Lipid Transfer Mediated by a Recombinant Pro-Sterol Carrier Protein 2

Jan 22, 2010 - New Insights into the Growth and Transformation of Vesicles: A Free-Flow ... Secondary Structure, and Cytotoxicity of a New Class of An...
0 downloads 0 Views 2MB Size
pubs.acs.org/Langmuir © 2010 American Chemical Society

Lipid Transfer Mediated by a Recombinant Pro-Sterol Carrier Protein 2 for the Accurate Preparation of Asymmetrical Membrane Vesicles Requires a Narrow Vesicle Size Distribution: A Free-Flow Electrophoresis Study Martin Holzer,* Joachim Momm, and Rolf Schubert Department of Pharmaceutical Technology and Biopharmacy, Albert-Ludwigs-University Freiburg, D-79104 Freiburg, Germany Received September 9, 2009. Revised Manuscript Received January 1, 2010 We applied protein-mediated lipid transfer using recombinant His-tagged pro-sterol carrier protein 2 (pro-SCP2) to prepare asymmetrical membrane vesicles (AMV) featuring an unequal transmembrane distribution of the negative phospholipid egg-phosphatidylglycerol (EPG). Pure egg-phosphatidylcholine (EPC) vesicles were used as the acceptor and EPC:EPG 90:10 mol % vesicles as the donor populations. The changes in surface charge during EPG transfer were used to quantify the degree of asymmetry by free-flow electrophoresis (FFE). The relative deflection in FFE correlated with EPG content in the outer monolayer (xEPG). The initial transfer rates and first order rate constants for the transfer process were determined. The addition of pro-SCP2 at a molar protein-to-lipid ratio RP/L of (15-20)  10-5 accelerated the EPG transfer to half-times of between 2 and 3 h. Thus, the transmembrane redistribution of EPG by flip-flop, which reduces the degree of asymmetry and occurs at half-times of tens of hours, was minimized during the transfer process. We investigated the influence of membrane curvature on the transfer rate using 50 and 100 nm vesicles with very low size distribution widths (RSD of 13-17%). Transfer occurred with a 55.7% higher initial rate between the smaller vesicles. The use of equally sized acceptor and donor populations of such narrow size distributions was shown to be important for the preparation of AMV with a uniform degree of asymmetry. Under these conditions, AMV were obtained after less than 3 h by preparative FFE separation. In the case of the acceptor vesicles, EPG transfer increased xEPG to 3 mol %, whereas it was reduced to 6 mol % in the donor vesicles.

1. Introduction Liposomes are widely used as drug carriers and also as a system to mimic biological membranes. Their structural similarity to natural membranes makes them a suitable tool for investigating various physiological processes such as membrane fusion,1 domain formation,2 cell-cell recognition, or cellular uptake mechanisms, as well as the study of protein activity and behavior in bilayers.3 Furthermore, several drug-membrane interactions (association, binding, transbilayer movement, partitioning, pore formation) can be analyzed4,5 in addition to the incorporation and activity of various membrane proteins or protein complexes.6,7 In order to provide an accurate liposomal model for biological membranes, three features that strongly affect the behavior of these systems must be considered: the supporting cytosolic membrane skeleton,8,9 the lipid composition of the membranes, *To whom correspondence should be addressed. Address: Institute of Pharmaceutical Sciences, Dept. of Pharmaceutical Technology and Biopharmacy, Hermann-Herder-Str. 9, D-79104 Freiburg, Germany. Telephone: þ49-761-203-4912. Fax: þ49-761-203-6366. E-mail: martin.holzer@ pharmazie.uni-freiburg.de. (1) Eastman, S. J.; Hope, M. J.; Wong, K. F.; Cullis, P. R. Biochemistry 1992, 31, 4262. (2) Scherfeld, D.; Kahya, N.; Schwille, P. Biophys. J. 2003, 85, 3758. (3) Poklar, N.; Fritz, J.; Macek, P.; Vesnaver, G.; Chalikian, T. V. Biochemistry 1999, 38, 14999. (4) Hellwich, U.; Schubert, R. Biochem. Pharmacol. 1993, 49, 511. (5) Avdeef, A.; Testa, B. Cell. Mol. Life Sci. 2002, 59, 1681. (6) Rigaud, J. L.; Levy, D. Methods Enzymol. 2003, 372, 65. (7) Waarts, B. L.; Bittman, R.; Wilschut, J. J. Biol. Chem. 2002, 277, 38141. (8) Gutmayer, D.; Thomann, R.; Bakowsky, U.; Schubert, R. Biomacromolecules 2006, 7, 1422. (9) Stauch, O.; Uhlmann, T.; Frohlich, M.; Thomann, R.; El-Badry, M.; Kim, Y. K.; Schubert, R. Biomacromolecules 2002, 3, 324. (10) Op den Kamp, J. A. Annu. Rev. Biochem. 1979, 48, 47.

4142 DOI: 10.1021/la903386d

and the lipid distribution between the two leaflets of the bilayer.10 Lipid asymmetry can influence various biophysical properties of the vesicles such as packing density of the monolayers, membrane permeability, surface charge, membrane potential, and viscoelasticity.11,12 Asymmetrical membrane vesicles (AMV) are of current interest in various fields of membrane research. The potential of AMV for studying the mechanisms of raft formation in the inner and outer monolayers of cell membranes has been previously recognized.13 AMV have been used to show that membrane asymmetry influences the shape and curvature of membranes.14 Soenen et al. successfully investigated the fate of outer and inner monolayer lipids during cellular uptake of asymmetrically labeled magnetoliposomes, and were thus able to distinguish between different potential uptake mechanisms.15 An asymmetric transbilayer lipid distribution has been shown to be present in the human erythrocyte membrane,16,17 the platelet plasma membrane,18 and the Golgi complex.19 Changes in the transbilayer lipid distribution for anionic phosphatidylserine (PS) have been recognized as cell signals involved in fundamental physiological processes such as blood coagulation,20 apoptosis, (11) Haest, C. W.; Oslender, A.; Kamp, D. Biochemistry 1997, 36, 10885. (12) Hill, W. G.; Zeidel, M. L. J. Biol. Chem. 2000, 275, 30176. (13) Simons, K.; Vaz, W. L. C. Annu. Rev. Biophys. Biomol. Struct. 2004, 33, 269. (14) Farge, E.; Devaux, P. F. Biophys. J. 1992, 61, 347. (15) Soenen, S. J.; Vercauteren, D.; Braeckmans, K.; Noppe, W.; De Smedt, S.; De Cuyper, M. ChemBioChem 2009, 10, 257. (16) Bretscher, M. S. Nat. New Biol. 1972, 236, 11. (17) Bretscher, M. S. J. Mol. Biol. 1972, 71, 523. (18) Chap, H. J.; Zwaal, R. F.; van Deenen, L. L. Biochim. Biophys. Acta 1977, 467, 146. (19) van Meer, G.; Stelzer, E. H.; Wijnaendts-van-Resandt, R. W.; Simons, K. J. Cell Biol. 1987, 105, 1623. (20) Zachowski, A. Biochem. J. 1993, 294, 1.

Published on Web 01/22/2010

Langmuir 2010, 26(6), 4142–4151

Holzer et al.

and membrane fusion as well as playing a role in a number of diseases, for example, bleeding disorders, diabetes, or cancer.21 In view of the physiological relevance of anionic PS, it would be desirable to have a membrane model at hand that provides an asymmetric membrane distribution of the anionic membrane component. In particular, such model systems could be used to attain a deeper insight into the role of PS asymmetry in the engulfment of apoptotic cells by macrophages,22 in a number of diseases, or in endocytosis.23 In this Article, we report on our efforts to create vesicles with an asymmetric distribution of egg-phosphatidylglycerol (EPG). EPG was used instead of PS as an anionic model phospholipid, since it has a much higher chemical stability and is more readily available than PS. Lipid asymmetry was generated using proteinmediated lipid transfer24,25 between pure egg-phosphatidylcholine (EPC) vesicles and EPC:EPG 90:10 mol % vesicles. Asymmetry that has been induced in membranes via lipid transfer is diminished by transmembrane lipid diffusion (flipflop). In order to achieve a high degree of asymmetry, EPG transfer was accelerated by adding small amounts of pro-sterol carrier protein 2 (pro-SCP2). Mature SCP2 is also known as nonspecific lipid transfer protein (ns-LTP) and facilitates the transfer of sterols, fatty acids, fatty acyl CoAs, and phospholipids.26 To make this protein conveniently available, we cloned pro-SCP2 from rat liver and expressed it as a His-tagged protein. In 1980, De Cuyper and co-workers introduced the method of free-flow electrophoresis (FFE) for analyzing the transfer of anionic phospholipids between two vesicle populations and for separating acceptors from donors.27,28 FFE is a preparative and analytical electrophoretic procedure which works continuously in the absence of a stationary phase or a solid support material such as a gel. This approach separates charged particles ranging from molecular in size to those with cellular dimensions. We used this technique for monitoring protein-mediated lipid transfer and evaluated its potential to quantify the asymmetry of vesicles by correlating the amount of EPG in the outer monolayer with deflection in FFE. With FFE analysis, it was possible to derive quantitative kinetic data on protein-mediated EPG transfer without the need for introducing fluorescent reporter groups or spin labels which might influence the transfer kinetics. The resulting AMV were quantitatively recovered by FFE separation of donor and acceptor. It has been pointed out that protein-mediated lipid transfer rates depend on the membrane curvature of the vesicles used. Machida and Ohnishi have reported a much faster transfer between sonicated vesicles than that between multilamellar large vesicle preparations,29 without specifying the size of these vesicle populations. We further investigated this issue by comparing the transfer between 50 and 100 nm vesicles of very narrow size distribution. Such vesicle populations were prepared using a procedure that was recently developed by our laboratory.30 We were (21) Zwaal, R. F.; Comfurius, P.; Bevers, E. M. Cell. Mol. Life Sci. 2005, 62, 971. (22) Schlegel, R. A.; Williamson, P. Cell Death Differ. 2001, 8, 551. (23) Devaux, P. F. Biochimie 2000, 82, 497. (24) Rothman, J. E.; Dawidowicz, E. A. Biochemistry 1975, 14, 2809. (25) Pagano, R. E.; Martin, O. C.; Schroit, A. J.; Struck, D. K. Biochemistry 1981, 20, 4920. (26) Gallegos, A. M.; Atshaves, B. P.; Storey, S. M.; Starodub, O.; Petrescu, A. D.; Huang, H.; McIntosh, A. L.; Martin, G. G.; Chao, H.; Kier, A. B.; Schroeder, F. Prog. Lipid Res. 2001, 40, 498. (27) De Cuyper, M.; Joniau, M.; Dangreau, H. Biochem. Biophys. Res. Commun. 1980, 95, 1224. (28) De Cuyper, M.; Joniau, M.; Dangreau, H. Biochemistry 1983, 22, 415. (29) Machida, K.; Ohnishi, S. I. Biochim. Biophys. Acta 1980, 596, 201. (30) Holzer, M.; Barnert, S.; Momm, J.; Schubert, R. J. Chromatogr., A 2009, 1216, 5838.

Langmuir 2010, 26(6), 4142–4151

Article

able to show a positive effect of vesicle curvature on transfer rate in the size range between 50 and 100 nm, and investigated the impact of size distribution width on AMV preparation. As a consequence, four characteristics were shown to be required of the vesicle populations. The ideal acceptor and donor populations should have a very narrow size distribution, be of equal size, and be unilamellar and spherical.

2. Experimental Procedures 2.1. Materials. The phospholipids egg-phosphatidylcholine (EPC) and egg-phosphatidylglycerol (EPG) were a kind gift of Lipoid GmbH (Ludwigshafen, GE). Cholesterol was purchased from Sigma-Aldrich (Steinheim, GE), buffer salts and n-octylβ-D-glucopyranoside (OG) were from Fluka (Buchs, CH), and sodium cholate (SC) was obtained from Carl Roth GmbH (Karlsruhe, GE). All experiments were performed using a filtered (pore size 0.22 μm) 5 mM phosphate buffer consisting of 3 mM Na2HPO4 3 2H2O and 2 mM KH2PO4 and adjusted to pH 7.0. Unless stated otherwise, all additional chemicals were from Carl Roth GmbH and of p.a. quality. 2.2. Unilamellar Acceptor and Donor Populations of Very Narrow Size Distributions. In order to yield essentially unilamellar and spherical vesicles, pure EPC acceptor and EPC: EPG 90:10 mol % donor vesicles were prepared by detergent removal31 using the two detergents SC and OG. The size distribution of the vesicles was minimized using preparative size exclusion chromatography (SEC). For vesicles with 100 nm diameter, initial mixed micelle solutions were prepared as follows: EPC, OG, and SC were mixed in a round bottomed flask from stock solutions containing methanol at a molar ratio of 20:100:3.33 (EPC:OG: SC) in the case of acceptor vesicles. For the donor vesicles, EPC, EPG, and OG were mixed at a molar ratio of 19:1:100 (EPC:EPG: OG). A thin lipid film was produced by gentle evaporation of the solvent using a rotary evaporator and subsequent drying under high vacuum for 1 h. The film was resuspended in 5 mM phosphate buffer to yield a final lipid concentration of 20 mM. Detergent removal was done in custom-made dialysis chambers (7.1 cm2 exchange area), using a 10 kDa cutoff membrane of very high permeability (Dianorm, Munich, GE) against a 200-fold excess of phosphate buffer with respect to the sample volume. The dialysis buffer was exchanged hourly during the first 5 h, and then dialysis was continued against a 500-fold excess of buffer for a total period of 24 h. Preparative SEC was done using a 190 mL packed volume of Sephacryl S 500 HR (GE Healthcare, Uppsala, Sweden) in a column of 1.6 cm diameter and 100 cm length. Briefly, samples were first concentrated by ultracentrifugation (3.5 h, 150 000g, 10 C, Optima LE-80 ultracentrifuge, 50.4 Ti rotor, Beckman Coulter GmbH, Krefeld, GE) to 100 mM total lipid. Afterward, they were applied to the SEC column at a flow rate of 1.0 mL 3 min-1. Fractions of 2.0 mL were collected, and lipid-containing fractions were concentrated by another ultracentrifugation step. The sizes and size distributions of the resulting liposome preparations were assessed by dynamic light scattering (Zetamaster S, Malvern Instruments, Malvern, U.K.) and cryotransmission electron microscopy (cryo-TEM). Total lipid concentration was measured by means of a phosphorus assay.32 All samples were corrected for the buffer phosphate content through the use of a buffer blank value determined in parallel. Besides the 100 nm vesicles, 50 nm vesicles were prepared in the same way but starting with mixed micelles composed of EPC:OG:SC = 20:100: 5.00 (acceptor) and EPC:EPG:OG:SC = 19:1:100:1.25 mol/mol (donor).

2.3. Analysis of the Size and Size Distribution of Acceptor and Donor Vesicles. The successful preparation of 100 or 50 nm vesicle populations was demonstrated by cryo-electron (31) Schubert, R. Method. Enzymol. 2003, 367, 46. (32) Bartlett, G. R. J. Biol. Chem. 1959, 234, 466.

DOI: 10.1021/la903386d

4143

Article

Holzer et al.

Figure 1. Cryo-electron microscopy pictures of liposomes prior to and after optimization of size distribution. Representative pictures showing the size distribution of nonoptimized acceptor vesicles (panel A) with a mean diameter of 76.6 ( 40.3 nm (RSD=52.5%), and of two different SEC fractions obtained thereof. Fractions of very narrow size distribution with mean diameters of 102.3 ( 13.5 nm (RSD=13.2%; panel B) and 53.5 ( 9.0 nm (RSD=16.8%; panel C) could be prepared. Similar results were achieved in the case of donor vesicles (not shown). Data on size and size distribution were obtained by picture analysis (see text). microscopy. Representative pictures of acceptor liposomes are shown in Figure 1. From panel A, it is evident that, prior to the SEC step, the sample consisted of a mixture of smaller and larger particles. A measurement of the particle sizes using the electron microscopy pictures indicated a mean diameter of 76.6 ( 40.3 nm (RSD =52.5%). From this sample, two fractions with vesicles of either 100 nm diameter (panel B) or of 50 nm diameter (panel C) were isolated using SEC. Average mean diameters were 102.3 ( 13.5 nm (RSD = 13.2%) and 53.5 ( 9.0 nm (RSD = 16.8%), respectively. Comparable separation results were achieved in the case of donor vesicles (data not shown). All liposomes were found to be unilamellar and spherical. Further details and data on the preparation of size-distribution-optimized vesicles have been published recently.30 2.4. Liposomes Used for FFE Calibration. EPC:EPG vesicles in varying molar ratios for FFE calibration purposes were prepared by extrusion. A thin lipid film was created by solvent removal from organic solution and redispersed in 5 mM phosphate buffer, yielding a total lipid concentration of 20 mM. The lipid dispersion was then extruded 21 times through a 200 nm and 51 times through a 80 nm polycarbonate membrane (Nuclepore, Whatman, Dassel, GE) using a LiposoFast extruder (Avestin, Ottawa, Canada). Characterization of vesicle size, size distribution, and lipid concentration was as described in section 2.2. By using dynamic light scattering, an average diameter of 97.0 ( 12.6 nm and an average polydispersity index of 0.047 ( 0.018 were determined for these vesicles. 2.5. Determination of Zeta Potential. The zeta potential of the vesicles was determined at 25 C in a Zetamaster S instrument (Malvern Instruments, Malvern, U.K.). Samples were diluted to about 200 μM total lipid using sterile filtered buffer, and 5 mL of the diluted liposome preparation was injected into the measurement cell through a 0.45 μm filter membrane. After allowing the temperature to adjust for 10 min, electrophoretic mobility was measured in the stationary layer of the measurement cell in an electric field of 30 V 3 cm-1 and converted into the zeta potential ζ using the Smoluchowski approximation. Results are mean values of at least 10 runs. 2.6. Free-Flow Electrophoresis. Free-flow electrophoresis (FFE) experiments were carried out using an Octopus PZE apparatus (Weber FFE, Kirchheim, GE) with an upright positioned separation chamber. The chamber walls were not precoated with bovine serum albumin (BSA), in order to avoid protein-liposome interactions.33 Within the separation chamber, a thin laminar film of separation buffer (5 mM phosphate buffer, (33) Dao, H. N.; McIntyre, J. C.; Sleight, R. G. Anal. Biochem. 1991, 196, 46.

4144 DOI: 10.1021/la903386d

pH 7.0) was running upward at a flow rate of 6.6 mL 3 min-1. The separation buffer was flanked by a stabilization buffer which was of a 10-fold higher concentration followed by the electrode wires. The latter were placed in channels through which electrode buffer (125 mM phosphate buffer, pH 7.0) was running and which were separated from the chamber by a non-ion-selective membrane. The higher concentrated buffers in proximity to the electrodes were used to yield a stable and homogeneous electrical field of 70 V 3 cm-1 perpendicular to the running buffers and samples, resulting in an approximate current of 135 mA. All separations were performed with a chamber temperature of 9 C, and 50 μL of liposome sample was continuously injected into the flowing separation buffer at a rate of 15 μL 3 min-1. Injection was done near the cathode (facing fraction 81), and the liposomes were then deflected in the electrical field toward the anode while running through the separation chamber. The dimensions of the separation volume were 500  100  0.50 mm3 (height  width  depth). After passing through the chamber, the buffer flow containing the samples was collected quantitatively in 96 fractions. Liposomecontaining fractions were identified as described below.

2.7. Detection of Liposome-Containing FFE Fractions. A diphenylhexatriene (DPH)/Brij-35-reagent was prepared by mixing 1.0 mL of an 11.5 mM stock solution of polyoxyethylene-23-laurylether (Brij-35, Sigma-Aldrich, Steinheim, GE) in purified water with 230 μL of a 30 mM stock solution of DPH (Fluka, Buchs, CH) in tetrahydrofuran and adding purified water to give a final volume of 100 mL. The dispersion was then treated in an ultrasonic bath for 15 min and used within 1 day (modified after refs 34 and 35). Liposome-containing fractions were identified by mixing 100 μL of each FFE fraction with 100 μL of the reagent. After a 10 min incubation period at room temperature, the fluorescence at 460 nm was measured in a plate reader (FluoroCount, CanberraPackard, Dreieich, GE) using an excitation wavelength of 360 nm.

2.8. Recombinant Expression of His6-Pro-Sterol Carrier Protein 2 (His6-pro-SCP2). Procedures introduced by Ossen-

dorp et al.36 were modified to allow for cloning and expression of rat liver pro-SCP2 (SCP2 including a presequence of 20 amino acids) as a His-tagged protein. Total rat liver cDNA was amplified by polymerase chain reaction (PCR) using the following reaction mixture: 1 μM primer I (50 -ACC-CTC-TAC-CAT-ATG-GGTTTT-CCC-GAA-30 , bold: Nde I restriction site), 1 μM primer II (34) London, E.; Feigenson, G. W. Anal. Biochem. 1978, 88, 203. (35) Caudron, E.; Zhou, J. Y.; Chaminade, P.; Baillet, A.; Prognon, P. J. Chromatogr., A 2005, 1072, 149. (36) Ossendorp, B. C.; Geijtenbeek, T. B.; Wirtz, K. W. FEBS Lett. 1992, 296, 179.

Langmuir 2010, 26(6), 4142–4151

Holzer et al. (50 -CGT-GGA-TTT-CTA-CGG-ATC-CAC-ACA-TCT-30 , bold: Bam H I restriction site), 40 ng of rat liver cDNA, 200 mM of each dNTP, 6 mM MgCl2, 5.0 μL of 10 PCR buffer (200 mM TrisHCl, 500 mM KCl, pH 8.4), and 2.5 U of Taq polymerase (GE Healthcare, Freiburg, GE) in a total volume of 50 μL. The mixture was covered with a layer of 50 μL of silicon oil. The PCR program was as follows: 7 min at 95 C, 5 cycles of 1 min at 95 C, 2 min at 55 C, 1.5 min at 72 C, 25 cycles of 1 min at 95 C, 2 min at 65 C, 1.5 min at 72 C, and finally 10 min at 72 C (Thermocycler, Biozym, Hess. Oldendorf, GE). In an analytical agarose gel, the amplified cDNA sequence was found to have a size of 500 bp, in agreement with the result of Ossendorp et al.36 After isolation of the amplified sequence using a preparative 2% low melting agarose gel and DNA extraction, it was cloned between the Nde I and Bam H I restriction sites of the plasmid vector pET-15b (Novagen, Madison, WI). Besides the cloned pro-SCP2 cDNA, this vector contained a sequence encoding the His6-tag. The construct pET-15bpro-SCP2 was then transformed into the E. coli EB21 strain (Novagen) by electroporation. For expression, the E. coli host strain EB21 containing the pET-15bpro-SCP2 vector was grown at 37 C in 1000 mL of lysogeny broth-medium containing 100 μg 3 mL-1 carbenicillin to an optical density A600 of 0.5. Isopropyl-β-thiogalactopyranoside (Fluka, Buchs, CH) was then added to a final concentration of 1 mM. After 6 h of cultivation, the cells were harvested by centrifugation and the pellet was resuspended in 30 mL of phosphate buffered saline (PBS) pH 7.4 (5.2 mM Na2HPO4 3 2H2O, 1.7 mM KH2PO4, 150 mM NaCl) containing 1 mM phenylmethylsulfonylfluoride (Fluka). After the addition of 50 mg of lysozyme (Fluka), the mixture was kept on ice for 30 min. The cells were then disrupted by ultrasonication, and the insoluble debris was removed by centrifugation. The supernatant was applied directly onto a Ni-NTA column (Qiagen, Hilden, GE) which had been pre-equilibrated with PBS. After a 15 min incubation step at 4 C, the crude protein extract was discarded and the loaded column was washed three times with 15 mL of PBS containing 25 mM imidazole. His6-pro-SCP2 was then eluted with PBS containing 400 mM imidazole. The protein was washed with PBS buffer and concentrated by ultrafiltration (Amicon 8010, Millipore, Eschborn, GE) using 10 kDa cutoff membranes (DiaFlo YM10, Millipore). Protein concentration was determined by a bicinchoninic acid assay (Uptima, Montluc-on, FR) using BSA as a standard.37

2.9. Transfer of the Anionic Phospholipid EPG and Kinetic Analysis. Acceptor (EPC 100 mol %) and donor (EPC: EPG 90:10 mol %) vesicles were mixed at a 1:1 molar ratio to yield a total lipid concentration of 14.0 mM. In the case of proteinmediated lipid transfer, recombinant His6-pro-SCP2 was added at molar protein-to-lipid ratios RP/L specified for the respective experiments. For assessing the spontaneous transfer rate, no protein was added. Incubation was done in a thermomixer (Eppendorf, Hamburg, GE) at 25 C and 700 rpm. Lipid transfer was followed by separating aliquots of the mixture by FFE (see section 2.6) after defined intervals of incubation. The amount of transferred EPG was calculated from the deflection in FFE. Kinetic analysis of the EPG uptake into acceptor membranes was done assuming first order kinetics in which xEPG(t), that is, the EPG content of the outer monolayer at time t, approaches an equilibrium value xEPG(¥). The latter was determined from the FFE deflection of the single peak to which acceptor and donor peaks had united by the late stages of EPG transfer. For initial EPG transfer, the reaction rate was defined as follows with kobs being the observed rate constant. d½xEPG ð¥Þ - xEPG ðtÞ ¼ -k obs ½xEPG ð¥Þ - xEPG ðtÞ dt (37) Smith, P. K.; Krohn, R. I.; Hermanson, G. T.; Mallia, A. K.; Gartner, F. H.; Provenzano, M. D.; Fujimoto, E. K.; Goeke, N. M.; Olson, B. J.; Klenk, D. C. Anal. Biochem. 1985, 150, 76.

Langmuir 2010, 26(6), 4142–4151

Article

Figure 2. Correlation of relative FFE deflection, zeta potential, and EPG content of the outer monolayer. The relative deflection ΔF in FFE was determined for EPC:EPG vesicles with increasing amounts of the negative phospholipid EPG (xEPG). A second order polynomial function was fit to the data points. Mean values of n = 3 independent experiments are displayed, and bars indicate SD. Inset: linear correlation between ΔF and zeta potential ζ (ΔF = -0.740ζ - 6.614; R = 0.9973; N = 6; p < 0.0001). The integrated form of this equation was used for the determination of kobs assuming xEPG(0) = 0 mol %.   xEPG ð¥Þ - xEPG ðtÞ ln ¼ -k obs t xEPG ð¥Þ Plotting of the left term versus time resulted in straight lines for the first 60% of the total transferrable EPG pool. The negative slope of these lines equals kobs. A monoexponential rate equation was used, since protein-mediated transfer occurred much faster than EPG flip-flop. Initial EPG uptake rates v0 (mol % 3 h-1) were determined from the slope of linear fits to the early time points of the xEPG(t) versus time plots.

3. Results 3.1. Correlation of EPG Content, Deflection in FreeFlow Electrophoresis, and Zeta Potential. Free-flow electrophoresis (FFE) was used for the separation of acceptor (EPC 100 mol %) and donor populations (EPC:EPG 90:10 mol %) and for monitoring the transfer of the anionic EPG which is accompanied by changes in the surface charge of both vesicle populations. In order to correlate the amount of EPG in the outer monolayer xEPG with the deflection in FFE, liposomes with a defined amount of between 0 and 10 mol % EPG were prepared by extrusion. The deflection in FFE was determined for each of these preparations, and the results are shown in Figure 2. The number of fractions ΔF between the peak maxima of a pure EPC liposome sample and of the investigated sample was used as the FFE signal. This “relative deflection” ΔF was very repeatable with a RSD of below 7%. The absolute deflection was influenced by a number of parameters that were difficult to control, such as the exact position of the sample injector within the flowing buffer, the exact geometry of the chamber used, and the ratio of the different buffer flow rates, or by the variability of the electroendoosmotic flow. However, the impact of the latter is largely reduced with the FFE device which was used in the present study, since its chamber walls are made of poly(methyl methacrylate) and poly(vinyl chloride). Zeta potential values ζ were also determined for each liposome preparation, and a similar correlation could be found between DOI: 10.1021/la903386d

4145

Article

Holzer et al.

xEPG and ζ. ΔF and ζ showed a linear correlation (see inset of Figure 2). An increasing xEPG was not followed by a linear increase in ΔF or ζ over the entire range of xEPG which was investigated. Whereas with up to 6 mol % EPG an almost linear correlation could be assumed, a marked negative deviation from linearity was observed with higher EPG content. Therefore, the data points were fit using a second order polynomial function. The equation of this function was used to calculate the EPG content of the outer monolayer xEPG from ΔF: ΔF ¼ 5:074xEPG - 0:1817xEPG 2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi xEPG ¼ 13:963 - 194:95 - 5:504ΔF Since only the outer monolayer phospholipid composition contributes to the surface potential, this equation allowed for the direct quantification of EPG asymmetry in the vesicle membrane. The calibrated FFE system made it possible to separate acceptor and donor vesicles in EPG transfer experiments and simultaneously quantify EPG asymmetry. Whereas extruded liposomes were used for FFE calibration, all EPG transfer experiments described in the following sections were performed using liposomes prepared by detergent dialysis to make them essentially unilamellar (see discussion in section 4.2). Although dialysis was performed over an extended time period and with frequent exchanges of the dialysis buffer, a very low amount of detergent might have remained in the membrane. To ensure that residual anionic SC, if present, does not influence the surface potential of acceptor and donor vesicles, we compared their zeta potentials with those of extruded samples (detergentfree). On average, acceptor and donor vesicles prepared by detergent dialysis (see section 2.2) had zeta potentials of -7.7 ( 2.0 mV (n= 36) and -51.0 ( 1.3 mV (n = 28), respectively. With extruded samples, we found an average zeta potential of -7.7 ( 1.5 mV (n = 3) in the case of the acceptor and -52.2 ( 0.4 mV (n = 3) in the case of the donor. Since detergent dialysis and extrusion yielded vesicles with the same zeta potential, the potential of acceptor and donor used in transfer experiments can be measured correctly using an FFE calibration obtained from extruded liposomes. 3.2. Expression and Purification of His6-Pro-SCP2. In order to achieve a maximum degree of asymmetry, lipid transfer must be faster than transmembrane lipid diffusion (flip-flop). It was shown that rat liver pro-sterol carrier protein 2 (pro-SCP2) can be used to accelerate lipid transfer.36 Since the isolation and purification of SCP2 from rat liver homogenate is labor-intensive, we developed an E. coli expression system that allowed for easy expression and purification of pro-SCP2 as a His-tagged protein (see section 2.8). A protein yield of about 7 mg was achieved from a cell culture of 1000 mL, and successful purification could be demonstrated using SDS-PAGE. This resulted in a single band with a molecular weight of 17 kDa, which is close to the theoretical total molecular weight of pro-SCP2, the spacer amino acids, and the His6-tag (17.3 kDa) (ref 36; Novagen pET-15b vector map). For convenience, the His6-pro-SCP2 is further referred to as “pro-SCP2”. 3.3. Impact of Vesicle Curvature on the Pro-SCP2Mediated EPG Transfer Rate. In the first set of experiments, we investigated how EPG transfer is influenced by the size and size distribution of the vesicles used as acceptor and donor populations. Size-distribution-optimized vesicles of 100 and 50 nm diameter were compared, and the transfer protein pro-SCP2 was 4146 DOI: 10.1021/la903386d

Table 1. Pro-SCP2-Mediated Transfer of EPG between Size-Optimized Vesicles of Different Diametersa outer monolayer EPG content xEPG (mol %) 100 nm vesicles incubation time (h)

acceptor

donor

50 nm vesicles acceptor

donor

0.47 ( 0.12 8.7 ( 0.82 0.67 ( 0.12 8.9 ( 0.31 0.0 0.5 0.88 ( 0.24 8.7 ( 0.82 1.0 ( 0.21 7.9 ( 0.26 1.0 1.2 ( 0.12 8.2 ( 0.75 1.5 ( 0.22 7.2 ( 0.41 1.5 1.5 ( 0.13 7.6 ( 0.43 2.5 ( 0.27 5.9 ( 0.20 2.0 2.0 ( 0.16 7.0 ( 0.28 2.9 ( 0.24 5.1 ( 0.18 3.0 3.3 ( 0.39 5.6 ( 0.19 4.6 ( 0.33 4.6 ( 0.33 4.0 4.8 ( 0.00 5.0 ( 0.16 4.6 ( 0.33 4.6 ( 0.33 6.0 4.6 ( 0.33 4.6 ( 0.33 4.5 ( 0.29 4.5 ( 0.29 a Total lipid concentration = 14.0 mM, molar pro-SCP2-to-lipid ratio = 15.0  10-5, and T = 25 C. Mean value and SD of n = 3 independent experiments. b At these high transfer rates, the deviation of xEPG from 0 mol % at t=0.0 h was probably observed due to transfer during the time needed to pump the aliquot into the FFE separation chamber, where transfer was stopped. b

Table 2. Kinetic Data of EPG Uptake in Acceptor Vesicles Using Different Molar Pro-SCP2-to-lipid Ratios RP/L and Different Vesicle Sizesa RP/L (10-5)

d (nm)

kobs (h-1)

τ1/2 (h)

v0 (mol % 3 h-1)

9.71 100 0.098 ( 0.0026 7.1 ( 0.19 0.34 ( 0.0091 15.0 100 0.22 ( 0.018 3.2 ( 0.27 0.75 ( 0.088 15.0 50 0.42 ( 0.058 1.7 ( 0.23 1.2 ( 0.12 19.5 100 0.34 ( 0.014 2.0 ( 0.08 1.4 ( 0.14 a EPG was transferred from donor vesicles with an initial EPG content of 10 mol % to pure EPC acceptors. Vesicle populations of very narrow size distributions and mean diameters d were used in a 1:1 molar ratio at a total lipid concentration of 14.0 mM, T=25 C. The first order observed rate constant kobs, half-time τ1/2, and initial transfer rate v0 were determined as described in the text. Fit quality: R=0.972-0.998, N = 5-7 data points, p < 0.01, n = 3 independent experiments.

added at a protein-to-lipid ratio RP/L of 15.0  10-5. The changes in the outer monolayer EPG content of acceptor and donor vesicles during transfer are shown in Table 1. With the smaller vesicles, transfer equilibrium was reached after 3 h, whereas over 4 h were needed with the 100 nm populations. Transfer clearly occurred faster between the smaller vesicles and thus depended on vesicle diameter. Quantitative kinetic data for the uptake of EPG in 100 nm compared to 50 nm vesicles were derived from the plots shown in Figure 5 (open and filled triangles). The results are shown in Table 2. A 93.5% increase in the observed rate constant kobs and a 55.7% increase in the initial transfer rate v0 were determined with the 50 nm vesicles as compared to the 100 nm populations. In view of the fact that transfer rate was found to increase with membrane curvature, the use of vesicle populations with very narrow size distributions seems to be advantageous. These were the types of populations which were used throughout this study. To demonstrate the importance of this aspect, protein-mediated transfer using a RP/L of 9.71  10-5 was followed by FFE between (a) vesicle populations that were not size-distribution-optimized by SEC (panel A in Figure 1) and (b) 100 nm vesicles of very narrow size distribution (panel B in Figure 1). The resulting FFE profiles are shown in Figure 3. With the vesicles that had a broad size distribution, a substantial peak broadening and an increase in the baseline between the two peaks occurred during EPG transfer (panel A). Under these conditions, the separation of acceptor and donor peaks became difficult at late stages of the transfer process. Clear FFE profiles with only minor broadening of the peaks and Langmuir 2010, 26(6), 4142–4151

Holzer et al.

Article

Figure 3. FFE profiles of lipid transfer between vesicles of different size distributions. Protein-mediated EPG transfer between (A) liposomes of broad size distribution (see panel A of Figure 1) and (B) liposomes of very narrow size distribution and 100 nm mean diameter (see panel B of Figure 1). Total lipid concentration was 14.0 mM, molar pro-SCP2-to-lipid ratio=9.71  10-5, and T=25 C. Numbers on peaks indicate the fraction with the highest fluorescence intensity. Position of anode and cathode was as indicated at the bottom. Langmuir 2010, 26(6), 4142–4151

DOI: 10.1021/la903386d

4147

Article

Holzer et al.

Figure 4. Spontaneous and pro-SCP2-mediated EPG transfer analyzed by FFE. Changes in the outer monolayer EPG content (xEPG) of acceptor (b) and donor vesicles (O) during spontaneous (A) and protein-mediated lipid transfer using a molar pro-SCP2-to-lipid ratio of 9.71  10-5 (B), 15.0  10-5 (C), and 19.5  10-5 (D). The two populations were separated by FFE at each time point, and the outer monolayer EPG content was then calculated from the deflection. Total lipid concentration was 14.0 mM, T=25 C. The size distribution of the vesicles was optimized using SEC, and the mean diameter was 100 nm. n = 3; bars indicate SD. Note the different x-axis scales.

increases in the baseline resulted in the case of the size-distributionoptimized 100 nm vesicles (panel B). In addition, transfer occurred faster with the inhomogeneous sample, which consisted to a large extent (53%) of vesicles e50 nm in diameter. Since we have demonstrated that transfer occurs faster between smaller vesicles, the observed increase in baseline and broadening of the peaks in FFE profiles is most likely due to the generation of vesicle species with intermediate surface potential and hence EPG content during transfer. With a vesicle preparation which is inhomogeneous, slower and faster EPG transfer processes occur simultaneously and create subpopulations of different EPG content and degree of asymmetry. Such vesicle populations are therefore less suitable for the preparation of well-defined AMV. All further experiments were done using size-distribution-optimized vesicles of 100 nm diameter. 3.4. Acceleration of EPG Transfer by Pro-SCP2 and Kinetic Analysis. The rate of EPG transfer between sizedistribution-optimized vesicles of 100 nm diameter was determined using different concentrations of pro-SCP2. The changes in xEPG of acceptor and donor vesicles during transfer are shown in Figure 4. The spontaneous (i.e., nonmediated) transfer of EPG between an equimolar mixture of acceptor and donor vesicles (panel A) was found to be a very slow process. Within a period of 8 h, the acceptor EPG content increased to only 0.47 ( 0.12 mol %. 4148 DOI: 10.1021/la903386d

Under our experimental conditions, an equilibrium EPG content of about 5 mol % was expected. Equilibrium was not reached within the time frame of the experiment. Even after 9 days (216 h) of incubation, the EPG content of the acceptor outer monolayer was never higher than about 2 mol %. EPG transfer was extremely accelerated with the addition of the recombinant pro-SCP2. This demonstrated that the transfer activity of the recombinant pro-SCP2 had not been abolished by the attached His6-tag. Whereas with a molar pro-SCP2-to-lipid ratio of RP/L =9.71  10-5 equilibrium was reached after between 20 and 26 h (panel B), only 3-4 h or 2-3 h were needed with RP/L =15.0  10-5 or 19.5  10-5, respectively (panels C and D). It should be noted that the time courses for EPG uptake into acceptor membranes and EPG release from donor membranes show that these two processes occurred at different rates. Particularly during the early time points when only small amounts of EPG had been transferred (panels B and C), a certain “lag-phase” was observed in the changes of donor EPG content, followed by faster release of EPG. This was not observed for the uptake of EPG by the acceptor, which initially happened linearly with time. Since there is no reason to doubt that all EPG that entered the acceptor in the early time points had been released from the donor, we attribute the apparent “lag-phase” to the polynomial correlation between xEPG and ΔF (see Discussion section). Langmuir 2010, 26(6), 4142–4151

Holzer et al.

Figure 5. Kinetic analysis of acceptor EPG uptake. EPG uptake into size-distribution-optimized 100 nm vesicles was fit to a first order equilibrium reaction. xEPG(t) and xEPG(¥) denote the outer monolayer EPG content at time t and after transfer has reached equilibrium, respectively. The observed rate constant kobs equals the negative slope of the lines. Total lipid concentration was 14.0 mM, T =25 C. Molar pro-SCP2-to-lipid ratios RP/L: (0) 9.71  10-5, (4) 15.0  10-5, and (O) 19.5  10-5. The influence of vesicle diameter on EPG transfer is shown using size-distributionoptimized vesicles with (4) 100 or (2) 50 nm diameter (RP/L = 15.0  10-5). n = 3; bars indicate SD.

Article

During EPG transfer, the surface potentials of donor and acceptor became more and more similar, and peaks which were initially separated moved toward each other. Membrane asymmetry is induced in both vesicle populations, since it has been shown in a number of studies that lipid transfer exclusively changes the lipid composition of the outer monolayer.24,40 At equilibrium, acceptor and donor had the same outer monolayer EPG content. If transfer occurred fast enough to minimize the influence of flip-flop, the highest degree of asymmetry was achieved at this point. However, in equilibrium, the two populations could no longer be separated. In order to recover acceptor and donor vesicles with a high degree of asymmetry, the two populations had to be separated directly before equilibrium was reached. The time courses in Figure 4 were used to determine these time points for each of the pro-SCP2-to-lipid ratios RP/L investigated. With RP/L of 15.0  10-5 and 100 nm populations, the latest possible separation was after 3 h of incubation and the resulting outer monolayer EPG contents were 3.3 ( 0.4 mol % in the acceptor vesicles and 5.6 ( 0.2 mol % in the donor vesicles. Using RP/L of 19.5  10-5, vesicles had to be separated after 2 h of incubation. Here, 3.0 ( 0.5 mol % and 5.9 ( 0.5 mol % were found in the outer monolayers of the acceptor and donor membrane, respectively. Using FFE, these AMV could be recovered quantitatively by pooling the respective peak fractions and were then available for further experiments.

4. Discussion Quantitative kinetic data of pro-SCP2-mediated EPG transfer were derived as described in section 2.9, assuming a first order equilibrium reaction. The late time points in the xEPG(t) versus time plots (Figure 4) were used to determine EPG content in the outer monolayer at equilibrium xEPG(¥). Linear plots of ln[(xEPG(¥) - xEPG(t))/xEPG(¥)] versus time were obtained as long as xEPG did not exceed 3.0 mol % (corresponding to 60% of the total transferrable EPG pool), and they are shown in Figure 5. The observed rate constants kobs and the half-times of initial EPG transfer τ1/2 = ln2/kobs were then calculated from these plots. The initial EPG transfer rate v0 was determined from linear fits to the xEPG(t) versus time plots, again limited to xEPG e 3.0 mol %. The results are summarized in Table 2. 3.5. Separation of Asymmetrical Acceptor and Donor Vesicles by FFE. The kinetic data of pro-SCP2-mediated EPG transfer were used to optimize the recovery of asymmetrical vesicle populations after FFE separation. This is demonstrated with the FFE profiles from the transfer experiments between sizedistribution-optimized 100 nm vesicles using a molar pro-SCP2to-lipid ratio of 9.71  10-5 (panel B in Figure 3). Initially, donor and acceptor vesicles were well separated by about 30 fractions (t = 0 h), with the donor being deflected more toward the anode which is located on the left-hand side of the diagrams. The neutral EPC vesicles were deflected over a distance of approximately 8-9 fractions in agreement with their zeta potential of -7.7 mV. This may be due to the adsorption of buffer phosphate ions onto the liposome surface, thus inducing some negative surface charge,38 or due to the orientation of the PC headgroups. At 25 C and with the low ionic strength of the separation buffer, the negatively charged phosphatidyl moiety of the headgroup is located close to the lipid-water interface as shown by Makino et al.39 (38) Dern, T. Ph.D. Thesis, Department of Pharmaceutical Technology and Biopharmacy, University of Freiburg, 2002. (39) Makino, K.; Yamada, T.; Kimura, M.; Oka, T.; Ohshima, H.; Kondo, T. Biophys. Chem. 1991, 41, 175.

Langmuir 2010, 26(6), 4142–4151

We investigated the generation of AMV by pro-SCP2mediated lipid transfer of the anionic phospholipid EPG between two vesicle populations and established the conditions for optimal results. Our study focused on (i) the establishment of a method for the fast and easy separation of donor and acceptor vesicles and for the quantification of asymmetry; (ii) the optimal characteristics for the vesicle populations used in transfer experiments, in particular concerning their size distribution; and (iii) the acceleration of lipid transfer through the use of pro-SCP2 to minimize the influence of transmembrane lipid diffusion (flip-flop). In addition, our approach made it possible to present quantitative kinetic data for protein-mediated transfer of unmodified EPG. It was not necessary to chemically modify EPG as is the case when transfer is followed using fluorescence or electron paramagnetic resonance measurements. The introduced reporter groups are likely to influence lipid transfer and flip-flop kinetics.41,42 4.1. Relative Deflection in FFE Is a Measure of the EPG Content of the Outer Monolayer. We evaluated the use of FFE for separating differently charged acceptor and donor vesicles and for quantifying the amount of EPG in the outer monolayer of the membrane. This method offers an alternative to the classical separation of neutral and charged vesicles by ion exchange chromatography.43 FFE features a comparatively high separation speed ( PC > PG > phosphatidylethanolamine.50 In our experimental setup, both EPC and EPG are therefore transferred between donor and acceptor. In this way, pro-SCP2 mediates a net transfer of EPG along the concentration gradient, as has been pointed out by Wirtz and Gadella.51 Frolov et al. investigated the interaction of SCP2 with fatty acids and fatty acyl CoAs of different chain length and number of double bonds using a displacement assay.52 From their results, it can be assumed that pro-SCP2 has a certain specificity for phospholipid species with unsaturated fatty acyl chains. The two phospholipid classes used in the present study have identical fatty acyl compositions, since the supplier prepares EPG from EPC by transphosphorylation. The predominant fatty acyl chains are C16:0 (30-33%), and C18:1 (30-38%) along with C18:0 (10-15%) and C18:2 (12-18%). The main species is the C16:0/ C18:1-substituted phospholipid (product information, Lipoid GmbH, Ludwigshafen, GE). It is likely that the unsaturated main species of EPC and EPG are preferentially transferred between donor and acceptor. Due to the identical composition of their acyl chains, this does not imply a preferred transfer of one of the two phospholipid classes. Without the presence of the transfer protein, EPG moved very slowly into the acceptor vesicles. Using pro-SCP2 at a molar protein-to-lipid ratio of (15-20)  10-5, transfer equilibrium was reached after between 3 and 4 h. The uptake of EPG in acceptor vesicles was analyzed based on a first order equilibrium reaction. Since equimolar mixtures of acceptor and donor vesicles were used, the back reaction of EPG could only be neglected at initial stages of transfer. Therefore, the transfer followed first order kinetics for only the first 60% of the transfer reaction. Depending on protein concentration, half-times of EPG transfer between 7.1 and 2.0 h were determined. Our analytical approach allowed for the determination of kinetic data using unmodified EPG (see Table 2), which are presented for the first time in this paper. The estimated half-times were compared with flip-flop half-times from the literature in order to estimate whether flip-flop impacts EPG membrane asymmetry during the transfer process. For phospholipids, flip-flop half-times between hours and days are reported. Homan et al. investigated the flip-flop of pyrenelabeled PG with acyl chain lengths of 8 and 12 carbon atoms at 37 C.53 A half-time of 69 h was determined for both chain lengths. From their experiments with PG thio-analogues, Ganong and Bell estimated a flip-flop half-time of over 8 days at room temperature.54 Regarding this range of flip-flop half-times, it is likely that with spontaneous EPG transfer the established membrane asymmetry (50) Nichols, J. W.; Pagano, R. E. J. Biol. Chem. 1983, 258, 5368. (51) Wirtz, K. W.; Gadella, T. W., Jr. Experientia 1990, 46, 592. (52) Frolov, A.; Cho, T.; Billheimer, J. T.; Schroeder, F. J. Biol. Chem. 1996, 271, 31878. (53) Homan, R.; Pownall, H. J. Biochim. Biophys. Acta 1988, 938, 155. (54) Ganong, B. R.; Bell, R. M. Biochemistry 1984, 23, 4977.

Langmuir 2010, 26(6), 4142–4151

Article

is influenced by flip-flop. We demonstrated that transfer halftimes using 100 nm vesicle populations can be reduced to less than 3 h when a molar pro-SCP2-to-lipid ratio of between 15  10-5 and 20  10-5 is used. Compared to the reported flip-flop half-times, this should be sufficient to avoid a significant transbilayer redistribution of EPG in the time frame of the transfer process. Preliminary results from our laboratory in which the zeta potential of liposomes with asymmetrical EPG distribution was measured also strongly suggest that flip-flop is negligible during the time frame of the transfer experiments. Under our conditions, the sum of the acceptor and donor xEPG should amount to 10 mol %. However, a negative deviation from this theoretical value was determined in some experiments. For example, we found a total acceptor and donor xEPG of 9.1 ( 0.46 mol % on average with the transfer experiment between 100 nm vesicles using RP/L of 9.71  10-5. In principle, movement of EPG to the inner monolayer would be a possible explanation, since EPG is not detectable in the inner monolayer with FFE. However, this appears very unlikely when comparing the halftimes of protein-mediated transfer and flip-flop. The observed loss of 10% of the total EPG can also hardly be explained by binding of EPG to pro-SCP2, since, with the highest RP/L used in these experiments, 1 mL contained approximately 4.2  1017 EPG molecules compared to only 1.6  1015 pro-SCP2 molecules. Therefore, the reason for this deviation remained unclear.

5. Conclusions and Outlook Using acceptor and donor populations that meet the requirements made on morphology and size distribution width, it was possible to prepare AMV that provide a higher EPG concentration on the outer monolayer (acceptor) or on the inner monolayer (donor). The degree of asymmetry was analyzed using the relative deflection in FFE measured during the separation of acceptor and donor populations. The transfer process was accelerated using recombinant pro-SCP2 to obtain half-times of a few hours. In this time frame, the impact of EPG flip-flop, which occurs with a halftime of tens of hours, on asymmetry can be excluded. After recovery through FFE separation, AMV were ready-to-use for further experiments. This asymmetrical membrane model can now be used to investigate the impact of EPG membrane asymmetry on cellular uptake, vesicle-vesicle interactions, and the release of drug substances. In further studies, we will analyze the flip-flop of unmodified EPG using this model system and apply our approach to other negatively charged phospholipids, in particular phosphatidylserine. The reconstitution of proteins in asymmetrical liposomes for studying the influence of asymmetry on membrane proteins is another promising direction for research in the future which develops from these results. Acknowledgment. We thank Lipoid GmbH, Ludwigshafen for the generous gift of lipids. We are grateful to the SFB 428 for financial support, to Sabine Barnert for preparing the cryo-TEM pictures, and to Marie Follo for reading the manuscript. Supporting Information Available: Cryo-TEM pictures of extruded liposomes used for FFE calibration (Figure S1). This material is available free of charge via the Internet at http://pubs.acs.org.

DOI: 10.1021/la903386d

4151