Membrane Protein Selectively Oriented on Solid Support and

Jan 30, 2007 - Alain R. Brisson,† and Olivier Lambert*,†. Laboratoire d'Imagerie Mole´culaire et Nano-Bio-Technologie, IECB, UMR-CNRS 5471, UniVe...
2 downloads 0 Views 1MB Size
Langmuir 2007, 23, 2647-2654

2647

Membrane Protein Selectively Oriented on Solid Support and Reconstituted into a Lipid Membrane Sylvain Tre´pout,†,# Ste´phane Mornet,†,# Houssain Benabdelhak,‡ Arnaud Ducruix,‡ Alain R. Brisson,† and Olivier Lambert*,† Laboratoire d’Imagerie Mole´ culaire et Nano-Bio-Technologie, IECB, UMR-CNRS 5471, UniVersite´ de Bordeaux1, 2 rue Robert Escarpit, F-33607 Pessac, France, and Laboratoire de Cristallographie et RMN Biologiques, UMR 8015 CNRS, Faculte´ de Pharmacie Paris 5, 4 aVenue de l’obserVatoire, 75270 Paris Cedex 06, France ReceiVed July 28, 2006 Mimetic functional membranes on solid support are now emerging for the development of membrane biosensor or for the study of membrane-mediated processes and should have an important impact on biodiagnostics. We established a method to reconstitute a membrane protein into a lipid membrane in a selective orientation on a solid support. Membrane protein OprM, a component of OprM-MexA-MexB multidrug efflux pump, solubilized in detergent was immobilized via its extracellular domain on aminosilane-modified silica surface. The oriented protein was reconstituted into a lipid membrane by detergent removal. The membrane protein reconstitution process carried out on silica nanoparticles and on planar silica surfaces was followed by cryo-electron microscopy (cryo-EM) and quartz crystal microbalance with dissipation monitoring (QCM-D) respectively. The selective protein orientation on aminosilanemodified silica surface was assessed by cryo-EM and was compared to the nonspecific protein deposition on silica surface. Finally, the binding of MexA, a periplasmic component of the tripartite efflux complex, was monitored with QCM-D on the oriented OprM protein monolayer. The large adsorbed mass gave a direct evidence of the high affinity of MexA with the periplasmic helical part of OprM.

Introduction Transmembrane proteins deposited on a solid support present a growing interest for the development of new model systems of biological membranes and for the design of nanobiomaterials with controlled properties.1-3 In a cell, the orientation of membrane proteins within the lipid membrane and their number are “naturally” controlled. Transposing this degree of control using supported membrane proteins (SMP) incorporated into a lipid bilayer is difficult mainly because of the amphiphilic nature of membrane protein.4,5 In recent years, the formation of supported lipid bilayer (SLB) by the spreading of liposomes from solution has emerged as a successful approach to deposit a continuous lipid membrane on solid support.6-11 Images of structural intermediates of the SLB formation have provided some insights into processes of vesicle deposition, vesicle decomposition, and membrane fusion, even though efforts are still needed for comprehensive understanding of the driving forces of the SLB-formation.9-11 Vesicle spreading technique has been extended to the deposition of vesicles containing membrane proteins to form an SMP.5,12-18 * To whom correspondence should be addressed. E-mail: o.lambert@ iecb.u-bordeaux.fr. † University of Bordeaux1. ‡ Faculte ´ de Pharmacie Paris 5. # These authors contributed equally to this work. (1) Sackmann, E. Science 1996, 271, 43. (2) Sackmann, E.; Tanaka, M. Trends Biotechnol. 2000, 18, 58. (3) Groves, J. T. Sci. STKE 2005, 301, pe45. (4) Tanaka, M.; Sackmann, E. Nature 2005, 437, 656. (5) Goennenwein, S.; Tanaka, M.; Hu, B.; Moroder, L.; Sackmann, E. Biophys. J. 2003, 85, 646. (6) Brian, A. A.; McConnell. H. M. Proc. Natl. Acad. Sci. U.S.A. 1984, 81, 6156. (7) Ra¨dler, J.; Strey, H.; Sackmann, E. Langmuir 1995, 11, 4539. (8) Reviakine, I.; Brisson, A. Langmuir 2001, 17, 8293. (9) Keller, C. A.; Kasemo, B. Biophys. J. 1998, 75, 1397. (10) Richter, R.; Mukhopadhay, A.; Brisson, A. Biophys. J. 2003, 85, 3035. (11) Reimhult, E.; Ho¨o¨k, F.; Kasemo, B. Langmuir 2003, 19, 1681.

Using this approach, the formation of an SMP appears less successful than the SLB and suffers from new constraints because of the presence of the membrane proteins. Indeed, after their adsorption to the support, proteoliposomes remained intact on the support reducing the efficiency of the process of vesicle decomposition into lipid bilayer as described for liposome.15,16 The use of potential fusogenic components for triggering the vesicle fusion and inducing SMP formation may overcome this problem.17,18 The major drawback of using proteoliposomes to form SMP concerns the orientation of proteins within an SMP. As the reconstitution process of the proteoliposomes hardly allows the tuning of the protein orientation within the membrane, proteoliposomes contain randomly oriented proteins preventing a selective orientation of the proteins within an SMP. More recently, a new approach has emerged and consists first in the specific binding of proteins on the surface and then its reconstitution into a lipid membrane during detergent removal.19,20 The presence of specific binding sites on the solid support is in favor of the control of a selective orientation of the protein within the lipid membrane. A surface modification has been built with the attachment of a nitrilotriacetic moiety (NTA) on gold surface or on silica surface and has been proposed a universal binding for His-Tag protein.19-21 The protein orientation is directly (12) Wagner, M. L.; Tamm, L. K. Biophys. J. 2000, 79, 1400. (13) Naumann, R.; Baumgart, T.; Gra¨ber, P.; Jonczyk, A.; Offenhausser, A.; Knoll, W. Biosens. Bioelectron. 2002, 17, 25. (14) Sinner, E. K.; Reuning, U.; Kok, F. N.; Sacca, B.; Moroder, L.; Knoll, W.; Oesterhelt, D. Anal. Biochem. 2004, 333, 216. (15) Puu, G.; Artursson, E.; Gustafson, I.; Lundstrom, M.; Jass, J. Biosens. Bioelectron. 2000, 15, 31. (16) Graneli, A.; Rydstrom, J.; Kasemo, B.; Ho¨o¨k, F. Langmuir 2003, 19, 842. (17) Zebrowska, A.; Krysinski, P. Langmuir 2004, 20, 11127. (18) Elie-Caille, C.; Fliniaux, O.; Pantigny, J.; Mazie´re, J. C.; Bourdillon, C. Langmuir. 2005, 21, 4661. (19) Giess, F.; Friedrich, M. G.; Heberle, J.; Naumann, R. L.; Knoll, W. Biophys. J. 2004, 87, 3213. (20) Ataka, K.; Giess, F.; Knoll, W.; Haber-Pohlmeier, S.; Richter, B.; Heberle, J. J. Am. Chem. Soc. 2004, 126, 16199.

10.1021/la062227z CCC: $37.00 © 2007 American Chemical Society Published on Web 01/30/2007

2648 Langmuir, Vol. 23, No. 5, 2007

Figure 1. (a) Scheme of the tripartite Pseudomonas aeruginosa multidrug efflux pump OprM-MexA-MexB. (b) Side view of OprM trimer (CR trace using the atomic coordinates from RCSB Protein Data Bank code 1WP1) composed of three monomers (colored in red, yellow, and green). OprM molecule forms a 14-nm-long hollow cylinder made of a 4-nm hydrophobic β barrel forming a pore within the membrane and of a 10-nm-long periplasmic R helices. The position of a C terminus is indicated by an asterisk.

Tre´ pout et al.

The selective attachment of OprM via its extracellular face on the solid support was achieved on a silica surface modified with an aminosilane. The formation of an SMP made of OprM molecules was performed on planar support and was monitored with quartz crystal microbalance with dissipation monitoring (QCM-D). The determination of the protein orientation with respect to the support was assessed with the use of silica nanoparticles combined with cryo-electron microscopy (cryoEM). This high-resolution imaging technique allows the observation of frozen hydrated and unstained samples at nanometer resolution range and was successfully applied to a recent study involving silica nanoparticles functionalized with supported lipid bilayer.27 While other imaging systems as atomic force microscopy allow visualization of sample surface (top view), cryo-EM gives access to the side view and in this case potentially reveals the location of the lipid bilayer with respect to the support (nanoparticles) and to the membrane proteins (OprM). Materials and Methods

related to the tag position within the protein that is supposed to vary the tag position for immobilizing the protein on the desired face (i.e., either cytoplamic or extracellular). However, Tag insertion within the protein gene by molecular genetic technique needs to be carefully evaluated as difficulties of protein expression and purification may be encountered and protein activity can be modified. Alternative methods still need to be developed without protein modification to extend the variety of available surface modifications for selective protein attachment. The ion exchange chromatography is a protein purification technique successfully used over decades. This technique is based first on the protein attachment to charged (positively or negatively) support via electrostatic interactions and then on the subsequent release of proteins with an ion exchange. This binding capacity is particularly of interest for membrane proteins, exposing hydrophilic loops on their intra- or extracellular faces. One of the faces may have a specific affinity for charged surface allowing the use of this selectivity for the SMP formation.22 This paper reports on an original method of membrane protein reconstitution on a solid support on the basis of the binding of the protein in a selective orientation via electrostatic interactions and its incorporation into a lipid membrane using detergent depletion. Membrane protein OprM is a component of the Pseudomonas aeruginosa multidrug efflux pump (OprM-MexAMexB). Although this tripartite pump is believed to form a complex channel that traverses inner and outer membranes of P. aeruginosa and allows the extrusion of drugs directly into the extracellular medium,23 the mechanism leading to its assembly is poorly understood (Figure 1a). OprM trimer has a cylindrical shape and is made of short extracellular loops, a hydrophobic β barrel, and long periplasmic helices24 (Figure 1b). Moreover, the architecture of OprM molecules incorporated into a lipid membrane remains as a trimer according to our previous structural cryo-EM study.26 OprM orientation of solid support is the keystone of the tripartite pump assembly for further functional studies. Therefore, OprM molecules must expose their periplasmic helices to the solvent. Unfortunately, the use of a His-Tag inserted at the C-terminus could not be considered because of its location on the side of the protein according to the atomic model (Figure 1b). (21) Rigler, P.; Ulrich, W. P.; Vogel, H. Langmuir 2004, 20, 7901. (22) Le´vy, D.; Chami, M.; Rigaud, J. L. FEBS Lett. 2001, 504, 187. (23) Zgurskaya, H. I.; Nikaido, H. Mol. Microbiol. 2000, 37, 219. (24) Akama, H.; Kanemaki, M.; Yoshimura, M.; Tsukihara, T.; Kashiwagi, T.; Yoneyama, H., Narita, S., Nakagawa, A.; Nakae, T. J. Biol. Chem. 2004, 279, 52816. (25) Yoshihara, E.; Maseda, H.; Saito, K. Eur. J. Biochem. 2002, 269, 4738.

Materials and Reagents. Dioleolylphosphatidylcholine (DOPC), dioleolylphosphatidylglycerol (DOPG), and dioleolyltrimethylammonium propane (DOTAP) were purchased from Avanti Polar Lipids (United States). Octyl-βD-glucopyranoside (βOG) was from Sigma. Tetraethoxysilane (TEOS) (98+%) and N-(trimethoxysilylpropyl)ethylenediamine (EDPS) (97%) were purchased from Aldrich. Preparation of OprM and MexA Proteins. The transmembrane protein OprM expression and purification were achieved as previously described.26 The MexA lipoprotein expression and purification was performed following a similar protocol. The membrane envelopes from broken Escherichia coli cells were solubilized in 20 mM TrisHCl pH 8, 10% glycerol (v/v), 15 mM imidazole, 2.5% β OG (w/v) overnight at 20 °C. The solubilized membrane proteins were loaded onto a Ni-NTA resin column and then were eluted with a linear gradient of imidazole (60-500 mM). The fractions containing the MexA protein were pooled and concentrated to 5 mg/mL. Finally, the mature lipoprotein MexA (as well as OprM) was exchanged for suitable buffer by dialysis in the presence of 20 mM Tris-HCl pH 8.0, 200 mM NaCl, 10% glycerol (v/v), and 1% βOG (w/v). Preparation of Small Unilamellar Vesicles. Lipids were dissolved in chloroform, were dried under a stream of nitrogen followed by drying in a vacuum desiccator overnight, were resuspended at 1-2 mg/mL final concentration, and were vortexed. Small unilamellar vesicles (SUVs) were obtained by sonication with a tip sonicator as described previously.10 Preparation of Untreated Silica Nanoparticles. Colloidal silica nanoparticles were prepared as previously described.27 Briefly, 18 mL of tetraethyl orthosilicate (TEOS) mixed with a solution containing 36 mL of 30% ammonia in 600 mL of dry ethanol was stirred overnight. Silica sol was prepared by evaporation of ammonia and ethanol and was dialyzed against ultrapure water (10 MΩ) and was stored at room temperature. Before use, silica nanoparticles were diluted to a concentration of 1 g/L in 20 mM Tris-HCl buffer at pH 7.4. Preparation of Amino-Functionalized Silica Nanoparticles. Surface modification of silica nanoparticles were carried out by adding N-(trimethoxysilylpropyl)ethylenediamine (EDPS) in the reactive medium after the TEOS condensation step. Assuming that the surface coverage of EPDS is nominally 55 Å2 per molecule, an excess of 10 equiv EDPS was added to provide 2-3 monolayer coatings of the silica nanoparticles.28 After an overnight reaction, 100 mL water was added and then ammonia and ethanol were evaporated at low boil (80 °C) under vacuum allowing transfer of modified nanoparticles in aqueous phase. An extra thermal treatment under vacuum at 100-110 °C was performed until sol flocculation (26) Lambert, O.; Benabdelhak, H.; Chami, M.; Jouan, L.; Nouaille, E.; Ducruix, A.; Brisson, A. J. Struct. Biol. 2005, 150, 50. (27) Mornet, S.; Lambert, O.; Duguet, E.; Brisson, A. Nano Lett. 2005, 5, 281. (28) Interfaces in Polymer Matrix Composites; Plueddeman, E., Broutman, L. J., Knock, R. H., Eds.; Academic Press: New York, 1974; Vol. 6, p 15.

Membrane Reconstitution of OprM on Solid Support

Langmuir, Vol. 23, No. 5, 2007 2649

Figure 2. Plots of ζ potential as a function of pH of bare SiO2 nanoparticles ([) compared with amino-functionalized SiO2 nanoparticles (2). The great difference between these ζ potential profiles attests the presence of an aminosiloxane network on EDPS modified silica nanoparticles.

Figure 3. Binding of OprM on aminated silica nanoparticles. (a-c) Cryo-EM view of OprM micelles interacting with aminated silica nanoparticles (a). On enlarged views of nanoparticles (b, c), stiff rods clearly visible at the particle surface correspond to OprM molecules (black arrows), as shown in (f). (d and e) In the presence of 300 mM NaCl, stiff rods are not observed on nanoparticle surfaces. Only a few proteins seem to absorb on the nanoparticle surfaces with a nonradial orientation (black arrow heads). Scale bars 50 nm. (about 2 h). To maintain a fixed volume, aliquots of water were periodically supplemented. The flocculated nanoparticles were then washed with ultrapure water three times by centrifugation to remove coproducts. Aminated silica nanoparticles were dispersed by adjusting pH 6 with HCl. The final concentration is equal to 12 g/L. Zeta Potential Measurements of Silica and Silica-Modified Nanoparticles. The zeta potential of silica nanoparticles was assessed by using a Zetasizer 3000HSA setup (Malvern Instruments). Measurements were performed for 20 s using a standard capillary electrophoresis cell. The dielectric constant of solvent (water) and the Smoluchowsky constant f(ka) were set to 80.4 and 1.5, respectively. Zeta potential value of modified and unmodified silica nanoparticles diluted at a final concentration of 0.1 mg/mL was measured for a pH range (1-12). Reconstitution of Membrane Protein on Nanoparticles. A micellar solution of OprM molecules at 30 µg/mL in 20 mM TrisHCl pH 7.4 in the presence of 50 mM βOG supplemented with a DOPC liposome suspension at 300 µg/mL formed ternary micelles (OprM-βOG-DOPC) in a final volume of 100 µL. Then, silica

nanoparticles (15 µg) were added to the ternary micelles and were maintained at 4 °C. After a 15-min incubation time, the detergent was removed with two successive additions of 50 mg/mL polystyrene beads (SM2 Biobeads, Biorad).29 After an overnight incubation at 4 °C under gentle stirring, the reconstituted material was pipetted off and was stored at 4 °C. Preparation of Amino-Functionalized QCM-D Sensor Surface. After a 45-min UV ozone treatment, a sensor crystal was plunged successively in pure water and absolute ethanol bath for 1 min. Then, it was immersed in a 2% EDPS solution (v/v ethanol) for 30 min at room temperature. After an extensive rinsing with ethanol, the sensor crystal immersed in a glycerol bath was submitted to a thermal treatment at 110 °C under primary vacuum for 30 min using a Bu¨chi Glass oven. Finally, the crystal was rinsed with ethanol, was dried under a gentle nitrogen flow, and was immediately used for QCM-D experiment. (29) Rigaud, J. L.; Mosser, G.; Lacape`re, J. J.; Olofsson, A.; Le´vy, D.; Ranck, J. L. J. Struct. Biol. 1997, 118, 226.

2650 Langmuir, Vol. 23, No. 5, 2007

Tre´ pout et al.

Figure 4. Reconstitution into lipid membrane of OprM bound to aminated surface. (a-d) Gallery of cryo-EM images of supported OprM molecules reconstituted into a lipid membrane after detergent removal. As the lipid membrane is close to the solid surface (white arrows), the molecules (black arrows) are bound via their extracellular faces, with their long periplasmic helices accessible to the solvent as shown in (e). (f and g) On large nanoparticles (175 nm in diameter), OprM molecules bind to the aminated surface with the same orientation. (h) Gallery of enlarged views of supported reconstituted OprM on low curved surface. Scale bars 50 nm (a-g) and 25 nm (h). Cryo-Electron Microscopy. A 5-µL sample was deposited onto a holey carbon-coated copper grid. The excess was blotted with a filter paper. Unstained samples were frozen into liquid ethane and the grids were mounted onto a Gatan 626 cryoholder, were transferred into the microscope, and were kept at a temperature of about -175 °C. Sample observations were performed with a Tecnai-F20 FEItransmission electron microscope, operating at 200 kV. Low-dose images were recorded at a nominal magnification of 50 000× with a 2k × 2k USC1000 slow-scan CCD camera (Gatan, CA). Quartz Crystal Microbalance with Dissipation Monitoring. The procedure of cleaning untreated silica-coated sensor was described previously.10 QCM-D measurements were performed with the Q-SENSE D300 system equipped with an axial flow chamber (QAFC 302). Briefly, upon interaction of (soft) matter with the surface of a sensor crystal, changes in the resonance frequency, f, related to attached mass (including coupled water), and in the dissipation, D, related to frictional (viscous) losses in the adlayer, are measured with a time resolution of better than 1 s. Measurements were performed in exchange mode10 which allows following processes of adsorption and surface adlayer changes in situ while sequentially exposing different solutions to the supports. Resonance frequency and dissipation were measured at several harmonics (15, 25, 35 MHz) simultaneously. The working temperature was 24 °C. Changes in dissipation and in normalized frequency (∆f ) ∆fn/n, with n being the overtone number) of the third overtone (n ) 3, i.e., 15 MHz) are presented. Adsorbed masses, ∆m, are calculated

according to the Sauerbrey equation,30 ∆m ) -C∆f, with the mass sensitivity constant C ) 17.7 ng‚cm-2‚Hz-1 for 5 MHz sensor crystals.

Results and Discussion In the present paper, our aim was the formation of a membrane protein layer specifically oriented on a solid support and incorporated into a lipid bilayer after detergent removal. Reconstitution of OprM membrane protein on solid support was carried out in three main steps: (1) silica surface modification with aminosilane, (2) specific binding of OprM membrane proteins via its extracellular face, and (3) lipid membrane reconstitution after detergent depletion. The SMP formation was performed on two complementary supports, planar sensor and nanoparticles and was characterized with QCM-D and cryo-EM. The latter approach successfully used for imaging functionalized silica nanoparticles with lipid bilayer has been extended to the determination of the membrane protein orientation on the support.27 1. Amino-Functionalized Silica Nanoparticles Characterization. As OprM has been purified on a DEAE ion exchange column,25 it has a strong affinity with a positively charged amino group. For this purpose, we prepared and characterized silica (30) Sauerbrey, G. Z. Phys. 1959, 155, 206.

Membrane Reconstitution of OprM on Solid Support

Figure 5. Nonspecific binding of OprM on silica nanoparticle and reconstitution into lipid membrane. (a) Cryo-EM view of OprM micelles interacting with silica nanoparticles. Annular features visible at the particle surface correspond to OprM molecules (black arrows), as shown in b. (c) OprM molecules reconstituted into a lipid membrane after detergent removal interact with silica particle with a nonselective orientation via their extracellular face (1) and their periplasmic face (2) as shown in (d). Scale bar 50 nm.

nanoparticles, modified with aminosilane. The zeta potential curve (Figure 2) of silica nanoparticles displayed an isoelectric point (IEP) at pH 2.6, in agreement with previous reports (IEPs of silica are equal to 2-3).31 After silanization of the silica surface, the IEP of aminated nanoparticles reached a pH value of 9.5 that corresponds to the pKa of the primary aminopropyl group. This result indicated a high density of amino groups exposed to the modified-nanoparticle surface.32 Because of the thermal treatment, the condensation of polysiloxane film is completed on the surface triggering the rupture of intramolecular interactions between silanol and the amino groups.33,34 Moreover, the presence of these protonable amino groups in neutral pH conditions ensures the electrostatic colloidal stability. 2. Binding and Reconstitution of OprM Molecules on Amino-Functionalized Silica Nanoparticles. Ternary micelles (OprM/detergent/lipid), which led to a successful reconstitution of OprM into lipid membrane,26 were mixed with these nanoparticles and were observed by cryo-EM (Figure 3). Cryo-images showed silica nanoparticles of about 80 nm in diameter presenting an overall spherical shape and heterogeneous electron-scattering density because of their porosity. After OprM addition, small rods of about 15-nm long were clearly visible (black arrows) and radiated from the nanoparticle periphery (Figure 3a-c). These rods correspond to OprM molecules attached on the amino surface of the nanoparticle and are (31) The Chemistry of Silica; Iler, R. K., Iler, R. K., Eds.; Wiley: New York, 1979; p 98. (32) Bayer, T.; Eichhorn, K. J.; Grundke, K.; Jacobasch, H. J. Macromol. Chem. Phys. 1999, 200, 852. (33) Chiang, C.-H.; Ishida, H.; Koening, J. L. J. Colloid Interface Sci. 1980, 74, 396. (34) Chemically Modified Surfaces; Vrancken, K. C., Van Der Voort, P., Possemiers, K., Grobet, P., Vansant, E. F., Peseck, J. J., Eds.; Royal Society of Chemistry: Cambridge, U.K., 1994; p 46.

Langmuir, Vol. 23, No. 5, 2007 2651

organized in a densely packed protein layer. Moreover, the length of these rods, in agreement with the OprM cylinder shape,24,26,35 revealed that OprM molecules were bound via one of their cylinder bases. Their bindings onto aminated surface were mediated via electrostatic interactions. Indeed, the aminated nanoparticules incubated with OprM molecules in the presence of 300 mM NaCl were not decorated with the stiff rods (Figure 3d and e). Faint protein densities at the nanoparticle surfaces seemed visible suggesting that few OprM molecules adsorb on the surface with a nonradial orientation. Similarly, the radial orientation of OprM on nanoparticle surface was also lost after addition of 300 mM NaCl. Nanoparticle surfaces exhibited similar features as those shown in Figure 3d and e. Then, an increase of ionic strength prevents the protein interaction via its cylinder base, in agreement with an electrostatic binding of OprM on the aminated surface. At this step, the precise OprM orientation could not be determined (Figure 3f). However, this difficulty has been overcome after detergent removal allowing the lipid membrane reconstitution around the hydrophobic part of the protein. Indeed, a fine layer of electron dense material appeared close to the particle (white arrow) (Figure 4a-d). This layer corresponds to the reconstituted lipid bilayer around the hydrophobic domain of OprM molecules. Moreover, the small distance between the lipid bilayer and the particle contour revealed that the extracellular face of OprM molecules interacted specifically with the aminated surface (Figure 4e). To evaluate the influence of curvature of solid surface on OprM orientation, binding and lipid reconstitution of OprM were performed on larger nanoparticles (175 nm in diameter). The protein orientation on these large particles was identical to those observed on the 80-nm nanoparticles (Figure 4f-h). Moreover, some of these particles were not perfectly spherical and showed some flattened edges (e.g., nanoparticle shown in Figure 4g). Enlargements of these areas clearly showed reconstituted OprM molecules on a unique orientation (lipid membrane close to the surface) on the aminated surface. In conclusion, OprM molecules form a dense protein layer on EDPS-modified surface and are attached in a selective manner via their extracellular face, exposing their long periplasmic helices to the solvent or to other components of the efflux pump. 3. Binding and Reconstitution of OprM Molecules on Untreated Silica Nanoparticles. OprM binding and reconstitution have also been performed on untreated silica nanoparticles that exhibited a negatively charged surface at pH 7.4. Ternary micelles of OprM/detergent/lipid mixed with silica nanoparticles were observed by cryo-EM (Figure 5). Electron-dense material appeared at the contour surface of the nanoparticles, revealing the adsorption of OprM molecules on the silica surface (Figure 5a). Annular structures with a 5-nm diameter were clearly visible at the nanoparticle surfaces corresponding to OprM molecules viewed along their long axis (black arrows in Figure 5b). This result indicated that OprM micelles were bound to the negatively charged surface using different molecular recognitions than those used on aminated surface. After detergent removal with polystyrene beads, silica nanoparticles were covered with OprM molecules reconstituted into a lipid membrane (Figure 5c). The location of the lipid membrane reconstituted around the OprM molecules and its distance from the surface revealed the protein orientation. The lipid membrane appeared either close to (case 1) or distant from 10 nm to the nanoparticle contour (case 2) indicating that OprM molecules were attached via their extracellular face or via their periplasmic (35) Koronakis, V.; Sharff, A.; Koronakis, E.; Luisi, B.; Hughes, C. Nature 2000, 405, 914.

2652 Langmuir, Vol. 23, No. 5, 2007

Tre´ pout et al.

Figure 6. QCM-D response for the binding and lipid membrane reconstitution of OprM and MexA interaction on silica surface and aminated surface. (A) Change in frequency ∆F on EDPS surface (-0-) and on silica surface (-O-) of the successive steps of OprM binding (6-64 min), membrane reconstitution (64-86 min, gray zone), and MexA interaction with OprM (90-140 min). (B) Change in dissipation ∆D on aminated surface (-0-) and on silica surface (-O-) of the successive steps described in A. From left to right, arrows indicate seven injections of lipid micelles (50, 30, 25, 20, 15, 10, and 5 mM βOG), a rinse with buffer, an injection of MexA solution, and a rinse with buffer without detergent.

face, respectively (Figure 5d). These two orientations of OprM molecules after detergent removal and lipid membrane reconstitution suggest an absence of selective orientation of OprM on silica nanoparticles. In conclusion, silica surface is not suitable to deposit OprM molecules in a unique orientation, even though it was densely covered with reconstituted proteins. 4. Adsorption of OprM Molecules on Aminosilane-Modified and Silica QCM-D Sensor. The binding of OprM molecules has been monitored with QCM-D on aminosilane-modified crystal sensors and on crystal sensors coated with untreated SiO2 (Figure 6). Aminosilane film was covalently linked to the silica surface of QCM-D sensor according to the procedure described in Materials and Methods. As a control of the positively charged surface sensor, pure DOTAP vesicles did not absorb while pure DOPG liposomes strongly interacted with the surface (data not shown). A micellar solution of OprM at a concentration of 13 µg/mL in the presence of 50 mM βOG was supplemented with a 30 µg/mL liposome suspension to form ternary micelles OprM/ detergent/lipid. After the injection of ternary micelles on aminated surface, the binding of OprM reached a plateau within a few minutes (∆F ) -83 Hz, ∆D ) 5 10-6) indicating a saturation of the surface. On untreated silica surface, frequency and dissipation revealed a different behavior. The adsorption of OprM molecules was slower and did not reach a plateau within a time lapse of 1 h. After rinsing, the final frequency shift and the final dissipation were ∆F ) -54 Hz and ∆D ) 5.2 10-6, respectively. The fast adsorption and the large amount of adsorbed OprM molecules on the aminated surface as compared to nonmodified silica surface correlate well with the cryoEM results.

To evaluate the contribution of lipid micelles for the adsorption signal, the adsorption of mixed micelles OprM/βOG was also performed on both surfaces. The attachment of OprM micelles led to frequency shifts ∆F ) -76 Hz and ∆F ) -50 Hz on aminated and nonmodified silica surfaces, respectively, similar to those obtained with ternary micelles (data not shown). This indicates that adsorption signal is dominated by the adsorption of OprM. 5. Reconstitution of OprM into a Lipid Membrane on Aminosilane-Modified and Silica QCM-D Sensor. Because of technical constraints (narrow capillaries and a chamber design uncompatible with the use of stirring device), the use of polystyrene beads for detergent removal was not possible for the QCM-D experiments. We first investigated a pathway of the formation of a supported lipid membrane from lipid/detergent micellar solutions (Figure 7). The adsorption of DOPC/βOG and DOPG/βOG micellar solutions on untreated silica and aminated surface, respectively, was followed by QCM-D. Several lipid/ OG mixtures containing a fixed lipid concentration and detergent concentrations of 30, 25, 20, 15, 10, and 5 mM βOG, respectively, were successively injected. The final frequency shift of -25 ( 2 Hz and the low dissipation of 0.5 10-6 corresponded to values obtained with small unilamellar vesicles and confirmed the formation of a lipid bilayer covering the support with minor defects.10,36 These results were in agreement with the formation of a supported lipid bilayer performed with the dodecylmaltoside/ lipid mixed micelles on silica surface.37 (36) Reimhult, E.; Ho¨o¨k, F.; Kasemo, B. J. Chem. Phys. 2002, 117, 7401. (37) Tiberg, F.; Harwigsson, I.; Malmsten, M. Eur. Biophys. J. 2000, 29, 196.

Membrane Reconstitution of OprM on Solid Support

Langmuir, Vol. 23, No. 5, 2007 2653

Figure 7. Lipid membrane reconstitution on silica surface and aminated surface using successive injections of lipid/detergent micelles. (A) Change in frequency ∆F on aminated surface (-0-) and on silica surface (-O-). (B) Change in dissipation ∆D on aminated surface (-0-) and on silica surface (-O-). Six lipid/OG mixtures containing a fixed lipid concentration (40 µM) and different detergent concentrations of 30, 25, 20, 15, 10, and 5 mM OG, respectively, are successively injected as indicated by arrows. A liposome solution is injected (V) and then rinses twice with buffer (B). Lipids injected on silica surface and on aminated surface are DOPC and DOPG, respectively. Table 1. QCM-D Measurements of SMP Formation on Untreated and Aminated Surface aminated suface ∆f (Hz) adsorption of OprM micelles reconstituted OprM (SMP) binding of MexA a

-83 ( 3a -73 ( 3a -95 ( 3b

∆D

silica surface (10-6)

7.3 ( 0.1a 0.8 ( 0.3a 6.2 ( 0.2b

∆f (Hz)

∆D (10-6)

-53 ( 1b -53 ( 4b -27 ( 7b

6 ( 1b 1.9 ( 0.5b 1.9 ( 0.2b

Averaged over five measurements. b Averaged over two measurements.

The next step after OprM molecule adsorption on crystal sensor was the lipid membrane reconstitution performed following this procedure of successive injections of lipid/detergent mixtures as described previously. Seven DOPC/βOG mixtures containing a fixed lipid concentration (40 µM) and detergent concentrations of 50, 30, 25, 20, 15, 10, and 5 mM βOG, respectively, were successively injected (Figure 5 gray zone). The reconstitution of OprM into a lipid bilayer on aminated surface showed a small effect on the frequency (∆Ffinal ) -72 Hz), and a large decrease of the dissipation ∆Dfinal (from 5 10-6 to 1.2 10-6). The reconstitution of OprM into a lipid membrane on untreated silica surface showed a similar behavior (∆Ffinal ) -49 Hz; ∆Dfinal ) 1.4 10-6). The high dissipation value is likely to be due to the irregular orientation of the protein molecules as shown in cryoEM results. The variations of frequency and dissipation may be mainly explained by the decrease in detergent concentration within the buffer solution. Indeed, as the QCM-D sensor was sensitive to a buffer composition change, an addition of 50 mM βOG to the buffer solution induced frequency and dissipation shifts of -10 Hz and +4.5 10-6, respectively. So, removing detergent had the

opposite effect and explained the variation of frequency and dissipation measured after OprM reconstitution into a lipid membrane. The protein density deposited on the aminated surface can be estimated. First, QCM-D measurements allowed to calculate the wet mass (mwet) of material deposited on the sensor according to Sauerbrey relation (i.e., mwet ) 73 × 17.7 ) 1292 ng) (Table 1). Second, an electron crystallography study of OprN, a protein closely related to OprM, showed two-dimensional crystals with a hexagonal packing and exhibited a unique orientation of the protein within the lipid membrane.26 So assuming this dense packing for OprM (i.e., each molecule occupied an area of 42 nm2), a hexagonal packing layer of OprM would represent a dry mass mdry of about 600 ng. The solvent contribution to the QCM-D signal has been reported for several proteins, ranging from 50% to 30%.38-40 Although the dependence of the solvent on the (38) Ho¨o¨k, F.; Vo¨ro¨s, J.; Rodahl, M.; Kurrat, R.; Bo¨ni, P.; Ramsden, J. J.; Textor, M.; Spencer, N. D.; Tengvall, P.; Gold, J.; Kasemo, B. Colloids Surf., B 2002, 24, 155. (39) Caruso, F.; Furlong, D. N.; Kingshott, P. J. Colloid Interface Sci. 1997, 186, 129. (40) Richter, R.; Brisson, A. Langmuir 2004, 20, 4609.

2654 Langmuir, Vol. 23, No. 5, 2007

protein shape is unknown, the measurement of adsorbed amount at saturation (1292 g.cm-2) suggests that OprM molecules form a densely packed monolayer in agreement with our cryo-EM data. 6. Interaction between OprM and MexA Monitored with QCM-D. To further establish the proposed orientation of OprM as well as the preservation of its genuine property of interaction with MexA,41-43 a solution of this protein at a 20 µg/mL concentration containing 20 mM βOG was injected at 90 min (Figure 6). Incubation of MexA with OprM led to adsorption of MexA, witnessed by the decreases in frequency. The adsorption of MexA reached rapidly a plateau on OprM molecules attached to aminated surface while it did not on OprM molecules deposited on silica surface (∆Fmexa ) -92 Hz for aminated surface and ∆Fmexa ) -20 Hz for silica surface, respectively). As a control, an incubation of MexA solution on an aminated surface led to a low protein adsorption as monitored by a frequency shift of -15 Hz (data not shown) providing evidence that MexA had a low affinity with the support. All these results demonstrate that MexA interacts strongly with OprM and specifically with its periplasmic domain that is exclusively exposed to the solvent when OprM is bound to the aminated surface. (41) Akama, H.; Matsuura, T.; Kashiwagi, S.; Yoneyama, H.; Tsukihara, T.; Nakagawa, A.; Nakae, T. J. Biol. Chem. 2004, 279, 25939. (42) Higgins, M. K.; Bokma, E.; Koronakis, E.; Hughes, C.; Koronakis, V. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 9994. (43) Yoneyama, H.; Maseda, H.; Kamiguchi, H.; Nakae, T. J. Biol. Chem. 2000, 275, 4628.

Tre´ pout et al.

Conclusion We show here a procedure to reconstitute membrane protein on solid support in a controlled orientation. We propose a chemical surface modification of silica support to selectively orient OprM membrane protein. The use of modified silica nanoparticles and of cryo-EM approach demonstrates the unique orientation of the lipid membrane-reconstituted protein attached to the support and reveals a dense protein layer attached to the surface. The characterization of the reconstituted protein layer on a planar solid support is achieved on the basis of frequency and dissipation measurements of QCM-D technique. With this development, the analysis of the interactions between OprM with the other components of OprM-MexA-MexB efflux pump becomes feasible. This is expected to give new valuable insights in their assembly mechanism. Furthermore, the assembly of such complexes of protein on solid support opens new perspectives for the development and the characterization of inhibitors preventing the assembly of tripartite pump.44 Acknowledgment. This work has been supported in part by EC grants NMP4-CT2003-505868- Nanocues and program ACI “Dynamique et re´activite´ des assemblages biologiques” DRAB04/ 136. The authors thank Jean-Christophe Taveau for critical reading and Jose´phine Lai Kee Him for technical assistance. LA062227Z (44) Nikaido, H. Science 1994, 264, 382.