Dynamics of the membrane lipid phase

Dynamics of the membrane lipid phase

PROSTAGLANDINSLEUKOTRIENES ANDESSENTIALFATTYACIDS Prastaglandms Leukotnene, and Ewntial D Longman Group UK Lrd 1993 Fatty Aads, 19931 3% 27-32 Dyna...

716KB Sizes 0 Downloads 22 Views

PROSTAGLANDINSLEUKOTRIENES ANDESSENTIALFATTYACIDS Prastaglandms Leukotnene, and Ewntial D Longman Group UK Lrd 1993

Fatty Aads,

19931 3% 27-32

Dynamics of the Membrane Lipid Phase S. Cribier, G. Morrot and A. Zachowski Institut de Biobgie


13 rue Pie/-r-e et Marie Curie. 75005 Par-is, FI-awe (Reprint requests to AZ)

ABSTRACT. The lipid bilayer of a membrane is sometimes seen as an inert hydrophobic phase allowing the ‘solubility’ of transmembrane proteins and acting as a barrier between two compartments. However, the bilayer is, in fact, a highly organized system subjected to many movements leading to a dynamically equilibrated structure. A lipid within a membrane experiences intramolecular motions (movement of some segments of the molecule) and moves or diffuses in and across each monolayer. In plasma membranes, transverse diffusion is either passive (cholinecontaining phospholipids, fatty acids. . .) or active via a carrier protein (aminophospholipids). The known asymmetric transverse distribution of phospholipids between the two plasma membrane leaflets is a stationary state resulting from all these motions, especially the active transport. Nevertheless, recent studies have shown that it is also possible to obtain an uneven distribution of some lipids (e.g. fatty acid, phosphatidic acid) across a membrane via a pH gradient. Lateral diffusion within a monolayer depends on the composition of the monolayer and not on the nature of the diffusing lipid. The phospholipid asymmetry, based on the polar head groups, exists also for the corresponding fatty acids, as the nature of the acyl chains differs according to the head group. A consequence is that the cytoplasmic leaflet of plasma membranes has a different ‘fluidity’ from that of the outer leaflet.

The lipid bilayer of a biological membrane is a mosaic of several molecular species. The basic brick of this edifice is the phospholipid. A phospholipid is based on a glyceride or a ceramide backbone. The fatty acyl chains are part of this structure through ester, ether or amide bonds. The polar head group consists of a phosphate residue which is (except for the phosphatidic acid) esterified by an alcohol. which itself may be modified by phosphorylation (phosphoinositide family) or by glycosylation (glycophospholipids). Some compounds, which derive from the phospholipids, may exist transiently in the bilayer: (free) fatty acids, lyso-phospholipids, di- and mono-acyl glycerides. Particularly in the plasma membrane of animal eukaryotic cells. cholesterol is the other main lipidic component of the bilayer. The lipid bilayer is subjected to various movements. A first category is represented by the intramolecular morions which occur at different levels of the lipid molecule. A good example of such a movement is the rotation of the C-C bonds of the aliphatic chains, allowing the acyl chains to ‘swing’. The other movements concern the lipid molecule as a whole in the bilayer. They are: l

I-otationul diffirsion of the lipid molecule around its main axis, i.e. around an axis which is roughly perpendicular to the membrane surface. This



motion, together with the intramolecular motions, reflect the microviscosity existing around a lipid molecule. exchange of the lipid between the two monolayers forming the bilayer. called rralzSl’el.se [email protected] or flipj7op. Such a movement is highly dependent on the nature of the lipid molecule and on the membrane in which it is embedded. later-al diffusion of a lipid within the plane of a monolayer.

Finally. one has to recall that a biological membrane also contains integral proteins. Within the lateral diffusion time-scale of lipids, proteins are immobile. Thus lipid molecules collide with proteins. and we must consider the exchange of the lipid between its position in the immediate vicinity of a protein (in the first lipid shell) and its position at some distance from the protein. This phenomenon has been the subject of a lengthy debate concerning lipid-protein interactions. All these motions have been studied using various biophysical approaches. Before reviewing some of the current knowledge on these movements, one has to recall that such studies are only meaningful if the characteristic time of the motion studied matches the characteristic time of the method applied. If that is not the case. the experimenter will be unable to quantify the movement taking place.





and Essential Fatty Acids


The method of choice to study this type of membrane dynamics is nuclear magnetic resonance (NMR). Unfortunately, due to the rather low sensitivity of this technique, experiments have been often carried out with artificial membranes which allow one to obtain the high lipid concentration required to get useful spectra. However, one can hope that the segmental movement of a lipid in an artificial bilayer is equivalent to that it experiences in a natural, protein-containing membrane. A first approach to the study of movement is to determine the spin-lattice (T,) relaxation time of 13C carbon atoms substituted at each position along the fatty acyl chain of a phospholipid (1, 2). This parameter is in fact only sensitive to very fast motions of the type experienced by these atoms. Such measurements show that the first 8-9 carbon atoms of the chain rotate with a characteristic time in the order of 0.10 ns. This time decreases when going further along the chain, being reduced by more than an order of magnitude at the level of the terminal methyl group. NMR studies can also be carried out with (per)deuterated lipids. Using solid-state spectrometers, each deuteron gives rise to a spectrum characterized by a quadrupolar splitting which can be related to a C-D segmental order parameter (3). These measurements also show that the upper part of the phospholipid chains is more ordered (the ‘plateau region’) than is the bottom part. The same results are obtained if the probe is a ‘free’ fatty acid intercalated in the bilayer. Similar values are obtained with labelled chains in position sn-1 and ~11-2 of the glycerol, with the exception of the first two carbons of the sn-2 chain, a consequence of its disposition within the monolayer: the first two C-C bonds are parallel to the interface, before a turn which allows the rest of the chain to be perpendicular to the interface (4). Note that the presence of a double-bond modifies the order parameter due to a tilting of the C=C bond from the normal to the bilayer. However, if one considers the molecular order parameter (which integrates the orientation of the bond), there is no such influence (5). If the chain contains two unsaturated bonds, the segment between the double bonds exhibits a very low molecular order parameter reflecting the possibility to jump between two conformations (6). With three (or more) unsaturated bonds, the jump seems unlikely. as it would orient the fatty acyl chain parallel to the membrane surface. NMR spectroscopy also allows one to study internal motions of the cholesterol molecule. All measurements (7, 8) indicate that the steroid ring is rigid. The tail of the cholesterol molecule is subject to motions, but its measured molecular order parameter is higher than that recorded for a fatty acyl chain at an equivalent depth in the monolayer. In a bilayer, in the fluid liquid-crystal state. cholesterol influence on the intramolecular dynamics of the phospholipid chains has been shown: cho-

lesterol increases the order in the upper part of the bilayer (at the level of its rigid steroid ring), but has almost no effect at the center of the structure (8). Studies of the motion of the different segments of a fatty acid, free or acylated to a phospholipid, have also been carried out with spin-labelled (9) and fluorescent (10) analogues. The results were similar to those described above, an upper region of the chain exhibiting rather constrained motions and an increase in freedom upon going further along the chain. One advantage of labelled molecules is that the associated techniques (fluorescence or electron spin resonance -ESR) are more sensitive than NMR, thus allowing measurements with biological samples. In this way. they have often been used to compare the effects of different experimental conditions on membrane behavior and to compare different membranes. For instance, labelling human erythrocyte membranes by spin-labelled analogues of phosphatidylcholine and phosphatidylserine (located in the outer and the inner half of the bilayer respectively, see below) demonstrated that local motion is easier in the inner (cytoplasmic) leaflet than in the outer one (11-13).

Rotation of the lipid molecule The velocity of the rotation of a lipid molecule around its long axis can be derived from various physical methods. In all cases, the values obtained are indicative of a very fast movement. Using NMR. determination of the spin-lattice correlation times allows one to calculate the characteristic time of this rotational diffusion. Correlation times of approximately 0.1 ns for cholesterol and from 0.1 ns to 10 ns for phospholipids have been reported (6,14,15). Results obtained when using another resonance technique, namely ESR, are in agreement: spectral analysis and simulation have given molecule rotations occuring in the ns range (15).



or flip-flop

Several methods have been used to follow the movement of lipids between the two membrane halves. As they have been recently reviewed (16), no details will be given here. Briefly, they consist in introducing a lipid probe (labelled with a radioactive isotope or bearing a fluorescent or paramagnetic reporter group) into one monolayer and in following its redistribution into the other monolayer as a function of time. All such studies led to comparable results. and three types of transmembrane movements are described: Actil>e tr-ansport In all the plasma membranes studied so far (and also in one type of intracellular membrane, the chromaffin granule), the translocation of aminophospholipids (PS and PE) from the outer (or intravesicular) leaflet to the cytoplasmic leaflet is catalyzed by a protein which

Dynamics of the Membrane Lipid Phase

concomitantly hydrolyzes ATP. The carrier system, aminophospholipid translocase, has been shown to be sensitive to sulfhydryl reagents, vanadate or fluoride, to be saturable by both the aminophospholipid and MgATP and to have approximately 10 times higher affinity for PS than for PE (17, 18). As a consequence, the inward movement of PE is somewhat slower than that of PS (‘l/2 of 40-60 min and <5 min respectively in human red blood cells at 37 “C), but much faster than that of the other phospholipids (see below). Recent investigations have shown that the stoichiometry of the transport is one ATP hydrolyzed per lipid translocated (Z. Beleznai, personal communication). The translocase does not recognize any other natural phospholipid found in plasma membranes. Using analogues modified either at certain head group residues or within the backbone and linked to various fatty acids, it was possible to characterize the translocated lipid: it should have a primary amine at the top of its polar moiety, a glycerol backbone esterified at the sn-2 position by a fatty acyl chain of at least 5 carbons; the nature of the bond at the sn-1 position is less crucial, as a di-acyl PE and a plasmalogen PE are transported equally well (13 and P. Fellmann, personal communication). The very fast inward movement of aminophospholipids. together with a very slow diffusion of the choline-containing phospholipids (see below) explains the transversal asymmetry of phospholipid distribution within the plasma membrane, where PS and PE are the main components of the cytoplasmic leaflet while PC and SM are found in the outer half. If one seeks the molecule(s) responsible for this specific transport. the first approach is to look for a vanadate-sensitive Mg-ATPase which has, as a second substrate. PS or PE. An activity, meeting these demands, has been described in the human erythrocyte membrane and in the bovine chromaffin granule membrane. two structures where the aminophospholipid translocase is present. This ATPase was identified. after labelling and polyacrylamide gel electrophoresis under denaturating conditions. as a I 16-120 kD protein ( 19). However, this ATPase may be only a sub-unit of the translocase system, where it could be associated with other peptides (notably a 32 kD component (30) according to one proposition).

This type of movement has been characterized in microsomal rat liver membranes. Experiments showed that there is a fast transbilayer movement of a great variety of phospholipids in isolated microsomes, and that this movement is modified after treatment of the membranes by sulfhydryl reagents or proteases (21, 22). This system is unable to accumulate a given lipid in one monolayer and does not require an energy source to ensure the redistribution within the bilayer. When assayed with various phospholipid analogues, the kinetic behaviour remains the same, but there is evidence for a competition between the molecules. One can hypothesize that non-


specific ‘pore’ exists, allowing the phospholipids, which are asymmetrically synthesized, to equilibrate between the two bilayer halves. One has to keep in mind that this system has been studied in isolated membranes and may be more complex in vivo. Simple diffusiotl A type of transmembrane movement which exists in artificial membranes, but which also occurs in natural membranes. It probably uses membrane defects (which can be enhanced by the presence of proteins spanning the bilayer) and is completely independent of the presence of metabolites. The movement is generally slow: for example, the diffusion of PC in the human erythrocyte membrane depends on the precise molecular species. but has characteristic times of the order of several hours (16). At first sight, this diffusion should be unable to produce an asymmetric distribution of lipids within a membrane. This is the general rule, but some exceptions can occur, as described by Cullis’ laboratory over recent years (23-25). This concerns weak bases (stearylamine for instance) or weak acids (free fatty acid. phosphatidic acid (PA), or phosphatidylglycerol (PG)). At any pH, these molecules exist in equilibrium between protonated and unprotonated forms, or. in other words, are either charged or neutral. The neutral species can flip-flop in the bilayer at very high rates due to the absence of electric charges (characteristic times of the order of a second), and can equilibrate between the two membrane halves. Moreover, at both membrane surfaces, these neutral species are in equilibrium with the charged ones. according to the local pH and the pK of the molecule. In consequence. if a pH gradient exists through the membrane. there will be an asymmetric repartition of such molecules (for instance, there will be more fatty acid molecules in the leaflet facing the more basic medium). The asymmetry exists as long as the pH gradient exists, and will be modified within minutes if the gradient changes. It is clear that the phospholipids which permanently bear ionized groups (PE, PS. PC) cannot behave in this way. Among the phospholipids which are weak acids, neither phosphatidylinositol nor diphosphatidylglycerol (or cardiolipin) behave like PA or PG. probably because even in the neutral form they remain highly hydrated. thus imposing a major thermodynamic barrier to a fast llip-flop. However, any lipid can be relocalized quickly in a bilayer if a scrambling event takes place. Its nature is not defined (‘non-bilayer’ structures’?) but it appears. for instance. when some amphiphilic compounds enter the membrane (26) or above few percents of diacylglycerols in a bilayer (27). and also during physiological phenomena such as platelet activation (28).

Lateral diffusion In most plasma membranes, phospholipids are asymmetrically distributed between the two bilayer halves.




and Essential Fatty Acids

This asymmetry is characterized by the polar head groups: schematically, in the outer leaflet the main components are the choline residues of PC and SM, while the primary amine moieties of PS and PE are very abundant in the inner leaflet. An important point to note is that the fatty acyl chain composition is not the same for all the phospholipid classes. A good example is the human erythrocyte membrane: its asymmetry is well determined and the chain composition of each lipid is quantified (29). Both monolayers contain approximately the same amount of saturated chains (48% outer, 43% inner), the outer layer contains more monounsaturated chains than the inner leaflet (42% vs 18%) the reverse being true for polyunsaturated species (10% vs 41%). One might expect that these differences lead to a more ‘fluid’ cytoplasmic monolayer. Several biophysical approaches can be used to measure lateral diffusion of lipids in a membrane. Two of the most common nowadays use fluorescent molecules. The ‘fluorescence recovery after photobleaching’ (FRAP) technique can be summarized as follows: fluorescent probes (for instance, NBD-labelled phospholipid analogues) are inserted into a monolayer and give an uniform fluorescence pattern. A fraction of this fluorescence is then irreversibly destroyed (bleached) in defined areas by an intense laser beam, thus creating an inhomogeneity of the labelling. The experimenter follows the redistribution of the probes between the bleached and the unbleached membrane moieties and has access to two parameters: the lateral diffusion coefficients (D) and the fraction of the probe (the so-called immobile fraction) which cannot diffuse freely (at least during the time of the experiment) within the membrane plane. The significance and interpretation of this latter parameter is often a matter of debate and has been recently reviewed (30). Here we shall only consider the measurement of lateral diffusion. FRAP experiments with human erythrocytes using NBD-labelled analogues of PS, PE and PC allow one to probe lateral phospholipid diffusion in each monolayer (31). due to the differential transverse diffusion of the molecules (see the chapter on transverse movements): PC will mainly be located in the outer leaflet. PS in the inner one and PE represented more homogeneously in both. From 4 “C to 37 “C, a consistently higher diffusion coefficient has been found in the cytoplasmic leaflet (8-9 x 10d9cm’ ss’ at 37 “C) than in the external one (1.5-l .8 x 10m9cmZs-i at 37 “C). It is clear that this difference is due to the location of the probe within one leaflet and not to its chemical nature: the diffusion coefficient in one leaflet, determined with the fraction of each analogue present in this leaflet, is identical. showing that the polar head group has no influence on this parameter, which is an intrinsic property of the host lipid. Moreover, a free fatty acid embedded in the erythrocyte membrane gives the same results, as does a fluorescent cholesterol analogue (32). The fact that this difference in lipid mobility is due to the different chains composing each leaflet was assayed with model mem-

branes. Phos-pholipids were extracted from the erythrocyte membrane and purified according to the polar head group, then bilayers were constructed with the composition of either one of the leaflets (33). Here also, the inner leaflet-like membrane exhibited a 5 times faster lateral diffusion. Addition of cholesterol slightly decreased the values of the diffusion coefficients, but did not affect the difference between them. Note that the latter system allows a faster diffusion than the erythrocyte membrane. showing that membrane proteins slow down the lipid diffusion, probably by molecular crowding within the membrane. This fluidity anisotropy has been described in other membranes, such as hamster fibroblast plasma membrane where it is even more pronounced. It is important to stress that a diffusion coefficient of lOm8cm’s_’ means that one molecule explores an area of 1 pm’ every second and the total surface of a spherical cell 10 pm in diameter in only 5 min. Another approach to probe the lateral diffusion of lipids in a bilayer is to use anthracene or pyrene derived molecules which exhibit different fluorescence when in monomeric form or dimeric (excimer) forms (30). Excimer formation is dependent on the collision rate of monomers which reflects the lateral diffusion of the probe in the membrane plane. This technique is complementary to FRAP as it gives information on short-range diffusion (0.1-l nm) while FRAP studies diffusion over distances in the pm range. Results obtained when erythrocyte membrane was doped with excimer-forming probes are however contradictory, as some authors reported a higher diffusion in the outer leaflet while others found the reverse.



One can wonder whether some proteins exhibit preferential interactions with a given lipid. Several laboratories have made extensive studies of this problem which is well documented in the literature. Typically, the lateral diffusion of lipids in a membrane predicts that when a lipid collides with an integral protein. the time of contact, in the absence of any interaction, will be approximately a nanosecond. A longer residence time will signal an interaction between these two molecules, and a hindrance of lipid motion. Due to the time scale, the method of choice for studying this question is ESR. Spectral analysis has shown that there is an exchange of lipids between a position at the protein periphery and a position not in contact with the protein. In the case of preferential interactions, the exchange time is longer. Data obtained with various lipid analogues in natural membranes or artificial proteo-liposomes have led to the conclusion that exchange proceeds at rates between 10’ to 10” per. indicating that there are no long-lived interactions (no ‘lipid annulus’). The Figure summarizes the data on various types of lipid dynamics discussed in this review.

Dvnamics of the Membrane Lioid Phase







16. 17.

18. Figure Characteristic times of the various movements by a phospholipid in a bilayer.

experienced 19.

Acknowledgments We are indebted to Dr P.F. Devaux for numerous helpful discussions and to Dr R. Lavery for careful reading of the manuscript. This work was supported by grants from the Centre National de la Recherche Scientifique (UA 526). the Institut National de la Sante et de la Rechere Medicale (900104) and Universite Paris 7.



References 22. 1. Lee AG. Birdsall NJM, Metcalf JC, Warren GB, Roberts



4. 5.




GCK. A determination of the mobility gradient in lipid bilayers by iiC nuclear magnetic resonance. Proc Roy Sot B 1976; 193: 253-274. Lepore SL, Ellena JF, Cafiso DS. Comparison of the lipid acyl chain dynamics between small and large unilamellar vesicles. Biophys J 1992; 61: 767-775. Seelig J. Seelig A. Lipid conformation in model membranes and biological membranes. Q Rev Biophys 1980; 13: 1941. Seelig J, Browning JL. General features of phospholipid conformation in membranes. FEBS Lett 1978.92: 4144. Seelig J, WaesepeSarcevic N. Molecular order in cis and tram unsaturated phospholipid bilayers. Biochemistry 1978: 17: 3310-3315. Baenziger JE, Jarrell HC, Smith ICP. Molecular motions and dynamics of a diunsaturated acyl chain in a lipid bilayer: implications for the role of polyunsaturation in biological membranes. Biochemistry 1992; 31: 3377-3385. De Kruijff B. ‘X? NMR studies on (4-l%) cholesterol incorporated in sonicated phosphatidylcholine vesicles. Biochim Biophys Acta 1978; 506: 173-182 Dufourc EJ. Parish EJ. Chitrakom S, Smith ICP. Structural and dynamical details of cholesterol-lipid interaction as revealed by deuterium NMR. Biochemistry 1984: 23: 6066607 1. Hubbell WL. McConnell HM. Molecular motion in spinlabeled phospholipids and membranes. J Am Chem Sot 1971; 93: 314-326. Tilley L, Thulbom KR, Sawyer WH. An assessment of the fluidity gradient of the lipid bilayer as determined by a set of n-(9-anthroyloxy) fatty acids (n=2,6,9.12,16). J Biol Chem 1979; 254: 2592-2594. Tanaka K-I. Ohnishi S-I. Heterogeneity in the fluidity of










intact erythrocyte membrane and its homogeneization upon hemolysis. Biochim Biophys Acta 1976; 426: 2 18-23 1. Sejgneuret M, Zachowski A, Herrmann A. Devaux PF. Asymmetric lipid fluidity in human erythrocyte membrane: new spin-label evidence. Biochemistry 1984; 23: 427 14275. Morrot G, HervC P. Zachowski A. Fellmann P. Devaux PF. Aminophospholipid translocase of human erythrocytes: phospholipid substrate specificity and effect of cholesterol. Biochemistry 1989; 28: 34563462. Yeagle PL. Cholesterol rotation in phospholipid vesicles as observed by 13C NMR. Biochim Biophys Acta 19X1: 640: 263-273. Moser M, Marsh D, Meier P. Wassmer K-H, Kothe G. Chain configuration and flexibility gradient in phospholipid membranes. Comparison between spin-label electron spin resonance and deuteron nuclear magnetic resonance, and identification of new conformations. Biophys J 1989; 55: 11 l-123. Zachowski A, Devaux PF. Transmembrane movements of lipids. Experientia 1990: 46: 644656. Seigneuret M. Devaux PF. ATP-dependent asymmetric distribution of spin-labeled phospholipids in the erythrocyte membranes: relation to shape changes. Proc Nat1 Acad Sci USA 1984; 81: 3751-3755. Zachowski A, Favre E, Cribier S. Herve P, Devaux PF. Outside-inside translocation of aminophospholipids in the human erythrocyte membrane is mediated by a specific enzyme. Biochemistry 1986; 25: 2585-3590. Morrot G. Zachowski A. Devaux PF. Partial purification and characterization of the human erythrocyte Mg’+ATPase. A candidate aminophospholipid translocase. FEBS Lett 1990; 266: 29-32. Schroit AJ, Bloy C, Connor J. Cartron J-P. Involvement of Rh blood group polypeptides in the maintenance of aminophospholipid asymmetry. Biochemistry 1990: 29: 10303-10306. Bishop WR. Bell RM, Assembly of the endoplasmic reticulum phospholipid bilayer: the phosphatidylcholine transporter. Cell 1985: 42: 5 I-60. Herrmann A. Zachowski A. Devaux PF. The protein mediated phospholipid translocation of the endoplasmic reticulum has a low lipid specificity. Biochemistry 1990; 29: 2023-2027 Hope MJ, Cullis PR. Lipid asymmetry induced by transmembrane pH gradients in large unilamellar vesicles. J Biol Chem 1987: 262: 43604366. Redelmeier TE, Hope MJ, Cullis PR. On the mechanism of transbilayer transport of phosphatidylglycerol in response to transmembrane pH gradients. Biochemistry 1990; 29: 30463053. Eastman SJ, Hope MJ, Cullis PR. Transbilayer transport of phosphatidic acid in response to pH gradients. Biochemistry 1991; 30: 1740-1745. Rosso J. Zachowski A. Devaux PF. Influence of chlorpromazine on the transverse mobility of phospholipids in the human erythrocyte membrane: relation to shape changes. Biochim Biophys Acta 1988; 942: 27 l-279. Siegel DP, Banschbach J, Alford D. et al, Physiological levels of diacylglycerols in phospholipid membranes induce membrane fusion and stabilize inverted phases. Biochemistry 1989; 28: 3703-3709. Schroit AJ, Zwaal RFA. Transbilayer movement of phospholipids in red cell and platelet membranes. Biochim Biophys Acta 1991; 1071: 313-329. Myher JJ. Kuksis A, Pind S. Molecular species of glycerophospholipids ans sphingomyelins of human eyrthrocytes: improved method of analysis. Lipids 1989: 24: 396407. Tocanne J-F. Dupou-Cezanne L, Lopez A. Toumier J-F. Lipid lateral diffusion and membrane organization. FEBS Lett 1989; 257: 10-16. Morrot G. Cribier S, Devaux PF et al. Asymmetric lateral mobility of phospholipids in the human erythrocyte membrane. Proc Nat1 Acad Sci USA 1986:





and Essential Fattv Acids

83: 6863-6867. 32. Schroeder F, Nemecz G, Wood WG, et al. Transmembrane distribution of sterol in the human erythrocyte. B&him Biophys Acta 1991; 1066: 183-192.

33. Cribier S, Morrot G, Neumann J-M, Devaux PF. Lateral diffusion of erythrocyte phospholipids in model membranes: comparison between inner and outer leaflet components. Eur Biophys J 1990; 18: 3341.