Portal de Periódicos da CAPES

AcrA, AcrB, and TolC of Escherichia coli Form a Stable Intermembrane Multidrug Efflux Complex

2004; Elsevier BV; Volume: 279; Issue: 31 Linguagem: Inglês

10.1074/jbc.m402230200

1083-351X

Elena B. Tikhonova, Helen I. Zgurskaya,

Drug Transport and Resistance Mechanisms

Many transporters of Gram-negative bacteria involved in the extracellular secretion of proteins and the efflux of toxic molecules operate by forming intermembrane complexes. These complexes are proposed to span both inner and outer membranes and create a bridge across the periplasm. In this study, we analyzed interactions between the inner and outer membrane components of the tri-partite multidrug efflux pump AcrAB-TolC from Escherichia coli. We found that, once assembled, the intermembrane AcrAB-TolC complex is stable during the separation of the inner and outer membranes and subsequent purification. All three components of the complex co-purify when the affinity tag is attached to either of the proteins suggesting bi-partite interactions between AcrA, AcrB, and TolC. We show that antibiotics, the substrates of AcrAB-TolC, stabilize interactions within the complex. However, the formation of the AcrAB-TolC complex does not require an input of energy. Many transporters of Gram-negative bacteria involved in the extracellular secretion of proteins and the efflux of toxic molecules operate by forming intermembrane complexes. These complexes are proposed to span both inner and outer membranes and create a bridge across the periplasm. In this study, we analyzed interactions between the inner and outer membrane components of the tri-partite multidrug efflux pump AcrAB-TolC from Escherichia coli. We found that, once assembled, the intermembrane AcrAB-TolC complex is stable during the separation of the inner and outer membranes and subsequent purification. All three components of the complex co-purify when the affinity tag is attached to either of the proteins suggesting bi-partite interactions between AcrA, AcrB, and TolC. We show that antibiotics, the substrates of AcrAB-TolC, stabilize interactions within the complex. However, the formation of the AcrAB-TolC complex does not require an input of energy. A distinctive feature of Gram-negative bacteria is the presence of the inner (IM) 1The abbreviations used are: IM, inner membrane; OM, outer membrane; His6 tag, six-histidine tag; MICs, minimal inhibitory concentrations; DSP, dithiobissuccinimidyl propionate; CCCP, cyanide m-chlorophenylhydrazone; PUR, puromycin; OLE, oleandomycin; PRO, proflavine; MC-207,110, Phe-Arg β-naphthylamide; WT, wild type; HMW, high molecular weight complexes; DTT, dithiothreitol; NTA, nitrilotriacetic acid; CyoA, cytochrome o ubiquinol oxidase subunit II. and outer (OM) membranes. The two membranes are separated by an aqueous compartment called the periplasm, which provides a microenvironment of small molecules that buffer the cell from changes in its local surrounding. In fulfilling this essential role, the periplasm is not static but a dynamic cell compartment that changes in response to external and internal stimuli (1Koch A.L. Crit. Rev. Microbiol. 1998; 24: 23-59Google Scholar). The dynamic structure of a periplasm creates a special problem for the transport of diverse molecules into and out of a bacterial cell. To avoid the diluting effect of the periplasm, all molecules destined to move into an external medium or into the cytoplasm must pass the two membranes in a coordinated manner. Yet the dynamic structure of the periplasm should not be compromised. An ultimate solution to the coordination of processes spatially separated into two different membranes is the existence of intermembrane multiprotein complexes that span both membranes and create a bridge across the periplasm. In this case, the transport is achieved without a periplasmic intermediate. Numerous transporters of Gram-negative bacteria involved in the extracellular secretion of protein and the efflux of toxic molecules were shown to operate by forming intermembrane complexes (reviewed in Refs. 2Andersen C. Hughes C. Koronakis V. Curr. Opin. Cell Biol. 2001; 13: 412-416Google Scholar and 3Zgurskaya H.I. Int. J. Med. Microbiol. 2002; 292: 95-105Google Scholar). In the latter case this feature is particularly important, because accumulation of periplasmic intermediates could lead to cell death. Although highly efficient in transport, the intermembrane complexes create tight links between the inner and outer membranes and, thus, could impede the plasticity of the periplasm. How such complexes operate without compromising the dynamic nature of the periplasm remains unclear. One possibility is that the intermembrane complexes are also dynamic structures and assemble only in the presence of specific substrates. Such a mechanism was proposed for the export of α-hemolysin by HlyBD-TolC complex from Escherichia coli. Here, the inner membrane complex HlyBD was shown to be transiently recruited by the outer membrane channel TolC to form an intermembrane channel (4Thanabalu T. Koronakis E. Hughes C. Koronakis V. EMBO J. 1998; 17: 6487-6496Google Scholar). The interactions between the IM and OM components were stabilized by the substrate, because the complex disintegrated after translocation. Consistent with the dynamic interactions within the complex is the finding that at least one component of the HlyBD-TolC transporter, the OM channel TolC, is shared by several other transport systems to achieve both an uptake and an efflux of diverse molecules (5Buchanan S.K. Trends Biochem. Sci. 2001; 26: 3-6Google Scholar). The major multidrug efflux pump AcrAB from E. coli cooperates with TolC to extrude out of the cell an extremely broad range of antimicrobial compounds, including antibiotics, detergents, dyes, and organic solvents (6Ma D. Cook D.N. Alberti M. Pon N.G. Nikaido H. Hearst J.E. J. Bacteriol. 1993; 175: 6299-6313Google Scholar). Located in the inner membrane, the drug:proton antiporter AcrB captures its substrates within the phospholipid bilayer (7Zgurskaya H.I. Nikaido H. Proc. Natl. Acad. Sci. U. S. A. 1999; 96: 7190-7195Google Scholar) and transports them into the external medium via the OM channel TolC (8Fralick J.A. J. Bacteriol. 1996; 178: 5803-5805Google Scholar). The cooperation between AcrB and TolC is presumably mediated by the periplasmic protein AcrA. All three components are required for efficient transport, because disruption of any of the three genes results in hypersusceptibility of E. coli to various substrates (9Okusu H. Ma D. Nikaido H. J. Bacteriol. 1996; 178: 306-308Google Scholar). A remarkable feature of this transporter is that the drug efflux proceeds directly into the external medium, bypassing the periplasm, suggesting the existence of an intermembrane complex. In this study, we investigate the architecture and assembly of the intermembrane AcrAB-TolC complex. Previous hydrodynamic studies suggested that the purified AcrA protein is a highly asymmetric molecule with a length of up to 200 Å (10Zgurskaya H.I. Nikaido H. J. Mol. Biol. 1999; 285: 409-420Google Scholar). Consistent with the hydrodynamic estimate, electron microscopy crystallography showed that AcrA molecules form a coiled structure with the length of 210 Å (11Avila-Sakar A.J. Misaghi S. Wilson-Kubalek E.M. Downing K.H. Zgurskaya H. Nikaido H. Nogales E. J. Struct. Biol. 2001; 136: 81-88Google Scholar). This length of AcrA is sufficient to span the periplasm and provide a connection between the IM and OM of E. coli. The N-terminal lipid moiety anchors this protein into the periplasmic leaflet of the IM. The interaction with the IM is further strengthened by specific binding to the AcrB transporter (12Zgurskaya H.I. Nikaido H. J. Bacteriol. 2000; 182: 4264-4267Google Scholar). Cross-linked invivo AcrA is an oligomer possibly a trimer. Although AcrB is not required for the AcrA oligomerization, only oligomeric AcrA cross-links to AcrB. Using the same approach, no cross-linked complexes containing TolC were identified (12Zgurskaya H.I. Nikaido H. J. Bacteriol. 2000; 182: 4264-4267Google Scholar). Purified and reconstituted into proteoliposomes AcrB was shown to cooperate with AcrA to transport fluorescently labeled phospholipids (7Zgurskaya H.I. Nikaido H. Proc. Natl. Acad. Sci. U. S. A. 1999; 96: 7190-7195Google Scholar). This transport activity of AcrAB was dependent on proton-motive force. Three charged amino acid residues Asp-407, Asp-408, and Lys-940 located in the transmembrane segments 4 and 10 were proposed to form a proton translocating pathway in AcrB and its close homologs (13Guan L. Nakae T. J. Bacteriol. 2001; 183: 1734-1739Google Scholar, 14Murakami S. Nakashima R. Yamashita E. Yamaguchi A. Nature. 2002; 419: 587-593Google Scholar). Mutations in these amino acid residues result in a complete inactivation of the transporter. In crystals, AcrB is a trimer (14Murakami S. Nakashima R. Yamashita E. Yamaguchi A. Nature. 2002; 419: 587-593Google Scholar, 15Yu E.W. McDermott G. Zgurskaya H.I. Nikaido H. Koshland Jr., D.E. Science. 2003; 300: 976-980Google Scholar) in concert with the trimeric arrangements of AcrA (12Zgurskaya H.I. Nikaido H. J. Bacteriol. 2000; 182: 4264-4267Google Scholar) and the OM component of this complex TolC (16Koronakis V. Sharff A. Koronakis E. Luisi B. Hughes C. Nature. 2000; 405: 914-919Google Scholar). The structural signature of AcrB is a large periplasmic domain, which was shown to play a central role in drug recognition and transport (17Tikhonova E.B. Wang Q. Zgurskaya H.I. J. Bacteriol. 2002; 184: 6499-6507Google Scholar, 18Elkins C.A. Nikaido H. J. Bacteriol. 2002; 184: 6490-6498Google Scholar). The top of this trimeric periplasmic domain forms a funnel, and the edge of the funnel has the dimensions that would fit with the tip of the periplasmic helical barrel of TolC (16Koronakis V. Sharff A. Koronakis E. Luisi B. Hughes C. Nature. 2000; 405: 914-919Google Scholar). This periplasmic barrel of TolC is about 70 Å long and could reach half the way across the periplasm. This observation prompted a model that AcrB could reach across the periplasm to bind to TolC (14Murakami S. Nakashima R. Yamashita E. Yamaguchi A. Nature. 2002; 419: 587-593Google Scholar). Bacterial Strains and Plasmids—E. coli strains and plasmids used in this study are listed in Table I. The affinity His6 tag was introduced by the PCR. All primer sequences are available upon request. The pUC151A plasmid was used as a template to construct pAHisB and pABHis, and the pTrc99A-TolC plasmid was used as a template for construction of pTolCHis. Native AcrB contains two C-terminal histidine residues. To construct the pAcrABHis plasmid, a 1331-bp PCR fragment of the acrB gene with four additional histidine residues was digested with HpaI and NsiI restriction enzymes and inserted into the pUC151A plasmid, treated with the same enzymes. The pAHisB plasmid was constructed by the replacement of an MscI-XbaI fragment of pUC151A with a PCR-amplified acrA containing the His6 tag. The resultant plasmid was then treated with XbaI and SphI restriction enzymes and ligated with the PCR-amplified acrB gene to produce pAHisB. The pTolCHis plasmid was constructed by the replacement of a BlpI-HindIII fragment of pTrc99A-TolC with the PCR-amplified tolC fragment encoding 63 C-terminal amino acid residues of TolC fused with His6 tag. To construct plasmid pAB408His a single D408A substitution was introduced into the acrB gene using the QuikChange™ XL site-directed mutagenesis kit (Stratagene) with pABHis as a template. Constructs were verified by DNA sequencing.Table IStrains and plasmidsStrain/plasmidDescriptionSource/referenceZK4Wild type MC410036Gilson L. Mahanty H.K. Kolter R. EMBO J. 1990; 9: 3875-3894Google ScholarZK796Tetr, same as MC4100 but tolC::Tn1036Gilson L. Mahanty H.K. Kolter R. EMBO J. 1990; 9: 3875-3894Google ScholarECM2112MC4100 but ΔacrAB::kan tolC::Tn10O. LomovskayaAG100AXargE3 thi-1 rpsL xyl mtl galK supE441 Δ(gal-uvrB)λ- ΔacrAB::kan ΔacrEF::spe37Miyamae S. Ueda O. Yoshimura F. Hwang J. Tanaka Y. Nikaido H. Antimicrob. Agents Chemother. 2001; 45: 3341-3346Google ScholarAG102MBargE3 thi-1 rpsL xyl mtl galK supE441 Δ(gal-uvrB)λ- ΔacrB::kanH. NikaidopUC18E. coli cloning vector, AmprpUC151ApUC18 vector carrying the acrAB genes6Ma D. Cook D.N. Alberti M. Pon N.G. Nikaido H. Hearst J.E. J. Bacteriol. 1993; 175: 6299-6313Google ScholarpUZ11pUC18 carrying the ompA-acrA-His6 fusion sequence under lac promoter10Zgurskaya H.I. Nikaido H. J. Mol. Biol. 1999; 285: 409-420Google ScholarpTrc99A-TolCpTrc99A vector, Ampr, carrying tolC38German G.J. Misra R. J. Mol. Biol. 2001; 308: 579-585Google Scholar, 39Ip H. Stratton K. Zgurskaya H. Liu J. J. Biol. Chem. 2003; 278: 50474-50482Google ScholarpBP184pACYC184 vector, Cmr, expressing acrB under native promoter35Borges-Walmsley M.I. Beauchamp J. Kelly S.M. Jumel K. Candlish D. Harding S.E. Price N.C. Walmsley A.R. J. Biol. Chem. 2003; 278: 12903-12912Google ScholarpAHisBHis6-tagged acrA derivative of pUC151A plasmidThis studypABHisHis6-tagged acrB derivative of pUC151A plasmidThis studypAB408HispABHis plasmid carrying acrB D408A mutationThis studypTolCHisHis6-tagged tolC derivative of pTrc99A-TolCThis study Open table in a new tab Media, Bacterial Growth, and MIC Determinations—All bacterial cultures were grown at 37 °C in Luria-Bertani (LB) broth or agar (10 g of Bacto-tryptone, 5 g of yeast extract, and 5 g of NaCl per liter). Antibiotics ampicillin (100 μg/ml), kanamycin (35 μg/ml), spectinomycin (50 μg/ml), tetracycline (25 μg/ml), and chloramphenicol (25 μg/ml) were used for the selection where indicated. Minimal inhibitory concentrations (MICs) of various antimicrobial agents were determined as described previously (17Tikhonova E.B. Wang Q. Zgurskaya H.I. J. Bacteriol. 2002; 184: 6499-6507Google Scholar). Protein Purification—Native AcrB was purified from E. coli ECM2112 containing pUC151A plasmid as described previously (7Zgurskaya H.I. Nikaido H. Proc. Natl. Acad. Sci. U. S. A. 1999; 96: 7190-7195Google Scholar). Unlipidated AcrAHis was purified from E. coli strain AG100AX containing pUZ10 plasmid as described (10Zgurskaya H.I. Nikaido H. J. Mol. Biol. 1999; 285: 409-420Google Scholar). For purification of native TolC without the His6 tag, the E. coli strain AG100AX containing pTrc99A-TolC plasmid was grown in 1 liter of LB medium with ampicillin (100 μg/ml) to A600 ∼ 0.6, then the expression of TolC was induced by 0.1 mm isopropyl-1-thio-β-d-galactopyranoside for 3 h. Preparation of membrane fractions and purification steps were done as described before (19Koronakis V. Li J. Koronakis E. Stauffer K. Mol. Microbiol. 1997; 23: 617-626Google Scholar) with minor modifications. Purified TolC was also used to raise polyclonal antibody in rabbits. Protein Assays and Analysis—Protein concentrations were determined using the DC Protein Assay (Bio-Rad) with bovine serum albumin as a standard. SDS-PAGE and immunoblotting were performed by standard techniques. When indicated in the figures, the reducing agent DTT and heat denaturation step were omitted from the sample preparation procedure. In all other cases, DTT was added to protein samples as a part of the SDS-sample buffer just prior to the electrophoresis. Proteins were visualized with anti-AcrA, anti-AcrB, and anti-TolC rabbit polyclonal antibodies and the alkaline-phosphatase-conjugated anti-rabbit antibody (Sigma) (7Zgurskaya H.I. Nikaido H. Proc. Natl. Acad. Sci. U. S. A. 1999; 96: 7190-7195Google Scholar). To determine the intracellular concentrations of AcrA, AcrB, and TolC and the amounts of co-purified proteins we used a quantitative immunoblotting approach. Whole cell lysates, membrane fractions or purified proteins were separated by the SDS-PAGE. AcrA, AcrB, or TolC were detected using immunoblotting with corresponding polyclonal antibodies. For each experiment, the respective purified proteins in increasing concentrations were loaded onto the same gels to serve as standards for the calibration. The alkaline-phosphatase-conjugated secondary antibody and the chromogenic reaction with 5-bromo-4-chloro-3-indolyl phosphate/nitro blue tetrazolium substrates were used to visualize protein bands. Quantification of immunoblots was done on a GS710 Calibrated Imaging Densitometer (Bio-Rad). Sucrose Density Gradient Fractionation—Cells were grown to A600 of 0.6–0.8 in 50 ml of LB medium with an appropriate antibiotic and collected by centrifugation. Disruption of cells by EDTA-lysozyme treatment was performed as described previously (20Osborn M.J. Gander J.E. Parisi E. Carson J. J. Biol. Chem. 1972; 247: 3962-3972Google Scholar). Spheroplasts were fragmented into membrane vesicles by sonication (Misonix Inc., XL 2020) for 45 s on ice. Then, the membrane vesicles in 20% sucrose solution containing 5 mm EDTA (pH 8.0) were layered as 1.6-ml fractions on a two-step gradient consisting of 3.2 ml of 60% sucrose and 7.5 ml of 25% sucrose. All subsequent steps for membrane purification and floatation were performed as described by Ishidate et al. (22Ishidate K. Creeger E.S. Zrike J. Deb S. Glauner B. MacAlister T.J. Rothfield L.I. J. Biol. Chem. 1986; 261: 428-443Google Scholar). Briefly, after centrifugation in a Beckman SW40 rotor at 40,000 rpm for 3.5 h, two adjacent bands at 25–60% sucrose interface were collected together by puncturing the side of the tube with a needle attached to a syringe. The resulting suspension (0.6–0.9 ml) was diluted with 5 mm EDTA and was then applied on the top of 11 ml of a 30–60% (w/w) sucrose gradient. Samples were centrifuged at 100,000 × g for 18 h, and fractions (0.1–0.2 ml) were collected from the bottom of the centrifuge tube. Separated IM and OM samples were further subjected to a floatation gradient centrifugation at 100,000 × g for 20 h. After centrifugation all fractions were analyzed by the SDS-PAGE and visualized by the silver staining technique or Western blotting. The NADH-oxidase activity was measured as before (20Osborn M.J. Gander J.E. Parisi E. Carson J. J. Biol. Chem. 1972; 247: 3962-3972Google Scholar). Sucrose concentrations (%) were converted into density values (g/ml) using tabulated values for sucrose in water at 5 °C (21Fasman G.D. Handbook of Biochemistry and Molecular Biology. CRC Press, Cleveland1975Google Scholar). Co-purification Experiments—For co-purification experiments, E. coli cells carrying the corresponding plasmids were grown in 0.5 liter of LB medium with ampicillin (100 μg/ml) to A600 ∼ 0.8–1.0, collected by centrifugation, and resuspended in 5 ml of buffer containing 10 mm Tris-HCl and 5 mm EDTA (pH 8.0). All centrifugation steps were carried out at +4 °C. We found that the cell lysis by French Press led to the disruption of the AcrAB-TolC complexes. Therefore, in all co-purification experiments we used EDTA-lysozyme treatment followed by osmotic shock to convert cells into spheroplasts. Then spheroplasts were sonicated for 45 s. Unbroken cells were removed by low speed centrifugation, and membranes were collected by centrifugation for 1 h at 250,000 × g. The membrane pellet was washed with 10 mm Tris-HCl (pH 8.0) to remove lysozyme and EDTA and resuspended in 1 ml of the column binding buffer (5 mm imidazole, 50 mm Tris-HCl (pH 8.0), 0.5 m NaCl). An equal volume of 10% Triton X-100 in the binding buffer was then slowly added to the membrane suspension and left on ice for at least 2 h. Insoluble material was removed by centrifugation for 30 min at 70,000 × g. Triton X-100 in the concentration of 5% efficiently solubilized all three AcrAHis, AcrBHis, and TolCHis proteins. However, the high detergent concentration interfered with the binding of the His6-tagged proteins to Ni+2-NTA resin (Novagen). To improve binding to the matrix we substituted Ni2+ ions with Cu2+. Solubilized membrane proteins were loaded onto a Cu+2-NTA column equilibrated with the binding buffer containing 0.2% Triton X-100 (equilibration buffer). The column was washed with equilibration buffer containing 100 mm imidazole (60 mm in the case of AcrAHis), and then proteins were eluted in the same buffer containing 500 mm imidazole. To analyze the effect of ionophores, substrates, and the inhibitor on the assembly of complexes, AG100AX E. coli cells containing pABHis or pAB408His plasmids were grown to A600 of 0.6–0.8 in 100 ml of LB medium with ampicillin (100 μg/ml). Cells were collected by centrifugation, washed in potassium-sodium phosphate buffer (pH 7.4), resuspended in the same buffer, and incubated with carbonyl cyanide m-chlorophenylhydrazone (CCCP) in concentrations of 5, 20, or 100 μm, puromycin (PUR) (64 μg/ml), oleandomycin (OLE) (250 μg/ml), proflavine (PRO) (12.5 μg/ml), and Phe-Arg β-napthylamide (MC-207,110) (5 μg/ml) for 30 min at 37 °C. After incubation, cells were collected by centrifugation, resuspended in buffer (10 mm Tris-HCl (pH 8.0), 5 mm EDTA (pH 8.0)), and subjected to the osmotic EDTA-lysozyme lysis. The His6-tagged proteins were purified as described above. Chemical Cross-linking—Chemical in vivo cross-linking was carried out as described previously (12Zgurskaya H.I. Nikaido H. J. Bacteriol. 2000; 182: 4264-4267Google Scholar). Dithiobissuccinimidyl propionate (DSP) in a concentration of 400 μm was used as a cross-linking agent. Reaction mixtures were incubated at +37 °C for 30 min. Cross-linking was terminated by adding Tris-HCl (pH 7.2) to the final concentration of 75 mm. For the analysis of the cross-linked complexes, reducing agents were omitted from the sample buffer. AcrA and AcrB Co-migrate with Both the Inner and Outer Membranes in Sucrose Density Gradient—To elucidate how the intermembrane AcrAB-TolC complex is assembled, we first determined the distribution of AcrA, AcrB, and TolC proteins in membrane fractions after separation of the IM and OM by sucrose density equilibrium centrifugation (Fig. 1). For this purpose, E. coli cells were converted into spheroplasts by EDTA-lysozyme treatment (20Osborn M.J. Gander J.E. Parisi E. Carson J. J. Biol. Chem. 1972; 247: 3962-3972Google Scholar). Crude membranes were first purified by sedimentation in a two-step sucrose gradient and then subjected to sedimentation through a 30–60% sucrose gradient, following the technique of Ishidate et al. (22Ishidate K. Creeger E.S. Zrike J. Deb S. Glauner B. MacAlister T.J. Rothfield L.I. J. Biol. Chem. 1986; 261: 428-443Google Scholar). Because of their different lipid compositions, the IM and OM migrate to different positions on a sucrose gradient, 1.16–1.18 and 1.22–1.26 g/ml, respectively (20Osborn M.J. Gander J.E. Parisi E. Carson J. J. Biol. Chem. 1972; 247: 3962-3972Google Scholar, 22Ishidate K. Creeger E.S. Zrike J. Deb S. Glauner B. MacAlister T.J. Rothfield L.I. J. Biol. Chem. 1986; 261: 428-443Google Scholar). We used three criteria to evaluate the efficiency of separation: (i) fractionation of known integral OM proteins (OmpA, OmpF/C, and TolC) at a density of 1.22–1.26 g/ml, which is a characteristic for the OM; (ii) fractionation of NADH-oxidase activity only at a density of 1.16–1.18 g/ml, which is characteristic for the IM; and (iii) floatation of membrane fractions separated by sedimentation into the same densities of sucrose (22Ishidate K. Creeger E.S. Zrike J. Deb S. Glauner B. MacAlister T.J. Rothfield L.I. J. Biol. Chem. 1986; 261: 428-443Google Scholar). As shown on Fig. 1A, vast amounts of OmpF/C and OmpA fractionated with the OM (Fig. 1A). Previous reports indicated that trimeric, fully assembled TolC is located exclusively in the OM (19Koronakis V. Li J. Koronakis E. Stauffer K. Mol. Microbiol. 1997; 23: 617-626Google Scholar). Similarly, using immunoblotting with anti-TolC antibody we found trimeric TolC only in the OM fractions (Fig. 1C). The NADH activity clearly co-migrated with the IM (Fig. 1B). This fractionation was not affected by the lack of AcrB or TolC proteins. To confirm that there was no cross-contamination between the separated membranes, regions of the sedimentation gradient corresponding to the IM and OM were further purified by the density floatation fractionation. For this purpose, the appropriate fractions were pooled together as indicated in Fig. 1C and analyzed by floatation gradient centrifugation. Consistent with the previous studies, the IM fraction was resolved into a single peak with apparent buoyant density of 1.18 g/ml. The OM fractionated into a single peak of density 1.23–1.26 g/ml (data not shown). Thus, the centrifugation in the sucrose density gradients efficiently separates the IM and OM. We estimate that after separation the nonspecific cross-contamination between the two membranes is only about 5–10%. AcrA is entirely located in the periplasm and postulated to interact with both the IM and OM. In three independent experiments, immunoblotting of the sucrose gradient fractions of membranes isolated from the wild type (ZK4) E. coli showed that 45 ± 4% of AcrA was present in the IM and 26 ± 1% of the protein co-migrated with the OM fractions (Figs. 1C and 2C). The rest of AcrA is distributed in the intermediate density fraction and on the top of the gradient. When the separated IM and OM were subjected to density floatation fractionation, AcrA co-floated with both membranes into 1.16–1.18 g/ml and 1.23–1.25 g/ml densities, respectively (Fig. 1D). Previous studies showed that AcrA and AcrB form a stable complex in the IM (12Zgurskaya H.I. Nikaido H. J. Bacteriol. 2000; 182: 4264-4267Google Scholar). If the OM-bound AcrA remains associated with AcrB, the latter should also be present in the OM fractions. We found that the majority of AcrB (70 ± 5%) co-fractionated with the IM on sucrose gradients (Fig. 1C). However, about 20% of AcrB co-migrated with the high density OM and the intermediate density fractions. The immunoblotting analysis of sucrose fractions without heat denaturation of proteins prior to electrophoresis showed that high density AcrB fractions contained several forms of protein: an AcrB monomer with an apparent molecular mass about 110 kDa and AcrB-containing high molecular weight (HMW) species with molecular masses of >200 kDa (Fig. 1C). To rule out protein aggregation as a possible cause of AcrB fractionation into high density regions of sucrose, the separated OM fractions were subjected to floatation centrifugation. We found that, similar to AcrA, AcrB remained bound to the OM during density floatation centrifugation (Fig. 1D). Furthermore, when OM fractions were subjected to the second round of sucrose density centrifugation in the presence of salts and chaotropic agents, both AcrA and AcrB still co-migrated with OM (data not shown). These results suggested that fractions of AcrA and AcrB could exist in a tight association with the OM. TolC Is Essential for the Co-fractionation of AcrAB with OM—To test if AcrB is required for AcrA association with the IM and OM, we fractionated AcrA in the AG102MB strain deficient in AcrB (Fig. 2A, delAcrB). We found that in the absence of AcrB the distribution of AcrA between the two membranes changed. Unlike AcrA in the WT strain, which fractionated 45%/26% with the IM and OM, respectively, in the delAcrB membranes 15 ± 3% of total AcrA co-migrated with the OM and the majority of AcrA (48 ± 1%) still co-migrated in the IM (Fig. 2C). In addition, the increased amounts of AcrA were present in the region of <1.16 g/ml sucrose gradient, which is mostly devoid of the IM proteins (Fig. 2A). This result suggested that AcrA is bound to the IM and OM even in the absence of AcrB. However, the interaction of AcrA with the OM is severed, and the protein is easily pulled away from the membranes. Because more AcrA co-fractionates with the OM in the presence of AcrB, the AcrB transporter could be the driving component of association with the OM. AcrB contains a large periplasmic domain, which might establish interaction with TolC even in the absence of AcrA (14Murakami S. Nakashima R. Yamashita E. Yamaguchi A. Nature. 2002; 419: 587-593Google Scholar). However, in the absence of AcrA we found only traces of AcrB in the OM fractions, even when AcrB and TolC were overproduced from the plasmids in the strain ECM2112 deficient of chromosomal copies of acrAB and tolC (Fig. 2B). Thus, AcrB alone cannot co-fractionate with the OM. Only the presence of both proteins, AcrA and AcrB, in the cytoplasmic membrane leads to the formation of specialized regions that co-fractionate in the sucrose gradient with the OM. The outer membrane channel TolC has been reported to be indispensable for the multidrug resistance phenotype of E. coli (8Fralick J.A. J. Bacteriol. 1996; 178: 5803-5805Google Scholar). However, the direct interaction between TolC and AcrAB has been elusive (12Zgurskaya H.I. Nikaido H. J. Bacteriol. 2000; 182: 4264-4267Google Scholar). Consistent with previous reports, we found that the stable TolC trimers fractionate on sucrose gradients only with the OM (Fig. 1C) (19Koronakis V. Li J. Koronakis E. Stauffer K. Mol. Microbiol. 1997; 23: 617-626Google Scholar). Neither AcrA nor AcrB affected the localization of TolC (data not shown). The distribution of AcrA between two membranes was not substantially affected by the lack of TolC (Fig. 2A, delTolC). In the delTolC mutant, the OM-bound AcrA constituted 21 ± 3% of total AcrA, whereas 48 ± 4% of AcrA were present in the IM fractions (Fig. 2C). However, in the delTolC membranes AcrA appears to be unstable and migrates in the SDS-PAGE as a shorter fragment with an apparent molecular mass of 43 kDa. This result suggests that, although TolC is not required for AcrA co-fractionation with the OM, the interaction with TolC stabilizes the AcrA structure. In contrast, we could not detect AcrB in the OM lacking TolC (Fig. 2B). Thus, AcrB associates with the OM mainly through the binding to TolC. In the absence of TolC, the association with AcrA alone is not sufficient to keep AcrB in the OM. We conclude that the co-fractionation of AcrB with the OM is a result of the tight association between AcrAB complex and TolC. AcrA, AcrB, and TolC Form a Stable Intermembrane Complex in Vivo—To