MMBR Figure table search 04
Home Help [Feedback] [For Subscribers] [Archive] [Search] [Contents]
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow Copyright Information
Right arrow Books from ASM Press
Right arrow MicrobeWorld
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Grkovic, S.
Right arrow Articles by Skurray, R. A.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Grkovic, S.
Right arrow Articles by Skurray, R. A.
Microbiology and Molecular Biology Reviews, December 2002, p. 671-701, Vol. 66, No. 4
1092-2172/02/$04.00+0     DOI: 10.1128/MMBR.66.4.671-701.2002
Copyright © 2002, American Society for Microbiology. All Rights Reserved.

Regulation of Bacterial Drug Export Systems

Steve Grkovic, Melissa H. Brown, and Ronald A. Skurray*

School of Biological Sciences, University of Sydney, New South Wales 2006, Australia

SUMMARY
INTRODUCTION
TRANSPORTER OVEREXPRESSION AND ANTIMICROBIAL RESISTANCE
    P. aeruginosa MDR Pumps and Multiantibiotic Resistance
    Drug Transporter Expression in Other Pathogens and Antibiotic Producers
    Regulation of N. gonorrhoeae MDR Transporters
CONTROL OF E. COLI acrAB MDR LOCUS BY GLOBAL ACTIVATORS
    MarA Global Activator
    MarR, Antimicrobial-Sensing Repressor of marRAB
    MarA Homologs SoxS and Rob
    Other Regulators of E. coli MDR Pumps
RELATED LOCAL AND GLOBAL ACTIVATORS CONTROL MDR GENE EXPRESSION IN B. SUBTILIS
    BmrR, Local Transcriptional Activator of bmr Expression
    Structure of Tripartite BmrR-Drug-Activated DNA Complex
    Binding of Antimicrobial Ligands by BmrR
    Global B. subtilis MDR Gene Regulator Mta
TetR, EXQUISITELY SENSITIVE REGULATOR OF TETRACYCLINE EFFLUX GENES
    Contribution of tet Promoters and Operators
    Structure of a TetR Dimer
    TetR Binding to tet Operator
    TetR-Ligand Interactions
    Conformational Changes in TetR That Transmit the Induction Signal
    QacR, REGULATOR OF S. AUREUS qacA/B MDR GENES
    QacR DNA Binding
    Versatile QacR Multidrug-Binding Pocket
    QacR Induction Mechanism
OVERVIEW
ACKNOWLEDGMENTS
REFERENCES

   SUMMARY
 Top
 Next
 References
 
The active transport of toxic compounds by membrane-bound efflux proteins is becoming an increasingly frequent mechanism by which cells exhibit resistance to therapeutic drugs. This review examines the regulation of bacterial drug efflux systems, which occurs primarily at the level of transcription. Investigations into these regulatory networks have yielded a substantial volume of information that has either not been forthcoming from or complements that obtained by analysis of the transport proteins themselves. Several local regulatory proteins, including the activator BmrR from Bacillus subtilis and the repressors QacR from Staphylococcus aureus and TetR and EmrR from Escherichia coli, have been shown to mediate increases in the expression of drug efflux genes by directly sensing the presence of the toxic substrates exported by their cognate pump. This ability to bind transporter substrates has permitted detailed structural information to be gathered on protein-antimicrobial agent-ligand interactions. In addition, bacterial multidrug efflux determinants are frequently controlled at a global level and may belong to stress response regulons such as E. coli mar, expression of which is controlled by the MarA and MarR proteins. However, many regulatory systems are ill-adapted for detecting the presence of toxic pump substrates and instead are likely to respond to alternative signals related to unidentified physiological roles of the transporter. Hence, in a number of important pathogens, regulatory mutations that result in drug transporter overexpression and concomitant elevated antimicrobial resistance are often observed.


   INTRODUCTION
 Top
 Previous
 Next
 References
 
Drug-resistant microorganisms are a major worldwide health issue, as a number of important human pathogens have now acquired mechanisms that make them largely resistant to all currently available treatment regimens. The action of antimicrobial compounds can be negated at a number of points, including enzymatic inactivation, the employment of alternative metabolic pathways to bypass their activity, sequestration, reduced uptake, and alteration of the target site to render it not susceptible to the effect of otherwise toxic substances (28). A further resistance mechanism that has become increasingly important involves membrane-bound efflux pumps that transport toxic antimicrobial compounds from the cell. ATP-binding cassette (ABC) transporters power this process via the hydrolysis of ATP, whereas secondary transporters utilize the transmembrane electrochemical gradient, typically the proton motive force, to drive drug efflux (161, 173).

The first drug transport proteins to be identified were a family of tetracycline efflux pumps, which provide widespread resistance to tetracycline antibiotics in both gram-negative and gram-positive bacteria (25). More recently, a large number of multidrug resistance (MDR) transport proteins have been found to be involved in the export of a wide range of antimicrobial compounds (161, 173, 185). In contrast to the narrow substrate range of most transporters, including the tetracycline efflux determinants, individual MDR pumps are capable of exporting compounds that have few structural similarities. MDR determinants appear to contribute to the emergence of drug-resistant microorganisms via two mechanisms. First, they can confer low-level protection that facilitates the initial survival of the organism and thus provides it with the opportunity to subsequently acquire one of the high-level specific resistance mechanisms listed above. Alternatively, the MDR transporters themselves can furnish protection against clinically relevant concentrations of many antimicrobial compounds.

In addition to providing many pathogenic bacteria with protection against antibiotics, antiseptics, and disinfectants (161), drug transporters pose a number of additional medical challenges. Host-encoded antimicrobial compounds that are produced to combat infections caused by organisms such as Escherichia coli (143) and Neisseria gonorrhoeae (203) are also exported by some MDR pumps; in Staphylococcus aureus, the presence of an MDR transporter has been correlated with resistance to a cationic antimicrobial peptide (17, 87). Treatment of immunocompromised patients undergoing long-term antifungal therapy is often complicated by development of resistance to antifungal agents by the pathogenic yeast Candida albicans as a result of MDR transporter overexpression (84). Additionally, in human cancer cells, overexpression of an MDR transporter, P-glycoprotein, can provide high-level resistance to antitumor drugs, a finding which has been linked to the failure of chemotherapy (45).

Although these problems have generated a substantial degree of interest in drug transporters, the close association of efflux pumps with cellular membranes has severely hampered efforts to define the exact molecular mechanisms by which these proteins function. Only recently have high-resolution structures for any membrane transport proteins been elucidated, the E. coli proteins MsbA (23) and BtuCD (103), although more limited structural information has also been gained by employing electron microscopy of two-dimensional crystals, e.g., the single-substrate TetA tetracycline exporter has been shown to function as a trimer (240), the MDR transporters EmrE (216) and YvcC (22) form dimers, and P-glycoprotein functions as a monomer (180, 181).

In comparison to the limited achievements in understanding the structure-function relationships of the drug transporters themselves, research into the regulatory pathways that govern the expression of drug transporters has progressed relatively rapidly, in particular for a number of bacterial regulatory proteins. The antimicrobial pumps which are known to be subject to regulatory controls typically belong to either the major facilitator superfamily (MFS) or resistance, nodulation and cell division (RND) superfamily (Table 1), both of which primarily employ the proton motive force to energize drug efflux (161, 173). The requirement for regulatory controls to prevent excessive production of an integral membrane protein that utilizes the proton motive force is demonstrated by the deleterious effect of constitutive expression of the gram-negative tetracycline/H+ antiporters TetA(B) and TetA(C), which, in the absence of tetracycline, place cells at a severe disadvantage when competing with nonconstitutively expressing strains (92, 142). Overproduction of TetA(B) can also be lethal to the cell if the gene is expressed from a strong promoter, resulting in nonspecific cation transport, loss of the membrane H+ potential, and cell death (31, 57).


View this table:
[in this window]
[in a new window]
 
TABLE 1. Transcriptional regulators that control expression of genes encoding bacterial drug efflux components

 
However, despite these observations, all members of a further family that utilize the proton motive force, the small multidrug resistance family, a subgroup of the drug/metabolite transport superfamily (67), do not appear to be subject to any regulatory controls that can alter the level at which these proteins are synthesized. It is therefore an intriguing question why some proteins that utilize the proton motive force to energize drug efflux are synthesized under strict regulatory controls, yet others appear to be expressed constitutively. This could well reflect currently unknown physiological roles for the unregulated pumps in normal cellular metabolism that require the constant low-level presence of such transporters. The supply of a small but continuous amount of a protein can be easily governed by mechanisms such as low-level production of the relevant mRNA and/or high turnover rates of the mRNA and/or the transport protein, without any need for additional, more complex regulatory controls.

For the bacterial drug efflux genes that are inducible, there are only a few documented cases for which translational controls have been shown to be the primary level at which expression is controlled. For example, translational attenuation has been proposed to modulate the synthesis of the gram-positive tetracycline resistance determinant TetA(K) (206), whereas a second gram-positive tetracycline resistance protein, TetA(L), is regulated by translational reinitiation (209). Experimental evidence suggests that the latter example involves tetracycline-induced stalling of ribosomes during the translation of a short leader peptide. This is likely to facilitate the transfer of ribosomes to the tetA(L) ribosome-binding site, an event which requires the presence of a stem-loop structure that is proposed to correctly orient the tetA(L) ribosome-binding site for ribosome transfer to occur (209).

In contrast to these determinants, expression of the majority of the bacterial drug transporter genes which are known to be subject to regulation is controlled by transcriptional regulatory proteins. Well-characterized proteins with a demonstrated role in controlling the expression of drug efflux genes encompass examples of both repressors and activators of target gene transcription, a process that can occur at either the local or global level. Local regulators of drug transporter genes include the Escherichia coli TetR repressor of tetracycline efflux genes (59) and three regulators of MDR transporter genes, the Bacillus subtilis BmrR activator (2), the Staphylococcus aureus QacR repressor (47), and the E. coli EmrR repressor (105). In some instances, local regulators appear to play only a modulating role, the principal factor controlling transcription instead being global regulatory proteins; e.g., increases in the expression of the E. coli acrAB MDR locus are mediated by the MarA, Rob, and SoxS global activators (7). Two-component regulatory systems are also increasingly being found to be associated with drug efflux genes (Table 1). Sequencing of entire bacterial genomes has identified a large number of additional MDR transporter homologs and their associated regulatory elements, although the functions of the majority of these systems remain to be experimentally investigated (162).

In general, the confirmed regulators of bacterial drug transporter genes belong to one of four regulatory protein families, the AraC, MarR, MerR, and TetR families, despite being from distantly related species, a classification which also shows little correlation with the family of the drug pump whose expression they control (Table 1). However, the assignment of these regulatory proteins to their respective families is based solely on similarities detected within their DNA-binding domains, which typically constitute only one third of each polypeptide. Like the majority of bacterial activators and repressors, the drug transport regulators identified to date all possess {alpha}-helix-turn-{alpha}-helix (HTH) DNA-binding motifs, which are embedded in larger DNA-binding domains that form a number of different structural environments, such as three-helix bundles and winged helix motifs (155). These serve to create a stable three-dimensional structure that buries the hydrophobic side chains of amino acids in the interior of the DNA-reading head but generally orients the second "recognition" helix of the HTH so that it fits into the major groove of B-DNA. Further detailed information on the structure and function of HTH motifs can be found in a number of excellent reviews (39, 54, 64, 155, 213).

Importantly, for four local regulators of drug efflux determinants, the portions of these proteins not involved in forming the DNA-binding domains have been demonstrated to be capable of directly binding substrates of their cognate pumps, which act as a signal to increase the synthesis of the relevant transport protein(s) in response to the presence of these toxic compounds. This finding has had important ramifications for the field of protein-drug interactions, since, in contrast to the membrane-bound transport proteins, which are notoriously difficult to purify and study in vitro, the soluble cytosolic regulators of drug resistance have provided much more amenable systems for the study of drug recognition and binding. For the TetR (60), QacR (198), and BmrR (245) regulatory proteins in particular, detailed X-ray crystallographic and biochemical data, combined with mutational studies, have been highly successful in providing a wealth of information on the molecular aspects involved in drug binding and the subsequent steps that result in the induction of target gene expression.

By concentrating principally on the better-characterized regulatory pathways, at both the local and global levels, this review is intended to illustrate the extremely varied nature of the transcriptional regulatory controls that act upon the genes encoding bacterial drug efflux systems. Particular emphasis is placed on the contributions that analysis of these regulatory pathways and their attendant proteins have made to the field of drug resistance as a whole. We also discuss the insights that have been gained from evolutionary and medical perspectives.


   TRANSPORTER OVEREXPRESSION AND ANTIMICROBIAL RESISTANCE
 Top
 Previous
 Next
 References
 
Almost all of the antimicrobial compounds, both synthetic and natural, which have been employed by humans to combat infectious bacteria are substrates of one or more drug efflux pumps, e.g., the fluoroquinolones; the antibiotics chloramphenicol, tetracycline, ß-lactams, and aminoglycosides; and the antiseptics benzalkonium, cetrimide, and chlorhexidine. Tetracycline-specific pumps such as TetA(B) possess regulatory controls that are sensitively attuned for adjusting expression levels in response to the presence of tetracycline. In contrast, the vast majority of bacterial MDR genes need to be expressed at a level substantially greater than that observed in the wild-type organism before significant efflux of clinically relevant antimicrobial compounds occurs. This finding has lent considerable support to the proposal that MDR pumps in general have preexisting physiological roles, such as protection against low levels of toxic hydrophobic molecules encountered in the natural environment of many organisms or the transport of specific metabolites. However, for a limited number of the drug transporters identified to date, such as the tetracycline pumps and perhaps also the plasmid-encoded S. aureus MDR determinant QacA, which exports an impressive array of antiseptics, disinfectants, and related compounds (21), analysis of the pumps and their regulatory controls indicates that the efflux of medically relevant antimicrobial compounds is now the primary function of these systems.

For clinical isolates of the important human pathogens E. coli, Neisseria gonorrhoeae, Pseudomonas aeruginosa, and S. aureus, a large number of regulatory pathway mutations that have resulted in the overexpression of various MDR determinants and a concomitant elevated resistance to antimicrobial compounds have been characterized. For example, it has been well established that low-level multiantibiotic resistance in E. coli can arise from overexpression of the AcrAB-TolC MDR efflux complex as a result of increased production of the MarA global transcriptional activator protein (112, 149, 232). MarA and a number of other E. coli MDR regulators are discussed in greater depth later in this review. Overexpression of MDR transporters has now been identified as a major source of antimicrobial resistance in an alarmingly large number of pathogenic species. The most striking example of this phenomenon is the serious opportunistic pathogen P. aeruginosa, for which the hyperexpression of MDR pumps has been particularly important in the emergence of multiantibiotic-resistant strains.

P. aeruginosa MDR Pumps and Multiantibiotic Resistance

Mutations leading to the overexpression of pump complexes such as MexAB-OprM have now been identified as the predominant cause for acquisition of elevated resistance to structurally unrelated antibiotics by strains of P. aeruginosa (167). The mexAB-oprM operon encodes a tripartite pump complex that comprises the MexB cytoplasmic pump component, the MexA membrane fusion protein, and the OprM outer membrane channel (Fig. 1). Thus, the MexAB-OprM complex can simultaneously transport antibiotics across both the cytoplasmic and outer membranes, which provides a highly effective resistance mechanism when combined with the notoriously low permeability of the P. aeruginosa outer membrane (167). MexAB-OprM together with MexCD-OprJ, MexEF-OprN, and MexXY (which also utilizes the OprM outer membrane channel) form a family of related P. aeruginosa multidrug efflux pump complexes that have an extremely broad substrate range (134, 168). Each of the operons for these four pump complexes has a local transcriptional regulatory protein that is encoded in the immediately adjacent upstream region (Fig. 1). In stark contrast to the high degree of relatedness between the individual components of the pump complexes, the proteins encoded by each of these upstream genes exhibit no homology to each other; instead, they belong to four distinct regulatory protein families (Table 1).



View larger version (63K):
[in this window]
[in a new window]
 
FIG. 1. Genetic organization of the mexR-mexAB-oprM, mexT-mexEF-oprN, nfxB-mexCD-oprJ, and mexZ-mexXY MDR loci from P. aeruginosa. Each operon contains genes (grey arrows) that encode a drug efflux complex and is regulated by the product of an upstream gene (black arrow) which either represses (-) or activates (+) operon expression, although this is yet to be confirmed for MexZ. The intergenic region separating the mexR and mexA genes is depicted in greater detail, with the positions of the binding sites for the two MexR dimers (black ovals) indicated relative to the -10 and -35 hexamers of the mexR promoter (PmexR), the mexA promoter (P2mexA), and a second potential mexA promoter (P1mexA). A schematic representation of the MexAB-OprM tripartite complex, which can efflux drugs simultaneously across both the cytoplasmic (CM) and outer (OM) membranes, is also shown. MexB, the RND component of the pump complex, transports drugs across the cytoplasmic membrane in exchange for protons (H+).

 
The regulation of the mexAB-oprM operon is the best-characterized example; the divergently encoded MexR, a member of the MarR family of proteins (Table 1), acts as a repressor of transcription of mexAB-oprM (170). MexR autoregulates expression of its own gene by repressing the mexR promoter (Fig. 1, PmexR) in addition to controlling transcription from the mexA promoter (Fig. 1, P2mexA) (32). A second promoter (Fig. 1, P1mexA) has been proposed to be responsible for the relatively high constitutive level of mexAB-oprM expression, which contributes substantially to the high intrinsic resistance to antibiotics that P. aeruginosa exhibits. However, recent experiments with S1 mapping and reporter gene fusions have suggested that P1mexA is not a functional promoter (189). A large number of mutant strains that contain alterations to the mexR coding region and produce an inactive MexR repressor have been described; all of these mutations result in overproduction of the MexAB-OprM pump complex and enhanced efflux of a broad range of antibiotics, such as ß-lactams, fluoroquinolones, tetracycline, and chloramphenicol (98, 186, 208, 246).

It has been proposed that the increased expression from the MexR-controlled promoter (P2mexA) occurring in these strains is responsible for the observed MexAB-OprM hyperexpression (32). In some instances, the inability of the MexR derivatives to repress P2mexA has been confirmed to be due to the production of MexR proteins which are either unstable or compromised in their ability to dimerize or bind DNA (1). However, none of the antibiotics to which these mutants exhibit increased resistance induce expression, suggesting that the MexAB-OprM pump complex has evolved for other physiological roles, which may include the export of secondary metabolites (166) as well as a demonstrated role in the active efflux of an N-(3-oxododecanoyl)-homoserine lactone autoinducer signal associated with quorum sensing (163).

A recent crystal structure for apo-MexR has suggested that the MexR ligand may be an acidic peptide signaling molecule or the C terminus of a protein ligand, either of which could insert between the two winged-helix DNA-binding domains of a MexR dimer and induce a significant reduction in their spacing, thereby rendering the protein incapable of binding DNA (101). Further mutations that alter mexAB-oprM transcription but are located outside of mexR or the intergenic region (171, 208, 246) support the existence of unidentified regulatory proteins that influence the expression of this MDR pump complex, as does the linking of mexAB-oprM expression to the growth phase of cells (33). Interestingly, the MDR systems of P. aeruginosa have also been found to be subject to coordinate regulation, the expression of some systems decreased in response to increased levels of another (99).

Additional multiantibiotic-resistant pseudomonal mutants with an NfxC phenotype display resistance to chloramphenicol, fluoroquinolones, and trimethoprim due to overexpression of the mexEF-oprN MDR efflux operon (81). The locations of the mutations which produce this phenotype are unknown, but the overexpression is dependent on the presence of a functional MexT protein, which is a LysR-type transcriptional activator (Table 1) encoded by an upstream gene transcribed in the same direction as the mexEF-oprN operon (Fig. 1) (82, 123). As for MexR, none of the known substrates of MexEF-OprJ have any effect on the activity of MexT or expression of mexEF-oprJ, although simple overexpression of the MexT activator is sufficient to instigate transcription of this operon (81, 148). However, because most LysR-type regulators become active upon binding a cognate effector molecule, it has been suggested that the NfxC phenotype may result from the overproduction of a MexT effector molecule that normally induces production of the MexEF-OprN pump complex in response to the presence of its physiological substrate(s) (81). An alternative explanation involving mutations in an unidentified suppressor of mexT expression has also been put forward (123).

MexT may possess both repressor and activator functions, as it has been implicated in the downregulation of oprD, a gene which encodes a porin involved in the uptake of the ß-lactam imipenem (148). Thus, the resistance to imipenem that nfxC-type mutants exhibit is due to reduced uptake of this antibiotic rather than its efflux by the overproduced MexEF-OprN pump complex. MexEF-OprN has also recently been proposed to influence the intracellular levels of Pseudomonas quinolone signal, a molecule involved in cell-to-cell signaling (83). This has been linked to the surprising finding that nfxC-type strains that overexpress MexEF-OprM, and hence have elevated antibiotic resistance, are actually less virulent due to a decrease in the production of extracellular virulence factors (83).

Overexpression of the mexCD-oprJ operon due to mutations in its divergently transcribed repressor of synthesis, NfxB (Fig. 1), can also confer multiantibiotic resistance on P. aeruginosa (169). NfxB autoregulates its own expression (204) and normally completely represses the mexCD-oprJ operon. Construction of P. aeruginosa strains in which three of the four mex operons are disrupted recently permitted the demonstration of mexCD-oprJ induction by acriflavine, tetraphenylphosphonium (TPP), ethidium bromide, or rhodamine 6G, all of which are substrates of this MDR pump complex (134). Although this result suggests a possible physiological role for MexCD-OprJ in the extrusion of toxic compounds, the mechanism by which the observed induction occurs needs to be identified before more concrete conclusions can be drawn.

Overexpression of the fourth P. aeruginosa multidrug resistance locus, MexXY, has been associated with aminoglycoside resistance (130, 233). In wild-type cells, MexXY also contributes to intrinsic antibiotic resistance by being inducible by the pump substrates tetracycline, erythromycin, and gentamicin, although again the mechanism by which this occurs is unknown (124). mexXY has been proposed to be regulated by MexZ, the product of a divergently transcribed gene (Fig. 1) that encodes a TetR family repressor (4). In addition to the four MDR transporter complexes discussed above, the extremely large P. aeruginosa genome appears to encode a number of additional potential efflux systems which show significant homology to known drug exporters (210).

Despite the recent demonstration that both MexCD-OprJ and MexXY are inducible by some antimicrobials, the majority of the currently available information indicates that the regulatory networks controlling the expression of MDR systems in P. aeruginosa are primarily intended to respond to physiological signals that are not related to the efflux of drugs. In support of this proposal is the identification of a number of Pseudomonas putida pump complexes that exhibit strong homology to MexAB-OprM. Although some of these systems are capable of exporting several structurally dissimilar antibiotics, the predominant physiological role of these efflux complexes appears to be the energy-dependent export of toxic organic solvents, which permits P. putida cells to grow in media containing relatively high concentrations of aromatic hydrocarbons (79, 175, 179). The identification of local and global regulatory proteins, both of which are likely to be involved in controlling the expression of P. putida pumps in response to the presence of toxic organic solvents, provides strong evidence that aromatic hydrocarbon efflux is the authentic function of these systems (30).

In contrast to the P. putida organic solvent pumps, the four P. aeruginosa pump complexes discussed above export an extensive and overlapping array of structurally dissimilar drugs, although none of their local regulatory proteins appear to be involved in responding to the presence of these pump substrates and no alternative induction mechanisms have yet been elucidated. Similarly, ArpABC, another P. putida RND pump which can export structurally unrelated antibiotics, such as carbenicillin, chloramphenicol, erythromycin, and tetracycline, but does not contribute to organic solvent tolerance is also not induced by the addition of antibiotics or solvents (78). However, pseudomonal MDR pump complexes continue to have a substantial impact on antimicrobial resistance due to the many clinical strains which overexpress MDR transporters as a result of regulatory mutations, a trend which is becoming increasingly frequent in other established and emerging human pathogens.

Drug Transporter Expression in Other Pathogens and Antibiotic Producers

Increased synthesis of a chromosomal S. aureus MDR gene which encodes the MFS transporter NorA is particularly important in the emergence of fluoroquinolone-resistant clinical isolates of this species (76, 239, 241). The flqB mutation, a single T-to-G substitution in the 5' untranslated region upstream of norA, has been shown to increase the half-life of norA mRNA 4.8-fold, resulting in NorA overproduction (37). It has been proposed that the increased mRNA stability is a consequence of changes to its secondary structure, which may affect an RNase III cleavage site involved in the degradation of this transcript (37). A single T-to-A change in the norA promoter region also appears to result in overexpression of NorA and subsequent fluoroquinolone resistance (75). Other mutations that produce increased resistance but lie outside the norA coding and promoter regions are currently uncharacterized (74, 136).

Although the regulation of norA expression is poorly understood, an unidentified 18-kDa protein has been demonstrated to bind to multiple sites in the DNA sequences upstream of the -35 region of the norA promoter (35). The binding of this 18-kDa protein is modified by ArlS, which appears to function as the transmembrane sensor of a two-component regulatory system, ArlR-ArlS. The first component of these systems is typically a transmembrane sensor protein that undergoes autophosphorylation upon binding a signal molecule present in the external environment (158). The phosphate is then transferred to the second component, a cytoplasmic response regulator which can be reversibly phosphorylated at a conserved aspartate residue (158). The activity of the response regulator is thereby altered so that it can then act to either stimulate or repress target gene transcription.

Although disruption of the gene for the ArlS transmembrane sensor resulted in upregulation of NorA expression and also altered the growth phase regulation of this transporter, it is not known if ArlS phosphorylates the 18-kDa protein directly or, alternatively, acts indirectly, e.g., via ArlR altering the transcription of a gene involved in the production of a ligand for the 18-kDa protein (35). The latter scenario could be related to the observation that S. aureus secretes a compound into the external medium that has the effect of lowering norA expression from the late logarithmic growth phase onwards (35). Although ArlR-ArlS also has a role in the regulation of virulence determinants in S. aureus (36), the expression of norA does not appear to be linked to that of known virulence determinants, suggesting that ArlR-ArlS modulates NorA expression in response to other physiological functions (35).

MDR pump overexpression has also made a substantial contribution to the emergence of many other bacterial species as serious human health threats. The intrinsic antibiotic resistance of the opportunistic pathogen Stenotrophomonas maltophilia has been attributed in part to MDR pumps (243), whereas elevated expression of the SmeDEF MDR pump in clinical isolates of this species has been correlated with increased levels of resistance to tetracycline, chloramphenicol, erythromycin, and quinolones (9). Although a second MDR pump homolog identified in this species, SmeABC, is not involved in drug efflux, the SmeC outer membrane component appears to be employed by another, as yet uncharacterized MDR pump which does transport antimicrobial compounds (100). The resistance of a clinical isolate of Acinetobacter baumannii to aminoglycosides and an extensive range of other antimicrobials has been confirmed to be due to an RND-type pump complex encoded by the adeABC gene cluster (111). While the basis for the increased contribution of the adeABC genes to drug efflux is unknown, three divergently transcribed genes which encode a putative two-component regulatory system (Table 1) and a transcriptional terminator-antiterminator have been proposed to be involved in the regulation of this MDR locus (111).

The inherent high-level resistance of the disease-causing bacterium Burkholderia pseudomallei to antimicrobial agents has also been partially attributed to an RND-type pump, AmrAB-OprA (133). Divergently transcribed from amrAB-oprA is the gene for AmrR, a TetR family protein proposed to be a transcriptional repressor of this operon. In the anaerobic pathogen Clostridium perfringens, the most widespread tetracycline resistance determinant is the plasmid-encoded tet(P) operon, comprised of two overlapping genes, which encode the TetA(P) tetracycline efflux pump and TetB(P), a protein likely to be involved in ribosome protection (72). Induction of this operon requires an unidentified, chromosomally encoded regulatory protein (73), whereas a transcriptional attenuation mechanism appears to prevent excessive production of TetA(P) in both the absence and presence of tetracycline (72). Although the mechanisms that lead to MDR overexpression and concomitant elevated antimicrobial resistance have yet to be identified in many of these species, it is immediately obvious that such processes represent one of the primary means by which bacteria subjected to drug exposure can acquire elevated resistance to these compounds.

For several antibiotic-producing bacteria, regulatory proteins have also been identified that control the synthesis of efflux pumps which are employed to provide protection against their own antibiotics. Streptomyces species, in particular, have been the object of considerable attention because these gram-positive bacteria are responsible for the production of the majority of commercially important antibiotics. Uncharacterized regulators designated Pip proteins have been proposed to be involved in the regulation of drug transport genes in a range of Streptomyces species (187, 188). Isolation of the Pip protein from the genetically well-defined Streptomyces coelicolor indicated that it was a TetR family repressor that regulates the expression of the MFS antiporter Ptr, which confers resistance to the antibiotic pristinamycin I (34).

The self-resistance of Streptomyces virginiae to the antibiotic virginiamycin S is provided by the efflux pump VarS, production of which is governed by complex regulatory controls that include the transcriptional repressor BarA, which responds to a {gamma}-butyrolactone autoregulator signal (138). Additionally, VarR, the product of a gene cotranscribed with varS, is a TetR family repressor that regulates varS transcription in a virginiamycin S-dependent manner (138). Another interesting example is the bacitracin peptide antibiotic-producing bacterium Bacillus licheniformis, which utilizes an ABC transporter, BcrABC, rather than a proton motive force-dependent exporter to provide self-resistance (164). An upstream two-component regulatory locus, bacRS, appears to control the expression of the bcrABC genes (139). The doxorubicin- and daunorubicin-producing organism Streptomyces peucetius also uses an ABC transporter, DrrAB, to confer self-resistance to these antibiotics (77). The role of activating expression of the synthesis and resistance operons for doxorubicin and daunorubicin in S. peucetius has been tentatively assigned to DnrI, which itself may be under the control of another positive transcriptional activator from the same gene cluster, DnrN (110).

Both the antibiotic biosynthesis genes and the self-resistance genes in all of these antibiotic-producing organisms appear to be under additional complex regulatory controls that respond to a number of environmental factors, in addition to linking their expression to differentiation events such as sporulation. In stark contrast to the majority of the drug pumps that confer antibiotic resistance to pathogenic bacteria, the expression of the transporters employed by antibiotic-producing organisms to provide self-resistance is intimately linked to the production, and hence the presence, of the relevant pump substrate(s). Understanding the function and regulation of these self-resistance mechanisms has considerable medical relevance, because in addition to aiding the exploitation of these species, their self-resistance mechanisms represent a pool of highly evolved determinants that can potentially be acquired by pathogenic bacteria.

Regulation of N. gonorrhoeae MDR Transporters

The mtrCDE operon of Neisseria gonorrhoeae encodes a multidrug efflux complex that is capable of exporting a range of MDR substrates that includes acriflavine, crystal violet, macrolides, and penicillin. However, the physiological role of this MDR pump is more likely to be reflected by its ability to extrude a range of structurally diverse hydrophobic agents, typically host-derived antimicrobial compounds such as fatty acids, bile salts, gonadol steroids, and antibacterial peptides, many of which coat mucosal sites colonized by N. gonorrhoeae (51). Transcription of mtrCDE can be increased by the addition of Triton X-100, an MtrCDE substrate which has a membrane-acting antimicrobial activity similar to that of the host-derived hydrophobic agents (184). Thus, in addition to being substrates of the efflux pump, hydrophobic agents are likely to act as natural inducers.

Analysis of the completed N. gonorrhoeae genome sequence identified the AraC family activator MtrA (Table 1), which was shown to be required for Triton X-100 enhancement of transcription (184). MtrA, like most AraC proteins (119), has an additional N-terminal domain which is likely to be involved in binding ligands, although it is currently unknown if this putative ligand-binding site plays a role in sensing the presence of toxic hydrophobic agents.

Divergently transcribed from mtrCDE is MtrR (156), a TetR family repressor (Table 1) that binds to a region which encompasses the mtrCDE promoter (106). Although MtrR appears to function solely as a modulator that is not involved in induction of mtrCDE expression (52, 106), mutations in either mtrR, the promoter for this gene, or the MtrR binding site can result in elevated mtrCDE expression and increased levels of resistance to hydrophobic agents (52, 106, 201, 202). Similar mutations have also been shown to produce increased resistance to the antibiotics azithromycin and erythromycin (242). However, disruption of MtrR binding alone does not fully explain the increases in mtrCDE transcription that result from some mutations that lie within the MtrR binding region and/or the mtrR promoter, suggesting the existence of other regulatory mechanisms and demonstrating the impact that cis-acting factors can have on efflux pump regulation (52).

A second N. gonorrhoeae efflux pump, FarAB, which utilizes the same outer membrane protein as the MtrCDE complex, MtrE (203), confers resistance to long-chain fatty acids (91). Surprisingly, although MtrR represses mtrCDE, it appears to enhance farAB expression (91), in addition to being either directly or indirectly responsible for controlling the expression of 14 other genes which may be linked to the establishment and/or maintenance of infections by N. gonorrhoeae (203). Overall, both the substrate range of the pumps and the available information on their regulation suggest that the function and expression of FarAB and MtrCDE are intimately linked to the virulence of N. gonorrhoeae.

The examples discussed to this stage have shown that the MDR systems of pathogenic bacteria are in general ill-adapted for the export of medically relevant drugs because of the requirement for mutations in their regulatory circuits before clinically significant resistance is observed. However, the following closer analysis of MDR regulatory controls from the model gram-negative and gram-positive organisms E. coli and B. subtilis, respectively, will demonstrate that this is not entirely the case. Additionally, the transcriptional repressors of tetracycline-specific pumps and the S. aureus plasmid-encoded QacA MDR determinant respond specifically to substrates of these pumps, and thus these regulatory proteins do not need to be bypassed before significant antimicrobial efflux can occur. These systems serve to illustrate in detail the better-characterized examples of proteins involved in the regulation of drug efflux genes, including examples of global activators and local activators or repressors.


   CONTROL OF E. COLI acrAB MDR LOCUS BY GLOBAL ACTIVATORS
 Top
 Previous
 Next
 References
 
The E. coli AcrB RND pump functions as part of a tripartite complex which also consists of the membrane fusion protein AcrA (108) and the outer membrane channel protein TolC (11, 38), which is encoded in a remote part of the chromosome (Fig. 2). This MDR efflux system confers resistance to a diverse range of antimicrobials, such as dyes, detergents, fluoroquinolones, and many other lipophilic antibiotics, e.g., ß-lactams, chloramphenicol, erythromycin, and tetracycline (108, 143, 150, 234). Immunoblotting has demonstrated that the AcrA protein was overexpressed in 9 of 10 E. coli clinical isolates which expressed high-level fluoroquinolone resistance (125). This has been attributed primarily to mutations that increase the production of the MarA global regulator, which modulates the expression of many genes, including acrAB and tolC (7, 149, 232). However, a number of insertions and single-amino-acid substitutions, duplications, and deletions that inactivate AcrR, a divergently transcribed local repressor of the acrAB operon (Fig. 2), have also been shown to result in enhanced expression of acrAB and increased fluoroquinolone resistance in clinical E. coli strains (71, 146, 230).



View larger version (27K):
[in this window]
[in a new window]
 
FIG. 2. Schematic representation of the known regulatory controls governing the expression of the E. coli acrAB and tolC genes. The AcrAB-TolC transport complex extrudes drugs across both the cytoplasmic (CM) and outer (OM) membranes (pale yellow shaded boxes). Excessive production of AcrA and AcrB is prevented (-) by the local dimeric repressor protein AcrR (green), whereas a regulatory protein involved in cell division, SdiA (grey), can increase (+) acrAB expression. However, activation of acrAB and tolC transcription occurs primarily because of the global regulatory proteins MarA, SoxS, and Rob (purple), any one of which can bind to a marbox (cyan) upstream of these genes. The intracellular level of MarA is controlled by MarR (red), a dimeric protein which binds to marO (orange) and represses (-) the expression of its own gene and the two others that constitute the marRAB operon. Binding of inducing compounds (diamonds) such as salicylate by MarR, in addition to the possible phosphorylation of MarR via a putative signal transduction pathway involving the periplasmic binding protein MppA, is proposed to transform MarR into a non-DNA-binding conformation, thereby permitting marRAB transcription to proceed. Hence, MarA protein is produced which can then bind as a monomer to the marO marbox upstream of the marRAB promoter, where it activates (+) transcription of marRAB and enhances the production of MarA. The ensuing highly elevated intracellular levels of MarA can then bind to marboxes adjacent to the promoters of mar regulon genes, such as acrAB and tolC, and activate their transcription. The MarA homologs SoxS and Rob can also bind to the marO marbox and activate marRAB transcription. The positive regulation by all three proteins on marRAB is enhanced by FIS (yellow), an accessory activating protein. SoxS is only produced upon conversion of the SoxR effector protein (magenta) into its active form (SoxR*) by superoxide-generating agents (O2-). Rob, like SoxS, in addition to mediating increases in MarA synthesis, can also directly activate the expression of some genes that belong to the mar regulon, such as acrAB. The regulatory-protein-binding sites within marO are shown in finer detail in the bottom left corner. The positions of the -35 and -10 hexamers of the marRAB promoter, the ribosome-binding site (RBS) and transcription start points (tsps) for the marRAB operon are indicated. See text for other details. (Modified with permission from reference 49.)

 
AcrR possesses a HTH DNA-binding domain that places it in the TetR family of repressors, together with the multidrug regulators MtrR from N. gonorrhoeae, QacR from S. aureus, and AcrS, a protein which is likely to control the expression of AcrEF (109, 156), an additional MDR transporter from E. coli that is homologous to AcrAB (Table 1). Although AcrR represses both its own and acrAB transcription, it is not involved in the induction of acrAB and acrR expression in response to general stress conditions, such as 4% ethanol, 0.5 M NaCl, and entry into stationary growth phase; instead, these increases in transcription were attributed to an unidentified regulatory protein (107). Thus, it appears that the primary function of AcrR is to modulate acrAB expression, thereby preventing excessive production of the AcrAB pump, whereas MarA and related global regulators are primarily responsible for the actual induction of acrAB and tolC (Fig. 2; described in detail below).

A recent study has demonstrated that AcrAB is also positively regulated by SdiA, a protein which regulates cell division genes in a manner dependent upon quorum sensing (174). The link between AcrAB and quorum sensing, combined with the observation that some signal molecules utilized by bacteria for quorum sensing are similar to known AcrAB substrates, viz., the fluoroquinolones, led to the suggestion that AcrAB may have a physiological role in the export of non-freely diffusible quorum-sensing signals (174). Therefore, increases in AcrAB expression as growth rates slow (107, 176) may be related to increased levels of the quorum-sensing signals produced by E. coli.

MarA Global Activator

Although not involved in the response to the general stress conditions described above, transcriptional activation of acrAB expression is the predominant cause of multidrug resistance in strains that overexpress MarA or the closely related global regulators SoxS and Rob (7). The mar (multiple antibiotic resistance) regulatory locus (Fig. 2) consists of the marRAB operon and the divergently transcribed marC, which encodes a protein of unknown function, as does marB (7). The intracellular levels of the MarA global activator are controlled by the product of the first gene of the marRAB operon, MarR. Both of these proteins bind to marO (7), a region of DNA that separates the two transcriptional units and contains a large number of regulatory protein binding sites, within and around the marRAB promoter (Fig. 2).

MarA, a member of the AraC family of transcriptional activators (Table 1), activates its own transcription and that of a large number of mar regulon genes by binding to 20-bp DNA sequences known as marboxes that are located in the vicinity of the promoters for the target genes. For example, the MarA-binding site within the marO regulatory region is 16 bp upstream of the -35 region of the marRAB promoter (Fig. 2) (118, 177). Importantly, the acrAB promoter is also adjacent to a marbox at which MarA has been demonstrated to bind and activate transcription (7). In addition, overexpression of MarA or its homologs SoxS and Rob has also been demonstrated to result in increased synthesis of the TolC component of the AcrAB-TolC pump complex, which, in combination with the identification of a putative mar/rob/sox-box upstream of the tolC gene, strongly suggests that tolC also belongs to the mar regulon (11).

The transcriptional activation functions of MarA have also been proven to be global in nature by the demonstration that MarA can promote the transcription of genes encoding proteins of diverse functions, both in vivo and in vitro (7, 68). Gene array analysis of a strain constitutively expressing MarA has indicated that more than 60 E. coli genes are differentially regulated by this protein (15), whereas a second study employing an inducible MarA expression system identified an additional 67 MarA-regulated genes (165). However, although the total number of promoters directly activated by MarA has recently been estimated to be less than 40 (122), an even later report has shown that MarA is capable of activating a gene that possess a marbox which diverges substantially from the consensus sequence (14), suggesting that 40 may be an underestimate of the number of mar regulon promoters. One well-documented example of MarA activity is activation of transcription of micF, which produces an antisense RNA that downregulates the expression of ompF, a gene encoding an outer membrane protein that is a site for drug entry (29). Thus, the reduction in the rate of drug influx via OmpF, combined with the increased production of the AcrAB and TolC drug efflux proteins, represents a highly effective mechanism by which MarA can act to coordinate a response to the presence of toxic antimicrobials.

MarA transcriptional activation has several unusual features: it binds DNA as a monomer, and its degenerate 20-bp marbox binding sites are asymmetric, lacking any of the inverted or direct repeats characteristic of bacterial regulatory sequences. Additionally, MarA and the closely related SoxS protein are significantly smaller than other AraC family activators, such as N. gonorrhoeae MtrA, because of the complete lack of a MarA or SoxS ligand-binding domain (119). The crystal structure of MarA in complex with the marbox from the marRAB promoter has been solved (Fig. 3), revealing a protein possessing two separate HTH DNA-binding domains linked by a long {alpha}-helix, another highly unusual arrangement for a prokaryotic regulator (177). Since a typical HTH motif is only capable of recognizing 6 bp, in contrast to the requirement of an operator sequence of at least 11 to 12 bp for a DNA-binding protein to recognize sites that do not occur by chance in a bacterial genome (207), most bacterial regulatory proteins acquire two HTH motifs by forming oligomers. However, for MarA, the presence of two HTH motifs in a single polypeptide chain explains how this protein can function as a monomer. In order to facilitate MarA simultaneously contacting two successive major grooves with the recognition helices from its two HTH motifs, the DNA in the MarA-marbox complex is bent by approximately 35° (Fig. 3) (177). Alanine-scanning mutagenesis has confirmed that the N-terminal HTH of MarA, which contacts the more highly conserved portion of the marbox consensus sequence, contributes the majority of the important MarA-DNA interactions (42).



View larger version (22K):
[in this window]
[in a new window]
 
FIG. 3. Structures of Rob and MarA proteins bound to micF and marRAB marboxes, respectively. The highly homologous DNA-binding domains of Rob and MarA are colored orange, with their respective HTH recognition helices in brown and the additional C-terminal putative ligand-binding domain of Rob in blue. The N-terminal Rob HTH is inserted into the major groove, whereas the C-terminal HTH makes contacts only to the DNA backbone. In contrast, MarA induces a significant bend in the marRAB promoter to facilitate the placement of both HTH motifs in successive major grooves of marRAB marbox DNA. (Reprinted with permission from reference 89; kindly provided by Tom Ellenberger.)

 
The orientations of marboxes and their distances from the -35 hexamers of mar promoters vary, but both are critical, presumably permitting the efficient formation of MarA-RNA polymerase-DNA ternary complexes (117). Mutagenesis has indicated that distinct solvent-exposed MarA residues are involved in the activation of transcription from each class of promoter, suggesting that MarA employs different mechanisms to promote RNA polymerase activity, depending on the orientation and location of each marbox (42). Utilization of nuclear magnetic resonance techniques to investigate the interaction of MarA with degenerate marboxes in solution revealed that portions of the DNA-bound form of MarA exist in a highly dynamic state (27). This led to the proposal that MarA possesses an inherent flexibility that grants it the ability to undergo significant rearrangements in order to accommodate variations in the DNA sequences of marboxes (27).

More recent evidence suggests that the primary mode of transcriptional activation by MarA first involves the formation of a complex between the activator and RNA polymerase, which can then scan the chromosome for mar regulon promoters more efficiently than RNA polymerase or MarA alone (116). The ability of a MarA-RNA polymerase complex to selectively identify marboxes that are adjacent to promoter sequences was suggested to be the mechanism by which MarA can distinguish real marboxes from the many marbox-like sequences present in the E. coli chromosome (116). A similar mechanism has also been proposed for the activity of the MarA homolog SoxS (see below) (46).

MarR, Antimicrobial-Sensing Repressor of marRAB

The MarR repressor, which is the product of the first gene in the marRAB operon, controls the intracellular levels of MarA and hence plays a crucial role in the MarA-mediated activation of mar regulon promoters (Fig. 2). MarR, like MarA, also binds within marO but at sequences distinct from the marbox, although a degree of competitive binding at marO between the two proteins does exist (118). MarR represses marRAB transcription by binding as a dimer to two distinct regions in marO, site I and site II, which are located downstream from the MarA binding site (Fig. 2) (120). MarR site I is positioned between the -35 and -10 regions of the marRAB promoter, an ideal binding site to prevent access by RNA polymerase, whereas site II does not appear to be required for repression or binding by MarR at site I (Fig. 2) (120).

A crystal structure obtained for MarR in the presence of the compound salicylate indicated that MarR contains a DNA-binding domain belonging to the winged-helix family (8), as was found for another MarR family member, the P. aeruginosa MexR protein (101). The MarR {alpha}3 and {alpha}4 helices constitute the HTH motif, whereas ß-sheets contribute to the formation of the "wings." The location of the predicted {alpha}4 recognition helix, which is likely to make contacts to the DNA major groove, is in good agreement with mutagenesis studies that had previously assigned this region a role in DNA binding (5). In contrast to MexR, for which the proposed induction mechanism was suggested to be facilitated by the inherently high degree of flexibility observed for this protein (101), MarR appears to be a much more rigid protein whose structure is stabilized by a number of salt bridges (8), indicating that these two family members are likely to function by distinct methods.

Overall, MarA activates expression of the mar regulon, including acrAB, tolC, and marRAB, whereas MarR acts to downregulate this response by repressing the synthesis of MarA. Although overexpression of MarA from a plasmid is sufficient to activate the mar regulon genes (40), the addition of the antibiotics tetracycline and chloramphenicol (50), weak aromatic acids, such as salicylate, and a structurally diverse range of other compounds, such as the uncoupling agent carbonyl cyanide m-chlorophenylhydrazone and the redox-cycling compounds menadione and plumbagin (6, 200), have all been shown to cause induction of mar regulon expression. Two independent mechanisms for mar regulon induction have been identified, each of which has been proposed to act by suspending the repressor abilities of MarR. First, salicylate, plumbagin, 2,4-dinitrophenol, and menadione have been demonstrated both to induce the marRAB operon in vivo and also to interfere with the binding of MarR to marO DNA in vitro (6). This suggests that MarR can directly bind a broad range of ligands, the outcome of which is its dissociation from marO, MarA synthesis, and, hence, mar regulon activation (Fig. 2).

Interestingly, in order to produce the aforementioned crystal structure of MarR, high concentrations of the inducer salicylate were found to be necessary. Salicylate was observed to be bound at two sites on the surface of each MarR subunit, in the vicinity of the proposed {alpha}4 recognition helix (8). Although these observations were highly suggestive of a potential influence on MarR-DNA interactions, it remains to be seen if binding of salicylate at these sites has any effect on protein conformation or is even physiologically relevant (8).

A second mechanism by which the repressor activities of MarR appear to be interrupted involves the binding of the cell wall component murein tripeptide and its transfer into the cytoplasm by MppA, a periplasmic binding protein (97). An mppA null mutant exhibits phenotypes consistent with derepression of the mar regulon, a process that requires a functional MarA but also alters the expression levels of other genes which are not part of the mar regulon (97). It has been suggested that a low level of murein tripeptide in the periplasm is an indicator of stress, sensed by MppA, which activates a signal transduction pathway that ultimately produces phosphorylated, presumably inactive, MarR (97). However, the suggestion that MarR contains a functional aspartyl phosphorylation-dephosphorylation sequence (97) has been cast in some doubt by the MarR crystal structure (196). Hence, MarR is capable of sensing the presence of deleterious substances and/or environmental conditions by at least one and possibly two independent mechanisms, both of which would cause marRAB to be derepressed and the ensuing activation of the mar regulon, including acrAB.

MarA Homologs SoxS and Rob

SoxS, the effector of the soxRS global superoxide response (sox) regulon, and Rob, which binds the E. coli chromosomal origin of replication, are MarA homologs which, in addition to activating marRAB transcription by binding to marO, have also been shown to directly activate expression from promoters of genes belonging to the mar and sox regulons in vitro and in vivo (Fig. 2) (69, 70, 129, 215). Thus, it is not surprising that the majority of the residues identified in the MarA DNA-bound crystal structure as being important for forming DNA contacts represent amino acids that are conserved in the SoxS and Rob proteins (177). Elevated levels of SoxS and Rob have also been shown to specifically increase the transcription of acrAB (107, 234), which is the most important feature of the mar multidrug resistance phenotype, since even for strains constitutively expressing marA, soxS, or rob, deletion of acrAB results in hypersensitivity to many antimicrobial agents (129, 150, 215).

For SoxS to have a significant effect on acrAB-tolC transcription in wild-type cells, the expression of the soxS gene must first be activated by superoxide-generating agents via the conversion of SoxR, a divergently transcribed local transcriptional activator, into an active form (Fig. 2) (121, 147). Rob, on the other hand, has only a modest effect on the basal level of expression from the marRAB promoter, despite being naturally expressed at a relatively high level (121) and its ability to activate acrAB expression in a mar deletion strain (215). However, unlike SoxS and MarA, Rob contains an additional C-terminal domain, which the crystal structure of Rob complexed with DNA (Fig. 3) has revealed may be involved in the binding of an uncharacterized effector molecule with the potential to alter Rob activity (89). The Rob C-terminal domain has recently been demonstrated to be required for the apparent binding of the compound dipyridyl in vitro and also for the activation of Rob transcriptional stimulation by dipyridyl in vivo, although it is currently uncertain if this occurs via a direct interaction with Rob (182).

An unexpected finding from the Rob-DNA crystal structure (in this case with micF marbox DNA) was that, whereas the N-terminal HTH of Rob was inserted into the DNA major groove, the C-terminal HTH made contacts only to the DNA backbone (Fig. 3) (89). As a result, the Rob-bound micF marbox DNA was not bent, as had been observed for the MarA-bound mar marbox DNA (Fig. 3), although Rob does appear to bend the promoter DNA of some mar regulon genes in solution (70, 119). Because the mar and micF promoters used in the MarA and Rob crystallization studies represent different marbox promoter classes, it is possible that both of these proteins possess the ability to bind DNA with or without their C-terminal HTH inserted into the major groove. Although the role of the Rob C-terminal HTH is still in doubt (119), the use of alternative DNA-binding modes may provide a means by which these proteins can vary the activation mechanisms they employ to suit the different classes of promoters that they bind.

Yet another protein involved in the regulation of acrAB-tolC is FIS, a nucleoid-associated global regulatory protein that modifies transcriptional activity in response to various growth conditions and can also bind to a site within marO just upstream of the marbox (Fig. 2). FIS is proposed to limit the overall level of negative superhelicity and also stabilize the local DNA architecture of certain promoters (225), which, for the marRAB promoter, provides an additional twofold stimulation to MarA-, SoxS-, and Rob-mediated activation of transcription (Fig. 2) (121).

Regulatory systems akin to mar appear likely to be widespread, with evidence suggesting that many bacterial species contain closely related genes (26, 66, 90, 126). For example, in a clinical isolate of Salmonella enterica, increased quinolone resistance has been attributed to a point mutation that produced a constitutively active SoxR protein, presumably leading to elevated expression of sox/mar regulon genes, such as acrAB and tolC (86). Enterobacter aerogenes, an increasingly important cause of respiratory tract infections, also contains a marRAB operon analogous to that of E. coli which has been implicated in antibiotic resistance (24). Altogether, the activation of acrAB and tolC expression by several global regulators in response to a broad range of stress conditions suggests that the primary physiological function of the AcrAB-TolC complex in E. coli is the export of a wide variety of stress-related toxic compounds encountered by this organism in its normal environment, such as a demonstrated role in the efflux of fatty acids and bile salts (108, 143, 221).

Other Regulators of E. coli MDR Pumps

Neither MarR nor MarA has any effect on the expression of another chromosomally encoded E. coli multidrug pump, EmrAB, which, despite being an MFS transporter, also appears to form a tripartite complex with the TolC outer membrane porin in a manner analogous to the function of RND-type pumps (96). Induction of emrAB is controlled by the local repressor EmrR, the product of the first gene of the emrRAB operon. Disruption of the emrR gene via a frameshift mutation has been demonstrated to be responsible for increased expression of EmrAB and elevated resistance to the antibiotic thiolactomycin (105). EmrR is a member of the MarR family of repressors (Table 1) and has been shown to be capable of repressing the marRAB promoter when overexpressed (211). Carbonyl cyanide m-chlorophenylhydrazone, carbonyl cyanide p-(trifluoromethoxy)phenylhydrazone, and 2,4-dinitrophenol, structurally unrelated substrates of the EmrAB pump which induce emrRAB expression (105), have been confirmed to be directly bound by EmrR, one ligand per repressor dimer (20).

EmrR acts by binding to an imperfect inverted repeat centered around the -10 region of the emrRAB promoter, an interaction that can be disrupted in vitro by addition of the ligands carbonyl cyanide m-chlorophenylhydrazone, 2,4-dinitrophenol, tetrachlorosalicylanilide, and nalidixic acid (238). Thus, like MarR, EmrR functions to sense the presence of toxic compounds, but differs in that EmrR exerts direct control over the expression of the EmrAB pump rather than acting indirectly through a global regulator. Interestingly, EmrR was originally described as MprA, the regulator of the plasmid-borne mcb operon, which encodes microcin B17 (104). The addition of compounds that inactivate EmrR repression of emrRAB transcription conversely resulted in EmrR-mediated repression of the main mcb operon promoter. Additionally, EmrR was required for positive regulation of a second mcb promoter that controls expression of microcin export and immunity genes, the products of which have also been implicated in the export of some fluoroquinolone antibiotics (104).

Two-component signal transduction systems involved in modulating the expression of several E. coli drug transporters have also been identified. Overexpression of the EvgA response regulator from the evgSA two-component system produces an elevated level of resistance to MDR substrates, including benzalkonium, crystal violet, deoxycholate, doxorubicin, erythromycin, rhodamine 6G, and sodium dodecyl sulfate (145). These increases have been attributed primarily to increased expression of the yhiUV MDR transporter (144), although the observed increase in resistance to deoxycholate is due in part to enhancement of the expression of another drug transport operon, emrKY, by EvgA (145). It is intriguing that yhiUV and emrKY encode an RND-type and an MFS transporter, respectively, both of which have been shown to require the TolC outer membrane channel (144).

An additional two-component regulatory system, BaeSR, has recently been demonstrated to activate the transcription of yet another E. coli MDR transporter, an RND-type pump complex encoded by the mdtABC operon, which confers resistance to bile salts, novobiocin, and deoxycholate (12, 137). Both mdtB and mdtC appear to encode RND pumps that interact, perhaps as heteromultimers, with the MdtA membrane fusion protein, whereas a fourth gene in the same operon, mdtD, encodes an MFS-type pump that is not required for drug resistance (12, 137). In addition to these E. coli examples, it is interesting that in several other species, such as Acinetobacter baumannii, Stenotrophomonas maltophilia, and S. aureus (Table 1), the expression of a number of MDR transporters also appears to be under the control of two-component regulatory systems which are designed to respond to specific external environmental stimuli (158). Thus, the two-component-regulated MDR transporters, by extension, are likely to perform specific export functions in response to these unknown signals.

A comparison of the nucleotide sequences upstream of known and hypothetical MDR transporters has suggested that the regulation of E. coli MDR efflux genes is likely to be considerably more complicated than the available experimental evidence indicates. Possible regulatory sequences that have been identified include a marbox and an AcrR-binding site upstream of acrEF, which encodes an AcrAB homolog, and an EmrR-binding site upstream of acrAB (178). Sequences upstream of the marRAB and emrRAB operons were also proposed to be bound by a hypothetical regulator (178). In combination with the available information on the local, global, and two-component regulatory systems known to control E. coli drug transport pumps, it would appear that MDR regulatory proteins in this organism are likely to be involved in a complex web of interactions governing the expression of their target genes. Although the individual pumps may have specific physiological roles, E. coli cells may be capable of coordinating the expression of at least some of these MDR transporters in response to multiple threats, making efficient use of their ability to extrude overlapping ranges of substrates.


   RELATED LOCAL AND GLOBAL ACTIVATORS CONTROL MDR GENE EXPRESSION IN B. SUBTILIS
 Top
 Previous
 Next
 References
 
The chromosome of the gram-positive bacterium Bacillus subtilis encodes two highly similar MFS MDR transporters, Bmr and Blt. These proteins show 51% amino acid identity and confer very similar levels of resistance to an identical range of toxic substances when overexpressed (3, 141). They are each regulated by the product of an adjacent gene encoding a transcriptional activator belonging to the MerR family, BmrR (2) and BltR (3), respectively (Table 1). Although these activators have similar N-terminal DNA-binding motifs, their C-terminal domains, which are known to be responsible for inducer binding in this family of proteins (113, 212), show little homology to each other or to any other protein sequences currently in the DNA sequence databases, suggesting that distinct ligands are bound by each protein. However, the ligands for BltR have yet to be identified, which, in combination with the inability of known Blt substrates to induce expression of the blt gene, suggests that the multidrug-transporting abilities of Blt are likely to be entirely fortuitous (3).

Interestingly, the second gene in the blt operon, SpAT (Fig. 4A; formerly known as bltD), encodes an enzyme that acetylates polyamines, such as the natural cellular constituent spermidine. Despite this finding, alterations of polyamine levels had no influence on expression of the blt operon (115). Thus, although spermidine is a known substrate of Blt (237), the suggestion that the efflux of polyamines may be the primary role of Blt remains uncertain.



View larger version (29K):
[in this window]
[in a new window]
 
FIG. 4. (A) Regulation of expression of B. subtilis MDR genes bmr and blt. A BmrR dimer concurrently bound to both bmr promoter DNA and drugs (D) can correctly orient the -10 and -35 hexamers of this promoter to facilitate the binding of RNA polymerase. White arrows indicate the locations of promoters and also the direction in which transcription occurs from these sequences. Because many of the substrates of the Bmr MDR transporter are also ligands of BmrR (red ovals), activation (+) of bmr expression can occur in response to the presence of these deleterious compounds, permitting drug efflux across the cytoplasmic membrane (pale yellow; CM) in exchange for protons (H+) to occur. The global regulatory protein Mta (purple ovals) and the local activator of the blt operon, BltR (green ovals), are likely to act in the same fashion as BmrR, although inducing ligands for these proteins have yet to be identified (?). Mta activates both the bmr and blt genes by binding to the same DNA sequence as the local regulators. The blt operon also encodes SpAT, a polyamine acetyltransferase, whereas Bmr expression can result from transcription initiated at either its own promoter or the promoter of bmrU, an upstream gene of unknown function. (B) The DNA sequence from the region indicated by an asterisk in A, which contains the bmr promoter (Pbmr). The -10 and -35 hexamers of Pbmr and the unusually large 19-bp spacing between these hexanucleotides are indicated, while the large arrows denote the imperfect inverted repeat (labeled -11 to +11) within Pbmr that constitutes the BmrR binding site (2). Bases protected from DNase I digestion by BmrR bound to Pbmr are highlighted in white. (C) DNA contacts made by BmrR to the half-site that encompasses positions -1 to -10 in B. Also shown is the thymine from position +1, which in the BmrR-DNA complex was observed to be no longer base-paired to its partner, whereas the adenine and thymine bases at position -1, although significantly displaced, still formed a distorted base pair. Thin dashed lines indicate hydrogen bonds, and thick broken lines indicate van der Waals interactions between BmrR amino acids and bases (grey boxes) or the phosphates (P) and deoxyribose rings (pentagons) that form the DNA backbone. Note that the sequence depicted in C differs at position -8 from that shown in B because the experimentally determined data best fit a GC base pair at this location rather than the AT that was actually present in the BmrR-DNA complex. (Panel A modified with permission from reference 49; panel C reprinted with permission from reference 245.)

 
BmrR, Local Transcriptional Activator of bmr Expression

In contrast to Blt, Bmr is expressed under normal growth conditions, and disruption of its gene therefore produces cells that are hypersensitive to MDR substrates. In addition to its own promoter, bmr can also be cotranscribed with bmrU, an upstream gene of unknown function (Fig. 4A), which suggests that Bmr and BmrU may have related but currently unknown physiological roles (115). However, more importantly from a drug resistance perspective, the local transcriptional activator, BmrR, can mediate increases in expression from the promoter immediately upstream of bmr after binding some of the synthetic substrates of the Bmr pump (Fig. 4A), such as astrazon orange, diethyl-2,4'-cyanine, rhodamine 6G, and TPP (2, 113, 114, 227). The drug-bound form of BmrR, in addition to possessing the ability to activate expression from the bmr promoter, also exhibits an improved affinity, in comparison to apo-BmrR, for bmr promoter DNA (2, 196, 227). Thus, it appears likely that the Bmr/BmrR system exists at least partially to provide protection against toxic, lipophilic, cationic substances that B. subtilis may encounter in its natural environment (2).

Like other MerR family members, BmrR binds as a dimer to an imperfect inverted repeat located between the -10 and -35 hexamers of a target promoter that exhibits an unusually large spacing of 19 bp (Fig. 4B) (2). This spatial arrangement places the -10 and -35 regions of the bmr promoter on opposite sides of the DNA helix, a conformation that is incompatible with RNA polymerase binding. In the case of blt, a 1-bp deletion that altered its promoter spacing to 18 bp was sufficient to cause a marked increase in expression (3). For promoters regulated by MerR and another family member, SoxR, 2-bp deletions in their spacer regions in each case produced a promoter that exhibited highly elevated transcription levels independently of their respective activator proteins (58, 157). Binding of the MerR activating ligand is known to have an effect similar to that of the 2-bp deletion, as it allows the protein to initiate the partial untwisting of promoter DNA, which suggested that activation by MerR involves the production of a promoter with a more favorable spacing (212).

Structure of Tripartite BmrR-Drug-Activated DNA Complex

The finer details of the mechanism by which MerR family proteins activate expression from promoters under their control have recently been elucidated by determination of the crystal structure for the tripartite complex of BmrR bound concurrently to DNA and a drug (245). BmrR was found to consist of a DNA-binding domain which is linked to the C-terminal two-thirds of the protein responsible for ligand binding by a long helix, {alpha}5 (Fig. 5) (245). The tight packing of the drug-binding domain from each polypeptide against the DNA-binding domain of the second polypeptide in a BmrR dimer, in combination with an antiparallel coiled coil that {alpha}5 forms with {alpha}5' from the other subunit, constitutes an extensive dimerization interface (Fig. 5). The BmrR DNA-binding domain belongs to the winged-helix superfamily and consists of an HTH ({alpha}1, the {alpha}2 recognition helix, and their connecting turn), and two additional wings, W1 and W2 (Fig. 5) (245). The combination of these DNA-binding elements ensures that BmrR makes a large number of contacts with bmr promoter DNA (Fig. 4C), however, at the same time, the promoter remains accessible for RNA polymerase binding, an essential feature for a transcriptional activator protein (245).



View larger version (50K):
[in this window]