Effect of temperature on the formation of liquid phase-separating giant unilamellar vesicles (GUV)

Chemistry and Physics of Lipids 165 (2012) 630–637

Contents lists available at SciVerse ScienceDirect

Chemistry and Physics of Lipids journal homepage: www.elsevier.com/locate/chemphyslip

Effect of temperature on the formation of liquid phase-separating giant unilamellar vesicles (GUV) Viktoria Betaneli a,1 , Remigiusz Worch a,b,∗,1 , Petra Schwille a a b

BIOTEC, Biophysics Research Group, Technical University Dresden, Tatzberg 47-51, 01307 Dresden, Germany Laboratory of Biological Physics, Institute of Physics, Polish Academy of Sciences, Al. Lotnikow 32/46, 02-668 Warsaw, Poland

a r t i c l e

i n f o

Article history: Received 29 March 2012 Received in revised form 12 June 2012 Accepted 13 June 2012 Available online 29 June 2012 Keywords: Giant unilamellar vesicles Protein reconstitution Lipid rafts Confocal microscopy Fluorescence correlation spectroscopy Lipid fluorescent probe

a b s t r a c t Giant unilamellar vesicles (GUVs) are widely used as model systems to study both, lipid and membrane protein behavior. During their preparation by the commonly applied electroformation method, a number of issues must be considered to avoid the production of artifacts due to a poor lipid hydration and protein degradation. Here we focus on the effect of preparation temperature on GUVs composed of the most commonly used domain-forming mixture dioleoylelphospatidylcholine/shingomyelin/cholesterol (DOPC/SM/chol) (2/2/1). Lower production temperatures are generally preferable when aiming at a functional reconstitution of proteins into the membrane. On the other hand, lower growth temperature is suspected to alter the lipid composition and the yield of phase-separating vesicles. By confocal imaging, we find that vesicles prepared significantly above and below the melting temperature Tm have the same overall morphology, similar size distributions of vesicles and a similar area coverage by liquidordered (Lo ) domains. However, a large population analysis indeed reveals a different overall yield of phase-separating vesicles. Two-focus scanning fluorescence correlation spectroscopy measurements did not show any divergence of lipid analog mobility in (Lo ) and (Ld ) phases in vesicles prepared at different temperatures, indicating that the lowered growth temperature did not influence the lipid organization within the two phases. Moreover, the expected advantages of lower preparation temperature for proteo-GUVs could be exemplified by the reconstitution of voltage dependent anion channel (VDAC) into DOPC/SM/chol GUVs, which aggregates at high, but not at low preparation temperatures. © 2012 Elsevier Ireland Ltd. All rights reserved.

1. Introduction Biological membranes are assemblies of various types of lipids and integral membrane proteins, but their complexity is a limiting factor for quantitative experiments. Therefore a variety of simpler artificial membrane models with controllable size, geometry and compositions were developed (for a review see Chan and Boxer, 2007). Giant unilamellar vesicles (GUVs) are especially useful for optical microscopy, mainly because of their cell-like size and curvature (for a recent review see Garcia-Saez et al., 2009 and Kahya, 2010). Apart from the studies devoted exclusively to lipid mixtures, GUVs are also used as a platform for membrane protein reconstitution from previously prepared proteoliposomes. This approach was

Abbreviations: DOPC, dioleoylphosphatidylcholine; SM, sphingomyelin; chol, cholesterol; RT, room temperature; FV, field of view; Lo , liquid-ordered; Ld , liquid-disordered; FCS, fluorescence correlation spectroscopy; 2fsFCS, two-focus scanning FCS; DiD, 1,1 -dioctadecyl-3,3,3 ,3 tetramethylindodicarbocyanine 4chlorobenzenesulfonate salt. ∗ Corresponding author. Tel.: +48 22 843 66 01x2204; fax: +48 22 843 09 26. E-mail address: [email protected] (R. Worch). 1 Both authors contributed equally to this work. 0009-3084/$ – see front matter © 2012 Elsevier Ireland Ltd. All rights reserved. http://dx.doi.org/10.1016/j.chemphyslip.2012.06.006

first advertized by Girard et al., successfully incorporating the Ca2+ ATPase and bacteriorhodopsin. Following similar protocols, several other proteins were successfully incorporated in the membranes of artificial vesicles (summarized in Kahya, 2010). However, GUV electroformation from proteoliposomes poses several limitations. One of them is the use of buffers with physiological salt concentration required for protein stability, which was a motivation for development of novel electroformation protocols (Pott et al., 2008; Shaklee et al., 2010). Another one is related with the preparation temperature, which is advised to be high enough to ensure melting and mixing of all lipids, followed by a slow cooling to achieve near-equilibrium states for phase behavior studies (Morales-Penningston et al., 2010). Indeed, it was reported that for phospholipids with high gel-to-liquid phase transition temperatures, the electroformation did not result in GUV production when performed at room temperature (Shimanouchi et al., 2009). The effect of growth temperature was shown to influence the miscibility transition temperature in DOPC/DPPC mixtures containing 35% of cholesterol (Veatch and Keller, 2005). However, even more drastic changes were observed when a too low amount of lipids in the film was used (Veatch and Keller, 2005). Therefore different experimental methods produce vesicles with slightly different

V. Betaneli et al. / Chemistry and Physics of Lipids 165 (2012) 630–637

distribution of compositions, estimated for ±2 mol% for each lipid species in the mixture (Veatch and Keller, 2005). The correspondence to the equilibrium thermodynamic phases in GUVs was carefully checked by determining the fraction areas of solid ordered and liquid disordered phases in GUVs composed of DLPC/DPPC binary mixtures (Fidorra et al., 2009). A confirmation of the lever rule was observed for the average values of domain area ratios obtained for 11–14 vesicles, however for certain molar fractions, standard deviations exceed 30% of the average values, reflecting the compositional heterogeneity at the level of single vesicles. As underlined by many researchers in the field, for the aforementioned reasons, one has to examine large populations of vesicles to conclude about the properties of the studied lipid mixture. Among different parameters influencing GUVs growth and composition, incubation at high temperatures seem to be potentially deleterious for membrane proteins reconstituted in proteoliposomes. Multi-span transmembrane proteins usually unfold irreversibly, showing aggregation at high temperatures in the most severe cases (Minetti and Remeta, 2006). This fact may be also a limiting factor for single-span transmembrane proteins, more abundant in higher eukaryotes (Worch et al., 2010), whose cytoplasmatic and extracellular domains are physically similar to multi-domain globular proteins prone to irreversible unfolding (Fitter, 2009). Therefore, the reconstitution of eukaryotic membrane proteins in GUVs must be optimized if the aim is to reconstitute these proteins in a physiologically relevant lipid environment. In recent years, great interest has arisen in specific nanodomains, enriched in sphingomyelin (SM) and cholesterol (chol), commonly named as membrane rafts (Simons and Ikonen, 1997), supposed to be a native membrane environment for proteins involved in a number of biological activities (Jacobson et al., 2007; van Meer et al., 2008; Lingwood et al., 2009). Ternary mixtures containing saturated sphingomyelin, cholesterol, and unsaturated phosphatidylcholine (PC) are typically used to mimic membrane domain formation, as they form large-scale ‘raft’-like domains of fluid liquid-disordered (Ld ) and liquid-ordered (Lo ) phases, (for a ˜ et al., 2008) and references therein). Creation of review see (Goni the Lo phase is related with abolishing a typical gel-fluid phase transition, as it was shown in details for DPPC (Vist and Davis, 1990), DPPC/DMPC mixture (Almeida et al., 1993), and DOPC or POPC/SM/chol mixtures (Veatch and Keller, 2005). Lipid dynamics in Lo and Ld phases has been studied by means of fluorescence correlation spectroscopy (FCS) (Korlach et al., 1999; Kahya et al., 2003; Kahya and Schwille, 2006; Carrer et al., 2009), single particle tracking and imaging (Dietrich et al., 2001), fluorescence recovery after photobleaching (FRAP) (Almeida et al., 1993) and pulse gradient NMR (Filippov et al., 2003). It was pointed out that cholesterol is a key factor regulating the diffusion in membranes, decreasing the lateral mobility of lipids in unsaturated glycerophospholipid membranes (Filippov et al., 2003; Kahya and Schwille, 2006), but slightly increasing the diffusion coefficient in vesicles composed of saturated phospholipids and 18:0 SM (Kahya and Schwille, 2006). In parallel to the aforementioned experiments, a number of simulation studies on PC/SM/chol ternary mixtures thermodynamics was carried out (Risselada and Marrink, 2008; Almeida, 2009; Feigenson, 2009; Almeida, 2011) for better understanding of lipid–cholesterol interactions, which is, however, still incomplete. In this article, we address the influence of temperature on GUV preparation for the DOPC/SM/chol (2/2/1) mixture, aiming at the characterization of this particular system as a platform assay for membrane protein reconstitution requiring lower temperatures. To illustrate the effect of high temperature on protein degradation, we perform a reconstitution of human voltage dependent anion channel (VDAC) at room temperature (RT) and 65 ◦ C. We compare the vesicles prepared at RT and at 65 ◦ C, which is 8 ◦ C higher than

631

the highest melting point of the components (SM 18:0) (Estep et al., 1980). We quantify a series of confocal fluorescence images with respect to morphology, size, Lo domain areas and the overall yield of phase separating liposomes observed at room temperature. To specifically examine the dynamic nature of Lo and Ld phases we collected a comprehensive data set of two-focus scanning fluorescence spectroscopy (2fsFCS) measurements. We discuss our results in the context of composition variations at the level of single vesicles, as well as benefits and limitations of using GUVs as a versatile platform for membrane protein reconstitution. 2. Experimental procedures 2.1. Materials 1,2-Dioleoyl-sn-glycero-3-phosphocholine (dioleoylphospatidylcholine; DOPC), n-stearoyl-d-erythrosphingosylphosphorylcholine (sphingomyelin 18:0; SM) and cholesterol (chol) were purchased from Avanti Polar Lipids (Alabaster, AL) as powders and used without further purifications by dissolving them in chloroform at 10 mg/ml concentration. The fluorescent lipid dye 1,1 -dioctadecyl-3,3,3 ,3 tetramethylindodicarbocyanine, 4chlorobenzenesulfonate salt (DiIC18(5); DiD) and Alexa 647 carboxylic acid, succinimidyl ester were purchased form Invitrogen. Atto655-carboxylic acid was purchased from Atto-Tec GmbH. 2.2. Preparation of GUVs from lipids Giant unilamellar vesicles (GUVs) were prepared by electroformation (Angelova and Dimitrov, 1986) using a custom-made teflon chamber with two electrodes made of Pt wires. The composition of the lipid mixture was DOPC/SM(18:0)/chol (2/2/1). 5 ␮l of the same batch of the lipid mixture (1 mg/ml) was spread on each wire of two teflon chambers which were used for electroformation at RT and 65 ◦ C. The concentration of DiD in the mixture was adjusted to obtain an optimal number of molecules in FCS measurements and was 0.005 mol% in the case of measurements in Ld phase and 0.15 mol% for FCS in the Lo phase and imaging. After evaporation of the solvent, the chambers were assembled and filled with 300 mM sucrose solution (at osmolarity 300 mOsm g−1 ). Electroformation was performed in parallel at 65 ◦ C and at room temperature by applying alternating electric current (2 V electric field corresponding to 400 V/m, 10 Hz) for 1.5 h. In the final step, the frequency was decreased to 2 Hz for 30 min to detach the vesicles from the electrodes. 50 ␮l of the solution were transferred to one well in a 8-well chamber (#1.5 cover slide, Nunc, MatTeck) containing 800 ␮l of equiosmolar phosphate-buffered saline (137 mM NaCl, 2.7 mM KCl, 4.3 mM Na2 HPO4, 1.4 mM KH2 PO4 , pH 7.4; PBS). The osmolarity of PBS was adjusted with sucrose to keep the difference not larger than 10 mOsm g−1 . The chambers were kept at room temperature for 30 min, to let the vesicles sediment and to equilibrate thermally prior to imaging and FCS measurements. For GUVs prepared at 65 ◦ C the cooling rate was equal to the rate of cooling to the ambient room temperature. 2.3. Preparation of GUVs with VDAC Reconstitution of VDAC labeled with Alexa 488 in GUVs was done as reported previously (Betaneli et al., 2012). In brief, small unilamellar vesicles (SUVs) were prepared from DOPC/SM18:0/chol (2/2/1) at a total lipid concentration of 4 mg/ml by sonication in 2 mM MOPS-Tris buffer, pH 7. SUVs were solubilized in LDAO at a lipid/detergent ratio of 2 w/w. VDAC in LDAO was added at the lipid/protein ratio of 40 w/w. For detergent removal, the solution was incubated with Bio-Beads (Bio-Rad Laboratories)

632

V. Betaneli et al. / Chemistry and Physics of Lipids 165 (2012) 630–637

(30 mg per 1 mg of detergent) were incubated for 4 h. Such proteoliposomes were deposited on indium tin-oxide coated glass slides in the form of several 2 ␮l droplets and dried for 30 min in a desiccator. For electroformation, performed in parallel at RT and 65 ◦ C, AC current of 400 V/m, 12 Hz was applied for 3 h across a home built chamber filled with 300 mM sucrose, 1 mM MOPS-Tris, 2 mM KCl, pH 7. After preparation, the buffer surrounding the vesicles was exchanged to 10 mM Tris, 150 mM KCl, pH 7.2. 2.4. Imaging and FCS Confocal imaging and two-focus scanning FCS were performed on a LSM Meta 510 system (Carl Zeiss, Jena, Germany) using a 40× NA 1.2 UV-VIS-IR C–Apochromat water immersion objective. For the domain area estimation typically 55–120 image stacks were acquired with a z-step of 230–420 nm and a x = y sampling rate of 36–54 nm, depending on the size of vesicles. For 2fsFCS, a home-built detection unit was assembled at the fiber output channel as described previously (Ries and Schwille, 2006). A bandpass filter HQ700/75 was used behind a collimating achromat lens to reject the residual laser and background light. Another achromat (LINOS Photonics, Goettingen, Germany) with a shorter focal length was used to image the internal pinhole onto the aperture of the fiber of the avalanche photo diode (Perkin-Elmer, Boston, MA). The photon arrival times were recorded in the photon mode of the hardware correlator Flex 02-01D (correlator.com, Bridgewater, NJ). All filters and dichroic mirrors were purchased from AHF Analysentechnik (Tuebingen, Germany). The intensity was measured by continuously scanning the vesicles perpendicularly to the membrane for 500 s. The temperature was checked in the chamber after the measurement (24.0 ± 0.1 ◦ C). Prior to the FCS measurements, the pinhole position and objective correction ring were carefully adjusted to ensure optimal fluorescence signal from the dye solutions (Alexa Fluor 647 or Atto 655) disposed in one of the chamber wells. 2.5. Data analysis Confocal layer 3-D projections were created in ImageJ (Rasband, 1997) without prior deconvolution. For Lo area estimation home developed software in Python was used. The area was calculated as an integral on a grid created on a contour determined by the boundaries of the Lo domain in the following way: the z coordinates of ∼40 contour points were calculated assuming a perfect sphere with a radius determined from the equatorial confocal slice (see Fig. 3D for illustrations) and the areas of small rectangles (100 in one ‘column’) were summed. Data analysis for scanning FCS was performed in software developed in Matlab (MathWorks) as described previously (Ries and Schwille, 2006). In brief, the continuous intensity signal was aligned according to the maximum value, reflecting the membrane position on the scan path. The maximum and adjacent pixels of each scan were averaged, and the resulting intensity trace was correlated over time. The corresponding cross-correlation curve Gx , resulting from a delayed detection of the photons along the second line at a distance d, was fitted by the equation: Gx () =

1 N



1+

4D w02

−1 

1+

4D w02 S 2

−1/2



exp

−

d2 w02 + 4D



(1)

where N denotes the average particle number, D the diffusion coefficient, w0 the 1/e2 radius of the detection area, S the structural parameter. A global fit of the two autocorrelation functions (corresponding to S → ∞ and d = 0 in (1)) and the cross-correlation function was used to improve the accuracy of the 2fsFCS analysis. The dye surface concentrations were calculated using the

maximum cross-section of the confocal volume during scanning perpendicular to the membrane (w02 S). 3. Results 3.1. Reconstitution of VDAC in GUVs To illustrate the potentially harmful effect of higher temperature on protein degradation, we first prepared GUVs from proteoliposomes containing VDAC and DOPC/SM/chol (2/2/1) at RT and 65 ◦ C, following a recently described protocol (Betaneli et al., 2012). The liposomes prepared at different temperatures showed a noticeable contrast (Fig. 1A and B). VDAC in vesicles prepared at RT was homogenously distributed in the membrane, showing no significant aggregations (Fig. 1A), in contrary to GUVs prepared at 65 ◦ C, with numerous small vesicles or aggregates (Fig. 1B). The protein fluorescence followed the fluorescence of spectrally distinct DiD, present in proteoliposomes at low concentration, indicating the partitioning of VDAC into Ld phase (Fig. 1C, D). Membrane organization resulting from the electroformation at 65 ◦ C excluded the possibility of mobility measurements, whereas VDAC reconstituted at RT showed free diffusion with the coefficient of 3.1 ± 0.6 ␮m2 s−1 in Ld phase, which is slightly slower to the diffusion in pure DOPC membrane reported previously (4.6 ± 0.5 ␮m2 s−1 ) (Betaneli et al., 2012). 3.2. The effect of growth temperature on DOPC/SM/chol GUVs To focus on the effect of growth temperature on GUVs, we analyzed a series of confocal images and performed diffusion measurements. The electroformation was always performed in parallel at room temperature and 65 ◦ C from the same amount of lipids and the same mixture to avoid any potential differences resulting from the slight changes in composition and amount of the deposited mixture. The yields of independently carried out electroformations at two temperatures were comparable. In the sections below we describe several aspects in details. 3.2.1. General morphology In both cases, the obtained vesicles were spherical, without any shape deformations (Fig. 2). Coexistence of Lo and Ld phases was observed due to preferential partitioning of a long-chain dialkylcarbocyanise dye (DiD) in the latter one. Regardless of the preparation temperature, both populations of liposomes showed visible phase separation. Fig. 2 illustrates three-dimensional image reconstructions of confocal z-scans of GUVs prepared at different temperatures. From hundreds of vesicles analyzed by imaging, we conclude that the lower preparation temperature did not affect the domain shapes, which were always circular. In some samples prepared at RT, we observed little vesicles or aggregates present inside, as well as outside, of the giant liposomes (Fig. 2A). Less frequently, we observed long fragments of tubulated membranes, often linking two liposomes or being located at the top surface of the vesicle (Fig. 2A). We noticed that such artifacts were also present in the samples prepared at higher temperature, however, to a substantially reduced amount. To assess whether the temperature of preparation may affect the size of the liposomes, we measured their diameters in the equatorial plane. To determine the boundaries of the vesicles, we analyzed the fluorescence intensity profiles and calculated the diameter from the pixel size (Fig. 3A and B). The distributions of GUV diameters prepared at low and high temperatures have a very similar non-Gaussian shape with a similar mean and SD values of 28 ± 10 ␮m and 25.3 ± 9.6 ␮m for RT and 65 ◦ C, respectively (Fig. 3C, Table 1).

V. Betaneli et al. / Chemistry and Physics of Lipids 165 (2012) 630–637

633

Fig. 1. Transmembrane protein (VDAC labeled with Alexa Fluor 488) reconstituted in GUVs grown at (A) RT, (B) 65 ◦ C. White arrows indicate aggregated VDAC. (C), (D) green (VDAC-A488) and red (DiD) channel, respectively of GUVs with VDAC reconstituted at RT in phase separating mixture used in the studies. Scale bar: 20 ␮m.

3.2.2. Percentage of Lo /Ld phase-separating vesicles and domain areas To see whether the GUVs prepared at different temperatures were similar regarding their Lo /Ld phase separation, we performed a quantitative analysis of confocal images. In each batch of electroformation, we scanned the bottom surface of a chamber containing GUVs with a field of view adjusted to observe few vesicles at the same time. We carefully checked the occurrence of separate phases in whole, single liposomes and calculated the ratio between the number of phase-separating vesicles and their total number. The above analysis was repeated to obtain statistically comprehensive data sets from hundreds of vesicles collected from three independent electroformations (Table 1). We did not notice any striking differences regarding these values between single experiments performed at the

same temperature. However, the average percentages of phase separating vesicles depended on the growth temperature. In the pool of liposomes prepared at 65 ◦ C we observed 79 ± 22% GUVs exhibiting Lo /Ld phase separation, whereas this value was reduced to 59 ± 28% when vesicles were prepared at room temperature. To address whether the preparation temperature affects the domain size in vesicles showing phase separation, we estimated the areas of Lo phase in 15 vesicles in each sample (see Section 2.4 for details). The average ratios of Lo area and the total GUV surface were about one third for both pools of vesicles (Table 1). Correspondence of the obtained ratios with the equilibrium conditions of phase coexistence is, however, not possible without a precise determination of exact tie-lines for this particular ternary mixture, what is discussed in Section 4.

Table 1 Size, phase separation percentage and diffusion coefficients in GUVs prepared at room temperature (RT) and 65 ◦ C. n – numbers of vesicles analyzed, FV-fields of view. Errors are given as standard deviation (SD). Temperature of preparation

Diameter [␮m] Percentage of phase separating GUVs [%] Percentage of GUV area covered by Lo

2

−1

Diffusion coefficient D [␮m s ] Dye surface concentration [␮m−2 ]

RT

65 ◦ C

28 ± 10 (n = 146) 59 ± 28 (n = 449; 94 FV) 32 ± 14 (n = 15)

25.3 ± 9.6 (n = 145) 79 ± 22 (n = 332; 79 FV) 36 ± 10 (n = 15)

Ld

Lo

Ld

Lo

5.3 ± 0.7 (n = 17) 7.3 ± 4.3

1.5 ± 0.3 (n = 23) 9 ± 13

6.0 ± 1.1 (n = 23) 12.7 ± 8.6

1.3 ± 0.2 (n = 25) 19 ± 19

634

V. Betaneli et al. / Chemistry and Physics of Lipids 165 (2012) 630–637

Fig. 2. Giant unilamellar vesicles prepared at (A) room temperature and (B) 65 ◦ C. Three-dimensional image reconstructions (upper panels) and single confocal slices (lower panels). Scale bars are 10 ␮m, except the lower right images in both (A) and (B) for which the bars are 20 ␮m.

3.2.3. Diffusion in GUV membrane Considering the above results, it was an obvious next step to address the influence of the preparation temperature directly on the lipid microenvironments. Thus, we performed a series of 2fsFCS measurements, to probe the local viscosities by determining the diffusion coefficients of a fluorescent dye (DiD) in Ld and Lo phases in GUVs prepared at different temperatures. Such measurements were realized by scanning two parallel laser lines through the equator of a liposome exhibiting Lo /Ld phase separation, which allows

Fig. 3. Quantification of GUV sizes and domain areas. (A) Image of a vesicle (scale bar 10 ␮m) with (B) a corresponding intensity profile used for the diameter determination. (C) Histograms of the diameters of phase separating vesicles for the room temperature preparation (RT) (light grey) and 65 ◦ C (dark grey). (D) Estimation of the Lo area by the sum of rectangles created by the grid points determined by the contour of Lo domain (white points). For clarity only one vertical ‘column’ of rectangles is shown.

measurements in coexisting phases on a single vesicle (Fig. 4A). Typical auto- and cross-correlation curves for the Lo and Ld phases in liposomes prepared at different temperatures are presented in Fig. 4 B and C, and the corresponding averaged fitting values of diffusion coefficients are summarized in Table 1. The values of diffusion

V. Betaneli et al. / Chemistry and Physics of Lipids 165 (2012) 630–637

635

Fig. 4. (A) Schematic comparison of ‘standard’ confocal single-point FCS and dual focus scanning FCS (2fsFCS) on GUV membrane. Typical correlation curves along with corresponding fits for a fluorescent lipid analogue diffusing in the Ld and Lo phases in the vesicles prepared at (B) room temperature and (C) 65 ◦ C. Autocorrelation curves are depicted in dark grey, cross-correlation curves between the two foci are in light grey.

coefficients for both disordered and ordered phases are in agreement for the two preparations. Thus, it may be concluded that the temperature of preparation did not affect the diffusion coefficients in fluid phases in GUVs, as an indicator of their viscosities. Apart from the measurements in phase-separating vesicles, we measured the mobility of DiD in homogenously looking vesicles with no visible phase separation. The diffusion coefficients averaged from 10 vesicles were comparable to these in Ld phases: 5.9 ± 0.7 ␮m2 s−1 and 5.7 ± 0.7 ␮m2 s−1 for the vesicles prepared at 65 ◦ C and RT, respectively. To quantify the diffusion coefficient reduction in the Ld phase of DOPC/SM/chol vesicles as compared to the fluid phase unmodified by the presence of cholesterol, we measured the mobility of DiD in GUVs composed exclusively from DOPC. The electroformation at room temperature guarantees the liquid state of DOPC, for

which the gel-liquid phase transition temperature is −20 ◦ C. In the same FCS measurements, the diffusion coefficient was determined as 10.7 ± 1.0 ␮m2 s−1 (average from n = 14 measurements), which is ∼2 times larger than the diffusion in the Ld phase and 7–8 times larger than the diffusion in the Lo phase of the studied tertiary mixture. It is worth mentioning that the relative error of the diffusion coefficient in the case of mono-component DOPC membrane did not exceed 10%, whereas for the tertiary mixture it was larger than 15%, for both phases (Table 1). 3.2.4. Dye surface concentration Fitting the autocorrelation curves allows also to determine the average number of fluorescent molecules in the focal volume. Since we performed the electroformation from the same amount of lipids in parallel at two temperatures, it is possible to compare directly

636

V. Betaneli et al. / Chemistry and Physics of Lipids 165 (2012) 630–637

the DiD surface concentrations, adjusted individually for FCS measurements in Ld and Lo (see Section 2.2 for details). In both phases we noticed ∼2 times higher dye concentration in vesicles prepared at 65 ◦ C, suggesting that higher temperature facilitates mixing with lipids (Table 1). However high standard deviations of the average values, present in all cases, indicate that there exist large variations of composition at the level of single vesicles. Surprisingly, it was also the case for GUVs made of DOPC (26 ± 24 ␮m−2 ), which unsaturated chains should be perfectly mixed with DiD at RT.

4. Discussion Giant unilamellar vesicle electroformation depends on many factors that need to be optimized to produce vesicles with desirable properties. They are becoming more widely used in applications involving reconstituted membrane proteins in a membrane with a composition of a choice. However, the example of aggregated VDAC, possessing relatively stable ␤-barrel fold, in vesicles formed at 65 ◦ C clearly shows that in some cases the preparation conditions used for GUVs composed only from lipids need to be modified when the goal is to reconstitute a functional membrane protein. Although the thermal effects on proteins are specific for each protein, it is likely that most of proteins will denaturate at 65 ◦ C, for example a transition temperature for a helical bundle is ∼45 ◦ C (Wen et al., 2012). Recently structural elements influencing thermal stability of a ␤-barrel membrane protein were studied, showing that the denaturation temperature for the wild type protein was higher than 70 ◦ C (Gessmann et al., 2011). Despite that fact, we observed aggregation of a similarly folded VDAC even at slightly lower temperature. Therefore we asked the question on the influence of lowered growth temperature on the properties of vesicles composed of DOPC/SM/chol (2/2/1), commonly used as a mimicry of membrane ‘rafts’, aiming at look for the conditions for bottom-up protein reconstitution. We observed similar general morphology and similar size distributions for the vesicles prepared at different temperatures. The corresponding size histograms had similar shapes and mean values to those obtained in former studies on GUVs made of egg phosphatidylcholine (EPC)/lysophospatidylcholine (LPC) mixtures (Mathivet et al., 1996) or from EPC/egg phosphatidic acid (EPA) with reconstituted Ca2+ -ATPase (Girard et al., 2004). Despite these similarities, the most pronounced difference between the vesicles prepared below and above the melting temperature is the ratio of vesicles showing Ld /Lo phase separation. This effect may be attributed to a globally altered composition of vesicles prepared at room temperature, however the coverage by Lo areas in phaseseparating vesicles is very similar. Taking into consideration the fact that in non-phase separating GUVs the diffusion coefficient was similar to the one measured in Ld phase, it is likely that some part of SM and cholesterol is not taken up from the wires during electroformation at RT. A decreased dye surface concentration in both fluid phases of GUVs prepared at RT, compared to vesicles grown from exactly the same lipid/DiI mixture at 65 ◦ C, is also consistent with this explanation. However, as indicated in Section 3.2.4, a wide distribution of the dye surface concentration is observed even for a single-component DOPC GUVs, indicating that the electroformation may results in compositionally varied vesicles even when performed at conditions assuring optimal mixing of components. As it was discussed in the literature, one of the problems related with the canonical raft mixture is the fact that slight changes of laboratory conditions may lead to the changes in lipid composition of GUVs (Veatch and Keller, 2003). We believe that it would be beneficial to determine the composition of Lo and Ld domains more precisely and check whether the growing conditions lead

to formation of vesicles under the thermodynamic equilibrium conditions. However, the determination of tie-lines for particular mixtures requires extra information, for example from other experimental techniques. The precision of determination is crucial, since slight changes in the shape of Lo /Ld coexistence region may lead to different molar fractions of the phases decomposed along the tie-line. Moreover, care has to be taken in the analysis of existing phase diagrams of the ‘raft mixture’, which itself is a wide term, since it is known that the substitutions of hightemperature melting lipid may be not intuitive, like the increase of the miscibility transition temperature for shorter sphingomyelin chains (discussed in the review by Veatch and Keller, 2005). To the best of our knowledge, the exact tie-lines for DOPC/SM18:0/chol (2/2/1) were not determined. We believe that future studies will result in a consistent description of ternary systems containing cholesterol and will reconcile the discrepancies between existing phase diagrams, often resulting from a slight difference in the interpretation of experimental data (see Veatch and Keller, 2005 and Fidorra et al., 2009 for discussions). When comparing two pools of vesicles, it is also likely that the difference in vesicles may be more pronounced when larger populations of vesicles are analyzed. In typical GUV studies, especially those related with protein reconstitution, tens of vesicles are analyzed, which is few percent of the (proteo-)liposome material used for electroformation. In our studies the most noticeable difference, which was the percentage of phase separating vesicles, appeared when hundreds, not tens, of vesicles were analyzed. However, we are convinced that in GUVs prepared at room temperature, which may be a condition required to avoid protein aggregation, it is possible to obtain at least some fractions of vesicles exhibiting Lo /Ld phase separation with homogenously incorporated protein. Nevertheless, it has to be kept in mind, that the pool of electroformed GUVs from mixtures has a wide distribution of composition, which is probably even higher when dealing with membrane proteins reconstituted in such systems. It is known that cholesterol is a pivotal factor tuning the membrane viscosity, which was a subject of numerous studies. Therefore we have used FCS to systematically analyze lipid mobility in Lo and Ld phases present in liposomes prepared at two different temperatures. By means of this technique, the lipid dynamics in GUVs composed of DOPC/SM/chol was characterized, exploring various molar fractions of the components (Kahya et al., 2003). The diffusion coefficient for the lipid dye in the Ld phase is in the range of 4.9–5.15 ␮m2 s−1 in the whole region of Lo /Ld coexistence, and in the Lo phase in the range of 0.1–0.8 ␮m2 s−1 , with a strong dependence on the cholesterol amount (Kahya et al., 2003). Pulse gradient solid state NMR studies in SM/chol mixtures showed diffusion coefficients of the same range, however highly depending on temperature, probably as a consequence of the broad-phase transition observed for SM (Filippov et al., 2003). The difference in lipid mobility between the Lo and Ld phase by a factor of 5 was also revealed by coarse-grained molecular simulations (Risselada and Marrink, 2008). Taking into consideration that diffusion coefficients in membranes are usually reported with at least 20% absolute experimental uncertainties, we can conclude that the values measured in this study agree well, although we observed slightly higher diffusion coefficients in the Lo phase. Most importantly, the mobilities in Lo and Ld phases were not affected by lower growth temperature, which indicates that the physical nature of the domains is the same. Since the presence of cholesterol changes the diffusion in the Lo phase quite dramatically (Kahya et al., 2003), our results may indicate, that its abundance in the vesicles prepared at different temperatures is comparable. However, we are aware of the fact that different molar fractions of DOPC and SM may lead to the similar diffusion coefficient values in the region of fluid phase coexistence.

V. Betaneli et al. / Chemistry and Physics of Lipids 165 (2012) 630–637

5. Conclusions In summary, our results indicate that the giant unilamellar vesicles prepared from DOPC/SM/chol (2/2/1) mixture at room temperature and at 65 ◦ C have very similar physical properties. We did not observe any substantial differences for the lipid microenvironments of ordered and disordered phases, as revealed by two-focus scanning FCS. The preparation of Ld /Lo phase-separating, ‘raft’mimicking GUVs from DOPC/SM/chol (2/2/1) mixture at room temperature can be therefore justified, especially in the cases of protein reconstitution, requiring lower temperatures for protein stability and biological activity, like the one illustrated by VDAC reconstitution. Acknowledgments We thank Dolores Carrer for critical reading of the manuscript. This work was funded by an Alexander von Humboldt fellowship to RW, a DIGS-BB fellowship by the German Research Foundation (DFG) to VB and a DFG grant (MicroLipo) to PS. References Almeida, P.F., 2009. Thermodynamics of lipid interactions in complex bilayers. Biochimica et Biophysica Acta 1788, 72–85. Almeida, P.F., 2011. A simple thermodynamic model of the liquid-ordered state and the interactions between phospholipids and cholesterol. Biophysical Journal 100, 420–429. Almeida, P.F., Vaz, W.L., Thompson, T.E., 1993. Percolation and diffusion in threecomponent lipid bilayers: effect of cholesterol on an equimolar mixture of two phosphatidylcholines. Biophysics Journal 64, 399–412. Angelova, M.I., Dimitrov, D.S., 1986. Liposome electroformation. Faraday Discussions in Chemical Society 81, 303–311. Betaneli, V., Petrov, E.P., Schwille, P., 2012. The role of lipids in VDAC oligomerization. Biophysical Journal 102, 523–531. Carrer, D., Kummer, E., Chwastek, G., Chiantia, S., Schwille, P., 2009. Asymmetry determines the effects of natural ceramides on model membranes. Soft Matter 5, 3279–3286. Chan, Y.H., Boxer, S.G., 2007. Model membrane systems and their applications. Current Opinion in Chemical Biology 11, 581–587. Dietrich, C., Bagatolli, L.A., Volovyk, Z., Thompson, N., Levi, M., Jacobson, K., Gratton, E., 2001. Lipid rafts reconstituted in model membranes. Biophysical Journal 803, 1417–1428. Estep, T.N., Calhoun, W.I., Barenholz, Y., Biltonen, R.L., Shipley, G.G., Thompson, T.E., 1980. Evidence for metastability in stearoylsphingomyelin bilayers. Biochemistry 19, 20–24. Feigenson, G.W., 2009. Phase diagrams and lipid domains in multicomponent lipid bilayer mixtures. Biochimica et Biophysica Acta 1788, 47–52. Fidorra, M., Garcia, A., Ipsen, J.H., Härtel, S., Bagatolli, L.A., 2009. Lipid domains in giant unilamellar vesicles and their correspondence with equilibrium thermodynamic phases: a quantitative fluorescence microscopy imaging approach. Biochimica et Biophysica Acta 1788, 2142–2149. Filippov, A., Orädd, G., Lindblom, G., 2003. The effect of cholesterol on the lateral diffusion of phospholipids in oriented bilayers. Biophysical Journal 84, 3079–3086. Fitter, J., 2009. The perspectives of studying multi-domain protein folding. Cellular and Molecular Life Sciences 66, 1672–1681. Garcia-Saez, A., Carrer, D., Schwille, P., 2009. In: Weissig, V. (Ed.), Fluorescence Correlation Spectroscopy in Model Membrane Systems. Humana Press. Girard, P., Pécréaux, J., Lenoir, G., Falson, P., Rigaud, J.-L., Bassereau, P., 2004. A new method for the reconstitution of membrane proteins into giant unilamellar vesicles. Biophysical Journal 87, 419–429.

637

Gessmann, D., Mager, F., Naveed, H., Arnold, T., Weirich, S., Linke, D., Liang, J., Nussberger, S., 2011. Improving the resistance of a eukaryotic ␤-barrel protein to thermal and chemical perturbations. Journal of Molecular Biology 413, 150–161. ˜ F.M., Alonso, A., Bagatolli, L.A., Brown, R.E., Marsh, D., Prieto, M., Thewalt, Goni, J.A., 2008. Phase diagrams of lipid mixtures relevant to the study of membrane rafts. Biochimica et Biophysica Acta, Molecular and Cell Biology of Lipids 1781, 665–684. Jacobson, K., Mouritsen, O.G., Anderson, R.G., 2007. Lipid rafts: at a crossroad between cell biology and physics. Nature and Cell Biology 9, 7–14. Kahya, N., Scherfeld, D., Bacia, K., Poolman, B., Schwille, P., 2003. Probing lipid mobility of raft-exhibiting model membranes by fluorescence correlation spectroscopy. Journal of Biological Chemistry 278, 28109–28115. Kahya, N., Schwille, P., 2006. How phospholipid–cholesterol interactions modulate lipid lateral diffusion, as revealed by fluorescence correlation spectroscopy. Journal of Fluorescence 16, 671–678. Kahya, N., 2010. Protein–protein and protein–lipid interactions in domainassembly: Lessons from giant unilamellar vesicles. Biochimica et Biophysica Acta Biomembranes 1798, 1392–1398. Korlach, J., Schwille, P., Webb, W.W., Feigenson, G.W., 1999. Characterization of lipid bilayer phases by confocal microscopy and fluorescence correlation spectroscopy. Proceedings of the National Academy of Sciences of the United States of America 96, 8461–8466. Lingwood, D., Kaiser, H.J., Levental, I., Simons, K., 2009. Lipid rafts as functional heterogeneity in cell membranes. Biochemical Society Transactions 37, 955–960. Mathivet, L., Cribier, S., Devaux, P.F., 1996. Shape change and physical properties of giant phospholipid vesicles prepared in the presence of an AC electric field. Biophysical Journal 70, 1112–1121. Minetti, C.A., Remeta, D.P., 2006. Energetics of membrane protein folding and stability. Archives of Biochemistry and Biophysics 453, 32–53. Morales-Penningston, N.F., Wu, J., Farkas, E.R., Goh, S.L., Konyakhina, T.M., Zheng, J.Y., Judy, Y., Webb, W.W., Feigenson, G.W., 2010. GUV preparation and imaging: minimizing artifacts. Biochimica et Biophysica Acta 1798, 1324–1332. Pott, T., Bouvrais, H., Méléard, P., 2008. Giant unilamellar vesicle formation under physiologically relevant conditions. Chemistry and Physics of Lipids 154, 115–119. Rasband, W.S., 1997. ImageJ. U.S. National Institutes of Health, Bethesda, Maryland, USA, http://rsb.info.nih.gov/ij/. Ries, J., Schwille, P., 2006. Studying slow membrane dynamics with continuous wave scanning fluorescence correlation spectroscopy. Biophysical Journal 91, 1915–1924. Risselada, H.J., Marrink, S.J., 2008. The molecular face of lipid rafts in model membranes. Proceedings of the National Academy of Sciences of the United States of America 105, 17367–17372. Shaklee, P.M., Semrau, S., Malkus, M., Kubick, S., Dogterom, M., Schmidt, T., 2010. Protein incorporation in giant lipid vesicles under physiological conditions. ChemBioChem 11, 175–179. Shimanouchi, T., Umakoshi, H., Kuboi, R., 2009. Kinetic study on giant vesicle formation with electroformation method. Langmuir 25, 4835–4840. Simons, K., Ikonen, E., 1997. Functional rafts in cell membranes. Nature 387, 569–572. van Meer, G., Voelker, D.R., Feigenson, G.W., 2008. Membrane lipids: where they are and how they behave. Nature Reviews Molecular Cell Biology 9, 112–124. Veatch, S.L., Keller, S.L., 2003. A closer look at the canonical ‘Raft Mixture’ in model membrane studies. Biophysical Journal 84, 725–726. Veatch, S.L., Keller, S.L., 2005. Miscibility phase diagrams of giant vesicles containing sphingomyelin. Physical Review Letters 94, 148101. M.R., Davis, J.H., 1990. Phase equilibria of cholesterol/ Vist, dipalmitoylphosphatidylcholine mixtures: 2H nuclear magnetic resonance and differential scanning calorimetry. Biochemistry 29, 451–464. Wen, J., Kelly, A., Chemmalil, L., Muzammil, S., Gabrielson, J., Jiang, Y., 2012. Applications of differential scanning calorimetry for thermal stability analysis of proteins: qualification of DSC. Journal of Pharmaceutical Sciences 101, 955–964. Worch, R., Bökel, C., Höfinger, S., Schwille, P., Weidemann, T., 2010. Focus on composition and interaction potential of single-pass transmembrane domains. Proteomics 10, 4196–4208.