11726
Langmuir 2007, 23, 11726-11733
Membrane Curvature Effects on Glycolipid Transfer Protein Activity Matts Nylund,† Christina Fortelius,† Elina K. Palonen,† Julian G. Molotkovsky,‡ and Peter Mattjus*,† Department of Biochemistry and Pharmacy, Åbo Akademi UniVersity, Turku, Finland, and The Shemyakin-OVchinnikoV Institute for Bioorganic Chemistry, Russian Academy of Sciences, Moscow, Russia ReceiVed June 28, 2007. In Final Form: August 27, 2007 The glycolipid transfer protein (GLTP) is monomeric in aqueous solutions, and it binds weakly to membrane interfaces with or without glycolipids. GLTP is a surface-active protein and adsorbs to exert a maximal surface pressure value of 19 mN/m. The change in surface pressure following GLTP adsorption decreased linearly with initial surface pressure. The exclusion pressure for different phospholipids and sphingolipids was between 23 and 31 mN/m, being clearly highest for the negatively charged dipalmitoyl-phosphatidylserine. This can be explained by electrostatic forces when GLTP is positively charged at neutral pH (isoelectric point ) 9.0) and by phosphatidylserine being negatively charged. If GLTP is injected under a palmitoyl-galactosylceramide monolayer above 30 mN/m, the presence of GLTP leads to a decrease in the surface pressure as a function of time. This suggests that GLTP is able to remove glycolipids from the monolayer without penetrating the monolayer. On the other hand, if phospholipid vesicles with or without glycolipids are also present in the subphase, no change in the surface pressure takes place. This suggests that GLTP in the presence of curved membranes is not able to transfer from or to planar membranes. We also show that transfer of fluorescently labeled galactosylceramide is faster from small highly curved palmitoyl-oleoyl-phosphatidylcholine and dipalmitoyl-phosphatidylcholine bilayer vesicles but not from palmitoyl-sphingomyelin vesicles regardless of the size.
Introduction Lipid transfer proteins possess the unique feature of bringing together interfacial and membrane related events, with the features of protein behavior in solution. Glycolipid transfer proteins (GLTPs, 24 kD, isoelectric point ) 9.0) specifically transfer glycolipids with β-linked sugars, including neutral glycosphingolipids and gangliosides between bilayer vesicles in vitro.1 GLTP activity has been detected in a wide variety of cell and tissue types, including mammalian spleen,2 brain,3 and liver.4 Recent findings show that the characteristics of bovine and human GLTPs are different from other known lipid transfer proteins5 and from other proteins that interact with sugars, such as sphingolipid activator proteins6 and lectins.7 Bovine GLTPs have been crystallized,8 and the crystal structure has been solved.5 Bovine GLTP with the bound glycosphingolipid resembles human GLTP that folds with a previously unknown two-layer all-Rhelical topology.9 The glycosphingolipid is anchored with the sugar-amide head group to the GLTP recognition center by hydrogen-bonding networks and hydrophobic contacts. Both acyl chains are encapsulated in a hydrophobic cavity.5,9 How GLTP interacts with membranes and what part of the protein that seems to recognize membrane interfaces is still unclear, as well as the * To whom correspondence should be addressed. Telephone: +35825154271. Fax: +35822154745. E-mail:
[email protected]. † Åbo Akademi University. ‡ Russian Academy of Sciences. (1) Yamada, K.; Abe, A.; Sasaki, T. Biochim. Biophys. Acta 1986, 879, 345349. (2) Metz, R. J.; Radin, N. S. J. Biol. Chem. 1980, 255, 4463-4467. (3) Abe, A.; Yamada, K.; Sasaki, T. Biochem. Biophys. Res. Commun. 1982, 104, 1386-1393. (4) Yamada, K.; Sasaki, T. J. Biochem. 1982, 92, 457-464. (5) Airenne, T. T.; Kidron, H.; Nymalm, Y.; Nylund, M.; West, G.; Mattjus, P.; Salminen, T. A. J. Mol. Biol. 2006, 355, 224-236. (6) Kolter, T.; Sandhoff, K. Annu. ReV. Cell DeV. Biol. 2005, 21, 81-103. (7) Ambrosi, M.; Cameron, N. R.; Davis, B. G. Org. Biomol. Chem. 2005, 3, 1593-1608. (8) West, G.; Nymalm, Y.; Airenne, T. T.; Kidron, H.; Mattjus, P.; Salminen, T. T. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2004, 60, 703-705. (9) Malinina, L.; Malakhova, M. L.; Teplov, A.; Brown, R. E.; Patel, D. J. Nature 2004, 430, 1048-1053.
potential structural changes taking place during the substrate recognition and binding events. Previously, bovine brain GLTP has been used as a tool to study the mixing properties of galactosylceramide (GalCer) in different lipid bilayer vesicle matrixes.10-14 GLTP has proven to be a sensitive tool that can be used to probe membrane interfaces regarding glycosphingolipid (GSL) mixing and their lateral organization. Langmuir surface monolayer studies at air/buffer interfaces provide data from a macroscopic model system for comparing the probable behavior of lipid transfer proteins at the membrane interface. A key benefit of the system is that the surface area, the lipid packing density and composition, as well as the protein concentration can accurately be controlled. The interaction of the protein molecule with a lipid monolayer is influenced by the physical state of the monolayer and the structure of the protein. The lipid packing density, that is, monolayer surface pressure, is a crucial factor, and, at sufficiently high pressure, the penetration of protein molecules into a lipid monolayer is prevented.15 By varying the vesicle preparation procedure, vesicles of different sizes can be produced. The curvature of these vesicles will differ according to their size. The smaller the vesicle, the greater the bending and the curvature stress will be. It has been shown that the spontaneous transfer of cholesterol is about 10 times faster from small unilamellar vesicles (SUVs) than from large unilamellar vesicles (LUVs).16,17 In biological membranes, the curvature usually is very low. However, the membranes of (10) Mattjus, P.; Kline, A.; Pike, H. M.; Molotkovsky, J. G.; Brown, R. E. Biochemistry 2002, 41, 266-273. (11) Mattjus, P.; Malewicz, B.; Valiyaveettil, J. T.; Baumann, W. J.; Bittman, R.; Brown, R. E. J. Biol. Chem. 2002, 277, 19476-19481. (12) Mattjus, P.; Pike, H. M.; Molotkovsky, J. G.; Brown, R. E. Biochemistry 2000, 39, 1067-1075. (13) Nylund, M.; Mattjus, P. Biochim. Biophys. Acta 2005, 1669, 87-94. (14) Nylund, M.; Kjellberg, M. A.; Molotkovsky, J. G.; Byun, H. S.; Bittman, R.; Mattjus, P. Biochim. Biophys. Acta 2006, 1758, 807-812. (15) Colacicco, G. Lipids 1970, 5, 636-649. (16) McLean, L. R.; Phillips, M. C. Biochemistry 1981, 20, 2893-2900. (17) McLean, L. R.; Phillips, M. C. Biochim. Biophys. Acta 1984, 776, 21-26.
10.1021/la701927u CCC: $37.00 © 2007 American Chemical Society Published on Web 10/04/2007
Membrane CurVature Effects on GLTP ActiVity
intracellular organelles have greater curvature, and recently there has also been much attention drawn to the Bin/amphiphysin/Rvs (BAR) domains and endophilins that sense and generate local membrane curvature.18,19 To further understand the behavior of GLTP, it is vital to know how the protein behaves and interacts with lipids at the air/water interface. Experimental Procedures Chemicals. 1-Palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC), 1,2-dipalmitoyl-sn-glycero-3-phosphocholine (DPPC), 1,2dipalmitoyl-sn-glycero-3-phosphoserine (DPPS), 1,2-dipalmitoylsn-glycero-3-phosphoethanolamine (DPPE), d-galactosyl-β,1-1′-Derythro-sphingosine (lyso-GalCer), N-oleoyl-D-erythro-sphingosylphosphorylcholine (OSM), and egg sphingomyelin (SM) were all from Avanti Polar Lipids (Alabaster, AL). Myristic acid, palmitic acid, and stearic acid were from Larodan Fine Chemicals (Malmo¨, Sweden), and bovine sulfatide was from SUPELCO (Bellefonte, PA). N-Palmitoyl-D-erythro-sphingosylphosphorylcholine (PSM) was purified from egg SM using reverse-phase high-performance liquid chromatography (HPLC) (Discovery C18 (Supelco, Bellefonte, PA), 5 µm particle size, 250 mm × 21.2 mm column dimensions) at 9 mL/min with methanol as the mobile phase. N-Palmitoyl-Derythro-dihydro-SM (PDHSM) was prepared from PSM by hydrogenation in ethanol using palladium as the catalyst.20 d-Galactosylβ,1-1′-N-myristoyl-D-erythro-sphingosine (MGalCer), d-galactosylβ,1-1′-N-palmitoyl-D-erythro-sphingosine (PGalCer), and d-galactosylβ,1-1′-N-stearoyl-D-erythro-sphingosine (SGalCer) were synthesized from lyso-GalCer and the respective fatty acids as described recently,21 based on earlier work,22 and purified with HPLC as described above for PSM. The fluorescent probe N-[(11E)-12-(9anthryl)-11-dodecenoyl]-1-O-β-galactosylsphingosine (AV-GalCer) was prepared as described earlier,23 DiO-C16 (3,3′-dihexadecyloxacarbocyanine perchlorate) was from Invitrogen (Carlsbad, CA), and Triton X-100 was from MP Biomedicals (Solon, OH), whereas dicyclohexylcarbodiimide, triethylamine, and palladium all were from Fluka (Buchs, Switzerland). The concentration of the different phospholipids was determined by the Bartlett method,24 and the glycolipids and probes were determined gravimetrically (MT5, Mettler-Toledo, Columbus, OH). Recombinant bovine GLTP was expressed and purified as described earlier.8 Ultrapure HPLC grade water was used for the buffers prepared for all monolayer experiments (Riedel-de Ha¨en, Seelze, Germany). Fluorescence Anisotropy of Anthryl-Vinyl-Galactosylceramide. Fluorescence anisotropy of anthrylvinyl labeled galactosylceramide was carried out on a PTI QuantaMaster 1 spectrofluorimeter (Photon Technology International, Lawrenceville, NJ) that operates in the T-format. The samples (probe sonicated vesicles) were excited at 370 nm, and the emission was recorded at 425 nm, with all the slits set to 6 nm. The scan (from 20 to 90 °C) was done in a quartz cuvette under constant stirring, and the temperature was controlled by a Peltier element. Protein Adsorption to Interfaces. All troughs and monolayer equipment were from KSV Instruments Ltd. (Helsinki, Finland). The measurements of the adsorption of protein to the air/buffer interface was done in a cylindrical Teflon trough (surface area ) 10.2 cm2, volume ) 16.0 mL) filled with a sodium phosphate buffer consisting of 10 mM sodium dihydrogen phosphate, 1 mM dithiothreitol, and 1 mM ethylenediaminetetraacetate (EDTA) at pH 7.4. The surface pressure was monitored with a Wilhelmy-type (18) Farsad, K.; Ringstad, N.; Takei, K.; Floyd, S. R.; Rose, K.; De Camilli, P. J. Cell Biol. 2001, 155, 193-200. (19) Peter, B. J.; Kent, H. M.; Mills, I. G.; Vallis, Y.; Butler, P. J.; Evans, P. R.; McMahon, H. T. Science 2004, 303, 495-499. (20) Schneider, P. B.; Kennedy, E. P. J. Lipid Res. 1967, 8, 202-209. (21) Bjo¨rkqvist, Y. J.; Nyholm, T. K.; Slotte, J. P.; Ramstedt, B. Biophys. J. 2005, 88, 4054-4063. (22) Cohen, R.; Barenholz, Y.; Gatt, S.; Dagan, A. Chem. Phys. Lipids 1984, 35, 371-384. (23) Molotkovsky, J. G.; Mikhalyov, I. I.; Imbs, A. B.; Bergelson, L. D. Chem. Phys. Lipids 1991, 58, 199-212. (24) Bartlett, G. R. J. Biol. Chem. 1959, 234, 466-468.
Langmuir, Vol. 23, No. 23, 2007 11727 platinum plate system (KSV Instruments Ltd, Helsinki, Finland). When the measurements where done with lipid monolayers, they were preformed by spreading a lipid solution in hexane/2-propanol (3:2, v/v) with a Hamilton syringe on an air/buffer surface. The solvent was allowed to evaporate, and the surface pressure was allowed to stabilize for several minutes, after which a solution of GLTP was added to the subphase of the reaction chamber to a final concentration of 95 nM. The penetration of GLTP to the lipid monolayer could then be monitored as a function of time. A more thorough description of this type of monolayer setup is described by Verger and de Haas.25 In Figure 3, where the activity of untreated GLTP was compared to GLTP from the interface and from the subphase, the protein concentration was verified with the Lowry method26 and 2 µg of GLTP was used for each experiment. In the experiments for the results in Figure 5, the lipid monolayer was spread as described above but to surface pressures of 34 to 37 mN/ m, and the glycolipid transfer from the monolayer was detected. The experiments done for Figure 6 were similar but with a 10-fold excess (compared to the monolayer) of POPC vesicles in the subphase when the monolayer was of PGalCer and with PGalCer/POPC 1:9 when the monolayer was of DPPC. Vesicle Preparation. The vesicles used in this study where either probe sonicated27 or extruded.28 In measurements where the size of the donor vesicles was varied, the acceptor vesicles were always probe sonicated and the donor vesicles were probe sonicated or extruded with a Lipex bench extruder (Northern Lipids Inc., Vancouver, Canada) through 50 , 100 , 200, 400, and 800 nm polycarbonate filters (Whatman, Kent, U.K.). On the contrary, when the acceptor vesicles were varied, the donor vesicles were always probe sonicated. In the results presented in Figure 9, both vesicle populations were varied accordingly. The donor vesicles consisted of POPC, DPPC, or PSM with 1% AV-GalCer and 3% DiO-C16. The lipids and the probes were mixed from stock solutions in hexane/ 2-propanol (3:2 v/v) and dried under nitrogen. They were redissolved in a sodium phosphate buffer (pH 7.4) containing 10 mM sodium dihydrogen phosphate, 1 mM dithiothreitol, and 1 mM EDTA. The suspension with a total concentration of 0.4 mM was sonicated for 10 min (donor vesicles) or 15 min (acceptor vesicles) on ice with a Branson 250 sonifier, and then centrifuged for 15 min at 15 000 rcf to remove the titanium probe particles and undispersed lipid (negligible amount). The final concentration of the donor vesicles per assay was 13.3 µM, and the final AV-GalCer concentration in each assay was 0.13 µM. The acceptor vesicles were used in a 10-fold excess. The vesicles used in the anisotropy experiments were made in a similar way except that the total concentration was 0.2 mM and the probe sonication was carried out for 6 min and the buffer excluded EDTA and DTT. Dynamic Light Scattering. Dynamic light scattering was done to determine the vesicle sizes. The measurements were carried out on a Malvern Zetasizer NanoS instrument equipped with a 633 nm He-Ne laser light source (Malvern Instruments, Worcestershire, U.K.). The vesicles were placed in a 10 mm plastic cuvette in the same buffer used for the transfer kinetic measurements. The temperature was set to 37 °C. The possible size value effect of 140 mM NaCl was checked, and it did not significantly differ from that of the measurements carried out without NaCl. Glycolipid Transfer Assay. The fluorescence method used for measuring the transfer of the glycolipid between two bilayer vesicle populations has been thoroughly described previously,13 and it is based on the method described earlier.29 Briefly, AV-GalCer transfer from probe sonicated donor vesicles consisting of 1% AV-GalCer, 3% DiO-C16 (nontransferable quencher), and 96% phosphatidyl(25) Verger, R.; de Haas, G. H. Chem. Phys. Lipids 1973, 10, 127-136. (26) Lowry, O. H.; Rosebrough, N. J.; Farr, A. L.; Randall, R. J. J. Biol. Chem. 1951, 193, 265-275. (27) Saunders, L.; Perrin, J.; Gammack, D. J. Pharm. Pharmacol. 1962, 14, 567-572. (28) Hope, M. J.; Bally, M. B.; Webb, G.; Cullis, P. R. Biochim. Biophys. Acta 1985, 812, 55-65. (29) Mattjus, P.; Molotkovsky, J. G.; Smaby, J. M.; Brown, R. E. Anal. Biochem. 1999, 268, 297-304.
11728 Langmuir, Vol. 23, No. 23, 2007
Nylund et al.
Figure 1. Fluorescence anisotropy of 1% AV-GalCer in 4% MGalCer and 95% POPC between 20 and 80 °C (gray trace). Excitation was detected at 370 nm, and emission was detected at 425 nm. choline (POPC) to probe sonicated pure POPC acceptor vesicles was started by injection of 2 µg of GLTP. The assay as well as the vesicle preparation were done in a sodium phosphate buffer containing 10 mM sodium dihydrogen phosphate, 1 mM dithiothreitol, and 1 mM EDTA at pH 7.4 and 37 °C. The transfer rates in the first minute after GLTP injection could be calculated by comparing the increase in the fluorescence intensity (unquenching) and comparing this value to the total fluorescence intensity obtained when treating the sample with Triton X-100 (final concentration ) 1%) subtracted with the Triton X-100 blank. For a more detailed explanation of the calculations, see previously published data.13
Results Anisotropy on AV-GalCer in POPC Matrices. The fluorescently labeled galactosylceramide AV-GalCer is thought to mimic the natural galactosylceramide well.30-32 The anthrylvinyl probe is hydrophobic in nature, is located within the hydrophobic core of the bilayer,23 and is not looping out to the interfacial region like, for instance, the more polar probe NBD.33-35 It has previously been shown that anthrylvinyl labeled galactosylceramide at low concentrations truly reflects the phase transitions of the host lipids.23 To examine if AV-GalCer is miscible with the glycolipid MGalCer in our system, we performed anisotropy experiments where 1 mol % AV-GalCer was added to a 4 mol % mixture of MGalCer in POPC. MGalCer was chosen because of its lower melting temperature compared to PGalCer. The main phase transition temperature of MGalCer is around 50 °C, whereas that for POPC is -3 °C.36 In Figure 1, we present the anisotropy values of AV-GalCer in MGalCer/POPC probe sonicated vesicles versus temperature. The melting temperature of MGalCer corresponds well with the change in the anisotropy value. The change in the anisotropy (30) Molotkovsky, J. G.; Imbs, A. B.; Bergelson, L. D. Bioorg. Khim. 1983, 9, 112-114. (31) Molotkovsky, J. G.; Manevich, Y. M.; Babak, V. I.; Bergelson, L. D. Biochim. Biophys. Acta 1984, 778, 281-288. (32) Johansson, L. B.; Molotkovsky, J. G.; Bergelson, L. D. Chem. Phys. Lipids 1990, 53, 185-189. (33) Chattopadhyay, A.; London, E. Biochemistry 1987, 26, 39-45. (34) Huster, D.; Muller, P.; Arnold, K.; Herrmann, A. Biophys. J. 2001, 80, 822-831. (35) Huster, D.; Muller, P.; Arnold, K.; Herrmann, A. Eur. Biophys. J. 2003, 32, 47-54. (36) Curatolo, W.; Sears, B.; Neuringer, L. J. Biochim. Biophys. Acta 1985, 817, 261-270.
was around 48 °C. We therefore conclude that, at low concentrations, such as those used under our experimental conditions, AV-GalCer disperses with MGalCer and reports its melting. Adsorption of GLTP at the Air/Buffer Interface. The surface activity of GLTP was first analyzed at the air/buffer (sodium phosphate) interface in the absence of lipids. Different amounts of GLTP (40, 60, and 95 nM final concentration) were added to the subphase. The change in surface pressure was monitored as a function of time (Figure 2). Increasing the GLTP concentration clearly caused the surface pressure to plateau in a much shorter time (Table 1). The small differences in the final surface pressures could be due to the longer equilibration times and slight subphase evaporation. After the protein had adsorbed to the interface, aliquots were taken from both the surface and the subphase and analyzed for glycolipid transfer activity. Transfer activity was found both in the subphase and in the aliquots taken from the surface (Figure 3). The protein taken from the surface has 71% activity compared to the untreated protein and 78% activity compared to the protein taken from the subphase. This confirms that GLTPs at least do not completely denature at the interface and that there is only a slight activity loss when they are exposed to the buffer/air interface for an extended time (90 min). Adsorption of GLTP to Different Lipid Monolayers. To determine how the presence of lipid regulates the interaction of GLTP with the interfaces, similar experiments as those described above were performed where the protein was injected under preformed lipid monolayers. The monolayers were made of the glycerophospholipids POPC, DPPC, DPPE, and DPPS and the sphingolipids PSM, OSM, PDHSM, PGalCer, SGalCer, and sulfatide. All monolayers were in the gel phase under the experimental conditions, except the POPC and OSM monolayers which were in the fluid phase. The lipids were spread to an initial surface pressure between 4 and 24 mN/m. This initial surface pressure was chosen to be significantly below the collapse surface pressures of the lipids but comparable to the maximum pressure exerted by GLTP in the absence of lipid (Figure 2). In this way, surface pressure increases should reflect the relative affinities of GLTP for the different lipid classes, that is, the ability of GLTP to penetrate the lipid monolayer. The change in surface pressure decreases linearly with initial surface pressure (Figure 4). The
Membrane CurVature Effects on GLTP ActiVity
Langmuir, Vol. 23, No. 23, 2007 11729
Figure 2. Time course for the adsorption of GLTP to the air/buffer interface. Adsorption was measured in the absence of lipids at three different protein concentrations at 37 °C under mild stirring.
Figure 3. AV-GalCer transfer activity of GLTP. The measurements were done with 1 µg of GLTP. The black trace represents control GLTP, the red trace represents protein taken from the subphase, and the green trace represents protein taken from the surface. Table 1. Reached Surface Pressure after GLTP Adsorption to a Clean Air/Buffer Interface GLTP (nM)
time (min)
surface pressure (mN/m)
40 60 95
60 30 15
17.4 18.0 19.0
data for DPPC, POPC, and DPPE are similar, slightly increasing in the given order, whereas the negatively charged phosphatidylserine can be penetrated by GLTP at much higher surface pressures. This displacement moves the intercept of the regression line with the abscissa giving an exclusion pressure (also called critical pressure) of about 31 mN/m for DPPS (Figure 4A). The data in Figure 4B show that the glycolipids PGalCer, SGalCer, and sulfatide are not more easily penetrated by GLTP than the other lipids, even though these lipids are substrates for GLTP. The negatively charged sulfatide has a slightly higher exclusion pressure than those of the neutral glycolipids. PSM has about the same exclusion pressure as that for DPPC, and OSM has a slightly
Figure 4. Surface pressure changes following GLTP adsorption to lipid monolayers. Monolayers of glycerophospholipids (A) and sphingolipids (B) were spread to different initial surface pressures, and GLTP was injected into the aqueous subphase to a final concentration of 98 nM. The lines represent least-square fits of the data obtained.
higher exclusion pressure, which is about the same as that for POPC (Table ). Adsorption of GLTP to PGalCer Monolayers. GLTP binding to PGalCer monolayers at different initial surface pressures was
11730 Langmuir, Vol. 23, No. 23, 2007
Nylund et al.
Figure 5. GLTP adsorption to monolayers as a function of time. (A) GLTP was added beneath a PGalCer monolayer to a final concentration of 0.49 µM at 34 or 37 mN/m. (B) GLTP adsorption to a DPPC monolayer as a function of time at a surface pressure of 36 mN/m. Table 2. Exclusion Pressures for GLTP for Different Phospholipid and Sphingolipid Monolayers lipid
exclusion pressure (mN/m)
POPC DPPC DPPE DPPS PSM OSM PDHSM PGalCer SGalCer sulfatide
25.9 24.7 27.0 30.7 24.7 26.1 25.3 24.5 23.4 26.4
studied next. Monolayers of PGalCer were spread to an initial surface pressure of either 34 or 37 mN/m. The initial pressure was chosen to be well above the measured exclusion pressure to prevent a possible penetration of GLTP into the PGalCer monolayer. The aim was to study if GLTP was able to extract PGalCer from the monolayer, and hence, a drop in the surface pressure would be seen. After the surface pressure had stabilized, GLTP was injected into the subphase in the reaction chamber of the zero-order through to a final protein concentration of 0.49 µM (Figure 5A). We examined the effects of the addition of GLTP to the subphase of a monolayer composed of DPPC at a surface pressure of 36 mN/m (Figure 5B). GLTP adsorption caused a decrease in the surface pressure as a function of time. This was caused by a loss of PGalCer from the monolayer due to GLTP binding and diffusion to the subphase. The pressure decrease continued until equilibrium between GLTPbound PGalCer and monolayer PGalCer was reached. In the control experiment, the monolayer was formed with DPPC instead of PGalCer, and no adsorption or GLTP mediated lipid removal
from the monolayer was detected (Figure 5B). In another control experiment, bovine serum albumin was added to the subphase under a PGalCer monolayer, but no change in the surface pressure over time could be detected (data not shown). In the following, we added beneath a PGalCer (100%) monolayer at a surface pressure of 36 mN/m GLTP (0.49 µM final concentration) and small unilamellar POPC acceptor vesicles (50 nm in diameter) in a 10-fold excess compared to the monolayer area (Figure 6, upper trace). No change in the surface pressure was detected, indicating PGalCer transfer from the monolayer to the POPC acceptor vesicles in the subphase. If the composition of the monolayer was changed to contain 10 mol % PGalCer in the POPC, DPPC, or PSM matrix, no additional transfer was detected (data not shown). Lowering the surface pressure did not result in any increase in the GLTP mediated PGalCer transfer; only penetration of GLTP and vesicles to the surface resulted. Reversing the transfer direction such that the monolayer was composed of DPPC (POPC or PSM data not shown) with the subphase containing GLTP and vesicles composed of various amounts of PGalCer in a phospholipid matrix (about 50 nm in diameter) did not result in any increase in the monolayer surface pressure (Figure 6, lower trace). The lower trace represents a DPPC monolayer at 32 mN/m, with the addition of GLTP and small unilamellar vesicles composed of 10 mol % PGalCer in POPC. No changes in the surface pressure indicating transfer occurred. AV-GalCer Transfer from Increasing Vesicle Sizes. Because of the inability of GLTP to transfer PGalCer from planar monolayers at physiological surface pressures in the presence of small POPC liposomes, we decided to investigate the transfer of AV-GalCer from vesicles with increasing diameter to small unilamellar probe sonicated acceptor vesicles. The different donor vesicle populations were prepared by extrusion through polycarbonate filters described in the Materials and Methods section. In Table 3, the derived average vesicle diameters and the standard deviations are given based on light scattering measurements. The data in Figure 7 are derived by using the described resonance energy transfer method at 37 °C and varying the donor vesicle sizes, with the probe sonicated vesicles as the acceptors. GLTP (2 µg) mediated AV-GalCer transfer was faster from POPC donor vesicles than from DPPC donor vesicles at all measured vesicle sizes. This difference in the transfer rate was most significant in small extruded vesicles. For POPC, an almost linear decrease in the AV-GalCer transfer rates was seen as a function of vesicle size increase. The data in Figure 7 are given as transfer rates versus donor vesicle surface area. For DPPC, the fastest transfer rate was seen from the smallest 49 nm diameter vesicles and with no significant difference for the rest of the analyzed donors regardless of the size. There was no transfer of AV-GalCer detected from any of the PSM donor vesicles regardless of the vesicle size (data not shown). In Figure 8, the AV-GalCer transfer rates are compared using small unilamellar probe sonicated donor vesicles, with a variation in the size of the acceptor vesicles. GLTP mediated (2 µg) AVGalCer transfer from donor vesicles to POPC acceptors of increasing size shows the most dramatic variation in the transfer rates with a 6 times faster transfer to small probe sonicated vesicles compared to the case of the biggest vesicles. The transfer to DPPC vesicles also differs significantly between the different sizes, although the curvature sensitivity is not as pronounced as for the POPC vesicle series. The transfer to PSM vesicles did not significantly differ with the vesicle size. Therefore, GLTP mediated AV-GalCer transfer to small vesicles (up to diameters of 100 nm) was slower to PSM acceptors compared to the cases
Membrane CurVature Effects on GLTP ActiVity
Langmuir, Vol. 23, No. 23, 2007 11731
Figure 6. GLTP and vesicle addition beneath PGalCer or DPPC monolayers. The upper trace represents a PGalCer monolayer with a subphase containing small unilamellar POPC vesicles and GLTP, and the arrow indicates the time of GLTP addition. The lower trace represents a DPPC monolayer with a subphase containing GLTP and small unilamellar vesicles composed of 10 mol % PGalCer in POPC, and the arrow indicates the time of vesicle addition.
Figure 7. AV-GalCer transfer from donor vesicles (13.3 µm) with increasing diameter to probe sonicated acceptor vesicles (133 µm) mediated by GLTP. The transfer was mediated by 2 µg of GLTP. The data are expressed as transfer rate versus vesicle surface area in nm2. Averages and standard deviations are of at least five measurements. Table 3. Derived Average Vesicle Diameters and Standard Deviations Determined by Light Scattering POPC
DPPC
PSM
vesicle preparation
size
SD
size
SD
size
SD
probe sonication extrusion, 50 nm filter extrusion, 100 nm filter extrusion, 200 nm filter extrusion, 400 nm filter extrusion, 800 nm filter
49.8 66.9 103 196 422 715
4.6 1.8 4.6 18.5 30.6 111
48.6 70.7 100 172 387 626
4.4 0.8 1.7 3.7 79.6 65.3
53.4 68.7 97.8 207 509 1126
1.5 0.75 1.0 24.4 26.9 102
of POPC and DPPC, but faster to PSM acceptor vesicles with diameters of 400 nm and above. The GLTP mediated AV-GalCer transfer from donor to acceptor vesicles of equal sizes was analyzed next, that is, the transfer from probe sonicated to probe sonicated and 50 nm filter extruded to 50 nm filter extruded. The theoretical value for the predicted transfer can be calculated from the values in Figures
Figure 8. GLTP mediated AV-GalCer transfer to acceptor vesicles (133 µm) with increasing diameter from probe sonicated vesicles (13.3 µm). The transfer was mediated by 2 µg of GLTP. The data are expressed as transfer rate versus vesicle surface area in nm2. Averages and standard deviations are of at least four measurements.
7 and 8. This is done by multiplying the fractional rates for the transfer from donors (Figure 7) with the fractional rates for the acceptor transfer values (Figure 8), and by multiplying this derived value with the experimentally obtained value for the probe sonicated value. Figure 9 shows the experimental data that correlate approximately with the theoretical values.
Discussion The lipid membrane properties embedding the glycolipid play an important role in GLTP lipid transfer activity.12,13,37,38 In the present report, the monolayer technique was used to study the interfacial behavior of GLTP as well as unilamellar vesicles of different sizes. In the vesicle transfer assay, the transfer of AVGalCer is gradually decreasing as the donor vesicle diameter is (37) Rao, C. S.; Chung, T.; Pike, H. M.; Brown, R. E. Biophys. J. 2005, 89, 4017-4028. (38) Rao, C. S.; Lin, X.; Pike, H. M.; Molotkovsky, J. G.; Brown, R. E. Biochemistry 2004, 43, 13805-13815.
11732 Langmuir, Vol. 23, No. 23, 2007
Figure 9. GLTP mediated AV-GalCer transfer from donor (13.3 µm) to acceptor vesicles (133 µm) of equal sizes. The theoretical value for the transfer was calculated from the values in Figures 7 and 8 by multiplying the values for the respective vesicles. The data are expressed as transfer rate versus vesicle surface area in nm2. Averages and standard deviations are of at least five measurements.
increasing. The results presented here show that the protein is surface-active and that the protein remains active at the interface. GLTP does seem to have the ability to transfer glycolipids from planar monolayer membranes, but only when there are no vesicles present in the subphase. The previous report that porcine GLTP isolated from brain tissue showed a minor transfer of glycolipids between two monolayers is contradictory with our findings.39 However, that report finds on the other hand no GLTP/glycolipid complex in the subphase beneath the radio labeled GalCer monolayer, which makes the interpretation and conclusions of GLTP mediated transfer of galactosylceramide complicated and questionable.39 It should be noted that there are some difficulties with the earlier studies that may relate to the extended purification and instability of the GLTP preparations,40,41 which is not highly abundant in animal tissues. This could also explain why Sasaki and Demel39 found that GLTP absorption to the clean air/water interface gave a much lower surface pressure of 11.2 mN/m compared to ours. Further, when they decreased the surface pressure of the monolayer or increased the GLTP concentration in the subphase, they did not detect any changes in the measured transfer rates, which could be an indication that the protein was not pure and partially inactive. The penetration ability of GLTP for different lipid monolayers balanced at different initial surface pressures indicated that the protein has a slightly greater ability to penetrate neutral phospholipids than the substrates for GLTP (the neutral glycolipids). The protein being positively charged at neutral pH penetrates negatively charged lipid films at higher surface pressures than those of neutral lipid films. However, DPPS is also more easily penetrated than the glycolipid sulfatide. The relatively larger penetration of GLTP into the lipid monolayers at low surface pressures does not necessarily indicate a direct interaction of the protein with the lipid molecules, as GLTP alone is surface-active and can spontaneously migrate from the subphase to a lipid free interface. The surface pressure limiting value for GLTP penetration was between 23 and 26 mN/m except for DPPS films where it was 31 mN/m. This would indicate that at higher pressures GLTP (39) Sasaki, T.; Demel, R. A. Biochemistry 1985, 24, 1079-1083. (40) Metz, R. J.; Radin, N. S. J. Biol. Chem. 1982, 257, 12901-12907. (41) Gammon, C. M.; Vaswani, K. D.; Ledeen, R. W. Biochemistry 1987, 26, 6239-6243.
Nylund et al.
cannot penetrate the monolayer. It should be noted that GLTP does not traverse the membrane, and it can only access glycolipids from the outer leaflet of vesicles;42,43 therefore, the lipid monolayer should be a valid system to also study the mode of GLTP action. Since no transfer was observed from the monolayer to the acceptor vesicles in the subphase, it appears that GLTP cannot and does not need to penetrate the interfacial region and bind the PGalCer molecule.5,9,38 This would confirm the finding that the rate-limiting step for GLTP action seems to be the process leading to the GLTP/glycolipid complex formation. The formation of a transition-state complex was also recently found to be predominantly enthalpy driven, whereas the net transfer of GSLs appears to be mainly entropy driven.38 The model for the course of interaction of GLTP with membranes has previously been worked out,37,38 and the first part starts with the association of the apo form of GLTP with the membrane interface, the second part is the scanning of the membrane surface and the formation of a GLTP/glycolipid complex at the interface, and the third part is the release of the complex from the interface. The data presented here as well as previous data37,38 suggest that GLTP interacts peripherally with the membrane and does not penetrate above 26 mN/m. For transfer to occur, GLTP must find and recognize the sugar moiety on the glycolipid5 and then form the complex that is released into the bulk. It has been shown previously that GLTP interacts with membranes with or without the glycolipid present.37,38,44 As we showed previously, GLTP cannot transfer AV-GalCer from probe sonicated and small extruded PSM vesicles, probably due to the sphingolipid-sphingolipid hydrogen-bonding networks,13 and here we show that this is also true for large vesicles and monolayers. The transfer rate of AV-GalCer from fluid POPC donors gradually decreases as the vesicles become larger, and for the planar monolayers we could detect a glycolipid transfer, though small, under the examined conditions. The small transfer of GalCer from the monolayers could be due to the relatively restricted vertical movement and fluctuation mobility of the glycolipid at the monolayer interfacial region, because one membrane leaflet is missing in monolayers. For the gel-like DPPC membrane matrix, there is no difference in the AV-GalCer transfer rates regardless of the donor vesicle size except for the smallest probe sonicated donors (49 nm in diameter) that are significantly higher. The transfer rate of AVGalCer from POPC donors at all vesicles sizes analyzed was about 2-fold higher than that from the DPPC system. We speculate that GLTP can access clustered glycolipids in a fluid environment such as POPC, but it would have a limited accessibility to glycolipids well dispersed into a more tightly packed environment such as that in DPPC and PSM.13,14 It is tempting to speculate that vesicles with higher curvature stress could allow for a different lateral distribution of the glycolipid than planar membranes, and this in turn would regulate GLTP accessing its substrate. However, we cannot rule out the possibility that GLTP would need a curved interface for the localization and binding events of glycolipids. AV-GalCer is transferred faster to highly curved POPC and DPPC vesicles than to large vesicles in an almost linear fashion. The transfer to the more tightly packing DPPC and PSM vesicles does seem to be dependent on the curvature to the same extent as for the case of POPC and shows an almost similar transfer rate of AV-GalCer regardless of vesicle size, although AV-GalCer (42) Wong, M.; Brown, R. E.; Barenholz, Y.; Thompson, T. E. Biochemistry 1984, 23, 6498-6505. (43) Brown, R. E.; Jarvis, K. L.; Hyland, K. J. Biochim. Biophys. Acta 1990, 1044, 77-83. (44) West, G.; Nylund, M.; Slotte, J. P.; Mattjus, P. Biochim. Biophys. Acta 2006, 1758, 1732-1742.
Membrane CurVature Effects on GLTP ActiVity
transfer to probe sonicated DPPC vesicles was slightly faster than that to the rest of the DPPC acceptor vesicles examined. The curvature does not seem to have a similar regulatory effect in terms of GLTP glycolipid release to gel-like acceptor membranes. For a fluid acceptor membrane such as POPC, the curvature however seems to be required for an efficient release of the transferred glycolipid. The results presented here would indicate that when only a planar membrane is present, GLTP binds and removes glycolipids. However, if curved membranes are present, GLTP prefers these and binds but cannot deliver the glycolipids to planar membranes, and consequently, we do not detect any transfer. These findings are supported by studies from the Brown laboratory37 that planar membranes are not a suitable environment for GLTP to extract glycolipids from or deliver to. This could also explain why the transfer is faster to small vesicles. The protein interacts more with the highly curved membranes than with the more planar vesicles, and therefore, the glycolipid is also more easily transferred to these vesicles. However, the hypothesis does not give an explanation why the transfer of glycolipids to sphingomyelin membranes is not sensitive to the curvature stress.
Langmuir, Vol. 23, No. 23, 2007 11733
As shown here, a planar membrane surface does not serve as a platform from which GLTP can transfer glycolipids. Consequently, membrane curvature is likely to play a crucial role in the release of the complex from the membrane interface. It is therefore tempting to speculate that the likely action of GLTP in a biological system takes place on curved membrane components. Such intracellular membrane structures would be the transport vesicles of the Golgi complex and plasma membrane invaginations such as caveolae. Acknowledgment. Dr. J. Peter Slotte is acknowledged for comments and for access to instrumentation. This work was supported by the ISB Graduate School, Academy of Finland, Sigrid Juse´lius Foundation, Magnus Ehrnrooth Foundation, Oscar O ¨ flund Foundation, Svenska Kulturfonden, Medicinska Understo¨dsfo¨reningen Liv och Ha¨lsa r.f., K. Albin Johanssons Stiftelse, Svensk-O ¨ sterbottniska Samfundet r.f., and Åbo Akademi University. LA701927U
