, to confer resistance to group A streptogramins and lincosamides in Staphylococcal spp.¹⁸³ A msr(A)-like gene, now called msr(D),¹⁴⁹ is invariably associated with the genetic element(s) that carry the mef(A/E) gene in S. pneumoniae,^149,153 S. pyogenes¹⁰² and group A Streptococcus,¹⁸⁴ though msr(D) (from S. pneumoniae) alone is sufficient for macrolide resistance.¹⁴⁹ Interestingly, Msr(D) also promotes resistance to telithromycin, a ketolide, but not streptogramins, distinguishing it from Msr(A).¹⁴⁹ msr(D) is also associated with mef(A) genes found in a variety of commensal Gram-negative bacteria.¹⁷⁰

Efflux determinants of lincosamide resistance have also been reported, including the chromosomal lsa gene of E. faecalis responsible for the characteristic intrinsic resistance of this organism to lincosamides and streptogramins A (LSA resistance phenotype).^173,185,186 Like Msr(A), this ABC family resistance protein lacks the usual transmembrane domains associated with ATP-dependent transporters and an efflux mechanism of resistance has yet to be proved. A related determinant (41% identical; 69% similar to Lsa) of low-level clindamycin resistance (cloned gene increased MIC of S. aureus 16-fold), Lsa(B), has been reported on a plasmid in Staphylococcus sciuri.¹⁸⁷ An uncharacterized macrolide efflux mechanism distinct from mef(A) and msr(A/D) and suggested to export 14-, 15- and 16-membered macrolides (and to a limited extent, ketolides) has been reported in S. pyogenes strains inducibly resistant to MLS antimicrobials.¹⁸⁸

Spontaneous lincomycin- and puromycin-resistant mutants of Bacillus subtilis showing elevated expression of a gene, lmrB, encoding a putative MF family multidrug exporter have been described.^189,190 The lmrB gene occurs in an operon with lmrA, which encodes probable repressor protein. A like-named gene has also been reported as a determinant of efflux-mediated lincomycin resistance in Corynebacterium glutamicum.¹⁹¹ Other multidrug exporters that accommodate MLS antimicrobials include the LmrP (MF family) and LmrA (ABC family) exporters of Lactococcus lactis, which export macrolides, lincosamides and streptogramins,⁸⁵ although this is of no clinical relevance, L. lactis being a non-pathogenic microorganism. The recently identified chromosome-encoded MF family multidrug exporter, MdeA, of S. aureus also accommodates macrolides (erythromycin), lincosamides (lincomycin) and streptogramins A (virginiamycin), and mutants overexpressing the mdeA gene show modestly increased (2-fold) resistance to lincomycin and virginiamycin.¹⁹² Moreover, homologues of this gene have been identified in Staphylococcus haemolyticus, Bacillus cereus and B. subtilis and all provide for modest (2- to 8-fold) increases in resistance to all three agents when expressed from plasmids.¹⁹²

The ABC family MacAB-TolC system of E. coli is specific for macrolides (14- and 15-membered macrolides) and is unique in being the only known chromosome-encoded ABC type efflux system in Gram-negative bacteria that operates with MFP and OMF components (a plasmid-encoded ABC exporter associated with fluoroquinolone resistance in a Pseudomonas spp. operates with a MFP and, probably, OMF component¹⁹³). Not surprisingly, given the broad substrate specificity of this family, many RND type multidrug exporters of Gram-negative bacteria accommodate macrolides^{74,77,145,194,195} and, where tested, lincosamides,¹⁹⁶ probably explaining, at least in part, the general insusceptibility of many of these bacteria to these agents. Loss of AcrAB-TolC in E. coli had a modest (4-fold) impact on ketolide (telithromycin) resistance although treatment of E. coli or E. aerogenes with the efflux inhibitor Phe-Arg-ß-naphthylamide (PAßN) had a much more marked impact on ketolide (and macrolide) resistance (128- to 512-fold decrease) indicating that additional, presumed efflux mechanism(s) of macrolide and ketolide resistance occur in these enteric organisms.¹⁹⁵ Expression of the RND family MtrCDE multidrug efflux system of N. gonorrhoeae has been reported in clinical isolates displaying reduced susceptibility to azithromycin and/or erythromycin,^197,198 indicating that this multidrug transporter can be a determinant of acquired macrolide resistance in Neisseria. Studies on macrolide resistance in Haemophilus influenzae also implicated this organism's three-component RND family multidrug transporter, AcrAB-TolC, as a co-determinant of intrinsic and acquired macrolide resistance,¹⁹⁹ including high-level macrolide resistance.²⁰⁰ A plasmid encoded resistance determinant showing substantial similarity (61–80% identity amongst the three components) to the RND family MexCD-OprJ multidrug efflux system of P. aeruginosa and providing resistance to the macrolides erythromycin and roxithromycin has been reported in an environmental Pseudomonas spp. This is the first example of a naturally-occurring plasmid-encoded RND family multidrug transporter.²⁰¹ The MdfA and VceB multidrug exporters of E. coli and V. cholerae, respectively, also accommodate macrolides.⁷⁴ Erythromycin resistance reversible by the efflux inhibitor PAßN has been reported in clinical Campylobacter spp. suggestive of an efflux mechanism of resistance,²⁰² and the CmeABC RND-type pump of C. jejuni has been shown to accommodate erythromycin.²⁰³

Fluoroquinolones

Fluoroquinolones are an evolving class of antimicrobial²⁰⁴ that enjoys a broad spectrum of activity against Gram-positive, Gram-negative and mycobacterial pathogens.^88,205–208 Unlike most other efflux mechanisms, which are agent- or class-specific and encoded by genes present on mobile genetic elements, often plasmids, efflux determinants of fluoroquinolone resistance are almost invariably multidrug transporters encoded by endogenous chromosomal genes.^74,209,210 In Gram-positive bacteria, the most significant efflux determinants of fluoroquinolone resistance are MF family efflux systems that are invariably homologues of the well-characterized NorA exporter found in S. aureus (Table 2) (see References ²⁰⁹ and ²¹¹ for more in-depth reviews of efflux-mediated fluoroquinolone resistance in Gram-positive bacteria). These exporters tend to provide modest resistance to older fluoroquinolones (e.g. norfloxacin, ciprofloxacin) but not newer agents of this class, although uncharacterized efflux mechanisms of broader range fluoroquinolone resistance have been reported in both S. aureus²¹² and S. pneumoniae.²¹³ Recently, a second NorA-like (30% amino acid similarity) multidrug exporter of fluoroquinolones, NorB, was described in S. aureus, providing resistance to a broader range of fluoroquinolones that included sparfloxacin and moxifloxacin⁹⁴ and may, in fact, explain the previously seen broad range fluoroquinolone efflux activity mentioned above. Unlike multidrug efflux determinants of fluoroquinolone resistance in Gram-negative bacteria, which export and provide resistance to multiple classes of clinically-relevant antimicrobials (see below), fluoroquinolone-exporting multidrug transporters of Gram-positive bacteria generally export fluoroquinolones as the lone clinically-relevant agent; most of their substrates not being classical antimicrobials.^74,209 Unlike resistance attributable to mobile, agent-specific exporters, where resistant strains typically acquire the resistance genes, fluoroquinolone resistance owing to endogenous multidrug transporters typically arises from increased expression of the efflux genes. Enhanced expression of the norA gene has been reported in fluoroquinolone-resistant laboratory^214,215 and clinical^216,217 isolates of S. aureus, possibly owing to mutations on the norA promoter²¹⁸ (see also Reference ²⁰⁹ for additional citations). Mutations in the 5' untranslated region of norA can be associated with enhanced stability of the mRNA, leading to increased steady-state levels of the message, effectively enhancing norA expression.^219,220 Resistance to fluoroquinolones^212,221,222 and non-fluorinated quinolones²²³ independent of NorA but attributable, at least in part, to an efflux mechanism has also been reported in S. aureus.

An efflux contribution to fluoroquinolone resistance has also been noted in S. pneumoniae,^224,225 including clinical strains,^{150,226–228} attributable in some instances to the MF family PmrA exporter (reviewed in Reference ²⁰⁹). Efflux-mediated fluoroquinolone resistance independent of PmrA has, however, also been reported in this organism.²¹³ Highlighting the significance of efflux vis-à-vis fluoroquinolone resistance, efflux was shown to enhance survival of S. pneumoniae in a ciprofloxacin-treated mouse model of pneumonia.²²⁹ Efflux-mediated fluoroquinolone resistance has also been seen in viridans group streptococci.^230,231 As in S. aureus, efflux-mediated resistance to fluoroquinolones in S. pneumoniae appears also to be limited to older fluoroquinolones.^224,229,232 The Lde gene product of L. monocytogenes, showing 44% identity with PmrA, was shown to contribute to fluoroquinolone resistance of clinical isolates.²³³ Wild-type strains of enterococci have been shown to efflux fluoroquinolones,²³⁴ and a gene encoding a NorA homologue, emeA, has been identified and shown to contribute to intrinsic fluoroquinolone resistance in E. faecalis.^235,235a Efflux is also implicated in resistance to ciprofloxacin and norfloxacin but not to newer fluoroquinolones in in vitro-isolated fluoroquinolone-resistant Bacillus anthracis,^236,237 and fluoroquinolone efflux mechanisms have been described in other Bacillus spp. [e.g. B. subtilis where 3 MF family exporters capable of accommodating fluoroquinolones have been described (Table 2)]. A gene encoding a homologue of one of these, BmrA, has been identified in the B. anthracis genome,²³⁸ although its contribution, if any, to fluoroquinolone resistance remains to be tested.

In addition to MF family fluoroquinolone exporters, a limited number of ABC family fluoroquinolone/multidrug exporters have been reported in Gram-positive bacteria, including the EfrAB ciprofloxacin/norfloxacin exporter of E. faecalis and the well characterized LmrA pump of L. lactis, the model bacterial ABC type multidrug transporter (Table 2) (reviewed in References ⁴⁰, ⁸⁵ and ²³⁹). While LmrA is not relevant clinically, occurring as it does in a milk spoilage organism, it is noteworthy for its very broad range of substrates that include, in addition to fluoroquinolones, numerous clinically-relevant antimicrobials,⁸⁵ and its functional similarly to P-glycoprotein, the mammalian multidrug transporter.²³⁹ An in vitro-selected ethidium bromide-resistant mutant of Mycoplasma hominis displaying a multidrug-resistant phenotype and reduced ciprofloxacin accumulation was shown to overexpress two genes encoding a putative ABC type efflux,^240,241 another example, then, of a bacterial ABC exporter accommodating fluoroquinolones.

The MF family LfrA exporter was the first mycobacterial efflux determinant of fluoroquinolone resistance to be identified although numerous efflux determinants of fluoroquinolone resistance (low level) have now been described in the mycobacteria²⁰⁹ (see also Reference ⁹⁵ for a review of mycobacterial efflux pumps, including those contributing to fluoroquinolone resistance) (Table 2). Most of these putative exporters are of the MF family, although a cloned SMR family exporter, Mmr, provided a modest (2-fold) increase in MIC to ciprofloxacin and norfloxacin,²⁴² and a cloned ABC exporter, Rv2686c-2687c-2688c, increased MICs 2-fold (norfloxacin, sparfloxacin and moxifloxacin) to 8-fold (ciprofloxacin).²⁴³ But while efflux has been implicated in fluoroquinolone resistance in, for example, M. smegmatis,²⁴⁴ evidence for a contribution of known fluoroquinolone exporters to resistance in laboratory or clinical isolates is generally lacking—contributions to resistance are typically observed in strains harbouring genes cloned onto multicopy plasmids. Still, a recent report highlighting the enhanced expression of putative MF exporter, Rv1258, in a clinical isolate resistant to ofloxacin is suggestive of a role in resistance,²⁴⁵ although previous studies of this exporter confirmed a role in low-level tetracycline and aminoglycoside resistance only.¹²⁹ In vitro-selected mutants of LfrA-deficient M. smegmatis showing a fluoroquinolone resistance/efflux phenotype have also been reported although the efflux determinant(s) remain to be identified.²⁴²

Efflux-mediated resistance to fluoroquinolones in Gram-negative bacteria, though only somewhat recently appreciated, was in evidence >20 years ago in P. aeruginosa, with examples of fluoroquinolone-resistant P. aeruginosa cross-resistant to other antimicrobial classes attributed (wrongly) to reduced permeability.²⁴⁶ While such mutants were later characterized by reduced fluoroquinolone accumulation,^247–249 and despite early indications of an efflux mechanism,²⁴⁸ the attendant changes in outer membrane protein profiles in such mutants led researchers to attribute resistance to outer membrane permeability defects. Clearly, however, fluoroquinolone-selected multidrug-resistant strains of P. aeruginosa owe their resistance to fluoroquinolones (and the other antimicrobials) to expression of endogenous, chromosome-encoded three-component multidrug efflux systems of the RND family (see References ²⁵⁰–²⁵² for reviews of RND family multidrug efflux systems in P. aeruginosa). Indeed, RND family exporters are commonly encountered determinants of fluoroquinolone resistance in laboratory and clinical isolates of this organism^250–254 and remain the most significant efflux determinants of fluoroquinolone resistance, not just in P. aeruginosa but in Gram-negative bacteria as a whole (reviewed in References ³⁷, ⁷⁴ and ²¹⁰).

Efflux-mediated fluoroquinolone resistance (where the selecting agent in vitro or in vivo was a fluoroquinolone or a fluoroquinolone resistance phenotype in particular was highlighted) has been reported in a number of Gram-negative pathogenic bacteria including Aeromonas salmonicida,²⁵⁵ Campylobacter spp.,^210,256,257 Citrobacter freundii,^210,536 Enterobacter spp.,^210,258 E. coli,^{196,210,259,260,537} Klebsiella spp.,^{76,210,261–263} Morganella morganii,²⁶⁴ Proteus vulgaris,²¹⁰ P. aeruginosa,^{74,210,265–269} Salmonella spp.,^75,270–275 Serratia marcescens,^276,276a Shigella dysenteriae,²¹⁰ S. maltophilia,^210,277 anaerobes such as Bacteroides spp.,^{74,210,278,279} and, possibly, N. gonorrhoeae.²⁸⁰ There are reports, too, of efflux-mediated resistance to nalidixic acid but not fluoroquinolones in Yersinia enterocolitica²⁸¹ and A. baumannii.²⁸² Moreover, AdeAB of A. baumannii is known to accommodate fluoroquinolones and has been shown to be unregulated in clinical isolates resistance to fluoroquinolones, though it appears to be important in these only for resistance to non-fluoroquinolones.²⁸³

Where identified, efflux is invariably determined by three-component efflux systems of the RND family^{74–76,210,256,259,263,271–273,276a,277} (Table 2) though not all RND family exporters accommodate and provide resistance to fluoroquinolones (e.g. the AmrAB-OprA efflux system of Burkholderia pseudomallei⁷⁴) and some RND family transporters known to accommodate these agents have yet to be implicated as primary (selected for in vitro or in vivo by fluoroquinolones) determinants of fluoroquinolone resistance [e.g. RND family efflux systems in Acinetobacter baumannii, Burkholderia cepacia (cenocepacia) and Proteus mirabilis (Table 2)].⁷⁴ In vitro-selected fluoroquinolone-resistant E. coli sometimes demonstrate a multiple antibiotic resistance (MAR) phenotype²⁸⁴ attributed to increased expression of the RND type AcrAB-TolC multidrug efflux system (reviewed in Reference ²⁸⁵). MAR strains,²⁸⁵ including clinical isolates resistant to fluoroquinolones,²⁸⁶ often carry mutations in the marRAB locus, and such mutations have now been described in laboratory-isolated, multidrug (including ciprofloxacin)-resistant E. coli O157:H7^287,288 suggesting that the AcrAB-TolC homologue⁷⁴ of this organism may similarly play a role in resistance to fluoroquinolones (and other agents) in this organism. Fluoroquinolone selection of multidrug-resistant S. maltophilia in vitro is also reminiscent of an efflux mechanism of the RND type.²⁸⁹

A chromosomally-derived determinant of a single component ABC family multidrug exporter capable of accommodating fluoroquinolones has been reported in V. cholerae²⁹⁰ although its significance if any regarding fluoroquinolone resistance remains unknown. Genes encoding an ABC family exporter associated with resistance to nalidixic acid and low-level ciprofloxacin resistance have also been identified on a large multiresistance plasmid in a Pseudomonas spp., encoding a probable three-component ABC-MFP-OMF drug exporter.¹⁹³ There is a single reported example of an MF family exporter that accommodates fluoroquinolones, MdfA of E. coli,^210,291 although there are no reports of resistance attributable to this efflux system. Intriguingly, whereas the MATE family of drug exporters is the least well characterized and includes the fewest number of characterized systems, those studied to date often accommodate fluoroquinolones, providing resistance at levels comparable to that seen, for example, for the RND family AcrAB-TolC efflux system of E. coli (Table 3). BLAST searches of available bacterial genome sequences also reveal that homologues of NorM from Vibrio parahaemolyticus, the prototypic MATE family multidrug exporter, are present in many Gram-negative (and some Gram-positive) organisms⁷⁴ and, as such, these may be important contributors to fluoroquinolone resistance in bacteria.

View this table: [in this window] [in a new window]   Table 3.. MATE family pumps exporting fluoroquinolones

 
Aminoglycosides

Relatively few bacterial drug efflux systems are known to accommodate aminoglycosides [e.g. the AmrAB-OprA⁷⁴ and BpeAB-OprB²⁹² multidrug efflux systems of B. pseudomallei, the AcrAD-TolC multidrug efflux system of E. coli⁷⁴ and the MexXY/OprM multidrug efflux system of P. aeruginosa⁷⁴ (Table 1)], although there are numerous AcrD homologues in other Enterobacteriaceae⁷⁴ suggesting that additional aminoglycoside-exporting efflux systems may be present in Gram-negative bacteria. The MexAB-OprM and EmrE pumps of P. aeruginosa have also been reported to provide a very modest contribution to intrinsic resistance to these agents though only in a low ionic-strength medium,²⁹³ making it unlikely that these will be significant determinants of aminoglycoside resistance in clinical isolates. Similarly, the LmrA multidrug exporter of L. lactis displays a weak ability to accommodate aminoglycosides, the cloned gene providing for modest increases in resistance to these agents.⁸⁵ The majority of known aminoglycoside exporters are RND family efflux systems (EmrE is a SMR family exporter and LmrA is an ABC exporter), highlighting once again the significance of this family of multidrug pumps vis-à-vis export of and resistance to clinically important antimicrobials in Gram-negative bacteria.⁷⁴ Still, only in P. aeruginosa is efflux a significant determinant of aminoglycoside resistance, with numerous reports of impermeability-type pan-aminoglycoside resistance in clinical isolates^294–305 characterized by reduced drug accumulation that is now attributable to efflux via MexXY/OprM^{269,306–309} (see Reference ³¹⁰ for a recent review of aminoglycoside resistance in P. aeruginosa, including efflux). In light of the demonstration that mexXY expression is inducible by aminoglycosides,^137,311 MexXY/OprM-mediated aminoglycoside efflux also appears to explain the long-known phenomenon of adaptive aminoglycoside resistance in P. aeruginosa.^311,312 Here, exposure of the organism to any aminoglycoside provides for enhanced pan-aminoglycoside resistance that is characterized by reduced drug accumulation but is, however, quickly lost in the absence of drug (adaptive resistance is reviewed in a recent comprehensive review of aminoglycoside resistance in P. aeruginosa³¹⁰).

ß-Lactams

While many of the three-component RND family multidrug exporters of Gram-negative bacteria can accommodate ß-lactams^74,313–320 (see also Reference ³²¹ for a review of ß-lactam resistance in bacteria including a discussion of efflux mechanisms) (Table 1), there are very few instances where these have been implicated in ß-lactam resistance in vitro or in vivo (i.e. selected by ß-lactams). In P. aeruginosa, overproduction of the MexAB-OprM system has been associated with clinical episodes of carbapenem (meropenem) resistance³²² and there are reports of clinical ticarcillin-resistant P. aeruginosa that overproduce this efflux system.^323,324 Resistance to imipenem that characterizes P. aeruginosa strains overproducing the MexEF-OprN multidrug efflux system^{80,250,252,325} is not, however, explained by efflux but rather by the concomitant decline in level of the outer membrane porin, OprD, in such mutants.^326,327 OprD is the major portal for entry of carbapenems into this organism and its loss is the most common cause of carbapenem resistance in mutant strains.^322,328,329 Overexpression of the MtrCDE multidrug efflux system of N. gonorrhoeae has also been highlighted as an important contributor to the high-level penicillin resistance of certain clinical isolates of this organism.³³⁰ A recent report, too, of high-level ampicillin resistance in a so-called ß-lactamase negative ampicillin-resistant (BLNAR) H. influenzae implicated the endogenous RND family exporter AcrAB-TolC as a co-determinant of this resistance,³³¹ and efflux has been implicated in the cefuroxime resistance of clinical isolates of E. coli although the efflux system has yet to be identified.³³² The broadly specific ABC type multidrug exporter LmrA of L. lactis shows a limited ability to accommodate ß-lactams but, as with its contribution to aminoglycoside resistance, this is noted when the pump is expressed from a multicopy vector.⁸⁵

Others

Efflux-mediated resistance to fosmidomycin (targets enzymes of isoprenoid biosynthesis) in E. coli has been reported.³³³ The observation that low-level in vitro-selected fluoroquinolone-resistant S. aureus sometimes demonstrate a multidrug resistance phenotype that includes resistance to fusidic acid suggests that a multidrug exporter may accommodate and provide resistance to this agent in S. aureus.³³⁴ Many RND family pumps also accommodate organic solvents (reviewed in References ³³⁵ and ³³⁶) and, indeed, co-resistance to solvents and antibiotics has been used to imply the presence of RND family multidrug efflux system in resistant strains.^337,338 Tributyltin, an antifouling agent found in marine paints, is exported by an RND family multidrug transporter of Pseudomonas stutzeri that also accommodates antibiotics.³³⁹ Three-component RND pumps are also known that export and provide resistance to heavy metals (reviewed in Reference ³⁴⁰).

	Efflux-mediated resistance to biocides

Top Abstract Introduction Efflux-mediated resistance to... Efflux-mediated resistance to... Evolution and natural function... Overcoming efflux-mediated... Concluding remarks References

 
Efflux determinants of biocide resistance display broad substrate specificity, accommodating a variety of structurally unrelated agents that can also include antibiotics (see References ²⁵ and ³⁴¹ for reviews of biocide resistance, including efflux mechanisms of resistance) (Table 4).

View this table: [in this window] [in a new window]   Table 4.. Efflux determinants of biocide resistance

 
Quaternary ammonium compounds

A number of efflux determinants of biocide resistance that accommodate quaternary ammonium compounds (QACs; e.g. benzalkonium chloride) have been described in Gram-positive bacteria, predominantly in Staphylococcus spp. (Table 4), including clinical isolates,^342–344 equine isolates,³⁴⁵ bovine isolates³⁴⁶ and food-associated isolates.^347–349 The majority of these efflux determinants are plasmid-encoded, SMR family exporters [e.g. Smr (QacC/D), QacE1, QacG, QacH, QacJ] although QacA/B is a MF family efflux system, and resistance arises from plasmid acquisition.

Chromosomal efflux determinants of QAC resistance, though uncommon in Gram-positive bacteria, have been described and include the S. aureus NorA multidrug transporter implicated in fluoroquinolone resistance,^{192,343,350,351} the recently reported MF family MdeA and MATE family MepA multidrug efflux systems also present in S. aureus^192,351a and the EmeA multidrug exporter of E. faecalis.^235,235a An as yet unidentified putative multidrug transporter distinct from these but able to contribute to QAC resistance in S. aureus has also been reported.³⁵³ Efflux has also been implicated in QAC (i.e. benzalkonium chloride) resistance in L. monocytogenes^354,355 although the efflux genes have yet to be identified.

Efflux systems able to accommodate biocides, including QACs, in Gram-negative bacteria (Table 4) are also multidrug transporters, with the exception of the Ag⁺-specific efflux systems³⁵⁶ (see below). In contrast to efflux-mediated biocide resistance in Gram-positive organisms, however, biocide exporters in Gram-negative bacteria are generally chromosomally-encoded (the exceptions being the qacE, qacE1, qacF and qacG genes associated with QAC resistance). Many of the latter are associated with potentially mobile integron elements, perhaps explaining the broad distribution of qacE and especially qacE1 (which is prevalent on class I integrons^357–359) in a wide variety of Gram-negative bacteria (Table 4). Still, the presence of the qacE determinants does not appear to correlate with resistance to QACs, at least in one study of Gram-negative bacteria with/without these genes³⁶⁰ suggesting that they may not be significant determinants of QAC resistance in these organisms. A number of the MATE (NorM of Neisseria spp., PmpM of P. aeruginosa) and RND (AcrAB-TolC, AcrEF-TolC and YhiUV-TolC pumps of E. coli; SdeXY pump of S. marcescens) family multidrug transporters implicated in antibiotic resistance have been shown to contribute to QAC resistance (Table 4) although since in many/most instances the contribution of Gram-negative multidrug transporters to QAC resistance has not been addressed, the numbers may be greater. The RND family exporter, MexCD-OprJ, a significant determinant of fluoroquinolone resistance in laboratory and clinical isolates,^250,252 is inducible by QACs, although its contribution, if any, to QAC resistance has not been addressed.³⁶¹ Recently, too, there have been a number of reports of strains of E. coli,³⁶² including E. coli O157:H7³⁶³ adapted to e.g. benzalkonium chloride in vitro showing a multiple antibiotic-resistant phenotype reminiscent of mutants expressing the RND family AcrAB-TolC exporters of these bacteria. Moreover, the benzalkonium chloride-selected multidrug-resistant E. coli also showed enhanced efflux activity (measured using ethidium bromide as a model efflux compound) although the actual efflux determinant was not identified.³⁶² In another study, several benzalkonium chloride-adapted S. enterica strains had the benzalkonium chloride resistance compromised by efflux inhibitors, consistent with an efflux mechanism of resistance, although, again, a specific mechanism was not identified.³⁶⁴ The SMR family EmrE multidrug exporter of E. coli also accommodates QACs.^365,366 A Pseudomonas fluorescens isolate contaminating a batch solution of benzalkonium chloride and showing high-level resistance to multiple QACs that was compromised by CCCP treatment has been reported and is indicative of an efflux mechanism of resistance.³⁶⁷

Chlorhexidine

Chlorhexidine is a hand washing disinfectant extensively used in hospitals and as such it is not surprising to find nosocomial pathogens exhibiting reduced susceptibility to this agent.³⁶⁸ While the specific mechanism(s) of chlorhexidine resistance in most instances are unknown, a gene, cepA, encoding a putative efflux mechanism has been cloned from a chlorhexidine-resistant clinical isolate of K. pneumoniae and shown to increase chlorhexidine resistance in E. coli transformants.³⁶⁹ Attempts to address the contribution of CepA to the chlorhexidine resistance of the clinical K. pneumoniae isolate were, however, unsuccessful, owing to an inability to construct a cepA knockout strain. A cursory search of available bacterial genome sequences identifies a number of bacteria capable of producing CepA-like proteins including Shigella flexneri (accession number AAP18773 [GenBank] 85% identity), E. coli K-12 (protein Yiip, accession number AAN83294 [GenBank] 85% identity), E. coli O157:H7 (protein Yiip, accession number AAG59108 [GenBank] 85% identity), S. enterica serovar Typhimurium (accession number AAL22901 [GenBank] 83% identity), Salmonella typhi (accession number AAO71063 [GenBank] 83% identity), Yersinia pestis (accession number AAM83655 [GenBank] 75% identity), V. cholerae (accession number AAF96831 [GenBank] 57% identity), Haemophilus ducreyi (accession number AAP96289 [GenBank] 55% identity), P. aeruginosa (accession number AAG07350 [GenBank] 50% identity) and Haemophilus somnus (accession number ZP_00132429, 47% identity) indicating that a chlorhexidine efflux mechanism might be somewhat widely distributed amongst Gram-negative pathogenic bacteria. The recently described MATE family MepA exporter of S. aureus also provides resistance to chlorhexidine.^351a

Although a direct role for RND family exporters in chlorhexidine resistance has not been specifically studied, reports of benzalkonium chloride-³⁶³ and triclosan-^363,370 adapted E. coli (including E. coli O157:H7) exhibiting a multidrug-resistant phenotype typical of overproduction of an RND family multidrug transporter and showing reduced susceptibility to chlorhexidine suggest that such pumps may, indeed, accommodate this biocide. Interestingly, too, like benzalkonium chloride, chlorhexidine has been shown to induce expression of the MexCD-OprJ multidrug efflux system of P. aeruginosa although, again, a role in chlorhexidine resistance was not examined.³⁶¹ Still, chlorhexidine-adapted E. coli O157:H7 in one study did not exhibit a multidrug-resistant phenotype, suggesting that this biocide does not readily select for RND-type drug exporter-producing mutant strains, at least in vitro. QacA/B implicated in QAC resistance in Gram-positive bacteria also promotes reduced susceptibility to chlorhexidine (Table 4).

Triclosan

Triclosan is a biocide increasingly prevalent in household products and for which mechanisms of resistance are known in a variety of Gram-negative bacteria. Indeed, many of the RND family pumps associated with resistance to clinically important antibiotics are also able to accommodate triclosan [e.g. several of the three-component Mex pumps of P. aeruginosa,³⁷¹ SdeXY of S. marcescens,³⁷² SmeDEF of S. maltophilia,³⁷³ AcrAB-TolC of E. coli,³⁷⁴ CmeABC and CmeDEF of C. jejuni^374a and AcrAB-TolC of S. enterica serovar Typhimurium (Table 4)] and triclosan readily selects for strains expressing/hyperexpressing these systems in vitro. RND/Mex efflux systems are, in fact, the major determinants of triclosan insusceptibility in P. aeruginosa.³⁷¹ Moreover, recent studies showing ready selection, in vitro, of multiple antibiotic-resistant mutants of E. coli O157:H7^363,370 and Salmonella spp.³⁷⁵ with triclosan are consistent with this biocide being accommodated by and selecting for multidrug efflux mechanism(s). An association between reduced triclosan susceptibility and increased resistance to multiple antibiotics in human and animal Campylobacter spp. isolates³⁷⁶ is also suggestive of the presence in Campylobacter spp. of an RND family multidrug exporter(s) that accommodates both triclosan and antibiotics.³⁷⁵ Indeed, the CmeABC and CmeDEF RND-type multidrug exporters of this organism are both able to accommodate and so provide resistance to triclosan.^374a Growth in the presence of triclosan has also been shown to increase the frequency with which multidrug-resistant mutants of S. enterica could be selected in vitro.³⁷⁵ Again, while efflux was not examined, the phenotype (co-resistance to triclosan, antibiotics and cyclohexane) was typical of strains overproducing an RND family exporter (e.g. AcrAB-TolC). In contrast, chronic in vitro exposure of the dental pathogen Porphyromonas gingivalis to triclosan failed to select triclosan-resistant mutants despite the presence, in this organism, of an RND family exporter, XepCAB, whose contribution to antibiotic resistance has been documented.³⁷⁷ Still, while not all RND exporters may be significant determinants of triclosan resistance, the ability of most RND family pumps known to contribute to antibiotic resistance in clinical strains (Tables 1 and 2) to also contribute to triclosan resistance has not been tested. Thus, it is not yet clear the extent to which these might contribute to triclosan resistance in Gram-negative pathogens. The recent demonstration that triclosan treatment up-regulates putative efflux genes in Mycobacterium spp. is suggestive of the presence of a triclosan efflux mechanism of resistance in these bacteria.³⁷⁸

Silver

Silver (Ag⁺) is a biocidal agent whose best known use is in topical creams where it is the preferred antimicrobial for the treatment of serious burns, although other uses are known and include Ag⁺-coated bandages and Ag⁺-impregnated polymers used in medical devices (e.g. in catheters and heart valves) to prevent biofilm formation.³⁵⁶ Ag⁺-resistance in bacteria is known, however, particularly in Gram-negative bacteria (e.g. Salmonella spp.) where two plasmid-encoded efflux mechanisms have been described (SilP, an ABC type transporter and SilABC, a three-component RND family transporter; reviewed in Reference ³⁵⁶). Interestingly, genes encoding homologues of the SilABC system have been identified in the chromosomes of E. coli K-12 and E. coli O157:H7 where they were shown to play a role in Ag⁺ resistance.³⁵⁶ Such determinants are also found on large multiresistance plasmids in S. marcescens^379,380 and K. pneumoniae^379,381 (in one instance present on a large virulence plasmid of a clinical isolate³⁸¹). Moreover, a recent hybridization study documented the presence of silABC homologous DNA in a variety of unnamed enteric bacteria of clinical origin.³⁵⁶

Others

Resistance to pine oil found in household cleaners has also been linked to the expression of RND family multidrug efflux systems, with in vitro-isolated pine oil-resistant E. coli showing overproduction of the AcrAB-TolC efflux system.³⁸² The NorB MF family multidrug exporter of S. aureus implicated in fluoroquinolone resistance has been shown to contribute to reduced cetrimide susceptibility⁹⁴ as have NorA and MepA (Table 4).

Biocide-antibiotic cross-resistance

While there is some debate in the literature regarding the real risks of selecting for biocide-resistant organism outside the laboratory, at biocide concentrations typically used³⁸³ (and, thus, the significance of resistance mechanisms, efflux or whatever), there is considerably more debate concerning the risk of biocide selection of antibiotic-resistant organisms.^384–389 This is particularly true given the existence of multidrug efflux systems that accommodate both classes of antimicrobial (e.g. NorA, NorB, MepA and MdeA systems of S. aureus; Mex systems of P. aeruginosa; AcrAB-TolC of E. coli; SdeXY of S. marcescens; SmeDEF of S. maltophilia)³⁷³ (Table 4). Studies of clinical S. aureus isolates showing reduced susceptibility to the QAC benzalkonium chloride have, for example, demonstrated enhanced expression of the NorA multidrug exporter in some of these, with an attendant increase in fluoroquinolone resistance.^343,351 Moreover, QAC-resistant S. aureus selected in vitro often showed cross-resistance to fluoroquinolones as a result of increased norA expression in these and, indeed, QACs seemed to more effectively select for NorA-expressing mutants than did fluoroquinolones.³⁵⁰ Recently, too, a second MF family multidrug transporter that accommodates both QACs and antibiotics, MdeA, has been identified in S. aureus and again QAC-resistant laboratory isolates overproducing this protein showed a modest cross-resist to several antibiotics.¹⁹² Finally, a mutant overexpressing NorB was shown to be co-resistant to fluoroquinolones and cetrimide although cetrimide was not the selective agent.⁹⁴

Several in vitro studies have shown that triclosan readily selects for multiple antibiotic-resistant P. aeruginosa,^390,391 S. maltophilia³⁷³ and E. coli³⁷⁴ expressing these multidrug efflux systems, and for multidrug-resistant Salmonella spp.,³⁷⁶ E. coli K-12³⁶² and E. coli O157:H7³⁶³ where multidrug efflux mechanisms are implicated. The correlation, too, between triclosan and multidrug resistance in human and animal isolates of Campylobacter spp.³⁷⁵ and Salmonella spp.³³⁷ also suggests that a common, presumed RND family efflux, mechanism exists in these organisms for triclosan and antibiotics, with an attendant risk that triclosan can select for antibiotic resistance in these organisms. Certainly, the RND family AcrAB-TolC and Cme multidrug exporters of S. enterica serovar Typhimurium and C. jejuni, respectively, provide resistance to both antibiotics and triclosan (Table 4). While not all three-component RND family multidrug efflux systems appear to accommodate and provide for resistance to triclosan (e.g. the MexXY/OprM system of P. aeruginosa) many of the known and predicted RND family exporters that are so widespread in Gram-negative bacteria⁷⁴ will probably accommodate triclosan and so permit triclosan selection of multidrug resistance. Still, there are as yet no reports of biocide selection of antibiotic-resistant organisms outside the laboratory (a recent examination of resistance phenotypes of bacteria isolated from homes that employed/did not employ biocide-containing products found no