Preparation and Utility of Asymmetric Lipid Vesicles for Studies of Perfringolysin O-Lipid Interactions.

Abstract

Studying the interaction of pore-forming toxins, including perfringolysin O (PFO), with lipid is crucial to understanding how they insert into membranes, assemble, and associate with membrane domains. In almost all past studies, symmetric lipid bilayers, i.e. bilayers having the same lipid composition in each monolayer (leaflet), have been used to study this process. However, practical methods to make asymmetric lipid vesicles have now been developed. These involve a cyclodextrin-catalyzed lipid exchange process in which the outer leaflet lipids are switched between two lipid vesicle populations with different lipid compositions. By use of alpha class cyclodextrins, it is practical to a wide range of sterol concentrations in asymmetric vesicles. In this article, protocols for preparing asymmetric lipid vesicles are described, and to illustrate how they may be applied to studies of pore-forming toxin behavior, we summarize what has been learned about PFO conformation and lipid in symmetric and in asymmetric artificial lipid vesicles.

1. Introduction

For membrane proteins, including cholesterol-dependent cytolysins, an important approach to investigate protein-lipid interactions and their impact upon structure and function are studies using purified protein mixed with artificial lipid membranes. To date, these studies have generally used symmetric lipid bilayers, in which the inner and outer lipid monolayer (leaflet) of the lipid bilayer have the same lipid composition. However, many natural membranes have lipid asymmetry, in which the lipid composition differs in the two leaflets. This is certainly true for the plasma membrane of eukaryotic cells, in which the outer leaflet is rich in sphingolipids (SM) and phosphatidylcholine (PC), while the inner leaflet is relatively enriched in phosphatidylethanolamine (PE) and anionic phospholipids such as phosphatidylserine (PS) [1]. The level of cholesterol asymmetry remains controversial [2, 3]. The extent to which lipid asymmetry affects membrane protein behavior is little studied. This is a question of importance for PFO, which acts at the plasma membrane, and perhaps internal membranes that also possess lipid asymmetry.

2. Studies of the Effect of Lipid Composition Upon PFO Binding and PFO Domain Localization Using Symmetric Vesicles.

Much information about the impact of lipid composition upon PFO behavior has been gained from studies using symmetric lipid vesicles. This includes a basic picture of the steps involved in membrane insertion [4]. These steps include lipid binding, which requires interaction of residues at the tip of domain 4 with cholesterol, then oligomerization into a 30–50-mer, followed by a large structural rearrangement in domains 2 and 3 in which two sets of helix-containing loops on domain 3 insert into the membrane, forming two hairpins with two TM beta-strands each. The final structure of the membrane-inserted segments is an oligomeric transmembrane (TM) beta-barrel with a large aqueous pore. Based on mutational and kinetics studies, the events during the membrane insertion process in symmetric vesicles are believed to take place in the order membrane binding, oligomerization, insertion, and pore-forming beta barrel assembly, with the last two steps perhaps being nearly simultaneous

The nature of PFO interaction with lipid has been examined in studies in which lipid composition was systematically varied. These studies have examined the effect of lipid structure upon both insertion into membranes, and the degree of association of PFO with membrane domains [5–11]. In both artificial lipid vesicles and plasma membranes liquid ordered (Lo) domains rich in sphingolipids. which tend to have saturated acyl chains, and cholesterol can co-exist with liquid disordered (Ld) domains, which are more enriched in phospholipids having one unsaturated acyl chain [12].

The requirement of cholesterol in order that PFO can bind to membranes is absolute. In membranes containing a mixture of sterol and phospholipids, a stronger interaction with sterol results in PFO binding to the membrane at a lower sterol concentration. The 3 beta OH group of the sterol is essential, but the details of the structure of sterol rings and aliphatic tail often have only a modest effect on binding [5]. This indicates that PFO mainly binds to the polar end of the sterol, and the remainder for the sterol remains in contact with the lipid bilayer rather than bound in a pocket on the PFO molecule. This model is also consistent with the effect of sterol structure on domain localization (see below).

It appears that, as a function of the structure of the phospholipids in a symmetric lipid vesicle, PFO binding to cholesterol increases with the degree to which cholesterol is exposed to aqueous solution [5, 6, 8]. The partial exposure of membrane-inserted sterol to aqueous solution corresponds to a high free energy state because small OH group cannot fully shield the hydrophobic sterol rings from water [13]. As a consequence, for two different membranes with equal cholesterol concentrations, but different physical state or phospholipid composition, PFO preferentially binds to membranes in which there is loose packing between sterol and phospholipid, i.e. the lipids are in the Ld state. In such membranes the headgroups of the lipids cannot come close enough to the sterol to fully shield cholesterol from water. Differences in binding to different sterols may often reflect differences in how closely they pack with membrane phospholipids [5]. Relative to membranes containing PC, PFO binding to and insertion into membranes is also promoted by anionic lipid (PS) [5]. This presumably reflects favorable Coulombic electrostatic interactions with anionic lipids in the case of PS, and the fact that PE has a small headgroup that is even less effective in shielding cholesterol from water than that of PC.

The discussion above might suggest that PFO would favor greater association with Ld domains than with Lo domains. However, this ignores the fact that the cholesterol concentration in Lo domains is higher than in Ld domains, a factor that should tend to compensate for differences in sterol exposure to water. Studies in symmetric lipid vesicles have examined PFO domain localization directly. It has been found that PFO can localize in both Lo and Ld domains, with localization highly dependent upon the degree of mismatch between the length of the transmembrane beta strands and the width of the hydrophobic core of the lipid bilayer in the Lo and Ld domains. The less the degree of mismatch, the stronger the association with that type of domain [9]. It should be noted that the possibility that PFO in Lo domains is surrounded by a thin ring of lipids in the Ld state cannot be ruled out [7]. Such nanodomains are below the resolution of microscopy. It should also be noted that the final domain localization of PFO is not necessarily the same as that to which PFO monomers initially bind.

The type of sterol bound to PFO also influences its affinity for Lo and Ld domains. Cholesterol partitions preferentially into Lo domains, and this promotes localization of PFO bound to cholesterol in Lo domains. In contrast, PFO binding to a sterol that favors Ld domains has been found to result in localization of PFO in Ld domains [10]. The dependence of PFO domain localization upon the preference of a sterol to localize in Lo or Ld domains suggests that the PFO-bound sterol remains largely in contact with the lipid bilayer. As noted above, exposure to the bilayer rather than localization in a deep pocket within the PFO molecule is also consistent with the modest effect of sterol ring and aliphatic tail structure upon binding to PFO. Sterol bound to a deep cleft within the PFO would lose contact with lipid and be unlikely to directly influence PFO domain localization.

3. Methods to Preparing Asymmetric Vesicles Using Cyclodextrins

One barrier to studies of membrane proteins in a natural lipid environment has been the inability to easily prepare stable artificial asymmetric lipid bilayers to mimic natural membranes. Methods have now been developed to prepare asymmetric lipid vesicles. At present, the most versatile approach to prepare such vesicles involves lipid exchange using a cyclodextrin (CD) as will be described below [14–20]. The basic principles are straightforward (Figure 1). In this method, two populations of lipid vesicles with different lipid compositions are mixed. The population that will provide the outer leaflet (the donor population, which donates what is the outer leaflet lipid in the asymmetric vesicle) is used in excess over the acceptor population which provides what is the inner leaflet lipid of the asymmetric vesicle (it is the population that accepts the outer leaflet lipid from the donor).

Figure 1: Schematic illustration of preparation of asymmetric vesicles.

From London, E. Membrane Structure-Function Insights from Asymmetric Lipid Vesicles. Acc Chem Res. 2019 52: 2382–2391. doi: 10.1021/acs.accounts.9b00300. © American Chemical Society.

CD is added at the appropriate time (see below). The CD pulls an individual lipid molecule out of the outer leaflet of a donor one vesicle (the donor) and then deposits it in the outer leaflet of another vesicle (an acceptor). This leads to a continuous exchange of lipids that results in equilibration of the outer leaflet lipid compositions. As a result of this and the use of excess donor, the desired acceptor population ends up with an outer leaflet lipid composition close to that of the donor before exchange. The exact composition of the outer leaflet after exchange depends in a complex fashion on the values of the many equilibria involved. These values can be dependent on the lipid composition of the donor and acceptor, which change during the experiment, as well as the lipid-CD binding constants. Under the usual experimental conditions, the composition of the outer leaflet of the asymmetric vesicle is not exactly the same as one would predict if the lipids in the donor and acceptor were randomized during exchange, although it is often close. Thus, especially we using donor vesicles with a mixture of lipids, it is necessary to empirically determine what ratio of lipids must be used in the donor to obtain the desired outer leaflet composition in the asymmetric vesicles after exchange. In some cases, two rounds of exchange can be useful to achieve nearly complete exchange (see below).

It should be noted that CD binding to lipid is a cooperative process [21], and so the concentration of CD used is critical. In addition, some CDs will dissolve lipid vesicles, that is completely destroy the vesicles and formation of lipid-CD complexes, at the high concentrations that are most useful for exchange (MβCD and MαCD). To minimize problems, donor lipid in excess of that which can be dissolved by the CDs is mixed with the CD in a first step. This saturates the CD with lipid and avoids or minimizes subsequent solubilization of acceptor vesicles by the CD. It may also increase the extent of exchange because lipids originally on the inner leaflet of donor vesicles load onto the CD.

The final step is the isolation of the lipid-exchanged acceptor vesicles, which are the desired asymmetric vesicles. This is carried out by centrifugation either initially preparing donor or acceptor (usually the acceptor vesicles) with trapped sucrose to impart different densities to them [15, 16, 20, 22]. It is important to note that the vesicle yield is very sensitive to the density of the vesicles and the media in which they are centrifuged, which must have density intermediate to those of the vesicles. Generally, we can achieve about 10% yields relative to the input of acceptor vesicles. What limits yield is unclear. In some cases, it may reflect incomplete pelleting. In other cases, there may be imbalanced exchange in which the difference in the number of lipids delivered to or extracted from the acceptor vesicles is large enough to cause vesicle lysis with loss of trapped sucrose and/or formation of smaller vesicles that are difficult to pellet.

For inclusion of cholesterol in asymmetric vesicles, at least two approaches are possible. One is to add cholesterol complexed with MβCD to asymmetric vesicles lacking cholesterol, using a low enough MβCD concentration that other lipids do not bind to the MβCD. However, a better method involves carrying out lipid exchange alpha class CDs (MαCD or HPαCD). These CDs have a cavity too small to accommodate cholesterol, but can bind phospholipids and sphingolipids because they can accommodate acyl chains. To obtain asymmetric vesicles with the desired amount of cholesterol, one only has to include the desired amount of cholesterol in the acceptor vesicles. No cholesterol is incorporated into the donor vesicles. In fact, if cholesterol were present in the donor vesicles and a CD that dissolves up lipid vesicles were used, the free cholesterol released after solubilization of donor vesicles might well stick to the acceptor vesicles, which would be difficult to control and thus very undesirable.

Which alpha CD is best for preparing asymmetric vesicles with cholesterol, MαCD or HPαCD? We have used both, and different researchers in the lab have different personal preferences. Protocols for both will be described below. Considerations include efficiency of exchange, amount of CD needed, and lipid specificity. In artificial lipid vesicles efficient exchange is possible with both HPαCD and MαCD. The ability of these two CDs to exchange lipids has not be compared in great detail to date. The amount of MαCD needed is less than the amount of HPαCD. With HPαCD we found we can increase the extent of exchange with two rounds of exchange, although yield decreases [23]. In some cases, the fact that HPαCD does not dissolve up donor vesicles may be an advantage. MαCD has been used to prepare asymmetric vesicles with various sterols [1, 17]. This has not been attempted yet with HPαCD.

4. Differences in Lipid Binding and Conformation of PFO in Symmetric and Asymmetric Vesicles.

In one study our group took advantage of the ability to prepare asymmetric vesicles to define the effect of lipid asymmetry upon PFO conformation [11]. We summarize that work as an example to show how asymmetric vesicles can be used to increase the understanding of pore-forming toxin behavior. Vesicles that mimic the regions of plasma membranes in the Ld state were prepared, with an outer leaflet composed of PC and an inner leaflet of PS and PE. Cholesterol levels were varied. PFO was then added to the external aqueous solution. Five parameters that respond to PFO conformation and/or lipid interaction were measured as a function of cholesterol concentration. The first was Trp fluorescence, which increases when PFO binds to membranes. The second was FRET using FRET donor and acceptor attached to point mutant A215C of PFO. The third was insertion into the bilayer, as measured by the change in fluorescence at acrylodan-labeled residue 215, a residue on a membrane-inserting segment forming a TM beta hairpin. The fourth parameter was formation of a complete beta-barrel, which results in formation of an SDS-resistant oligomer which can be detected on SDS gels. The last parameter measured was pore formation, measured by release of biocytin trapped within the vesicles by PFO (11).

These parameters were measured for PFO added to asymmetric vesicles, and compared to those for PFO added to three types of symmetric vesicles, in each case as a function of cholesterol concentration. The three types of symmetric vesicles were: 1. Scrambled, with the same overall composition as the asymmetric vesicles, 2. With a composition equivalent to that in the inner leaflet of the asymmetric vesicles, and 3. Equivalent to that in the outer leaflet of the asymmetric vesicles. As shown in Figure 2, the behavior of PFO is different in each type of vesicle. In scrambled and outer leaflet compositions, all five changes occur to a similar extent at each cholesterol concentration as cholesterol concentration is increased (Figure 2B and 2D). This indicates once the protein binds to membranes it tends to undergo all the changes that convert it to the pore-forming state. In contrast, in the presence of asymmetric vesicles or symmetric vesicles with the composition of the inner leaflet, a structural intermediate was found to dominate at intermediate cholesterol concentrations (Figure 2A and 2C). This intermediate had undergone oligomerization and deep insertion, but did not form the SDS-resistant pore-forming beta-barrel state.. Additional experiments confirmed that the intermediate conformation was TM, as determined by the observation that PFO biotinylated at a residue at the distal tip of the beta hairpin that forms upon TM insertion reached the solution on the trans side of the membrane (opposite the side of the membrane from which PFO inserted).

Figure 2. Change in PFO (PFO A215C) behavior as a function of cholesterol concentration in asymmetric vesicles and in symmetric vesicles at room temperature.

A, POPCo/POPE:POPSi/cholesterol asymmetric vesicles. B, scrambled vesicles. C, 1:1 POPE/POPS symmetric vesicles (mimicking the inner leaflet of asymmetric vesicles). D, 90:5:5 POPC/POPE/POPS symmetric vesicles (mimicking the outer leflet of asymmetric vesicles). The parameters measured are shown in the key to the figure (within panel A). Trp fluorescence (Trp F) detects binding to lipid vesicles and acrylodan fluorescence (of labeled residue 215) detects insertion of the beta hairpin. Oligomerization was detected by FRET, and is distinguished from SDS-resistant oligomers, which are believed to reflect formation of a complete beta-barrel. the American Society for Biochemistry and Molecular Biology. This research and figure was adapted from one originally published in the Journal of Biological Chemistry. Qingqing Lin and Erwin London. The influence of natural lipid asymmetry upon the conformation of a membrane-inserted protein (perfringolysin O). J Biol Chem. 2014; 289:5467–5478 © the American Society for Biochemistry and Molecular Biology.

In terms of formation of a stable TM intermediate that did not convert to the pore-forming state, the behavior of PFO in the inner leaflet lipid mixture was similar to that in the asymmetric vesicles. However, both the insertion to form the TM intermediate, and conversion to the pore-forming state required lower concentrations of cholesterol than in the asymmetric vesicles. This may reflect the increased exposure of cholesterol due to their high concentration of PS and PE, consistent with the effect of PS and PE in symmetric vesicles [5]. This would promote interaction of PFO with the inner leaflet vesicles at lower cholesterol concentrations.

The difference of PFO behavior in symmetric outer leaflet vesicles and asymmetric vesicles which have the same outer leaflet but a different inner leaflet raises the question: how can the headgroups of the inner leaflet lipids influence insertion when PFO only can sense them once TM sequences reach the distal surface in TM insertion is irreversible? The likely answer is that formation of the intermediate TM inserted state is not irreversible, and it is in equilibrium with non-inserted state. This is supported by the observation that removal of cholesterol from PFO forming the TM intermediate reversed PFO conformational changes and even binding to vesicles. In contrast, formation of the pore-forming state was irreversible.

Figure 3 summarizes potential pathways for PFO membrane insertion and assembly. The upper pathway from the aqueous monomer to final pore-forming PFO structure in membranes (via “a”) shows the species detected in symmetric vesicles with a composition similar to the outer leaflet of plasma membranes, and is the traditional pathway in which any TM inserted intermediate is too transient or at too low a concentration to easily detect. The lower pathway (via “b”) includes the TM intermediate. A hybrid process (via “c”) is also possible. It should be noted that the latter pathways are not kinetically-defined, but rather involve structural intermediates that can be trapped lipid compositions unable to carry out the complete insertion process. There is always the potential that structural intermediates are off-pathway in terms of the kinetic events that occur during insertion in vivo. Nevertheless, the formation of structural intermediates shows they are sufficiently low free energy to form under reasonable conditions, and might do so during insertion in vivo.

Figure 3. Schematic diagram of the possible steps in PFO pore formation. Pathways a, b, and c represent different possible steps in the assembly process.

This research and figure was adapted from one originally published in the Journal of Biological Chemistry. Qingqing Lin and Erwin London. The influence of natural lipid asymmetry upon the conformation of a membrane-inserted protein (perfringolysin O). J Biol Chem. 2014; 289:5467–5478 © the American Society for Biochemistry and Molecular Biology.

One important general lesson of these experiments is clear. Lipid asymmetry can influence the conformational behavior of a membrane protein, and studies that take this into account may give a more complete picture of the influence of lipid upon membrane protein structure and function.

5. Remaining Questions About the Behavior of PFO in Membranes

Our understanding of PFO behavior in asymmetric membranes is by no means complete. First, as noted above, the experiments in asymmetric membranes only examined conformations that represent structural intermediates. The evolution of changes in the conformation of PFO vs. time after addition to lipid vesicles is needed to detect if the structural intermediates in asymmetric vesicles are also kinetic intermediates that form during the assembly of the pore-forming structure. A second unanswered question concerns the structure of the transmembrane intermediate. Spectroscopically, it appears identical to the pore-forming structure in terms of how it is inserted in the bilayer. Does it contain a partially formed beta-barrel with an incomplete ring of beta-strands? If so, it is difficult to visualize what type of interactions form between the hydrophilic face of the beta sheet, which normally faces an internal aqueous pore, and the lipid bilayer. Contact between such a hydrophilic surface and the acyl chains of lipids seems improbable. C-shaped partial rings piled against one another back to back are one possibility. Third, the lipid compositions studied to date only roughly mimic plasma membranes. It is possible the behavior of PFO could be substantially different in an asymmetric membrane with physiological levels of sphingolipids included in the outer leaflet. In this regard, whether PFO localizes in Lo or Ld domains in asymmetric membranes is also unanswered.

Finally, the effect of PFO upon lipid asymmetry is unknown. That is, we do not know if lipid asymmetry changes after PFO inserted into the membrane. This is a particularly difficult question to answer because pore formation makes it impossible to use many of the simpler methods to assess lipid asymmetry, such as whether a labeled lipid incorporated into just one leaflet of an asymmetric vesicle is or is not restricted to that leaflet after PFO insertion. The problem is that pore formation allows reagents added externally (to define lipid sidedness) to pass through the lipid bilayer and interact with lipids on the inner leaflet as well as the outer leaflet. Nevertheless, it seems unlikely that there is complete loss of lipid asymmetry because the behavior of PFO added to asymmetric membranes is so different from that of PFO added to symmetric membranes with the same lipid composition.

Given the many remaining questions, and the fact that other cholesterol-dependent cytolysins have not been studied in asymmetric membranes, it is clear that asymmetric membranes should have an important role in future studies of this class of proteins. It is hoped that the detailed protocols below will help the interested researcher prepare asymmetric lipid vesicles for such studies. The first protocol shown uses HPαCD. The second uses MαCD. Note that we recently extended the latter method to use of cationic lipids, which although unnatural, should allow a wider range of asymmetries to be prepared [18].

6. Asymmetric large unilamellar vesicle (aLUVs) preparation procedure using HPαCD

6.1. Materials

• 6.1.1.Palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC), and palmitoyl-2-oleoyl-sn-glycero-3-phosphoethanolamine (POPE), palmitoyl-2-oleoyl-sn-glycero-3-phosphoserine (POPS), and cholesterol (CHOL), egg sphingomyelin (egg SM), purchased from Avanti Polar Lipids (Alabaster, AL).
• 6.1.2.Randomly 2-hydroxypropylated-α-cyclodextrin also called (2-hydroxypropyl)-α-cyclodextrin (HPαCD), purchased from Sigma-Aldrich (St. Louis, MO). Average molecular weight is ~1180 Da.
• 6.1.3.10X phosphate-buffered saline (PBS), purchased from Bio-RAD (Hercules CA). When diluted to 1X concentration is 10 mM sodium phosphate, 150 mM NaCl (pH 7.8 ± 0.2).
• 6.1.4.Thin-layer chromatography (HP-TLC) plates (Silica Gel 60), purchased from VWR International (Batavia, IL).
• 6.1.5.Sucrose, crystal, purchased from J.T Baker, Avantor (Center Valley, PA).
• 6.1.6.Dry ice or liquid nitrogen.
• 6.1.7.Acetone, if dry ice is used, purchased from Fisher Chemical (Fair Lawn, NJ).
• 6.1.8.100 nm pore polycarbonate membrane, purchased from Whatman.

6.2. Equipment

• 6.2.1.Microdispensers, 25 μl, 50 μl, and 100 μl, and corresponding glass bores, purchased from Drummond Scientific Co, (Broomall, PA) (see Note 8.1).
• 6.2.2.Water bath.
• 6.2.3.Beckman 640 spectrophotometer or suitable alternative spectrophotometer and quartz cuvettes (Beckman Instruments, Fullerton, CA).
• 6.2.4.Beckman L8–80M Ultracentrifuge and SW 60 rotor (Beckman Instruments, Fullerton, CA).
• 6.2.5.Multitube vortexer (VWR, Westchester, PA). (see Note 8.2).
• 6.2.6.Convection oven (GCA Corp, Precision Scientific, Chicago, IL).
• 6.2.7.1 ml syringe equipped Lipid Extruder (Avanti Polar Lipids, Alabaster, AL).
• 6.2.8.Epson 1640XL (Epson America, Long Beach, CA), or similar scanner.
• 6.2.9.ImageJ 1.48v software (Wayne Rasband, National Institutes of Health, Bethesda, MD).
• 6.2.10.

  Cahn C-33 Microbalance (Thermo Electron Corporation) or other microbalance.

• 6.2.11.

  Vacuum Pump and Vacuum Desiccator.

• 6.2.12.

  Hot plate or high temperature oven for heating TLC plates.

• 6.2.13.

  Low speed centrifuge.

• 6.2.14.

  TLC tank.

• 6.2.15.

  Glass tubes.

• 6.2.16.

  Nitrogen gas.

6.3. aLUVs preparation protocol using HPαCD.

• 6.3.1.
  Stock solution preparation

  • 6.3.1.1.
    Non-fluorescent Lipids are dissolved in chloroform and stored at −20 °C. The concentrations of lipids are measured by dry weight.
  • 6.3.1.2.
    385 mM HPαCD is dissolved in distilled water and stored at 4 °C. At this high concentration the HPαCD powder greatly affects the volume of the solution. The total volume of the solution after the powder dissolves has to be measured carefully. (see Note 8.3).
  • 6.3.1.3.
    10–15 and 25–28 % (w/w) sucrose dissolved in distilled water or 1X PBS.
• 6.3.2.
  Preparation of asymmetric LUVs (aLUVs) using lipid exchange catalyzed by HPαCD.

  • 6.3.2.1.
    Donor multilamellar vesicles (MLVs) preparation and mixture with HPαCD. The desired amounts of phospholipids (total 10.4 μmoles) are taken from the appropriate stock solutions using the microdispenser and combined in a glass tube. For example, consider preparation of aLUV with egg SM and POPC in the outside leaflet, POPC in the inside leaflet, and ~25mol% cholesterol in both leaflets. In this case, Donor MLVs might be made with total 10.4 μmoles lipid having the desired mole ratio of egg SM : POPC. (see Note 8.4). There is no need to include cholesterol in the donor lipid mixture as it is not transferred between vesicles by HPαCD. The lipids are dried under nitrogen until visually dry. The lipid should be dispersed as a thin film on the walls of the glass tube. Gently warming the sample helps dry the solvent more quickly. The dried lipids are further dried by placing them in a vacuum desiccator, and subjecting them to a high vacuum for 1 h. The dried lipids mixtures in glass tubes are placed in a 70°C water bath for hydration by adding 177 μl of 385 mM HPαCD. The lipid-HPαCD mixture is incubated at 70°C for 2–3 min and vortexed several times until most lipids are evenly mixed. Then 473 μl 1X PBS is added. The final MLV lipid concentration is 16 mM, HPαCD concentration is 105 mM and the volume is 0.65 ml. The lipid-HPαCD mixture is incubated in the water bath at 70 °C for another 2–3 min and vortexed until the dried lipid film dispersed evenly (a few times over a period of 2–3 min). The dispersed lipid-HPαCD mixture is further vortexed for 2h at 55 °C. To do this vortexing is carried out in a multitube vortexer placed in an oven (see Note 2).
  • 6.3.2.2.
    Acceptor large unilamellar vesicles (LUVs) preparation. Acceptor LUVs are made with 5.5 μmoles of lipid. For the example noted above, they would contain POPC :CHOL = 75 : 25. The desired lipid mixture are dried in the same fashion as for the donor lipid samples. (see Note 8.5) Dried lipid mixtures in glass tubes are placed in a 70 °C water bath and hydrated by adding 550 μl of 28% (w/w) sucrose (see Section 6.3.2.4), dispersing the lipids 70 °C and vortexing several times over 5 min to form crude MLVs. The samples are then vortexed using a multitube vortexer placed in an oven at 55 °C for 15 min. To prepare LUVs from MLVs, the MLVs dispersions are subjected to seven cycles of freezing- thaw using a mixture of dry ice and acetone, or liquid nitrogen, to freeze the MLVs, and then warm water to thaw them, followed by extrusion 11 times through using a 1 ml extruder equipped with 100 nm-pore polycarbonate membranes. To wash away the untrapped sucrose, the resulting ~0.45 ml of acceptor LUVs (~0.1 mL is lost during extrusion) in 28 (w/w) % sucrose is mixed with 3.3 mL 1X PBS and the LUVs pelleted by ultracentrifugation at 190,000 × g for 30 min at 23°C (using a Beckman L8–80 M ultracentrifuge with an SW 60 rotor). LUV pellets are resuspended to 0.45 ml in 1X PBS.
  • 6.3.2.3.
    Lipid exchange step. Lipid exchange step is performed at a concentration of 8 mM donor lipid and 4 mM acceptor lipid in a total volume of 1 ml. To do this, 500 μl of the donor MLVs-HPαCD mixture is combined with 400 μl of acceptor LUV and 100 μl of 1X PBS and then vortexed for 30 min at 55°C using the multitube vortexer. During this step, lipid exchange occurs and the acceptors become converted to asymmetric vesicles. Of course, donor vesicles also pick up some acceptor vesicle lipid. After cooling to near room temperature, the donor-acceptor vesicle mixtures were overlaid on 2.8 ml 15 (w/w) % sucrose (see Section 6.3.2.4) and then centrifuged at 190,000 × g for 50 min at 23°C After centrifugation the resulting pellet is loose, and to avoid disturbing it remove about 3.3 ml of supernatant carefully, leaving the lowest ~ 400–500 μl of the supernatant, and then add 3.3 ml of 1X PBS to the centrifuge tube, mix, and centrifuge again at 190,000 × g for 30 min at 23°C. This time the pellet is tight, and the entire supernatant can be carefully removed. The final pellet is resuspended in 500 μl of 1X PBS. Lipid yield determined by HP-TLC is typically 10% of the theoretical based on the initial amount of acceptor lipid used (see below).
  • 6.3.2.4
    Optimizing sucrose concentrations needed. The sucrose concentration inside the vesicles is chosen to maximize yield while minimizing contamination of the asymmetric vesicles with donor lipids. For vesicles with an outer leaflet composed of egg SM and POPC, an inner leaflet composed of POPC, and 25 mol % CHOL we load the vesicles with 28 mol % sucrose and centrifuge over 15 w/w % sucrose, as described above. However, the amount of sucrose needed can depend upon lipid composition. For some cases we trapped 25% (w/v) sucrose/1X PBS and centrifuged over 10 (w/v) % sucrose (11). To optimize sucrose concentrations, the protocol described above is carried out in samples lacking either donor lipid or acceptor lipid. Conditions in which there is no lipid pellet in the sample lacking acceptor lipid and a maximal pellet in the sample lacking donor lipid are chosen for the lipid exchange experiments.

6.4. Analysis of asymmetric LUV lipid composition and yield by HP-TLC.

The lipids from 200 μl of asymmetric vesicle sample is extracted by methanol and chloroform in a ratio of 0.5:1:1 aLUVs preparation : methanol : chloroform (v/v). The methanol is added to the vesicles first, vortexed briefly to homogenize, and then the chloroform is added. The sample, depending on lipids used, can be milky at this point, and can be centrifuged at low speed for a few min to help the upper and lower layers separate. The lower chloroform layer contains the lipids. Carefully remove the upper aqueous layer by pipette and discard. Most sucrose is in the upper layer. The lower layer is then dried under a nitrogen stream. In the meantime, an HP-TLC plate is warmed at 90–100°C for 20 min and them cooled to room temperature (about 23°C) before use. This helps activate the silica gel by driving off water and gives more reproducible TLC lipid mobilities. The dried asymmetric vesicle lipids are re-dissolved 40–50 μl of 1:1 chloroform : methanol (v/v) and 20–30 μl are loaded on the TLC plate. Pure lipid standards are combined and various amounts (2– 10 μg ~ 3–14 nmoles of each lipid) are loaded onto the same HP-TLC plate as the asymmetric vesicle samples to be quantified. For vesicles containing egg SM, POPC and CHOL, the plated were developed in the solvent system 6:8:2:2:1 chloroform/acetone/methanol/acetic acid/water (v/v) to separate the lipids. Samples containing POPC, POPE, POPS and cholesterol were generally chromatographed in two solvents. The first solvent system was 50:38:8:4 chloroform : methanol: water:acetic acid (v/v). After the solvent front migrated about halfway up the plate, the plate was air-dried for 5 min. Then the plate was rechromatographed in 1:1 hexane : ethyl acetate (v/v) until the solvent front migrated to near the top of the plate. When the solvent migrates to ~ 80% of the full height of the plate, the plate is removed and dried for at least 30 min in a fume hood. To visualize lipid bands, TLC plates are evenly sprayed with 3% (w/v) cupric acetate and 8% (v/v) phosphoric acid solution, dried for at least 30 min, and charred on a hot plate or in a high temperature oven at 180–200°C for 5–10 min. For best charring, the charring time should be maximized to get the darkest lipid bands before the remainder of the plate starts to char and darken. Charred HP-TLC plates are scanned and the intensity of lipids bands are quantified by ImageJ software. The amount of each lipid in the asymmetric vesicle samples is calculated by comparing band intensity to that in the standard bands. To quantify the lipids the bands of interest are highlighted, and their intensity is measured. A background intensity of a region without lipid on the same plate, but of the same size as the bands being quantified should be subtracted. Alternatively, a single size can be used for all the bands of interest. In this case, thin bands will have more background area than thicker bands, but this does not affect the quantitation. A third method is to choose one lane of the TLC with the lipid of interest, and quantify using the WAND tool.

7. aLUVs preparation procedure using MαCD

7.1. Materials

• 7.1.1.Palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC), 1-palmitoyl-2-oleoyl-sn-glycero-3-ethylphosphocholine (chloride salt) (POePC), 1,2-dioleoyl-3-trimethylammonium-propane (chloride salt) (DOTAP), 1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-(1’-rac-glycerol) (sodium salt) (POPG), 1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-L-serine (sodium salt) (POPS), 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphate (sodium salt) (POPA), and cholesterol (CHOL), purchased from Avanti Polar Lipids (Alabaster, AL).
• 7.1.2.Randomly methylated-α-cyclodextrin, also called methyl-α-cyclodextrin (MαCD), purchased from AraChem Cyclodextrin Shop (Tilburg, the Netherlands). The degree of substitution (number of methyl groups) is 10–12.
• 7.1.3.Other materials listed in 6.1.

7.2. Equipment

• 7.2.1.Equipment in Section 6.2.
• 7.2.2.Refractometer.

7.3. Asymmetric large unilamellar vesicle (aLUVs) preparation procedure using MαCD.

• 7.3.1.
  Stock solution preparation

  • 7.3.1.1.
    Lipids are dissolved in chloroform and stored at −20 °C. The concentrations of lipids are measured by dry weight.
  • 7.3.1.2.
    285 mM MαCD is dissolved in distilled water and stored at 4 °C. If a standard curve of refractive index vs. concentration is constructed first, the concentration of the MαCD solution can be determined subsequently by comparison of its refractive index to the standard curve. But see Note 8.3 for correctly preparing standard samples to account for volume changes when CD is added to water
  • 7.3.1.3.
    10X phosphate-buffered saline (PBS) diluted to 1X or 4X. The final concentration at 1X is 10 mM sodium phosphate, 150 mM NaCl (pH 7.8 ± 0.2).
  • 7.3.1.4.
    7 and 23 % (w/w) sucrose dissolved in 4X PBS and 1X PBS, respectively.
• 7.3.2.
  Preparation of aLUVs using lipid exchange catalyzed by MαCD

  • 7.3.2.1.
    Donor vesicles (MLVs) preparation and mixture with MαCD. The desired amounts of lipids are taken out of the appropriate stock solutions using the microdispenser and combined in a glass tube. Total 8 μmoles lipid. (see Note 8.6). There is no need to include CHOL in the donor lipid mixture as it is not transferred between vesicles by MαCD. The lipids are dried under nitrogen until visually dry. The lipid should be dispersed as a thin film on the walls of the glass tube. Gently warming the sample helps dry the solvent more quickly. To remove residual chloroform the dried lipids are placed in a in a vacuum desiccator, and subjected to a high vacuum for 1 h. The dried lipids mixtures in glass tubes are placed in a 70°C water bath for hydration by adding 430 μL of 1X PBS. The sample is incubated at 70 °C for 2–3 min and vortexed several times until the dried lipid film disperses evenly. Then 70 μl of 285 mM MαCD is added and the mixture will be vortexed several times for 2–3 min. The final MLVs lipid concentration is 16 mM, MαCD concentration is 40 mM and the volume is 0.5 ml. The dispersed lipid-MαCD mixture is further vortexed for 2h at 55 °C. To do this vortexing is carried out in a multitube vortexer placed in an oven (see Note 2).
  • 7.3.2.2.
    Acceptor large unilamellar vesicles (LUVs) preparation. 4 μmols of lipid containing the desired lipid mixture are dried in the same fashion as for the donor lipid samples. (see Note 8.7). Dried lipid mixtures in glass tubes are placed in a 70 °C water bath and hydrated by adding 500 μL of 23% (w/w) sucrose, dispersing the lipids 70 °C and vortexing several times over 5 min to form crude MLVs. The samples are then vortexed using a multitube vortexer placed in an oven at 55 °C for 15 min. To prepare LUVs from MLVs, the MLVs dispersions are subjected to seven cycles of freezing- thaw using a mixture of dry ice and acetone or liquid nitrogen to freeze the samples, and then warm water to thaw them, followed by extrusion 11 times through using a 1mL extruder equipped with 100 nm-pore polycarbonate membranes. To wash away the untrapped sucrose, the resulting ~0.5 ml (0.01~0.03 ml is lost during extrusion) of acceptor LUVs is mixed with 3.5 mL 1X PBS and the LUVs pelleted by ultracentrifugation at 190,000 × g for 30 min at 23°C (using a Beckman L8–80 M ultracentrifuge with an SW 60 rotor). LUV pellets are resuspended to 0.5 ml in 1X PBS.
  • 7.3.2.3.
    Lipid exchange step. Lipid exchange step is performed at a concentration of 8 mM donor lipid and 4 mM acceptor lipid in a total volume of 1 ml. To do this, 500 μl of the donor MLVs-MαCD mixture is combined with 500 μL of acceptor LUV then incubated for 45 min at 37°C using the incubator with shaking. During this step, lipid exchange occurs and the acceptors become converted to asymmetric vesicles. Of course, donor vesicles also pick up some acceptor vesicle lipid. After cooling, the donor-acceptor vesicle mixture is overlaid on 3 ml of 7 (w/w) % sucrose in 4X PBS (see Section 7.3.2.4) and then centrifuged at 190,000 × g for 45 min at 23°C. After centrifugation the resulting pellet is loose. We remove about 3.5 ml of supernatant carefully, leaving the lowest ~ 400–500 μl of the supernatant, add 3.5 ml of 1X PBS to the centrifuge tube, mix, and centrifuge again at 190,000 × g for 30 min at 23°C. This time the pellet is tight, and the entire supernatant can now be carefully removed. The final pellet is resuspended in 500 μl of 1X PBS. Lipid yield determined by HP-TLC is typically 10% of the theoretical based on the initial amount of acceptor lipid used (see below). The extent of exchange, is determined by HP-TLC (see below) and can vary significantly depending on the lipids used [18].
  • 7.3.2.4
    Optimizing sucrose concentrations needed. The sucrose concentration inside the vesicles is chosen to maximize yield while minimizing contamination of the asymmetric vesicles with donor lipids. This can depend on lipids chosen. To optimize sucrose concentrations, the protocol described above is carried out in samples lacking either donor lipid or acceptor lipid. Conditions in which there is no lipid pellet in the sample lacking acceptor lipid and a maximal pellet in the sample lacking donor lipid are chosen for the lipid exchange experiments.

7.4. Analysis of asymmetric LUV lipid composition and yield by HP-TLC.

The lipids from 200 μL of asymmetric vesicle sample is extracted by 1:2:2 of deionized water: methanol: chloroform (v/v). The 1mL of deionized water is added to the vesicles solution first, vortexed briefly to homogenize, and then methanol is added, vortexed briefly to homogenize. Lastly, the chloroform is added, and the mixture is vortexed until milky. It can be centrifuged at low speed for a few min to help the separation of upper and lower layers. The lower chloroform layer contains the lipids. Carefully remove the upper aqueous layer by pipette and discard. (To get more complete extraction, a second round of extraction can be carried out, the upper layer solution can be moved to a tube and then dried by nitrogen flow with gentle heating. After extracting as described above, the lower layer can be combined with the lower layer solution from the first round.) The lower layer is then dried under a nitrogen stream. In the meantime, an HP-TLC plate is warmed at 90–100°C for 20 min and then cooled to room temperature before use. This helps activate the silica gel by driving off water and gives more reproducible TLC lipid mobilities. The dried aLUVs lipids are re-dissolved 50–100 μL of 1:1 chloroform : methanol (v/v) and 10–40 μl are loaded on the TLC plate. Pure lipid standards are combined and various amounts (2– 10 μg ~ 3–14 nmoles) of each lipid) are loaded onto the same HP-TLC plate as the asymmetric vesicle samples to be quantified. A typical solvent system we used for TLC was 50:18:18 chloroform : methanol : acetic acid (v/v). After the solvent front migrated almost to the top of the plate, the plate removed from the TLC tank, and air-dried for at least 30 min in a fume hood. TLC bands were visualized as described in Section 6.4.

8. Notes

• 8.1:It is essential to use glass bore positive displacement (solid plunger) dispensers/pipettors. You cannot accurately pipet with and ordinary air displacement pipettors using plastic tips, because of volatility of the solvent will cause inaccurate measurements and components in the plastic may contaminate the sample.
• 8.2:A multitube vortexer is very convenient as it can take multiple samples and set to always on position, avoiding having to hold the sample being vortexed in place by hand for long periods.
• 8.3:If the volume is large it can be measured in a graduated cylinder, but if it is small, we use an adjustable air-displacement pipette to measure the total volume of the solution.
• 8.4:The investigator may want to use different donor lipids than the POPC used as donor in reference [22]. If a lipid mixture is used, the ratio of lipids in the donor mixture and the ratio of what is exchanged into the acceptor will not be identical and the outer leaflet composition must be calculated from the final lipid composition of the asymmetric vesicles, which can be determined from HP-TLC as described here and references [18, 22]. Also, in addition to the unlabeled lipids, we often wish to include a fluorescent lipid. Details are given in [24].
• 8.5:The investigator may want to use different donor lipids than the 1:1 mol:mol mixture of POPE, POPS and cholesterol, used as donor in [22].
• 8.6:The investigator may want to use different donor lipids than the SM, POPC, and the 3:1 mixtures of POPC with DOTAP, POPG, POPA, POPS, POePC, used as donor in references [17, 18, 22]. If a lipid mixture is used, the ratio of lipids in the donor mixture and the ratio of what is exchanged into the acceptor will not be identical and the outer leaflet composition must be calculated from the final lipid composition of the asymmetric vesicles, which can be determined from HP-TLC as described here and reference [18, 22].
• 8.7:The investigator may want to use different donor lipids than in [18, 24] In these papers, the acceptors included dioleoylphosphatidylcholine (DOPC), or POPC, or the 3:1 mixtures of POPC with the same lipids noted in the donor samples, as well as CHOL, generally 25–40 mol%. Note that we found yield of aLUVs often decreased at lower CHOL concentrations using our protocols.