INTRODUCTION

Top Abstract Introduction Materials & Methods Results & Discussion References

Pseudomonas aeruginosa is an opportunistic human pathogen characterized by an innate resistance to a variety of antimicrobial agents. Although this property had generally been attributed to a highly impermeable outer membrane (31), which limits antibiotic uptake and, thus, access to cellular targets, it is now clear that drug efflux plays a crucial role as well (24, 36). One particular drug efflux system, encoded by the mexAB-oprM operon (11, 35, 36), effluxes a range of antibiotics, including tetracycline, chloramphenicol, quinolones, novobiocin, macrolides, trimethoprim, and apparently, -lactams and -lactamase inhibitors (11, 16, 19, 20, 36). The recently successful reconstitution of the MexAB-OprM system in Escherichia coli supports the involvement of MexAB-OprM in -lactam export (40), although the periplasmic location of the cellular targets of these agents raises questions about the mechanism of -lactam recognition and export since the majority of antibiotics exported by MexAB-OprM act within the cytoplasm. The -lactam specificity of MexAB-OprM is not determined by the outer membrane constituent (41), indicating that OprM is not the primary -lactam recognition and export component and, indeed, OprM alone failed to provide resistance to -lactams or any other agent when reconstituted in E. coli (40). Thus, -lactams are unlikely to exit the periplasm simply via OprM.

Expressed in wild-type cells (37, 41), where it contributes to intrinsic multidrug resistance (19, 36), the mexAB-oprM operon is hyperexpressed in nalB mutants (37), producing elevated levels of resistance to those antibiotics which are substrates for MexAB-OprM (11, 16, 19, 36). Homologous efflux systems encoded by the mexC-mexD-oprJ (34) and mexE-mexF-oprN(17) operons have also been described. Not expressed at detectable levels during growth under normal laboratory conditions, these systems are expressed in nfxB (34) and nfxC (17) multidrug-resistant mutants, respectively. Mutant nfxB strains are resistant to chloramphenicol, tetracycline, quinolones, macrolides, novobiocin, and cephems such as cefepime and cefpirome but display hypersusceptibility to most -lactam antibiotics (13). Mutant nfxC strains demonstrate resistance to chloramphenicol, trimethoprim, quinolones, and carbapenems including imipenem, although the latter arises from the loss of the outer membrane channel-forming protein OprD in these mutants and not from overexpression of MexEF-OprN (9, 17). The tripartite MexAB-OprM efflux pump consists of an inner membrane component (MexB), which exhibits homology to a resistance-nodulation-division (RND) family H⁺ antiporter (30, 38); an outer membrane, proposed channel-forming component (OprM) (24, 32); and a so-called membrane fusion protein predicted to link the membrane-associated efflux components (MexA) (24, 32).

Previous studies have shown that mexA, mexB, and oprM insertion mutations differentially affect drug susceptibility, with oprM mutants being more compromised with respect to drug resistance (19, 48). One explanation for this is that mexA or mexB mutants are not entirely OprM deficient and that OprM may contribute to resistance independent of MexA and MexB. To test this we examined, in greater detail, the influence of mexB and oprM mutations on both drug susceptibility and production of OprM as well as the influence of oprM on drug resistance in the absence of mexAB. We report here that OprM contributes to resistance to a variety of agents in the absence of MexAB, apparently via an energy-dependent efflux mechanism.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results & Discussion References

Bacterial strains, plasmids, and growth conditions. Strains and plasmids used in this study are listed in Table 1. K879 was constructed by disruption of the mexB gene of K372 with an Hg interposon by a previously described protocol (48). Plasmid pQZ05 is a pMMB206 derivative carrying the oprM gene on a 1.6-kb EcoRI-HindIII fragment derived from pKPM-2. Plasmid pQZ06 was constructed by cloning a ca. 5-kb SacI fragment of pRS19 carrying the mexAB genes into the SacI site of pDSK519. Plasmids pRSP43 and pRSP44 were constructed by cloning the 5' upstream region of oprM on a 1.6-kb BamHI fragment from pRSP09 (one BamHI site is present in the oprM gene while the second occurs in the plasmid multicloning site) into the lacZ fusion vector pMP190 in both orientations. All plasmid constructions were carried out in E. coli prior to their introduction into P. aeruginosa. Strains were cultivated in Luria-Bertani (LB) broth (Difco) at 37°C. Plasmid-containing P. aeruginosa was selected on medium containing 16 µg of chloramphenicol/ml (pMMB2106, pQZ05) or 100 µg of kanamycin (pQZ06)/ml following transformation as previously described (3). P. aeruginosa strains carrying pMP190 derivatives were cultured in the presence of 200 µg of chloramphenicol/ml. For propagation of various plasmids in E. coli, antibiotics were included in growth medium at the following concentrations: tetracycline, 10 µg/ml; chloramphenicol, 30 µg/ml; ampicillin, 100 µg/ml; and kanamycin, 50 µg/ml.

Purification of OprM and generation of a polyclonal antiserum. To facilitate purification of OprM, the oprM gene was first cloned into the His-Tag vector pET-21d(+) (Novagen) in-frame with a polyhistidine-coding sequence at the 3' end of the gene, thereby generating an OprM protein with six histidine residues at its C terminus. The oprM gene was amplified from the pT7-7 derivative pKPM-1 where it had been cloned such that the ATG start of the gene was optimally spaced 8 bp downstream of the ribosome binding site (RBS) present upstream of the multiple cloning site of pT7-7 (44). By using PCR and primers oprM-3 (5'-CGACTCACTATAGGGAGACC-3'), which anneals upstream of the RBS in pKPM-1, and oprM-4 (5'-AGTCAAGCTTTCCCGCCCTCTTTTGGCAG-3'), which anneals downstream of oprM, the oprM gene complete with the optimally spaced upstream RBS was amplified with Vent DNA polymerase (NEB). Reaction mixtures (100 µl) contained 20 ng of pKPM-1, 1 µM (each) primer, 200 µM (each) deoxynucleoside triphosphate, 4 mM MgSO₄, 10% (vol/vol) dimethyl sulfoxide, and 1 U of Vent polymerase in 1× reaction buffer. Mixtures were heated at 94°C for 2 min before being subjected to 30 cycles of 94°C for 1 min, 56°C for 1 min, and 72°C for 1.7 min, before finishing with 5 min at 72°C. PCR products (10 µl) were subsequently analyzed using agarose gel (0.8% [wt/vol]) electrophoresis and were purified by using the QiaQuick PCR Purification Kit (Qiagen). Following digestion with XbaI and HindIII, the ca. 1.6-kb oprM PCR product was cloned into XbaI-HindIII-restricted pET-21d(+) (which had been prepared from dam strain GM2163 to enable XbaI digestion) to yield pXZL6. The latter vector was then transformed into E. coli BL21(DE3) carrying the pLysS plasmid (K113). Overnight cultures of pXZL6-carrying K113 in LB medium containing appropriate antibiotics were diluted 1:49 into the same medium (500 ml) and incubated for 4 h, at which time isopropyl--D-thiogalactopyranoside was added (0.2 mM final concentration). Two hours later, cells were harvested by centrifugation (8,000 × g), washed once with 75 ml of Tris-HCl (pH 8.0), and resuspended in 15 ml of Tris-HCl (pH 8.0). Cell envelopes were subsequently prepared as previously described (28) and solubilized in 4 ml of Sarkosyl (N-lauroyl sarkosine, sodium salt) (1.5% [wt/vol] in Tris-HCl [pH 8.0]; 30 min at 23°C). Following centrifugation at 10,000 × g for 30 min, the OprM-containing supernatant was recovered, diluted 1:1 with an equal volume of 20 mM Tris-HCl (pH 8.0)-200 mM NaCl, and loaded onto a 2-ml (bed volume) TALON (Clontech Laboratories, Inc., Palo Alto, Calif.) metal affinity column (40 by 10 mm) equilibrated with 20 mM Tris-HCl (pH 8.0)-100 mM NaCl-0.1% (wt/vol) Sarkosyl (buffer A) as per the manufacturer's instructions. The column was then washed with 15 ml of buffer A, and bound proteins were eluted (300- to 400-µl fractions) with buffer A containing 50 mM imidazole. OprM-containing fractions were pooled and dialysed against 20 mM Tris-HCl (pH 8.0)-100 mM NaCl-0.1% (wt/vol) Sarkosyl, and the purified OprM was then used to raise antibodies in rabbits (L. Mutharia, University of Guelph).

SDS-polyacrylamide gel electrophoresis and Western immunoblotting. Cell envelopes were prepared on overnight cultures of P. aeruginosa grown in LB broth following disruption with a French pressure cell and harvesting of the cell membrane fraction by centrifugation as previously described (29). Cell envelope proteins were electrophoresed on sodium dodecyl sulfate (SDS)-polyacrylamide gels (with 11% [wt/vol] acrylamide in the running gel) as previously described (23). Electrophoretically separated proteins were blotted onto an Immobilon PVDF Transfer membrane (Millipore) at 12 mV constant voltage for 16 h at 4°C by using a previously defined protocol (45). Membranes were processed as previously described (5) with the exception that 10% (wt/vol) skim milk powder (Difco) replaced bovine serum albumin in the initial blocking step and a rabbit anti-OprM (diluted 1/5,000) and a horseradish peroxidase-coupled donkey anti-rabbit immunoglobulin G (Amersham) (diluted 1/10,000) were employed as the primary and secondary antibodies, respectively. Blots were developed by using the Enhanced Chemiluminescence (ECL) system (Amersham) according to the manufacturer's protocol. Prestained molecular weight markers (Bio-Rad) were coelectrophoresed and blotted to permit estimation of the sizes of the proteins visualized by immunoblotting.

Triparental matings. Introduction of plasmids pDSK519, pMMB206, and pMP190 and their derivatives into P. aeruginosa required a triparental mating procedure employing the helper vector pRK2013 (7). Briefly, overnight cultures (100 µl each) of plasmid-containing E. coli DH5, pRK2013-containing E. coli MM294, and P. aeruginosa was pelleted together in a microcentrifuge tube, resuspended in 25 µl of L broth, and spotted onto the center of an L agar plate. Following incubation overnight at 37°C, bacterial growth was resuspended in 1 ml of L broth and appropriate dilutions were plated on L agar containing 500 µg of streptomycin/ml (for PAO6609-derived recipients) or 10 µg of tetracycline/ml (for ML5087- and PAO1-derived recipients) to counterselect the E. coli strains and 16 µg of chloramphenicol/ml (for pMMB206 and its derivatives), 100 µg of kanamycin/ml (for pDSK519 and its derivatives), or 20 (for strain K1032) to 100 (for strain PAO6609) µg of chloramphenicol/ml (for pMP190 and its derivatives). Plasmid DNA was prepared from P. aeruginosa recipients by using the miniprep procedure to confirm successful plasmid transfer.

Assays. MIC determinations were carried out by using the broth dilution technique and an inoculum of 5 × 10⁵ organisms/ml as previously described (19). In some instances drug susceptibility was also assessed by the agar diffusion method. Briefly, bacteria (100 µl of an overnight culture) were added to 3 ml of molten top agar (0.7% [wt/vol] Bactoagar; Difco) and spread over L agar (L broth solidified with 1.5% [wt/vol] Bactoagar). After solidification of the top agar, sterile concentration disks (0.25-inch diameter; Difco) were impregnated with 2 to 10 µl of antibiotic solution and placed on the surface of the agar plates. Following overnight incubation of the plates, the diameters of the zones of bacterial growth inhibition surrounding the filter disks were measured. The relative susceptibilities of different strains to the various antibiotics tested were correlated with the sizes of the zones of inhibition, with increased zone size reflecting increased susceptibility. Subtle differences in susceptibility were also examined by performing growth assays in L broth supplemented with antibiotics, whereby the increase in cell density of bacterial cultures (measured as A₆₀₀) was monitored over time, looking for either differential rates of growth or lack of growth versus growth over the 6 to 8 h of the assay. -Galactosidase assays were carried out on log-phase cells (A₆₀₀ = 1.0) cultivated in LB broth in the presence of chloramphenicol as previously described (26). Accumulation of [³H]tetracycline (NEN/Dupont) was assayed exactly as described previously (19). When indicated, carbonyl cyanide m-chlorophenylhydrazone (CCCP) was added either 5 min before the accumulation assay was initiated (at 300 µM) or 10 min into the assay (at 200 µM). The protein assay has been previously described (22).

	RESULTS AND DISCUSSION

Top Abstract Introduction Materials & Methods Results & Discussion References

OprM is expressed independently of MexAB. Disruption of the mexA, mexB, or oprM gene of P. aeruginosa increases the susceptibility of this organism to many antibiotics, with disruption of oprM often rendering cells more susceptible than mexA or mexB mutations (19, 48). Using mutants carrying Hg insertions in the mexB (strain K879 in this study) or oprM (strain K613 in this study) gene, we reexamined this differential susceptibility and extended the range of antibiotics tested. In addition to confirming previous reports (19, 48) that oprM mutants were more susceptible than, e.g., mexB mutants to quinolones such as ciprofloxacin and norfloxacin and to tetracycline (K613 was twofold more susceptible than K879 as measured by using the broth dilution assay; these results were confirmed by using the agar diffusion assay and growth assays of the mutants with defined concentrations of each antibiotic), we also observed that such mutants were more susceptible to erythromycin as well as to the cephems cefepime and cefpirome (K613 was twofold more susceptible than strain K879, which result was also confirmed by agar diffusion and growth assays). In contrast to published results (48), however, no difference in susceptibility to chloramphenicol was observed between the mexB (K879) and oprM (K613) mutants. The differential effect of a mexB versus an oprM mutation can be explained by the mexB::Hg mutant still expressing oprM, whose product is then capable of contributing to antibiotic resistance independent of MexAB (see below). Indeed, examination of cell envelopes for OprM revealed that the mexB mutant K879 expressed markedly decreased but still detectable levels of the protein (Fig. 1), although substantial overloading of the gel was necessary in order to detect OprM in K879. Intriguingly, disruption of the mexB gene in a previous study actually increased production of OprM over that seen in the parental strain while disruption of mexA had no effect on OprM levels (48). Since Hg possesses transcription and translation stop signals in both orientations (6), insertional inactivation of mexB with Hg was expected to exhibit a polar effect on downstream oprM expression, but only if expression of this gene originated solely with the promoter upstream of mexA. These data suggest, therefore, that a promoter element may exist upstream of oprM (and downstream of the site of Hg insertion in K879) which provides for additional oprM expression independent of the mexAB genes. To test this directly, the 5' upstream region of oprM, from the PstI site in mexB (downstream of the site of Hg insertion in K879, some 825 bp upstream of the oprM start codon) to the BamHI site in oprM (777 bp downstream of the oprM start codon), was cloned into pMP190, upstream of the resident promoterless lacZ gene of this vector, yielding pRSP43. Strains of P. aeruginosa harboring pRSP43 demonstrated -galactosidase activity that was four- to fivefold greater than that observed for cells carrying the pMP190 vector alone or pMP190 with the PstI-BamHI fragment in the opposite orientation (pRSP44) (see Table 3), indicating that a weak oprM promoter resides upstream of oprM, within the coding region of mexB. This promoter is markedly less active than the one present upstream of mexA (>5,000 Miller U for a mexA-lacZ fusion) (37), however, indicating that the bulk of OprM expression in MexAB OprM⁺ cells in a rich medium, at least, is derived from the promoter upstream of mexA. This is consistent with the observation that OprM levels in the mexB mutant strain K879, where oprM expression is uncoupled from the mexA promoter, are markedly lower than the OprM levels seen in the parent strain K372, where expression of oprM is not uncoupled from this promoter (Fig. 1). The failure of Yoneyama et al. to see any decline in OprM levels upon disruption (by insertion of a tet cartridge) of mexA or mexB (48) might be explained by a failure of the tet cartridge to exhibit transcriptional polarity. In any case, examination of the region upstream of oprM failed to reveal any strong ⁷⁰ promoter candidates, although a region with weak homology to the canonical 10/35 sequences of this class of promoter, separated by 17 bp, was identified 47 bp upstream of the oprM coding sequence (TctACgtggcggtcagcacgctgTtcAAg; putative 10/35 regions are in boldface while uppercase letters indicate homology to consensus 35/10 sequences of ⁷⁰ promoters). In spite of the fact that an influence of the cloned oprM gene on resistance is only seen in a mexAB-oprM background (introduction of the gene into wild-type strains fails to alter drug susceptibility [47]), expression from the oprM promoter was the same in a MexAB OprM⁺ or a MexAB OprM strain (Table 2). The potential for expression of oprM independent of mexAB supports a role for this protein independent of MexAB.

View larger version (26K): [in this window] [in a new window]   FIG. 1.   Western immunoblot of cell envelopes of K372 (lane 1), K613 (lane 2), and K879 (lane 3) probed with an antiserum to OprM. All lanes were equally but substantially overloaded (as determined by Coomassie blue staining of a duplicate gel) in order to detect OprM in K879. Arrow indicates position of OprM. Molecular weights (in thousands) are at left.

View this table: [in this window] [in a new window]   TABLE 2.   Expression of an oprM-lacZ fusion in MexAB OprM+ and MexAB OprM P. aeruginosaa

OprM contributes to antibiotic resistance independent of MexAB. The observation that strain K879 expressing oprM but not mexB (and thus lacking a functional MexAB-OprM efflux system) is more resistant to certain antibiotics than the oprM::Hg strain K613 suggests that OprM can function independently of MexB and thus, the MexAB-OprM efflux system, in contributing to intrinsic antibiotic resistance. To assess this directly, a mexAB-oprM strain (K1032) was constructed and the influence of oprM alone on antibiotic resistance was examined. Elimination of mexAB-oprM markedly increased susceptibility of strain K1032 to a variety of antibiotics as expected (Table 3). Introduction of oprM into this strain on plasmid pQZ05 increased resistance to some (but not all) agents to which susceptibility was increased by the elimination of mexAB-oprM, including cefepime, cefpirome, norfloxacin, ciprofloxacin, erythromycin, and tetracycline (Table 3). These were the same agents to which the oprM::Hg mutant strain K613 was more susceptible than the mexB::Hg mutant K879. In contrast to results presented here, Yoneyama et al. (48) reported that OprM production in the absence of MexAB facilitated resistance to chloramphenicol, suggesting that a MexAB-independent chloramphenicol resistance mechanism involving OprM exists in P. aeruginosa. While we cannot explain this discrepancy, it likely reflects differences in the strains being used. Nonetheless, we noted that introduction of the oprM vector pQZ05 into three different mexAB-oprM strains (Table 3; see below) failed to promote any increase in resistance to chloramphenicol, suggesting that any contribution of OprM to resistance to this agent independent of MexAB is not very widespread.

OprM-mediated antibiotic efflux independent of MexAB. Given the involvement of OprM with a known efflux system (MexAB-OprM), it was likely that the contribution of this protein to antibiotic resistance independent of MexAB was also via an efflux mechanism. To assess this, drug (tetracycline) accumulation assays were performed with strain K1032 (MexAB OprM) carrying pMMB206 or pMMB206::oprM (pQZ05). As seen in Fig. 2A, K1032 carrying pMMB206 accumulated substantially higher levels of tetracycline than did the MexAB OprM⁺ parent strain PAO1, consistent with the absence of the MexAB-OprM efflux system in K1032. Tetracycline accumulation was markedly reduced upon the introduction of the oprM vector pQZ05 into K1032 (Fig. 2A). Addition of CCCP after 10 min of the assay increased drug accumulation in PAO6609 and K1032(pQZ05) to between 350 and 400 pmol of tetracycline/mg of protein within 10 min (data not shown), consistent with OprM functioning in an efflux capacity independent of MexAB. Moreover, pretreatment of all strains with CCCP prior to the addition of tetracycline yielded no differences in antibiotic accumulation between any of the strains (approximately 300 pmol/mg of protein after 5 min), indicating all differences seen in Fig. 2 are energy dependent and, thus, efflux related. Still, the observation that pQZ05-carrying K1032 accumulated more antibiotic than the MexAB OprM⁺ parent strain PAO6609 (Fig. 2A) indicates that this mechanism is less effective, in the case of tetracycline at least, than MexAB-OprM-mediated efflux or that there are fewer of the MexAB-independent, OprM-dependent efflux systems. These experiments were repeated with PAO1 and its mexAB-oprM derivative K1119 (with or without pQZ05) and similar results were obtained (Fig. 2B). That the MexAB-independent operation of OprM involves drug efflux strongly supports the involvement of additional efflux components, since outer membrane proteins are unable to function alone in energy-dependent transport processes.

View larger version (14K): [in this window] [in a new window]   FIG. 2.   Accumulation of [3H]tetracycline (tet) by P. aeruginosa PAO6609 () and K1032 (mexAB-oprM) carrying pMMB206 () or pQZ05 (pMMB206::oprM) () (A) and by P. aeruginosa PAO1 () and K1119 (mexAB-oprM) carrying pMMB206 () or pQZ05 (pMMB206::oprM) () (B). Radiolabelled tetracycline (5 µM) was added to log-phase cells and cellular drug accumulations were determined as a function of time. Data shown are representative of two separate experiments in which duplicate samples were taken at each time point.