This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Copyright Information Books from ASM Press MicrobeWorld Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Piddock, L. J. V. Search for Related Content PubMed PubMed Citation Articles by Piddock, L. J. V.

Clinically Relevant Chromosomally Encoded Multidrug Resistance Efflux Pumps in Bacteria

Antimicrobial Agents Research Group, Division of Immunity and Infection, The Medical School, University of Birmingham, Birmingham, United Kingdom, B15 2TT

SUMMARY

INTRODUCTION

CLASSES OF MDR EFFLUX PUMPS, GENOMICS, AND STRUCTURAL BIOLOGY

Classes and Organization of Efflux Pump Systems

Structural Biology

SUBSTRATES OF MDR EFFLUX PUMPS

CLINICAL RELEVANCE OF MDR EFFLUX PUMPS

Inherent Resistance of Gram-Negative Bacteria to an Entire Class of Agents

Inherent Resistance of Some Species of Gram-Negative Bacteria to Specific Agents

Efflux Pumps of Clinically Relevant Bacteria That Confer MDR via Overexpression

Gram-negative bacteria. (i) Pseudomonas aeruginosa.

(ii) Escherichia coli.

(iii) Salmonella enterica.

(iv) Campylobacter spp.

(v) Acinetobacter baumannii.

(vi) Neisseria gonorrhoeae.

(vii) Other gram-negative bacteria.

Gram-positive bacteria.

(i) S. aureus.

(ii) S. pneumoniae.

Mycobacteria.

Evidence for Resistance in Clinical Isolates Mediated by Enhanced Efflux

REGULATION OF EFFLUX PUMPS IN CLINICAL ISOLATES

Mutations in the Local Repressor Gene

Mutations in Global Regulator Genes

Mutations in the Promoter Region of the Gene Encoding the Transporter

Insertion Sequences

MDR EFFLUX PUMPS AND DEVELOPMENT OF ANTIBIOTIC RESISTANCE

NATURAL ROLES OF MDR PUMPS

Bile Tolerance of Enteric Bacteria

Colonization, Invasion, and Survival in the Host

INHIBITORS OF MDR EFFLUX PUMPS

BIOCIDES

CONCLUDING REMARKS

ACKNOWLEDGMENTS

REFERENCES

Top Next References

INTRODUCTION

Top Previous Next References

 
Efflux is the pumping of a solute out of a cell. Efflux pump genes and proteins are present in both antibiotic-susceptible and antibiotic-resistant bacteria. Some systems can be induced by their substrates so that an apparently susceptible strain can overproduce a pump and become resistant. Antimicrobial resistance in an efflux mutant is due to one of two mechanisms: either (i) expression of the efflux pump protein is increased or (ii) the protein contains an amino acid substitution(s) that makes the protein more efficient at export. In either case, the intracellular concentration of the substrate antimicrobial is lowered and the organism becomes less susceptible to that agent. Efflux pumps may be specific for one substrate or may transport a range of structurally dissimilar compounds (including antibiotics of multiple classes); such pumps can be associated with multiple drug (antibiotic) resistance (MDR). Resistance in this context does not necessarily mean resistance to those agents that would be used to treat an infection by a particular species, or even resistance to clinically achievable concentrations of these drugs, and so the clinical relevance of efflux-mediated resistance is species, drug, and infection dependent. Genes encoding efflux pumps can be found on the chromosome or on transmissible elements such as plasmids (e.g., tet and qac genes); this review focuses on chromosomally encoded MDR efflux pumps.

In addition to the role of enhanced efflux in antimicrobial resistance, it has also been suggested that increased expression of efflux pump genes may be the first step in the bacterium becoming fully resistant (144). It is also thought that increased efflux decreases the intracellular concentration of the antimicrobial, thereby allowing bacterial survival for a greater length of time, such that bacteria containing mutations in other genes, such as those that encode target proteins (e.g., fluoroquinolones and topoisomerase genes), can accumulate. More recently, there have been several publications on different species of bacteria, suggesting natural physiological roles of MDR efflux pumps in addition to export of antimicrobials.

There are still many questions that must be addressed, as genomics has revealed that efflux pumps are ubiquitous throughout nature in all types of cells, from eukaryotic to prokaryotic. One important question is for which species MDR efflux pumps confer clinically relevant resistance, i.e., in which the MIC is greater than the recommended breakpoint concentration of a drug for a particular bacterial species, and specifically with reference to antimicrobials that are used in human or veterinary medicine for an infection caused by that bacterium. Which of the clinically relevant agents, then, are typically affected by bacteria that overexpress MDR efflux pumps?

There have been several keynote publications in this field in recent years describing the various classes of efflux pumps and their substrates, structures, and functions (19, 63, 96, 155, 166, 236). This article focuses on those pumps and bacterial species considered to be of current clinical relevance, the putative natural roles of efflux pumps, and inhibitors.

CLASSES OF MDR EFFLUX PUMPS, GENOMICS, AND STRUCTURAL BIOLOGY

Top Previous Next References

 

View larger version (37K): [in this window] [in a new window]   FIG. 1. Diagrammatic comparison of the five families of efflux pumps. (Courtesy of Melissa Brown; reproduced by kind permission.)

View larger version (20K): [in this window] [in a new window]   FIG. 2. Comparison of the genetic organization of E. coli acrRAB with S. enterica serovar Typhimurium acrRAB, C. jejuni cmeABC, and P. aeruginosa mexAB-OprM. This diagram shows the similarities between the RND MDR efflux pump genes of different bacterial species.

MATE MDR efflux pumps have been described for various bacteria, including Vibrio parahaemolyticus (NorM), Vibrio cholerae (VcrM; VcmA), Bacteroides thetaiotaomicron (BexA), Haemophilus influenzae (HmrM), P. aeruginosa (PmpM), Clostridium difficile (CdeA), and S. aureus (MepA). Two energy sources have been identified for MATE efflux pumps: the PMF and the sodium ion gradient. MATE pumps transport some of those agents also transported by RND pumps. However, a key distinguishing feature is that while RND pumps are tripartite, MATE pumps are not (Fig. 1).

Although no ABC transporters that give rise to clinically relevant MDR in human or animal pathogens have so far been described, ABC transporters are present in the genomes of pathogenic bacteria, and it may be predicted that at least one will confer antimicrobial drug resistance in the same way that P glycoprotein (PgP) confers resistance to anticancer agents. It has already been shown that Lactococcus lactis LmrA, an ABC transporter, confers MDR in this organism (223). ABC transporters have structural characteristics that differ from RND and MFS pump proteins, in that there are typically only six transmembrane-spanning regions (178). There are also three signature motifs within ABC transporter proteins; these include the Walker A, Walker B, and ABC signature motifs. In contrast to the RND, MSF, and MATE families of transporters, efflux by ABC transporters is driven by ATP hydrolysis.

To date, DNA sequencing data have not revealed significant variation between the sequences of the efflux pump genes of strains of the same species; however, recent work with C. jejuni cmeB has indicated some sequence diversity (27; B. Guo, J. Lin, D. Reynolds, and Q. Zhang, Abstr. 105th Gen. Meet. Am. Soc. Microbiol., abstr. A-001, 2005). Sequence variations occur throughout cmeB, but no data are available to indicate what effect, if any, these may have on efflux. Kim et al. (82) explored genomic data for similarities between LmrA and MexB (examples of the ABC and RND superfamilies, respectively). They observed that there are many hydrophobic, hydrophilic, and semipolar residues conserved in both proteins and that these are important for structure and/or function in both families.

An important structural feature of efflux pump proteins is that hydropathy plots typically reveal 12 transmembrane-spanning regions. Such predicted structural information from genomic data has been useful in identifying proteins involved in multidrug efflux.

View larger version (63K): [in this window] [in a new window]   FIG. 3. Model of the assembled tripartite drug efflux pump. This possible model of an RND-class drug efflux pump is based on the open-state model of TolC (red) forming a minimal contact interface with the six hairpins at the apex of AcrB (green). A ring of nine MexA molecules (blue) is modeled to form a sheath around AcrB and the -barrel of TolC (MexA is a close homologue of AcrA, the natural partner of AcrB/TolC). Variants of the model might include a lower-order oligomer of MexA (4) and more extensive interaction between AcrB and TolC. IM, inner membrane; OM, outer membrane. (Reprinted from reference 42 with permission from Elsevier and V. Koronakis.)

The crystal structure of AcrB was first resolved at 3.5-Å resolution (133). The functional unit of AcrB is a homotrimer, with 12 membrane-spanning helices and a large periplasmic domain. The structure also revealed that AcrB has a "headpiece" that opens like a funnel, thereby facilitating contact with TolC. Three helices form a pore to the funnel, with a central cavity located at the bottom of the headpiece. The cavity has three "vestibules" (small pockets made from gaps around the three protomers of AcrB) at the side of the headpiece that lead into the periplasm. In the transmembrane region, each protomer has 12 transmembrane helices. The structure suggested that substrates are transported from the cytoplasm via the transmembrane region and also from the periplasm via the vestibules. Substrates are then actively transported through the pore into TolC. Yu et al. (235) further suggested that the binding of substrates in AcrB is due to the partial binding of the compound to the phospholipid bilayer of the cytoplasmic membrane, depending on the lipophilicity and charge of the molecule. From there, the substrate must diffuse through the membrane toward the AcrB vestibules. It is then bound to the wall of the central cavity and from there is pumped out with the proton gradient (235). The structure of AcrB has since been further resolved to 2.7 Å (170). Two transmembrane domains are thought to be critical in proton translocation: TM4, which contains Asp401 and Asp 402, and TM10, which contains Lys940. Interestingly, there seems to be only one attachment between monomers, a long loop that protrudes into the center of the monomer in a counter-clockwise direction. Recent evidence with AcrD suggests that RND pumps act as a "periplasmic vacuum cleaner" and capture their substrates from the periplasm and export them into the external medium (2).

Gerken and Misra (51) showed that it is possible to cross-link AcrA and TolC without the presence of AcrB and so proposed that AcrB does not mediate or influence binding in vivo of TolC-AcrA interactions. In experiments with mutants lacking AcrAB, protein levels of TolC were unaffected and receptor functions were still present. Therefore, the authors proposed that TolC can stably insert into the outer membrane without AcrA or AcrB and still perform receptor functions.

MexA is found in Pseudomonas aeruginosa and is homologous to AcrA in E. coli. However, AcrA functions in complex with AcrB, whereas MexA and MexB form a homologous system. MexA and AcrA are anchored to the inner membrane by a single transmembrane helix, or in some cases by a lipid modification on the N terminus of the protein. Higgins et al. (59) have resolved the crystal structure of MexA to 3.0-Å resolution and have resolved amino acids 29 to 259 (a total of 230). MexA is 360 residues long in vivo, so part of this protein was absent. The MexA monomer was shown to be 89 Å long and 35 Å wide and to take the form of a ß-barrel/lipoyl domain/-helical structure, which is normally associated in other structures with ligand binding. It was proposed that, in vivo, the monomers form a homomonomer and make a "sheath" around both TolC (used because at that time the crystal structure of OprM had not been solved) and AcrB. Akama et al. (4) also proposed a similar structure for MexA, although they resolved more of the protein, 241 residues in all, from residues 23 to 274. They also proposed the same ß-barrel/lipoyl domain/-helical hairpin structure as did Higgins et al. (59). However, Akama et al. (3) proposed two in vivo structures, one based upon the sheath, or "seal," model proposed by Higgins et al. (59) and one based upon a threefold dimer model. As MexA was found in the inner membrane fraction and the N terminus is fatty acid modified, Akama et al. (4) concluded that MexA is highly likely to be anchored to the inner membrane at least once. They also concluded that the -helical hairpin is directed toward OprM/TolC and the ß-barrel toward MexB/AcrB. A domain swapping experiment was also carried out, and they concluded that the C-terminal region of MexA/AcrA is very important in the interaction with MexB/AcrB.

SUBSTRATES OF MDR EFFLUX PUMPS

Top Previous Next References

 
The phenotype of an "efflux mutant" (those that phenotypically resemble mutants that overexpress an efflux pump) is that the strain is typically MDR, i.e., resistant (less susceptible) to antimicrobials from at least three different classes of antibiotics, disinfectants (biocides), dyes, and detergents. The agents typically include a quinolone (nalidixic acid, ciprofloxacin, norfloxacin), tetracycline, chloramphenicol, ethidium bromide, acriflavine, sodium dodecyl sulfate (SDS), Triton X-100, and triclosan, but the agents considered to be substrates of each pump are slightly different depending on the pump and bacterial species. Poole (166, 167) lists details available for specific bacterial efflux pumps. The MICs of these substrates for strains overexpressing an efflux pump are typically two- to eightfold higher than those for a typical representative susceptible strain of that species or an isogenic parent strain (see Tables 1 to 8). Likewise, in those mutants in which an efflux pump gene has been disrupted or deleted, the MICs of substrates of that pump decrease, giving rise to a hypersusceptible strain. In most cases, the increases in MICs due to overexpression of an MDR efflux pump do not confer the same magnitude of increase as other mechanisms of resistance, such as ß-lactamases, aminoglycoside modifying enzymes, or topoisomerase substitutions.

Top Previous Next References

The Enterobacteriaceae and other gram-negative bacteria are also generally considered to be less susceptible to many antimicrobials than gram-positive bacteria. Historically, this too has been attributed to the bacterial cell envelope conferring a "permeability barrier" (i.e., preventing uptake into the cell); however, it is increasingly recognized that this "barrier" is often due to efflux pumps acting alone or in concert with decreased expression of porins. Nonetheless, decreased transport of drugs into the bacterial cell is still a factor to be considered.

Recently, a MATE transporter, PmpM, in P. aeruginosa has been described (58). PmpM transports fluoroquinolones, benzalkonium chloride, ethidium bromide, acriflavine, and tetraphenylphosphonium chloride. This system uses utilizes hydrogen ions, but not sodium ions, as an energy source.

(ii) Escherichia coli. In E. coli, the AcrAB-TolC system is highly homologous to the MexAB-OprM RND system in P. aeruginosa (138). AcrA is a 397-amino-acid protein that interacts with AcrB, a much larger protein, 1,048 amino acids. TolC, a 506-amino-acid protein, is also associated with AcrA (234). The substrate profile of the AcrAB-TolC pump includes chloramphenicol, lipophilic ß-lactams, fluoroquinolones, tetracycline, rifampin, novobiocin, fusidic acid, nalidixic acid, ethidium bromide, acriflavine, bile salts, short-chain fatty acids, SDS, Triton X-100, and triclosan (46, 139, 149, 216, 230). In E. coli, acrD and the acrEF operon also encode efflux pumps (188, 236), and AcrD has been shown to efflux aminoglycosides (140, 188). AcrE and AcrF are 80 and 88% similar to AcrA and AcrB, respectively (110). While recognized as a commensal organism, E. coli is also the most common cause of urinary tract infections, and treatment is usually with a fluoroquinolone, trimoxazole or nitrofurantoin. Enteropathogenic E. coli and enterotoxigenic E. coli are a common cause of diarrhea in developing countries and for travelers to these locations, and if antimicrobial therapy is indicated, the same agents are often used as for the treatment of urinary tract infections. In children and the immunocompromised, E. coli can cause more serious infections, associated with higher morbidity and mortality. For these patient groups, antimicrobial therapy is required; treatment may be with a broad-spectrum cephalosporin (e.g., ceftriaxone) or a fluoroquinolone. While some of these agents are substrates of the AcrAB-TolC system, overexpression alone is unlikely to give rise to clinical levels of resistance (Table 2). For fluoroquinolones, a mutation(s) in a topoisomerase gene is also unlikely to give rise to clinical levels of resistance; however, when combined with enhanced efflux, such isolates are resistant to the breakpoint concentration of ciprofloxacin (e.g., see references 45, 124, 125, and 145). Of current concern is the increasing number of isolates of E. coli expressing an extended-spectrum ß-lactamase, in particular a CTM enzyme. Infections with such E. coli isolates are often treated with second- and third-line agents, which are often substrates of efflux pumps; therefore, the selective pressure on this species toward selection of highly MDR strains is increasing.

S. enterica serovar Typhimurium AcrA and AcrB are very similar to AcrA (94%) and AcrB (97%), respectively, of E. coli (40) (Fig. 1). Mutants of S. enterica serovar Typhimurium that lack various efflux pump genes have been constructed (15, 25, 40, 185). Mutants lacking AcrB were hypersusceptible to quinolones, tetracycline, chloramphenicol, bile salts, SDS, Triton-X100, acriflavine, ethidium bromide, cetyltrimethylammonium bromide (CTAB), and triclosan. Overexpression of AcrB has also been associated with MDR in human clinical and veterinary isolates (and laboratory mutants) of S. enterica serovar Typhimurium (14, 56, 165). The MICs of nalidixic acid, tetracycline, and chloramphenicol for an AcrB-overexpressing strain were above the recommended breakpoint concentrations (Table 3). The MIC of ciprofloxacin is usually 0.5 µg of ciprofloxacin/ml for an AcrB-overexpressing strain, below the CLSI (Clinical and Laboratory Standards Institute) and BSAC (British Society of Antimicrobial Chemotherapy) recommended breakpoint concentrations for this agent for this organism. However, serovars of S. enterica with mutations in gyrA are inhibited by 0.25 µg of ciprofloxacin/ml, but such strains have been shown to fail therapy with a fluoroquinolone (130, 160, 226). There has been considerable discussion in the literature that the recommended breakpoint concentration of ciprofloxacin should be lowered to 0.25 µg of ciprofloxacin/ml. If this were the recommended value, then the MIC of ciprofloxacin for an AcrB-overexpressing strain would be above this concentration and so would be deemed clinically resistant.

(iv) Campylobacter spp. In 2003, two teams independently showed that CmeABC mediated efflux in C. jejuni and conferred MDR (100, 175). CmeA and CmeB have some similarity to AcrA (51%) and MexA (49%) and to AcrB (63%) and MexB (62%), respectively, of E. coli and P. aeruginosa. Deletion of cmeB revealed that the substrates of CmeABC include ciprofloxacin and erythromycin, both common first-line agents should antimicrobial treatment be warranted to treat a human campylobacter infection. In addition, overexpression of CmeB confers resistance to ciprofloxacin, ampicillin, tetracycline, and chloramphenicol and decreased susceptibility to triclosan, bile salts, SDS, and Triton X-100 (Table 4) (102, 177, 178). A second efflux pump system, CmeDEF, has also been identified, but this system does not appear to confer resistance to ciprofloxacin or erythromycin (176).

MDR pumps of the MATE family have been described for several gram-negative bacteria: V. parahaemolyticus (NorM) (132), B. thetaiotaomicron (BexA) (129), V. cholerae (VcmA, VcrM) (65, 66), Brucella melitensis (NorMI) (21), N. gonorrhoeae (NorM) (189), H. influenzae (HmrM) (231), and P. aeruginosa (PmpM) (58). The substrate profile typically includes a fluoroquinolone (norfloxacin and ciprofloxacin), DNA-intercalating dyes, and detergents. However, the clinical relevance of these systems has not been established.

Gram-positive bacteria. It has been known for over a decade that Bacillus subtilis possesses an MDR efflux pump, Bmr, which belongs to the MFS family of efflux pumps (134, 135). While there is little clinical significance of this efflux pump in human and veterinary medicine, it has been shown that the NorA pump of S. aureus and PmrA of S. pneumoniae bear significant similarity and identity at the DNA and amino acid levels (55, 136). Therefore, a considerable number of analogies have been made between the properties of Bmr and of NorA (136) and PmrA (55).

(i) S. aureus. S. aureus NorA has been shown to have 44% amino acid identity and 67% similarity with Bmr. Bmr and NorA are structurally similar to the plasmid-encoded efflux proteins TetA, TetB, and TetC, with 24 to 25% sequence identity with these proteins (134). Overexpression of both Bmr and NorA confers MDR to fluoroquinolones, chloramphenicol, antiseptics, dyes, and disinfectants (Table 7) (77, 134, 136, 137, 232). NorA is present in both methicillin-sensitive S. aureus (MSSA) and methicillin-resistant S. aureus (MRSA). The agents used to treat infections by MSSA include flucloxacillin, nafcillin, and ciprofloxacin. The agents used to treat MRSA include ciprofloxacin (where the MRSA strain has been shown to be susceptible), vancomycin, and linezolid (8). The MICs of nafcillin and vancomycin are unaffected by overexpression of NorA (G. W. Kaatz, personal communication).

(ii) S. pneumoniae. Over the last decade or so, considerable effort has been expended by pharmaceutical companies to develop antipneumococcal agents, so there has been considerable focus on S. pneumoniae and the presence of efflux pump proteins that could confer MDR, including to new agents. In 1999, Gill et al. identified PmrA. This protein has 43% amino acid similarity with NorA and 42% similarity with Bmr. These workers showed that when norfloxacin resistance was transformed from a clinical isolate into strain R6 (widely used by geneticists, as it is highly transformable and nonencapsulated, i.e., nonpathogenic; 40 kb of the genome is deleted compared with the parent strain from which it originates, and this removes the capsule locus), the MIC of norfloxacin for R6 increased to 16 µg/ml for the construct, R6N. Gill et al. (55) also introduced the cat gene into R6N to construct strain R6N-cat. Insertion of cat into pmrA gave rise to the same susceptibility to norfloxacin, ciprofloxacin, ethidium bromide, and acriflavine in R6N-cat as in R6 (Table 8). The MIC of ciprofloxacin for strain R6N is within one doubling dilution of the recommended breakpoint concentration for this agent. No data are available as to whether overexpression of pmrA has any effect on the MIC of penicillin or cephalosporins. These workers also investigated the activity of fluoroquinolones for pneumococci in the presence of reserpine, an inhibitor of Bmr (see "Inhibitors of Efflux Pumps," below). Due to the synergistic effect of reserpine and antimicrobial agents for B. subtilis, this agent has been widely used in MIC studies of a variety of different antimicrobials with S. pneumoniae as a means of identifying those isolates that overexpress an efflux pump (e.g., see references 11, 22-24, and 164). Gill et al. (55) also demonstrated that the MIC of norfloxacin for strain R6N was reduced fourfold in the presence of reserpine; they interpreted these data to indicate that reserpine interacted with PmrA to inhibit its efflux activity, so that the activity of norfloxacin was potentiated. However, to date there is no biochemical evidence showing a direct interaction between reserpine and the PmrA protein. Piddock and Johnson (162) examined the concentration of fluoroquinolones accumulated by strain R6N compared with strain R6. They showed that R6N did indeed accumulate significantly lower concentrations of norfloxacin than did R6, supporting the hypothesis that PmrA transported norfloxacin. However, the addition of reserpine at the same concentration as used in MIC studies did not significantly affect the concentration of norfloxacin accumulated. It may be that reserpine interacts with another protein, hence giving the observed synergy in MIC studies. In addition, the concentrations of 10 other fluoroquinolones accumulated by strain R6N were similar to those accumulated by strain R6, suggesting that PmrA did not transport these other drugs. Recently, Marrer et al. (120) identified an ABC transporter associated with ciprofloxacin resistance. Robertson et al. (186) showed that deletion of this transporter conferred multidrug susceptibility.

Mycobacteria. Several efflux pumps of different classes have been described for Mycobacterium tuberculosis and/or M. smegmatis (e.g., see references 10, 36, 98, 154, 204, and 205). Several have also been shown to be involved in the transport of several different antibiotics, including fluoroquinolones, aminoglycosides, tetracycline, rifampin, and possibly isoniazid and ethambutol. However, it is unclear which of these are associated with antibiotic resistance in clinical isolates.

Ziha-Zarifi et al. (237) showed that 11 patients had MDR P. aeruginosa that overexpressed the MexAB-OprM pump. Oh et al. (146) showed that 17 of 20 fluoroquinolone-resistant clinical isolates of P. aeruginosa expressed high levels of MexB, MexD, MexF, or MexY. Hocquet et al. (62) also showed that 14 of 18 isolates overexpressed the MexAB-OprM pump. Wolter et al. (230) showed that six of seven gentamicin-resistant isolates overexpressed the MexXY system. These data indicate that when resistant clinical isolates are investigated, a considerable number overexpress one or more of the Mex pumps that have been associated with clinically relevant levels of MDR.

In 22 of 36 fluoroquinolone-resistant isolates of E. coli, enhanced efflux was suggested (44). Webber and Piddock (227) confirmed that 11 isolates overexpressed acrB. Mazzariol et al. (123) showed that 9 of 10 clinical isolates of E. coli overexpressed AcrA. Both cyclohexane tolerance and multiple antibiotic resistance have been attributed to overproduction of the AcrAB-TolC complex in E. coli, mediated by overexpression of a global regulator (marA or soxS) (6, 229). It has been observed that in E. coli there is a close correlation between organic solvent tolerance and low-level resistance to multiple antibiotics (9). Tolerance to this organic solvent has been used as a marker for multiple antibiotic resistance (MAR) and used to screen clinical isolates. Randall et al. (180) observed a similar association for different serovars of S. enterica. Kallman et al. (80) showed that cyclohexane-tolerant clinical isolates of E. coli were inhibited by significantly higher MICs of cefuroxime, suggesting that efflux contributed to resistance to this agent in E. coli. Schneiders et al. (201) showed that 5 of 10 organic solvent-tolerant MDR K. pneumoniae isolates overexpressed AcrA.

In 2000, overexpression of acrB was shown in three MDR human clinical isolates of S. enterica serovar Typhimurium (165). Phenotypic data supported the role of overexpression as being a primary mechanism of resistance. Efflux via AcrAB-TolC has also been shown to be important in the MDR of isolates of S. enterica serovar Typhimurium DT104 from cattle (14).

Pumbwe et al. (176) examined 32 isolates of C. jejuni from both humans and poultry that had an MDR phenotype. Nine isolates overexpressed cmeB, and three of these also overexpressed cmeF. All isolates that overexpressed cmeB accumulated low concentrations of ciprofloxacin, suggesting that the overexpression of cmeB was involved in the MDR.

Recently, Higgins et al. (60) investigated an outbreak of MDR A. baumannii in a German hospital and showed that a pretherapy isolate, U10247, was susceptible to netilmicin and meropenem but that a subsequent isolate, U11177, became resistant due to overexpression of AdeABC, such that the MICs were above the breakpoint concentrations for gentamicin and cefotaxime, both agents that are used clinically. Two of the outbreak strains were clones and expressed 20-fold-more adeB mRNA than a strain obtained earlier in the outbreak (60).

Kaczmarek et al. (79) showed that four high-level, ampicillin-resistant clinical isolates of H. influenzae contained frameshift insertions in acrR which were associated with MDR.

Few studies have examined many clinical isolates of S. aureus for the prevalence of overexpression of NorA; however, where investigated it has been shown that some norfloxacin-resistant clinical isolates overexpress norA (70, 77, 142). However, other studies have found little or no relationship between overexpression of norA and fluoroquinolone resistance (147, 200). This may be because NorA does not transport the agents investigated. Since 1990, fluoroquinolones have been increasingly used as a treatment for infections caused by MRSA. Many MRSA strains then evolved to become resistant to fluoroquinolones and became widely disseminated, such that for some countries the predominant clones of MRSA are often fluoroquinolone resistant (39). While most of this fluoroquinolone resistance has been deemed to be due to mutations in the genes encoding the target protein (either grlA or gyrA), overexpression of NorA may also play a role. To date there is still no clear evidence as to the cause of the rapid increase and clonal spread of MRSA in many hospitals in developed countries, despite the availability of the sequences of the genomes of seven different strains of S. aureus, including three MRSA strains; there are many hypotheses for the epidemic spread of MRSA and problems in eradication. It may be that overexpression of NorA, with its concomitant effect on biocide activity, plays a role.

Piddock et al. (163) also determined the prevalence of pmrA overexpression in clinical isolates of S. pneumoniae from several geographically distinct areas. The isolates were divided into four categories: (i) those isolates inhibited by 16 µg/ml norfloxacin and for which reserpine lowered the MIC of norfloxacin fourfold and where the MIC suggested that these isolates had a phenotype similar to that of strain R6N; (ii) isolates that were susceptible to norfloxacin but for which reserpine also lowered the MIC of norfloxacin; (iii) norfloxacin-resistant (MIC 16 µg/ml norfloxacin) isolates for which reserpine had no effect; and (iv) norfloxacin-susceptible isolates for which reserpine had no effect. Isolates from groups i and iii also contained mutations in topoisomerase genes (163). The level of expression of pmrA mRNA was measured by Northern blotting and quantitative competitive RT-PCR, and it was shown that there were isolates in all four groups that overexpressed pmrA. Three isolates that were phenotypically similar to R6N also had no detectable expression of pmrA. These data indicate that pmrA overexpression is not exclusively associated with MDR S. pneumoniae isolates, despite their MIC phenotype suggesting otherwise.

Taken together, all of these studies indicate that overexpression of an efflux pump is often found in antibiotic-resistant clinical isolates and therefore impacts the therapeutic options available.

REGULATION OF EFFLUX PUMPS IN CLINICAL ISOLATES

Top Previous Next References

 
Although there have been many studies on the mechanisms of regulation of efflux pumps in laboratory-derived mutants, the mechanisms giving rise to increased efflux in clinical isolates have been shown to fall broadly into four groups: (i) mutations in the local repressor gene, (ii) mutations in a global regulatory gene, (iii) mutations in the promoter region of the transporter gene, and (iv) insertion elements upstream of the transporter gene.

Comparison of the mutations in the local repressor genes of various gram-negative species reveals that the majority of the substitutions in the proteins occur in the predicted helix-turn-helix motif involved in DNA binding to the target structural gene, i.e., the efflux pump gene, such as acrB (Table 9). Large deletions have also been observed and are predicted to render the repressor inactive.

The AdeABC efflux pump of A. baumannii is regulated by a two-component regulatory system encoded by AdeS and AdeR. Inactivation of AdeS gives rise to lower aminoglycoside MICs. Spontaneous gentamicin-resistant mutants contained substitutions in AdeS and in AdeR; these led to constitutive expression of the pump and MDR (115).

Expression of norA in S. aureus is via two systems, one a two-component regulatory system, ArlRS, the other MgrA (NorR) (219). Recently, MgrA has been shown also to regulate Tet38 and NorB (218). A MarR-type regulator, MepR, regulates expression of MepA (74). To date, no data for clinical isolates have been published, so it is not known what the contributions of these mechanisms are, if any, in conferring the overexpression of norA in clinical isolates.

Rob is also a member of the same AraC/XylS family of transcriptional regulators as MarA and SoxS (17). Overexpression of rob in E. coli produces both increased organic solvent tolerance and low-level resistance to multiple antimicrobial agents, due to increased expression of the AcrB-TolC system (229).

Other species of the genus or of the family Enterobacteriaceae possess homologues of the AcrAB-TolC MDR efflux pump, and so expression is also considered to be regulated as in E. coli. However, evidence is accumulating to suggest that the global regulator may not always be MarR or SoxR but RamA. Overexpression of this protein has been associated with MDR in S. enterica serovar Typhimurium, E. aerogenes, and K. pneumoniae (29, 49, 201, 222).

There are few global regulatory genes on the genome of P. aeruginosa, and regulation of the Mex pumps in P. aeruginosa is considered to be due to a mutation(s) in one of the local regulatory genes, such as mexR. Recently, however, SoxR, but not SoxS, in P. aeruginosa has been described (153). SoxR regulates a six-gene regulon including genes encoding putative efflux pumps and a pump involved in quorum-sensing signal homeostasis. Evidence that may support a global regulator involved in efflux-mediated antibiotic resistance in P. aeruginosa is provided by several studies with clinical isolates. First, an isolate for P. aeruginosa from a chronic obstructive airways disease patient overexpressed not only the MexAB-OprM efflux pump but also the MexEF-OprN pump and had decreased expression of a porin protein (OprF), plus a mutation in a topoisomerase gene (161, 173, 174). By complementing various genes, the role of each mechanism was determined. The net result was to yield a broadly resistant organism (Table 11). Le Thomas et al. (93) isolated a similar mutant from a surgical site. Llanes et al. (105) also showed that clinical isolates of P. aeruginosa can express two efflux pumps simultaneously and provided evidence for additional genes that regulate expression of mexAB-oprM and mexXY.

Although N. meningitidis also possesses MtrCDE, expression is not regulated by MtrR or MtrA. Analysis of 12 isolates revealed the presence of an insertion sequence (otherwise known as a Correia element) in all isolates (191). One isolate also contained IS1301. The authors concluded that the Mtr efflux system of N. meningitidis was regulated by integration host factor and posttranscriptional regulation by cleavage in the inverted repeat of the Correia element.

MDR EFFLUX PUMPS AND DEVELOPMENT OF ANTIBIOTIC RESISTANCE

Top Previous Next References

 
For some antibiotics in clinical use, the clinical relevance of overexpression of an efflux pump is that this confers MDR. However, there is also evidence to suggest that increased expression of efflux pumps may be the first step in which a bacterium becomes resistant to clinically relevant antimicrobials. It has been shown that reserpine (see "Inhibitors of MDR Efflux Pumps," below) suppresses the in vitro emergence of norfloxacin-resistant S. aureus (117) and ciprofloxacin-resistant S. pneumoniae (116, 118). Lomovskaya et al. (106) also showed that the efflux pump inhibitor MC-207-110 suppressed the emergence of levofloxacin-resistant P. aeruginosa. This phenomenon has been confirmed and extended for S. pneumoniae and gemifloxacin and moxifloxacin (Garvey and Piddock, unpublished data). It was proposed that inhibition of efflux gave rise to high intracellular concentrations, such that for S. aureus the MICs afforded by mutations in the gene encoding the target topoisomerase are too low to allow survival (117).

It is also hypothesized that increased efflux of antimicrobials decreases the intracellular concentration, such that the bacterium can survive longer than may have been predicted from the MIC for that organism. During this time and within this population of bacteria, spontaneous mutants that contain mutations in genes encoding the target protein occur (e.g., gyrA of E. coli). Evidence in support of this hypothesis is that deletion of acrAB in E. coli with two mutations in gyrA resulted in the bacterium becoming hypersusceptible to fluoroquinolones, suggesting that a functional AcrAB MDR efflux pump is essential for resistance. Baucheron et al. (12) inactivated acrB in MDR S. enterica serovar Typhimurium DT204 (also containing a mutation[s] in topoisomerase genes), giving rise to low MICs, some of which were below the recommended breakpoint concentrations for these agents and this species. Luo et al. (108) obtained similar data when cmeB was deleted in C. jejuni containing a substitution in GyrA. Furthermore, S. enterica serovar Typhimurium strains that lack tolC do not give rise to ciprofloxacin-resistant mutants in vitro (185). S. enterica serovar Typhimurium lacking AcrB only gave rise to ciprofloxacin-resistant mutants when the concentration of bacteria was high (>10¹¹), and even so the frequency of mutation to resistance was very low, at 10¹³ (Randall et al., unpublished data). Underlining the importance of the AcrAB-TolC pump in the development of ciprofloxacin resistance in S. enterica serovar Typhimurium, the strain lacking AcrB gave rise only to mutants containing a substitution in GyrA, whereas the parent strain containing AcrB gave rise to MDR mutants and those with a substitution in GyrA. These data indicate that the AcrAB system of E. coli and S. enterica serovar Typhimurium and the homologous system, CmeABC, in C. jejuni are important in the development of resistance to fluoroquinolones and some other agents. Use of fluoroquinolones in veterinary medicine has been a controversial issue for some years, as there are data to indicate that such use allows the selection of resistant bacteria that are transmitted to humans via the food chain. Use of efflux pump inhibitors concomitant with the use of antimicrobials may reduce the selection pressure, thereby allowing this valuable class of antimicrobials to continue to be used in animals.

NATURAL ROLES OF MDR PUMPS

Top Previous Next References

 

Over the last 10 years there have been sporadic reports implicating components of efflux pumps in pathogenicity. Stone and Miller (213) showed that S. enterica serovar Enteritidis tolC Tn5 insertional mutants were avirulent for mice. Urban et al. (220) found that the phytopathogenic fungus Magnaporthe grisea requires the up-regulation of specific ABC transporters for pathogenesis. In 2002, Hirakata et al. (61) concluded that the P. aeruginosa MexAB-OprM efflux system exports virulence determinants that contribute to the virulence of this organism. Jerse et al. (68) reported that a functional MtrCDE efflux system in N. gonorrhoeae enhanced bacterial survival in a female mouse model of genital tract infection. These authors suggested that this was due to the bacteria being able to survive in the presence of hydrophobic mucosal substances. Lin et al. (101) demonstrated that the efflux pump CmeB confers resistance of C. jejuni to bile. They also suggested that resistance to bile improves C. jejuni survival in vivo in poultry and concluded that inhibition of CmeABC function not only may be involved in antibiotic resistance but also may prevent in vivo colonization by Campylobacter. Recently, Burse et al. (26) reported