The MexAB–OprM drug efflux pump is central to multidrug resistance of Pseudomonas aeruginosa. The ability of the tripartite protein to confer drug resistance on the pathogen is crucially dependent on the presence of all three proteins of the complex. However, the role of each protein in the formation of the intact functional complex is not well understood. One of the key questions relates to the (in)ability of MexB to act independently of its cognitive partners, MexA and OprM. In the present study, we have demonstrated that, in the absence of MexA and OprM, MexB can: (i) recruit AcrA and TolC from Escherichia coli to form a functional drug-efflux complex; (ii) transport the toxic compound ethidium bromide in a Gram-positive organism where the periplasmic space and outer membrane are absent; and (iii) catalyse transmembrane chemical proton gradient (ΔpH)-dependent drug transport when purified and reconstituted into proteoliposomes. Our results represent the first evidence of drug transport by an isolated RND (resistance–nodulation–cell division)-type multidrug transporter, and provide a basis for further studies into the energetics of RND-type transporters and their assembly into multiprotein complexes.
- multidrug resistance
- Pseudomonas aeruginosa
- resistance–nodulation–cell division transporter (RND transporter)
Pseudomonas aeruginosa is a pathogen that easily infects immunocompromised hospitalized patients. It is also the main cause of mortality of patients with cystic fibrosis and is characterized by an innate resistance to a wide spectrum of antibiotics . Multidrug-resistant strains of Gram-negative organisms such as Ps. aeruginosa rely on tripartite protein assemblies that span both the inner and outer membranes to pump cytotoxic compounds from the cell. Ps. aeruginosa expresses several such multidrug-transporter complexes of which MexAB–OprM is the predominant system involved in intrinsic drug resistance. In the tripartite complex, MexB is the IMP (inner membrane protein), OprM is the OMP (outer membrane protein), and MexA is the periplasmic MFP (membrane fusion protein). AcrAB–TolC from Escherichia coli is a homologous system, with MexB and AcrB sharing 70% identity, MexA and AcrA sharing 57% identity, and OprM and TolC sharing 19% identity.
Crystal structures have been reported for MexA [2–4], MexB  and OprM . The crystal structure of MexA reveals that the protein comprises a membrane proximal domain, a β-barrel domain, a lipoyl domain and an α-helical hairpin domain. OprM is a trimeric channel with an outer-membrane-spanning pore-forming β-barrel domain and a periplasmic α-helical domain. MexB is an inner-membrane-spanning protonmotive-force-dependent xenobiotic efflux protein and was found to be a homotrimer using MS  and X-ray crystallography . Each monomer of MexB consists of 12 transmembrane α-helices constituting the inner membrane domain. The connecting loops between TM (transmembrane domain) 1 and TM2 as well as TM7 and TM8 fold into two large periplasmic domains, which form the pore domain and OprM-docking domain. The overall architecture of the proteins from the MexAB–OprM complex closely resembles that of the proteins in the AcrAB–TolC complex from E. coli [4,8–11].
The structure of an assembled tripartite complex has not yet been determined and, despite the available crystal structures of the individual components, the details of assembly remain largely unknown. The interaction of a tight complex between all three pump components in vivo was confirmed as the tripartite MexAB–OprM complex could be purified from overproducing E. coli or Ps. aeruginosa cells without the need for chemical cross-linking . Genetic and biochemical studies suggested further that the interaction between MexA and OprM occurs through the α-helical hairpin of MexA and α-helical barrel of OprM below the equatorial domain [13–16]. Also, MexB most probably interacts with the periplasmic tip of OprM via its top β-hairpin regions . The details of the interaction of MexB with MexA are much less clear, but several studies have implied the importance of the β-barrel domain in MexA for the interaction with MexB [17,18]. Cross-linking data for AcrA with AcrB suggest extensive contact between the lipoyl, β-barrel and membrane proximal domains of AcrA and the periplasmic domain of AcrB .
Most studies indicate that the MFPs and IMPs form highly specific interactions that do not allow interchange between homologous pumps. For instance, MexA and MexB cannot be functionally interchanged with any of the other Mex systems in Ps. aeruginosa [12,19]. MexAB is also strict in its requirement for OprM. Although MexAB can form a partially functional complex with OprJ , it cannot form a functional complex with OprN from Ps. aeruginosa  or TolC from E. coli [16,21,22]. However, recently, functional interaction of MexB with AcrA and TolC has also been reported , raising questions about the exchangeability of tripartite components from different organisms.
Another question that also remains is whether or not RND (resistance–nodulation–cell division) multidrug transporters can function independently of their MFP and OMP. The RND multidrug transporters AcrB and AcrD from E. coli have been purified and reconstituted into lipid vesicles. For both proteins, the transport activity was dependent on the presence of the MFP AcrA [23,24].
In the present study, we have investigated the role of the MFP and OMP in drug efflux by IMPs by (i) expressing MexB in E. coli in the absence of MexA and OprM, (ii) expressing MexB in the Gram-positive organism Lactococcus lactis where the periplasmic space and outer membrane are absent, and (iii) purifying and reconstituting MexB in proteoliposomes and measuring drug transport in the absence of the other protein partners from the tripartite complex.
Cloning of MexB and MexA in E. coli
MexB was cloned in E. coli as described previously  to yield plasmid pMexBH.
The mexA gene was amplified from genomic DNA of Ps. aeruginosa strain PAO1 by PCR by using Pfu Turbo polymerase (Stratagene) and primers 5′-GGAATTCCATATGTCCGGAAAAAGCGAGGCGCCGCC-3′ (forward) and 5′-CGCGGATCCGCGTCAGCCCTTGCTGTCGGTTTTCG-3′ (reverse). The primers were designed to amplify residues 24–383, thereby omitting the periplasmic signal sequence at the N-terminus. In addition, Cys24 was mutated to serine, to prevent attachment of a fatty acid. This PCR produced a DNA fragment of 1080 bp with unique restriction sites at the 5′ end (NdeI) and the 3′ end (BamHI). After restriction enzyme digestion, the fragment was ligated into NdeI/BamHI-restricted pET28b(+) (Novagen). This protocol generated a construct containing a gene for the Δ(1–23) C24S mutant of MexA with an N-terminal His6 tag attached via a thrombin cleavage site and was designated pHMexAΔN. The cloned PCR product was sequenced to ensure that only the intended changes were introduced.
Construction of the D407N MexB mutant
The D407N mutation was introduced in the mexB gene in the E. coli vector pMexBH by PCR using KOD Hot Start DNA polymerase (Novagen) and the forward primer 5′-CTTGCTGGTGAACGACGCCATCGTGGTG-3′ and reverse primer 5′-GATGGCGTCGTTCACCAGCAAGCCGATGGC-3′. The mutant mexB gene was then subcloned into pET41a+ as an NdeI/XhoI fragment to yield pMexBH D407N. The mutated mexB gene was sequenced to ensure that only the intended changes were introduced.
Creation of the E. coli TolC-knockout mutant
The tolC gene on the chromosome of E. coli BW25113 was deleted using the method described by Datsenko and Wanner . The primers used were 5′-TAATTTTACAGTTTGATCGCGCTAAATACTGCTTCACCACGTGTAGGCTGGAGCTGCTTC-3′ and 5′-GTATCTTTACGTTGCCTTACGTTCAGACGGGGCCGAAGCCCATATGAATATCCTCCTTAG-3′ which contained priming sites 1 or 2 from pDK4 with 40 nt before or after the tolC gene respectively. The deleted tolC locus was marked by genes conferring chloramphenicol resistance. The antibiotic resistance gene was removed by the pCP20-encoded recombinase . Positive clones were verified by PCR and drug-susceptibility tests. The BW25113 strain was converted into a (DE3)-lysogen, but all cytotoxicity and drug transport assays employed basal levels of expression without induction.
MexB cloning in L. lactis
MexB was cloned into L. lactis by digesting pMexBH and pMexBH D407N with NdeI and XhoI (NEB). The non-His6-tagged fragments of MexB and D407N MexB that were released were then cloned into the NcoI/XhoI-restricted lactococcal pNZ8048 expression vector .
E. coli BW25113 (DE3) cells with deletions of AcrB, AcrAB or TolC were used to propagate the control (pET41a+) or the MexB-expressing (pMexBH) plasmids. All experiments employed basal levels of expression without induction. Cells were grown to a D660 of 0.2 in LB (Luria–Bertani) medium containing kanamycin (25 μg/ml). Cytotoxic drugs were added to the cell suspensions at increasing concentrations. For the determination of viable cell counts, cells were incubated at 37 °C for 3 h. The bacterial cultures were then serially diluted up to 109-fold and plated on LB agar plates containing kanamycin. Colonies were counted and the survival ratios were expressed as a percentage of the colony-forming units observed at a given concentration of antibiotic over that observed without added antibiotics. For determining the inhibitory effect of drugs on the growth rate of cells, the cell densities were monitored by measuring the D660 of the cultures every 10 min for 6 h in a VersaMax plate reader (Molecular Devices). The relative growth rates were determined, and the IC50 values were calculated. The IC50 is the concentration of drug necessary to reduce the growth rate of cells by 50%. For oxacillin, the final D660 was used instead of the growth rate. Results are means±S.E.M. for three independent experiments, each carried out in duplicate.
Cytotoxicity assays on L. lactis NZ9000 ΔlmrA ΔlmrCD cells were carried out in the same way as for the E. coli cells with the following changes. Cells were grown in M17 medium (Difco) containing 0.5% glucose and chloramphenicol (5 μg/ml) to a D660 of 0.3, and protein expression was induced by the addition (1:2000) of the supernatant of the nisin-A-producing L. lactis strain NZ9700  (containing approx. 5 ng of nisin-A/ml), before ethidium bromide was added at increasing concentrations.
Substrate transport in E. coli cells
LB Broth Miller (Formedium) containing 25 μg/ml kanamycin was inoculated (1:50 dilution) with an overnight culture of E. coli BW25113 (DE3) cells with deletions of AcrB, AcrAB or TolC, propagating various constructs. Cells were incubated with shaking at 37 °C until a D660 of 0.5 was reached. The cells were harvested by centrifugation at 3000 g for 10 min at 4 °C, then washed twice by resuspension in 50 mM potassium phosphate buffer (pH 7.0) containing 5 mM MgSO4 and sedimented by centrifugation at 3000 g for 10 min at 4 °C. The cells were then resuspended in the same buffer to a D660 of 0.5 and incubated for 3 min at room temperature (25 °C) in the presence of 25 mM glucose to energize the cells. The fluorescence measurement was started, and 30 s later 2 μM ethidium bromide, 0.125 μM Hoechst 33342 or 0.25 μM TMA-DPH [1-(4-trimethylammoniumphenyl)-6-phenyl-1,3,5-hexatriene p-toluenesulfonate] was added. The fluorescence was followed as a function of time in a PerkinElmer LS 55B fluorimeter. Excitation and emission wavelengths and excitation and emission slit widths were 500, 580, 5 and 10 nm respectively for ethidium bromide; 355, 457, 10 and 4 nm respectively for Hoechst 33342; and 350, 425, 5 and 5 nm respectively for TMA-DPH.
Cross-linking in E. coli cells
E. coli BW25113 (DE3) cells with genomic deletions in either AcrB or TolC and propagating pMexBH were grown in LB medium containing kanamycin (25 μg/ml). When the D660 reached ~0.6, protein production was initiated by the addition of IPTG (isopropyl β-D-thiogalactoside) (1 mM) and cells were grown for another 2 h. Cells were harvested and resuspended in 2 ml of cross-linking buffer (50 mM potassium phosphate buffer, pH 7.4, containing 150 mM NaCl) and energized by the addition of 0.5% glucose or de-energized by incubation with the protonophore CCCP (carbonyl cyanide m-chlorophenylhydrazone) (10 μM). DSP [dithiobis(succinimidyl propionate)] (5 mM) was added to the cell suspensions, and the cells were incubated for 30 min before the reaction was quenched by the addition of 50 mM Tris/HCl (pH 7.5). After 15 min at room temperature, inside-out vesicles were prepared and the His6-tagged MexB was purified as described in the ‘Protein purification’ section. Proteins were separated by SDS/PAGE and then transferred on to PVDF membranes. The blot was probed with anti-AcrA antibody and visualized by an HRP (horseradish peroxidase)-conjugated anti-goat antibody.
Ethidium bromide transport in L. lactis cells
L. lactis strain NZ9000 ΔlmrA ΔlmrCD [27,28] was used as a host for pNZ8048-derived plasmids. The cells were grown at 30 °C in sterile M17 medium (Oxoid) supplemented with 0.5% glucose and containing chloramphenicol (5 μg/ml). Overnight cultures of L. lactis NZ9000 ΔlmrA ΔlmrCD harbouring the pNZMexB or control plasmids were inoculated into fresh medium by 50-fold dilution. When a D660 of ~0.5 was reached, protein expression was induced by the addition (1:1000) of the supernatant of the nisin-A-producing L. lactis strain NZ9700  (containing approx. 10 ng of nisin-A/ml). The cells were grown for another 1 h to allow protein expression and were then harvested, washed and assayed for transport as described in the ‘Substrate transport in E. coli cells’ section.
Preparation of inside-out membrane vesicles
M9 medium (25 mM KH2PO4, 40 mM Na2HPO4, 10 mM NaCl, 20 mM NH4Cl, 2 mM MgSO4·7H2O, 0.2 mM CaCl2 and 0.5% glucose) containing kanamycin (25 μg/ml) was inoculated with an overnight culture of BW25113 ΔAcrAB propagating pMexBH or pMexBH D407N cells. Cells were grown at 30 °C until a D660 of 0.7–0.8 was reached when gene transcription was induced by the addition of 1 mM IPTG. The cells were harvested 2 h later by centrifugation at 13000 g for 15 min at 4 °C and washed by resuspension in 100 mM potassium phosphate buffer (pH 7.0) followed by centrifugation at 13000 g for 15 min at 4 °C. The cell pellet was resuspended in 100 mM potassium phosphate buffer (pH 7.0), and DNase (10 μg/ml) was added. The cells were lysed by three passages through a Basic Z 0.75 kW benchtop cell disruptor (Constant Systems) at 0.138 MPa. The lysate was subjected to low-speed centrifugation at 13000 g for 10 min at 4 °C to remove cell debris, and the resulting supernatant was subjected to high-speed centrifugation at 40000 rev./min for 40 min at 4 °C using a 50.2 Ti rotor to collect the inside-out membrane vesicles. The pellet, containing the inside-out membrane vesicles was resuspended in 50 mM potassium phosphate buffer (pH 7.0) containing 10% (v/v) glycerol to a protein concentration of approx. 50 mg/ml and stored in liquid nitrogen. The protein concentration of the inside-out membrane vesicles was determined by the DC Protein Assay (Bio-Rad Laboratories) with BSA as a standard.
MexB was purified essentially as described by Mokhonov et al. . Inside-out membrane vesicles containing MexB were solubilized by shaking at 4 °C for 1 h in a volume of solubilization buffer [10 mM Bis-Tris, pH 7.4, 20% glycerol, 500 mM NaCl, 2% (v/v) DDM (β-D-dodecyl maltoside) and 10 mM imidazole] sufficient to produce a final protein concentration of 5 mg/ml. Unsolubilized protein was removed by high-speed centrifugation at 50000 rev./min for 30 min at 4 °C using a TLA 110 rotor. Ni-NTA (Ni2+-nitrilotriacetate) resin (Sigma) was equilibrated by washing three times with 5 resin volumes of deionized water, each time sedimenting the resin by centrifugation at 1000 g for 3 min at 4 °C, and then twice with 5 resin volumes of buffer A (10 mM Bis-Tris, pH 7.4, 10% glycerol, 500 mM NaCl, 0.1% DDM and 10 mM imidazole), each time sedimenting by centrifugation as before. Solubilized protein was added to the equilibrated resin and allowed to bind at 4 °C for at least 1 h. Then the resin was transferred to a column (Bio-Rad Laboratories), and the unbound protein was allowed to drain through. The resin in the column was washed with a total of 30 resin volumes of buffer A and then with a total of 30 resin volumes of buffer B (10 mM Bis-Tris, pH 6.0, 10% glycerol, 500 mM NaCl, 0.1% DDM and 10 mM imidazole). His8-tagged protein was eluted in 3 resin volumes of elution buffer (10 mM Bis-Tris, pH 7.4, 10% glycerol, 200 mM NaCl, 0.1% DDM and 300 mM imidazole), of which the first 0.5 resin volume was discarded. Total purified protein was assayed using the colorimetric DC Protein Assay with BSA as a standard.
For MexA purification, E. coli BL21(DE3) cells propagating the MexA-overexpressing pHMexAΔN plasmid were grown in LB medium at 37 °C. When the D660 reached ~0.7, protein production was initiated by the addition of 1 mM IPTG. Cells were harvested and prepared as described in the ‘Preparation of inside-out membrane vesicles’ section. The supernatant after the high-speed spin was directly applied to Ni-NTA resin, and MexA was purified as described for MexB, but without any DDM in the buffers.
Hoechst 33342 binding to purified MexB
Hoechst 33342 binding to purified protein was carried out in 2 ml reaction mixtures containing 25 μg of purified protein in 10 mM Bis-Tris (pH 7.4). Hoechst 33342 was added to the solution in a stepwise fashion to a final concentration of 1.5 μM, when no major changes in fluorescence were detected. Measurements were performed in an LS-55B luminescence spectrometer at excitation and emission wavelengths of 355 and 457 nm respectively, and slit widths of 10 and 5 nm respectively. The binding constants Bmax and Kd were determined using the relationship B=Bmax[S]/(Kd+[S]), in which drug binding is represented by B, the drug concentration by [S], the maximal binding by Bmax and the drug concentration yielding 1/2Bmax by the dissociation constant Kd.
Reconstitution of purified MexB in proteoliposomes
For the preparation of liposomes, commercially available E. coli total lipid extract (Avanti Polar Lipids) was purified further by extraction with acetone/ether to yield E. coli polar lipids. E. coli total lipid extract (100 mg) was dissolved in 2 ml of chloroform and then slowly dripped into 10 ml of ice-cold nitrogen-flushed acetone containing 2 μl of 2-mercaptoethanol. The mixture was stirred at 4 °C overnight and centrifuged in glass tubes at 3000 g for 15 min. The pellet was dried under nitrogen gas and dissolved in 10 ml of nitrogen-flushed diethyl ether containing 2 μl of 2-mercaptoethanol for 10 min at room temperature. Following centrifugation at 1600 g for 10 min, the supernatant was evaporated to dryness in a rotary evaporator. The residue was weighed and dissolved in chloroform, and egg PC (phosphatidylcholine) (Avanti Polar Lipids) was added to the polar lipids at a PC/E. coli lipid ratio of 1:3 (w/w). Once mixed, the lipids were dried as a thin film under nitrogen gas and stored at −20 °C until required. For the reconstitution of purified MexB into Triton X-100-destabilized liposomes, the lipids were hydrated at a concentration of 20 mg/ml in 20 mM potassium phosphate buffer (pH 7.0) containing 100 mM potassium acetate and 1 mg/ml sonicated calf thymus DNA. The lipids were subjected to two freeze–thaw cycles and extruded 11 times through a 400 nm polycarbonate filter using a 1 ml LiposoFast-Basic extruder (Avestin). The resulting liposomes were diluted to a concentration of 4 mg/ml in 20 mM potassium phosphate buffer (pH 7.0) containing 100 mM potassium acetate and destabilized by the addition of 2–3 mM Triton X-100 at 0.25 mM increments until the D540 of the liposome suspension reached a maximum . Purified MexB in elution buffer was then added to a protein to lipid ratio of 1:100 (w/w). After incubation for 30 min at room temperature, 80 mg of hydrated polystyrene Bio-Beads (Bio-Rad Laboratories) were added per ml of liposome suspension to remove the detergent [34,35]. Following incubation for 2 h at 4 °C, these beads were replaced twice by fresh Bio-Beads (80 mg/ml), which were incubated at 4 °C for 2 h and overnight respectively. Before use, Bio-Beads were hydrated by one wash in methanol, followed by one wash in ethanol and five washes with ultrapure water. The proteoliposomes were collected by centrifugation at 55000 rev./min for 30 min using a TLA 110 rotor and resuspended to 1 mg of membrane protein/ml in 20 mM potassium phosphate buffer (pH 7.0) containing 100 mM sodium acetate and 2 mM MgSO4. Proteolipsomes were immediately used in transport assays.
Hoechst 33342 transport in proteoliposomes
To impose diffusion gradients , proteoliposomes in 20 mM potassium phosphate buffer (pH 7), containing 2 mM MgSO4, 100 mM potassium acetate and 10 nmol of valinomycin [to prevent the formation of a reverse ΔΨ (membrane potential)] per mg of protein were diluted 1:100 (to 10 μg of protein per ml) into the same buffer (no gradient) or into 20 mM potassium phosphate buffer (pH 7) containing 2 mM MgSO4 and 50 mM K2SO4 to generate a ΔpH (interior alkaline). After 30 s, 0.125 μM Hoechst 33342 was added and the fluorescence of the Hoechst–DNA complex formed inside the proteoliposomes was followed over time in an LS-55B luminescence spectrometer at excitation and emission wavelengths of 355 and 457 nm respectively, and slit widths of 10 and 5 nm respectively.
MexB can functionally complement AcrB in E. coli cells lacking AcrB
In order to gain an insight into the activity of MexB when not associated with its cognitive partners, MexA and OprM, plasmid pMexBH encoding His8-tagged MexB was expressed in E. coli strain BW25113 ΔAcrB, which lacked the MexB homologue, AcrB. Surprisingly, expression of MexB alone confers resistance on the cells to the toxic compound ethidium bromide. The IC50 of the survival ratio increased from 19±3 μM for the non-expressing control cells to 67±11 μM for MexB-expressing cells (Figure 1A). The resistance conferred by MexB on its own was also compared with that of the tripartite MexAB–OprM pump. When all three components of the MexAB–OprM pump were expressed, the IC50 of the survival ratio increased to 395±35 μM (Figure 1A). Similarly, when the effect of increasing the concentration of ethidium bromide on the maximum specific growth rate was studied, MexB-expressing cells exhibited a significant resistance to ethidium bromide with IC50 values of 38±4, 110±14 and 335±21 μM for non-expressing, MexB expressing and MexAB–OprM-expressing cells respectively (Figure 1B and Table 1). The observed resistance could be explained if MexB was able to interact with the periplasmic-binding protein AcrA and outer membrane pore TolC from E. coli and so complements the missing AcrB. The role of AcrA and TolC in MexB function was investigated further by expressing MexB in E. coli BW25113 ΔAcrAB which lacked both AcrA and AcrB and E. coli BW25113 ΔTolC, which lacked TolC. Deletion of TolC from the genomic DNA of strain BW25113 resulted in an increased drug-sensitive phenotype; cytotoxicity assays had to be carried out at lower drug concentrations than that used for strains with the AcrB or AcrAB deletion. Cytotoxicity assays revealed that MexB in E. coli could confer resistance on cells to a range of compounds such as the antibiotics oxacillin and novobiocin, the toxic compounds ethidium bromide and TPP (tetraphenylphosphonium) and the detergent SDS (Table 1). The observed resistance was dependent on the presence of both AcrA and TolC, as no increase in resistance was observed when MexB was expressed in cells which lacked either AcrA or TolC. The expression of MexB was equal in all of the BW25113 (DE3) E. coli cells used (results not shown). As a negative control, MexB with the inactivating D407N mutation  was also expressed in E. coli strain BW25113 ΔAcrB. No resistance was detected for the D407N mutant of MexB, which confirmed that the observed resistance was originating from MexB, as this mutant was expressed equally to wild-type MexB (results not shown). The cells expressing only MexB displayed the same substrate specificity as the tripartite MexAB–OprM pump. However, the IC50 values for MexB-expressing cells were severalfold lower than that of the MexAB–OprM-expressing cells for all drugs tested (Table 1). The resistance to ethidium bromide conferred by the tripartite MexAB–OprM in E. coli is similar to that of the native AcrAB–TolC complex. However, differences were observed in the conferred resistance to other compounds, with the MexAB–OprM complex being able to confer a higher level of resistance to novobiocin than the native AcrAB–TolC, but a lower level of resistance to SDS and TPP (Table 1).
MexB displays AcrA- and TolC-dependent efflux of fluorescent drugs from E. coli cells
To test whether drug extrusion from the cell is the underlying mechanism of drug resistance in E. coli ΔAcrB expressing MexB, we measured the extrusion of fluorescent substrates from the E. coli strains BW25113 ΔAcrB, BW25113 ΔAcrAB and BW25113 ΔTolC. Hoechst 33342 is a toxic compound that is virtually non-fluorescent in aqueous solution, but has a high fluorescent quantum yield when present in the hydrophobic environment of the lipid bilayer or when bound to DNA. Therefore Hoechst 33342 efflux from cells can be measured as a reduction in fluorescence. When cells were pre-energized by the addition of glucose, the uptake of Hoechst 33342 in cells expressing MexB was significantly lower than that observed in the control cells without MexB and was only slightly higher than cells expressing the tripartite MexAB–OprM complex. Initial influx rates of 18.9±0.2, 5.7±0.5 and 3.6±0.3 a.u. (arbitrary units)/min were obtained for the non-expressing control cells, the MexB-expressing cells and the MexAB–OprM-expressing cells respectively (Figure 2A). The most efficient reduction in Hoechst 33342 uptake was observed for the AcrAB–TolC complex with an initial influx rate of 0.9±0.2 a.u./min (Figure 2A). No difference in Hoechst 33342 influx was observed between the control and MexB-expressing cells in strains where AcrA or TolC has been deleted (results not shown). In addition, no Hoechst 33342 efflux was observed for the D407N mutant of MexB, again confirming that Hoechst 33342 efflux was indeed dependent on the presence of MexB (results not shown). These results demonstrate that MexB can partner with AcrA and TolC to form a functional efflux pump. The fluorescence of ethidium bromide and TMA-DPH, which is enhanced in hydrophobic environments, was also used to assay their transport from the cells. Similar to the results obtained for Hoechst 33342 transport and in accordance with the cytotoxicity assays for ethidium bromide, MexB was able to efflux both ethidium bromide (Figure 2B) and TMA-DPH from cells (Figure 2C). In both cases, the passive influx was reduced (higher efflux) in the MexAB–OprM-expressing cells compared with the cells which only express MexB (Figures 2B and 2C). No significant difference was observed between the transport of ethidium bromide and TMA-DPH by MexAB–OprM or the native AcrAB–TolC complex (Figures 2B and 2C).
To investigate whether a physical interaction between MexA and MexB occurs, His8-tagged MexB was purified using Ni-NTA-affinity chromatography from cells lacking AcrB or lacking TolC as indicated in Figure 2(D). Purified fractions were subjected to SDS/PAGE and blotted on to a PVDF membrane. The blot was developed with anti-AcrA antibody. A high-affinity complex forms between AcrA and MexB as AcrA co-purified with MexB even in the absence of cross-linker. In the presence of the cross-linker DSP, AcrA is observed as a higher oligomer complex. Formation of the AcrA–MexB complex is not dependent on the cells being energized, as the complex was observed in the absence of glucose and the presence of the protonophore CCCP. Formation of the AcrA–MexB complex was dependent on the presence of TolC, as no complex was observed in the cells lacking TolC (Figure 2D). AcrA did not bind non-specifically to the Ni-NTA-affinity resin as no AcrA was observed when the purification was carried out with cells propagating the non-expressing control plasmid (Figure 2E). A homogenous preparation of MexB is obtained from these preparations as indicated by the Coomassie Brilliant Blue-stained gel (Figure 2F).
Independent drug transport by MexB in a Gram-positive bacterium
The surprising promiscuity of MexB that apparently allows it to form a functional complex with the E. coli proteins AcrA and TolC raises questions as to the exact role of the periplasmic-binding protein and the outer membrane pore in MexB functionality. Are binding of these proteins to MexB necessary for the formation of functional MexB or is complex formation just a necessity to move drugs across the periplasm and outer membrane? We have addressed this question by expressing MexB in the Gram-positive organism L. lactis. In this way, there is no periplasmic space or outer membrane to traverse in order to transport drugs out of the cell. MexB and D407N MexB were expressed in an L. lactis strain in which genomic multidrug transporter genes lmrA and lmrCD were deleted . Both MexB proteins were expressed equally well in L. lactis cells (Figure 3A). Surprisingly, reduced ethidium bromide uptake was observed for the MexB-expressing cells in comparison with the D407N MexB-expressing and non-expressing control cells with initial influx rates of 19.8±0.9, 37.8±0.9 and 40.1±2.1 a.u./min for MexB-expressing cells, D407N MexB-expressing cells and control cells respectively (Figure 3B). Ethidium bromide transport by MexB in L. lactis was also compared with that of the LmrP and LmrCD drug-efflux proteins from L. lactis. LmrP transports ethidium bromide very efficiently as hardly any ethidium bromide is entering the cells (Figure 3B). The initial influx rate of ethidium bromide into the cells is higher for the LmrCD-expressing cells compared with the MexB-expressing cells, but then ethidium bromide accumulates at a lower steady-state level than in the MexB-expressing cells (Figure 3B). However, the higher expression level of LmrCD compared with that of MexB could also be responsible for the lower level of ethidium bromide accumulation in the LmrCD-expressing cells (Figure 3A). The expression of MexB also confers resistance to ethidium bromide on the L. lactis cells. In correlation with the transport assays, expression of LmrP conferred the highest level of resistance followed by LmrCD and then MexB (Table 2).
Independent drug transport by purified and reconstituted MexB
To confirm further the ability of MexB to interact with transported ligands independently of other tripartite components, we studied the properties of purified MexB in detergent solution and when functionally reconstituted into proteoliposomes. First, MexB and D407N MexB were purified, and the ability of the D407N MexB transport-negative mutant to interact with Hoechst 33342 was assessed in a drug-binding assay that utilized the unique fluorescent properties of the dye. As Hoechst 33342 is not fluorescent in aqueous solution, the binding of Hoechst 33342 to the hydrophobic binding pocket can be observed as an increase in fluorescence. Equimolar amounts of MexB and D407N MexB were titrated with increasing concentrations of Hoechst 33342. The Hoechst 33342 fluorescence was corrected for non-specific binding, and the results were plotted against the Hoechst 33342 concentration. The data were fitted with a one binding site hyperbola and yielded r2 values of 0.979 and 0.949 for MexB and D407N MexB respectively. No significant differences were obtained between the binding parameters of MexB and D407N MexB (Figure 4A and Table 3), indicating that D407N MexB retains the ability to bind Hoechst 33342 similar to the wild-type protein, even though transport activity is abolished. As another control, drug binding to purified MexA was measured. No drug binding to MexA could be observed, as the Hoechst 33342 fluorescence was equal to that of the buffer alone (Figure 4A).
MexB and D407N MexB were reconstituted into proteoliposomes and the Hoechst 33342 transport activity was measured. The (proteo)liposomes were loaded with DNA and potassium acetate inside, and the reconstitution method used should yield proteins that are oriented in a right-side-out fashion . Therefore, when the proteoliposomes are diluted into a buffer that does not contain potassium acetate, to allow the outward diffusion of acetic acid and the subsequent generation of a ΔpH (interior alkaline), an in vitro model for intact cells is created containing DNA and a basic internal pH. The generation of a ΔpH activates MexB which is a drug/H+ antiporter and hence would efflux Hoechst 33342 when a basic inside pH is generated. The D407N MexB mutant was purified and reconstituted into proteoliposomes to the same level as the wild-type protein (Figure 4B). Imposition of the ΔpH in the proteoliposomes reduced the passive diffusion of Hoechst 33342 in the MexB-containing proteoliposomes, but not in the empty liposomes or the proteoliposomes containing the transport-negative mutant D407N MexB. Initial influx rates for Hoechst 33342 of 6.64±0.91, 1.74±0.32 and 7.24±0.78 a.u./min were observed for the control liposomes and the proteoliposomes containing MexB or D407N MexB respectively (Figure 4C). Addition of purified MexA to the MexB-containing proteoliposomes had no effect on the Hoechst 33342 transport. In the absence of the ΔpH, no Hoechst 33342 transport was observed in the MexB-containing proteoliposomes (Figure 4D). These results indicate that Hoechst 33342 transport is dependent on MexB and that MexB can mediate Hoechst 33342 efflux by itself in a ΔpH-dependent manner.
Drug-resistant Gram-negative bacteria need to transport drugs over two membranes and the periplasmic space. This is normally affected by tripartite assemblies of an IMP, an OMP and an MFP, such as the MexAB–OprM protein complex from Ps. aeruginosa. Previous studies indicated that MexB could only form a functional complex with MexA and OprM . However, we have found that MexB is promiscuous and can partner with AcrA and TolC from E. coli to form an active complex. MexB–AcrA–TolC displays the same substrate-recognition profile as the MexAB–OprM complex, but is less efficient in conferring resistance to the toxic substances (Figure 1 and Table 1). Our results correspond well with that of another study that also found that MexB expressed in E. coli cells can confer resistance to some drugs . Our complementary study now demonstrates MexB–AcrA–TolC-mediated efflux of the fluorescent compounds Hoechst 33342, ethidium bromide and TMA-DPH. Consistent with the resistance data, MexB–AcrA–TolC was less efficient than the MexAB–OprM or native AcrAB–TolC complexes in transporting these compounds (Figure 2). Even though MexB–AcrA–TolC and the MexAB–OprM complex can efflux Hoechst 33342 from cells at 0.125 μM, these complexes do not confer resistance on the E. coli cells to Hoechst 33342. This is probably due to the inability of MexB in E. coli to transport Hoechst 33342 at the concentrations at which it becomes toxic to the BW25113 ΔAcrB cells (2 μM).
OprM and TolC share only a modest sequence homology (approx. 20% identity, compared with 70% for MexB and AcrB and 58% for MexA and AcrA). It is therefore not surprising that TolC cannot support MexB activity through functional complementation of OprM in the presence of MexA [16,22]. Models of the AcrAB–TolC complex predict a sparse interaction between the β-hairpins in the AcrB crown and the periplasmic end of TolC [4,33]. Hence a combination of MexB and TolC is tolerated as long as the third protein in the complex is AcrA. In this sense, MexAB is different from MexXY and MexCD, both of which can act together with TolC [21,34]. The question remains whether an intact tripartite assembly is needed for function of the individual IMP proteins or whether this is only a requisite for bridging the two membranes of the Gram-negative organisms. To answer this question, we have first expressed MexB in a Gram-positive bacterium that lacks the outer membrane. Surprisingly, energy-dependent drug efflux was still observed for the L. lactis ΔLmrA ΔLmrCD cells expressing MexB (Figure 3) and MexB conferred ethidium bromide resistance on the cells (Table 2). As RND proteins and MFP analogues are found in Gram-positive organisms [35–37], it cannot be ruled out that MexB activity in L. lactis is dependent on an endogenous MFP. In view of the absence of OMPs in L. lactis, the observations of MexB-dependent drug transport in this organism provide evidence that MexB can work alone.
To avoid the possibility of MexB interactions with a hitherto unidentified MFP in L. lactis, we have taken advantage of the E. coli expression system to purify sufficient amounts of protein at a very high purity to functionally reconstitute MexB in proteoliposomes. Transport of Hoechst 33342 was observed in the presence of a ΔpH (inside alkaline), but not in the absence of the imposed chemical proton gradient. Transport was also not observed for D407N MexB, in which the aspartate residue that is predicted to be involved in H+ coupling was mutated to neutral asparagine. The addition of MexA had no effect on MexB-mediated Hoechst 33342 transport in the proteoliposomes (Figure 4). This is the first time that drug transport is reported for a RND transporter reconstituted in proteoliposomes in the absence of MFP and OMPs. Previously, the MexB homologue, AcrB, had been reconstituted in lipid vesicles, but drug transport was not measured directly. Instead, the increase in the fluorescence of NBD lipids was measured as it was removed from donor vesicles to acceptor vesicles and this lipid transport was dependent on the presence of the MFP AcrA . Similarly, AcrD, another RND drug-transport protein from E. coli, has also been reconstituted and the proton flux upon addition of aminoglycosides was measured. This proton flux, as well as gentamicin uptake, was dependent on the presence of AcrA . In neither of these cases was a transport-negative mutant included in the measurements. Our results showing independent ΔpH-driven drug transport by an RND multidrug transporter in proteoliposomes make a useful contribution to our understanding of the transport mechanism of RND transporters. Moreover, it allows new opportunities to study the molecular mechanism of RND-type drug transporters and their associated MFPs.
The ability of MFPs to stimulate drug transport or ATPase activity of integral membrane transporter proteins has led to the speculation that the MFPs are actively involved in the transport process , and drug-binding sites have been identified in some MFPs [39,40]. However, we did not observe Hoechst 33342 binding to MexA, which would refute a direct involvement of this MFP in Hoechst 33342 transport at least. Similar to other MFPs [23,24,41], MexA might increase the drug-binding affinity of MexB, which remains to be investigated. Interestingly, Hoechst 33342 binds the D407N mutant and the wild-type protein with the same affinity. This observation agrees with the notion of Asp407 in MexB and AcrB being involved in proton coupling [32,42] rather than interacting directly with substrates. Yu and co-workers observed a drug-binding site formed by the loop between TM3 and TM4 in wild-type AcrB [43,44] that was collapsed in the D407N mutant, leading to the suggestion that the D407N mutant would be unable to bind drugs. The Hoechst 33342 binding we observed might have resulted solely from binding to a site in the periplasmic domain, where no large structural alterations were observed for the D407N mutant .
To our knowledge, this is the first time that independent drug transport was observed for an RND transporter in the absence of the outer membrane pore or the MFP. It is also the first time that drug transport in proteoliposomes has been verified by showing that no drug transport can be observed for a transport-negative mutant protein, even though this mutant binds the substrate with equal affinity to that of the wild-type protein. The functional reconstitution and independent activity of MexB opens up exciting new possibilities for further mechanistic studies on RND-type transporters and their functional regulation by MFP and OMP partners in a tripartite complex. It will also allow investigations on the mechanism of coupling between drug transport and H+ transport, and determination of drug/H+ stoichiometries.
Henrietta Venter designed research; Alexander Welch, Chidiebere Awah, Shiheng Jeng and Henrietta Venter performed research; Alexander Welch and Henrietta Venter analysed data; Henrik van Veen and Henrietta Venter wrote the paper.
This work was funded by a Royal Society Dorothy Hodgkin grant to H.V. and a Royal Society Research Grant.
We are grateful to Dr Martin Welch (Department of Biochemistry, University of Cambridge, Cambridge, U.K.) for the gift of Ps. aeruginosa strain PAO1 genomic DNA, Professor Martin Pos (Institute of Biochemistry, Goethe-University Frankfurt am Main, Germany) for the gift of E. coli strains BW25113 ΔAcrB and BW25113 ΔAcrAB, Professor Helen Zgurskaya (Department of Chemistry and Biochemistry, University of Oklahoma, Norman, OK, U.S.A.) for the plasmid expressing the MexAB–OprM complex, and Dr Markus Seeger (Institute of Biochemistry, University of Zurich, Zurich, Switzerland) for the gift of BW25113, pDK4 and pCP20, and for stimulating discussions.
Abbreviations: a.u., arbitrary units; CCCP, carbonyl cyanide m-chlorophenylhydrazone; DDM, β-D-dodecyl maltoside; DSP, dithiobis(succinimidyl propionate); IMP, inner membrane protein; IPTG, isopropyl β-D-thiogalactoside; LB, Luria–Bertani; MFP, membrane fusion protein; Ni-NTA, Ni2+-nitrilotriacetate; OMP, outer membrane protein; PC, phosphatidylcholine; RND, resistance–nodulation–cell division; TM, transmembrane domain; TMA-DPH, 1-(4-trimethylammoniumphenyl)-6-phenyl-1,3,5-hexatriene p-toluenesulfonate; TPP, tetraphenylphosphonium
- © The Authors Journal compilation © 2010 Biochemical Society