Functional properties of biological membranes: a physical-chemical approach.

Prog. Biophys. Molec. Biol. 1975.Vol. 29. No. 1. pp. 3 56. Pergamon Press. Printed in Great Britain.

FUNCTIONAL PROPERTIES OF BIOLOGICAL MEMBRANES" A PHYSICAL-CHEMICAL APPROACH A. G. LEE Department of Physiology and Biochemistry, University of Southampton, Southampton S09 3TU

CONTENTS I. INTRODUCTION

5

II. PHOSPHOLIPIDS

6 6 7 14

A. The Structures of Phospholipid- Water Mixtures B. The State of Lipid Fatty Acid Chains C. The Conformation of the Lipid Head Group Ill. LATERAL DIFFUSION OF LIPIDS IN BILAYERS

17

IV. THE E~FECrS OF TEMBERATUREON LIPIDS

21 21 24

A. The Phase Transition in Phosphatidylcholines B. Pre-meltin# and Pre-freezin# Phenomena V. LIPID MIXTURES

25

VI. THE PERME~a31LITVOF LIPID BILAYERS

30

VII. THE EFFECTS OF METAL IONS ON L1PIDS

34

VIII. BIOLOGICALMEMBRANES

A. B. C. D.

The The The The

Structural Organization of Biological Membranes State of Lipids in Biological Membranes Mobility of Macromolecules in Biological Membranes Effects of Temperature on Biological Membranes

IX. THE BIOLOGICAL MEMBRANE: CONCLUSIONS REFERENCES

38 38 40 42 43 45 49

FUNCTIONAL PROPERTIES OF BIOLOGICAL MEMBRANES: A PHYSICAL-CHEMICAL APPROACH A. G. LEE Department of Physiology and Biochemistry,University of Southampton, Southampton S09 3IU "For an answer which cannot be expressed the question too cannot be expressed. The riddle does not exist. If a question can be put at all, then it can also be answered. For doubt can only exist where there is a question, a question only when there is an answer, and this only when something can be said" (Wittgenstein).

I. I N T R O D U C T I O N

Membrane biology has grown rapidly to become a plant of impressive size. But how to find the fruit amongst all the foliage; that is now the problem. The search is especially difficult because the recent growth in membrane studies can be attributed in no mean part to the application of a variety of spectroscopic techniques, and before such studies can provide facts of biological relevance, it is necessary to establish the limits of applicability of each technique. Since this is no trivial task for systems as complex as biological membranes, many studies have created more problems than they have solved. There are already a large number of published reviews covering the applications of particular physical techniques to the study of membranes: enough to prompt many a reflection like those of Ecclesiastes, "Of making many books there is no end, and much study is a weariness of the flesh". The reason for adding one more is to try to put together some of the scattered observations that have been made about biological membranes, to see whether any coherent picture can be made to emerge. The most useful spectroscopic techniques for studies of membranes are X-ray diffraction and the magnetic resonance methods, nuclear magnetic resonance (nmr) and electron spin resonance (esr). In X-ray diffraction studies, data is collected over a long period of time, so that a time-averaged picture of the membrane is obtained: such studies therefore provide information about the basic, structural organization of the membrane. On the other hand, the magnetic resonance methods can provide information about the dynamic processes in membranes, since nmr and esr data are sensitive both to the chemical environment of particular groups in a molecule, and to the motion of those groups. In esr, a free radical such as a nitroxide group (see, for example, II, p. 8) is introduced into the membrane, and it is the unpaired electron of this free radical which is observed. In nmr, magnetic nuclei such as ~H and 13C are observed, so that it is possible to study membranes completely unperturbed by the presence of any probe molecules. Studies of the membrane as a dynamic structure require spectroscopic methods sensitive to motions in the range from 10-11 to a few seconds and this is the range spanned by the magnetic resonance techniques. Problems associated with the application of particular techniques to membrane structure will not be discussed here at any length. Reviews detailing these problems include the following: X-ray diffraction (Levine, 1973); electron spin resonance (Jest et al., 1971 ; Keith et al., 1973); nuclear magnetic resonance (Horwitz, 1972; Lee et al., 1974a); fluorescence (Radda and Vanderkooi, 1972); general techniques (Chapman, 1968, 1971 ; Levine, 1972). Most of these studies have been concerned with the lipid component of membranes; the study of the protein component requires highly simplified systems which have only recently become available. This means that only a rather one-sided picture of the membrane can be presented at the present time. Nevertheless, even such a picture is worthy of attention, since, as has long been recognized, lipids play a crucial role in determining membrane function: "Phosphatides (phospholipids) are the centre, life, and chemical soul of all bioplasm whatsoever, that of plants as well as animals. Their chemical stability is greatly due to the 5

6

A.G. LEE

fact that their fundamental radicle is a mineral acid of strong and manifold dynamicities. Their varied functions are the result of the collusion of radicles of strongly contrasting properties. Their physical properties are, viewed from a teleological point of standing, eminently adapted to their functions. Amongst these properties none are more deserving of further inquiry than those which may be described as their power of colloidation" (Thudichum, 1884). II. PHOSPHOLIPIDS

A. The Structures of Phospholipid-Water Mixtures Phospholipids can adopt a variety of structures in the presence of water, characterized by a long-range organization periodic in one, two or three dimensions (Luzzati, 1968). The commonest of these structures for the phosphatidylcholines (I) is the lamellar phase which

R--CH2--i--iH2 R--CH2--C--CH

II o

o IL CH 2 - - 0 - - P - - 0 - - CH 2 - C H 2 - - N + (CH~) 5

I

O-

(I)

consists of bimolecular leaflets of lipid lying with their planes parallel and separated by regions of water (Fig. 1). The bimolecular leaflet is often referred to as a lipid bilayer, and contains a central core of fatty acid chains surrounded by two planes of hydrated polar groups in an aqueous environment. Aqueous dispersions of phosphatidylcholines contain fragments of the lamellar phase dispersed in excess water. These fragments are referred to as liposomes, and consist of spherically concentric bimolecular leaflets, with water trapped between the bilayers (Bangham and Horne, 1964). On irradiating suspensions of liposomes with ultrasonic (high frequency) sound, the large particles are broken down to Water

Water

Water

FIG. 1. The lamellarphase adoptedby aqueousdispersionsof phospholipids.

Functional properties of biological membranes

flip

flopL~

. x '. •

slow ~ exchange ta diffusion v

~)

FIG. 2. The Singer-Nicolson model for the biological membrane, showing possible lipid motions (adapted from Singer and Nicolson, 1972).

give smaller more symmetrical aggregates. Prolonged ultrasonic irradiation produces vesicles which are closed spheres about 250 A in diameter, each composed of a single continuous lipid bilayer membrane enclosing a volume of aqueous solution (Huang, 1969). Lipid bilayers can also be formed across a small hole in a Teflon diaphragm immersed in an aqueous solution, by painting a small amount of a solution of lipid in n-decane across the hole. The film which is formed gradually thins until it becomes optically "black", at which point the membrane consists essentially of a bimolecular layer of oriented lipid molecules. Such black lipid films have been much used in studies of electrical conductivity (Lauger, 1972). Similar bilayers can be formed in the absence of the n-decane, but they tend to be very fragile (Montal and Mueller, 1972). Lipids also form other, more complex phases in water, but these tend only to occur in the high-temperature and low water concentration regions of the phase diagram. More interestingly, it has also been shown that transitions from one phase to another can be caused by the addition of metal ions. Thus addition of Ca 2 ÷ to an aqueous dispersion of cardiolipin causes a phase transition from a lamellar to a hexagonal phase (Rand and Sengupta, 1972). However, in mixtures of cardiolipin with phosphatidylcholine, the lamella phase persists in the presence of Ca 2 ÷. There is as yet no evidence for the presence of any lipid phase other than the lamella in biological membranes. The most commonly accepted picture of the biological membrane is of a lipid bilayer, with some protein molecules partially embedded in the lipid matrix and some completely penetrating the bilayer (Fig. 2) (Singer and Nicolson, 1972). In this review, we will be very largely concerned with the dynamic, motional, properties of the lipid component of the membrane. Figure 2 illustrates the types of motion that can be expected. Firstly, there is the possibility of rapid internal motion within each lipid molecule. Secondly, lipid molecules might diffuse laterally in the plane of the bilayer, although the rates of this motion could be very different for those lipids associated with proteins and those lipids which make up the bulk of the membrane. Thirdly, lipid molecules may be able to rotate rapidly as a whole about their long axes. Lastly, there is the possibility of transfer of lipid molecules from one side of the bilayer to the other--a motion referred to as "flip-flop". Associated with these lipid motions, there are also various possible motions for the protein molecules: rotation and lateral diffusion in the plane of the membrane, and rotation through the membrane. Evidence has now accumulated for all these motions except the last. B. The State of the Lipid Fatty Acid Chains The physical state of phospholipids depends markedly oI~ temperature. On heating, phospholipids undergo an endothermic transition at a temperature well below the true melting point. At this temperature, a change of state occurs from the crystalline (or gel)

8

A . G . LEE

state to the liquid crystalline state, which is associated with increased conformational freedom for the lipid fatty acid chains. The hydrocarbon chains are said to have "melted". The transition temperature. T,, increases with increasing length of the fatty acid chains, and decreases with increasing unsaturation in the chain; it also depends on the chemical nature of the lipid polar head groups (Ladbrooke and Chapman, 1969; Hinz and Sturtevant, 1972b). Below the transition temperature, in the gel phase, the phospholipids adopt a bilayer structure in which the fatty acid chains are packed in highly ordered hexagonal arrays. At maximum hydration in dipalmitoyllecithin, the long axes of the chains are tilted about 30 ° from the normal to the plane of the bilayer, with the conformation about all C--C bonds close to trans (Levine, 1973). Motion of the fatty acid chains in the gel phase is highly anisotropic and restricted (Salsbury and Chapman, 1968). It is clear from X-ray diffraction studies (Levine, 1973) that phospholipids in the liquid crystalline phase still adopt a bilayer structure, but that the lipid fatty acid chains are considerably more disordered than in the gel. This disorder is due to rotation about C--C bonds. A potential energy diagram for rotation about the central C--C bond of an isolated butane molecule is shown in Fig. 3. The minimum potential energy occurs in the trans conformation, but there are also two other, higher energy minima, corresponding to the gauche conformations. The energy of the gauche conformation is ca. 0.5 kcal/mole above that of the trans conformation, and the energy barrier between the trans and gauche conformations is ca. 3.6 kcal/mole (Hagele and Pechold, 1970). The relative sharpness of the minima and the quite high barriers to rotation mean that the molecule will remain for most of the time in the vicinity of the minima, carrying out torsional oscillations. Typically, C--C bond oscillation frequencies are 100-200 cm- 1 or 3-6 x 1012 sec- ~. The energy barrier preventing rotation between trans and gauche conformations in butane is ca. 3.6 kcal/mole. The rate at which the molecule jumps from one conformation to the other is related to the height of the barrier by the equation (1)

v - ~-kk-T e x p ( - E / R T )

where E is the activation energy for the jump, and 3( is the transmission coefficient for passage over the barrier. Generally it is assumed that ~ = 1 (Heatley, 1974). This then leads to a jump frequency at room temperature of ca. 101° sec- 1. Thus there are approximately 500 vibrations within a given conformation for each jump between conformations. When considering rotation about C--C bonds in the lipid fatty acid chains, it is necessary to take into account the very strong steric interactions between adjacent chains. Simple calculations show that the steric hindrance to rotation about a C--C bond will vary throughout the length of the fatty acid chain (Rothman, 1973). If the chain is in a fully extended, all trans conformation, then rotation of one of the C--C bonds out of its trans position will cause the length of the chain between that C--C bond and the terminal 6 5 U 4

(m°-~'~el) 3 2 I

0 1

0

I

I

60

1

i

i

120

1

I

i

180

I

!

!

240

i

!

1

300

I

i

i

360

Rotation Angle FIG. 3. The potential energy of butane as a function of the rotation angle of the centre bond C 2 H s - - C z H s (from Wunderlieh, 1973).


9

100,

10 ¸

I

I I , , I I , , , i , , , , I , , ,

0

5

10

15

Carbon number FIG. 4. The estimated maximum torsional angle @about C--C bonds in a phospholipid fatty acid chain.

methyl group to rotate along the locus of a cone. If rotation about the C--C bond is too great, then the distal part of the chain will come within the van der Waars radius of a neighbouring chain, and this will be energetically highly unfavourable. If it is assumed that only one bond per chain can be in a non-trans position at any one time, and if jt is also assumed that the nearest-neighbour chains are in the all-trans conformation, then the maximum sterically allowable torsional angle for a given C--C bond can be calculated. The results obtained (Rothman, 1973) are plotted in Fig. 4 for a 16-carbon chain, where bond 1 connects the carbonyl carbon (carbon 1) to carbon 2, bond 2 connects carbon 2 to carbon 3, etc. The allowable amplitude of oscillation away from the trans position increases markedly towards the end of the chain. There are two possible mechanisms for reducing steric interactions and thus allowing increased disorder and increased motion towards the begknning of ~ - c h a i n ! botchare prob' ably important in the liquid crystalline state. In the first, motion in neighbouring chains is correlated, so, that intramolecular rotation in neighbouring chains occurs to a Similar extent. In the second, motions about several C--C bonds within a single chain are correlated, so that the regular packing of neighbouring chains is not greatly disrupted. This second form of correlation has been discussed in particular in connection with the formation of gauche conformations within the fatty acid chains. The presence of gauche conformations within lipid fatty acid chains has been detected using Raman spectroscopy since the band at 1I00 cm- 1, characteristic of the all-trans conformation, is absent for dipalmitoyllecithin in the liquid crystalline state (Lippert and Peticolas, 1971, 1972). To minimize the steric interactions caused by the presence of these gauche conformations, it has been suggested that, rather than a single gauche rotamer, gauche rotamers form about two C--C bonds, to give a so-called 2gl kink (Trauble and Haynes, 1971). Such a kink is formed from an all-trans fatty acid chain by rotating about one C--C bond by an angle of 120° and then rotating about either of the two next nearest neighbouring C--C bonds by - 120° (Fig. 5). The result is a decrease in chain length by one CH2 unit length (1.27 A)and an ~ncrease in volume of ca. 25-50 A 3. The number of kinks present in the liquid crystalline phase can then be estimated from changes in bilayer thickness and volume at the transition. In the gel phase of dipalmitoyllecithin, the thickness of the lipid bilayer is ca. 42 A and the cross-sectional area of one lipid molecule is 45 A 2. At the phase transition the thickness of the bilayer decreases by about 5 A (Chapman et al., 1967; Levine, 1973). At the same time, there is an increase in volume of the lipids which can be detected by measurement of the volume of an aqueous lipid dispersion over the transition temperature. If this volume change is attributed to the fatty acid component of the lipid, then it correg~onds to about a 2.5% increase in volume. This volume is probably a lower limit because it ignores any possible increase in the amounts of water bound to the lipid at the transition

10

A.G. LI-:~

(Trauble and Haynes, 1971 : Melchior and Morowitz. 1972; Sheetz and Chan, 1972; Nagle, 19731. From the change in bilayer thickness and volume, it has bccn estimated that the number of kinks per fatty acid chain above the thermal transition is ca. 0.6 2.0 (Trauble and Haynes, 19711. These figures for the number of kinks arc, however, very approximate because some of the changes in bilayer thickness and volume could be associated with changes in the tilt of the fatty acid chains at the transition (see later). The formation of such kinks will be a dynamic phenomenon, with kinks forming and disappearing at any one position in thc chain. However, the probability of kink formation is unlikely to be equal throughout the chain. It is to be expected that the terminal methyl ends of the chains will exhibit increased disorder and increased motional freedom. The motional gradient along the fatty acid chains has been determined from L3C nmr experiments and the degree of ordering along the chain has been determined by esr experiments. The t3C data for phosphatidylcholines in the liquid crystalline phase indicate the presence of a very marked motional gradient within the lipid molecule: motion increases from the glycerol backbone of the lipid both towards the terminal methyl of the fatty acid chain and towards the - N M e 3 of the head group (Levine et al.. 1972a). The 13C data for the lipid fatty acid chains have been interpreted in terms of diffusion coefficients D~ for rotation about C C bonds (Levine et al., 1972b; Lee et al., 1974b). In the liquid crystalline phase the rate of motion about C---C bonds at the glycerol backbone end of the fatty acid chains is considerably slower than at the terminal methyl group end. For dimyristoyllecithin up to at least carbon 7 of the chains (where carbon 1 is the carbonyl carbon) rotational diffusion coefficients are equal with D~ = 1 x 10 9 sec- t. At the terminal methyl end of the chain, however, motion is faster with D,,..: = 6 × 109 sec- t and D,,, = 8 x 10 l° sec- 1. It should perhaps be emphasized that although the diffusion coefficients for rotation about the C C bonds in the first part of the chain are equal, this does not imply equal m o t i o n s for the carbons in this part of the chain. The motion in space of any one carbon is the resultant of the motion about all the C - - C bonds between it and the glycerol backbone, so that motion will increase along the chain because of the increasing number of bonds about which rotation is possible. As well as the motion about C C bonds within the fatty acid chains, t3C nmr data provides evidence for an axial rotation of the lipid fatty acid chains as a whole. This motion could be due to a rotation of the chain with respect to the lipid glycerol group. It is, however, more likely to be due to a rotation of the whole lipid molecule in the plane

all -t r'ans-chain

2g 1 -

k in k

FIG. 5. The formation o f a 2gl-kind in a fatty acid chain ffrom Lagaly and Weiss, 19711.


11

TABLE 1. COMPARLSONOF ROTATIONALDWFUSIONCOEFFICmNTS(SCC- 1) FOR DIMYRISTOYLLECITHINAT 52°C AND FOR n-ALKANESAT 31 °C

Dimyristoyllecithin c'~ n-alkanes °')

D.,i,i

D~

D,o- |

D~o

7 × 109 1 x 10 ~1

1 × 109 1 x 101°

6 x 109 1.2 x 101°

8 x 101° 6.2 x 101°

t'~ Data from Levine et al. (1972b), Lee et al. (1974b). tb) Average data for n-alkanes, C,2H26, Cl,,H3o and C16H34, from Birdsall et al. (1973), Levine et al. (1974).

of the bilayer. The diffusion coefficient for axial rotation is

Daxia t =

7

x

10 9

SeC- 1 (Lee et

al., 1974b).

The data for dimyristoyllecithin can be usefully compared with that for liquid n-alkanes (Table 1). Although the rotational diffusion coefficients at the terminal methyl ends of the lipid chains are similar to those in n-alkanes, nearer the middle of the chain the diffusion coefficients are a factor of ten smaller. The rotational diffusion coefficient for axial motion of the lipids is a factor of 15 less than in the alkanes. The only motion to be grossly restricted in lipid bilayers in comparison to the liquid alkanes is rotation about an axis perpendicular to the long axis of the molecule: in lipid bilayers this motion is the so-called "flipflop" motion in which a lipid moves from one side of the bilayer to the other, characterized by a half-time of several hours (Kornberg and McConnell, 1971). Data for didecanoyllecithin, dilauroyllecithin and dipalmitoyllecithin are similar to that for dimyristoyllecithin. Values for D,o and Do_ ~ are equal to those in dimyristoyllecithin, although at the glycerol backbone end of the lipids there is probably a threefold increase in Di for the shorter chains. Motion in the fatty acid chains of dioleoyl lecithin is comparable to that in lipids with saturated fatty acid chains, except for the olefinic carbons where motion is restricted in comparison with neighbouring C--C bonds (Lee et al., 1974b). It has been suggested (Schindler and Seelig, 1973) that the microviscosity at a particular depth within the bilayer can be estimated from the "effective diffusion coefficient" Daf using the Stokes equation, (2)

tl = k T / 8 n r 3 Deff

where r is the effective radius of the molecule and ~/is the viscosity. To the extent that microviscosity is a useful concept in a case where more than one diffusion coefficient is requiredto describe the motion of a group in a molecule, it is clear that the microviscosity will decrease markedly towards the centre of the bilayer. The microviscosities estimated from eq. (2) using the ~3C data are given in Table 2. The microviseosities for the n-alkanes can be compared with the kinematic viscosities of ca. 1-3 x 10- 2 p. In dimyristoyllecithin, the microviscosity at the terminal methyl end of the fatty acid chains is comparable to that in the n-alkanes. For the first half of the chain, however, ~/i is approximately 15 times greater than for an n-alkane. TABLE 2. "MICROVISCOSITIES"(Poise) ESTIMATEDFROM 13C nmr DATA System Dimyristoyllecithin(52°C) n-alkanes(31°C)

~

To - t

240 × 10 -2 30 x 10 -2 15 x 10 -2

~I,o 3 × 10 -2 3 x 10 -2

Although these microviscosities and diffusion coefficients were obtained for sonicated aqueous lipid dispersions, they should apply equally well to lipids in unsonicated lipos o m e s (Lee et al., 1974b); 13C nmr data are consistent with very similar motions for the lipid molecules in the two types of dispersion (lee et al., 1974b). Attempts to attribute a single rnicroviscosity coefficient to the bilayer from fluorescence measurements ignore the gradient present in the bilayer (Cogan et al., 1973; Pownall and Smith, 1973). However, the studi~ with qgg l~ithin gave a microviscosity of ca. 50 × I0- 2 P at 45°C (Cogan et al. 1973~ which is ~mtistent with our estimate for the microviscosity towards the centre of the bilayer. Spin label studies of dimyristoyllecithin at 21.5°C (just

12

A.G. LEE

below the thermal transition temperature) have also been interpreted in terms of a microviscosity of ca. 50 × 10 : P (Schindler and Seelig, 1973), but here the impurity effect of the spin label might be expected to be particularly important (see later). Although too much weight should not be attached to these figures, the very marked increase in microviscosity from the centre of the bilayer to the glycerol backbone region is probably genuine. The importance of these observations to the problem of the diffusion of a small molecule across the bilayer is described on p. 30. The results of esr experiments involving the incorporation of spin labelled fatty acids (II) or lipids (III) into lipid bilayers have been interpreted in terms of an order parameter CH3 (CH2)m/C~H2) n COOH O--N 0 (1"r) CH3 (CH 2)2 COOiH2

CH3(CH2)mC (CH2)nCOOCH ÷

O('fir)

S.. This order parameter is related to the average orientation of the nitroxide radical by the equation S, = 1/2(3cos20,- 1).

(3)

Here n is the number of carbon atoms between the carbonyl carbon and the labelled carbon, 0, is the angle between the nitroxide 2p~ orbital and the normal to the plane of the bilayer at some time, and the bar denotes that the time average of this angle is taken. If the C--C bonds preceding the labelled carbon are all trans and perpendicular to the plane of the bilayer, S, = 1. If the motion of the spin label is isotropic, then S, = 0. For phospholipid spin labels (III) incorporated into bilayers of dipalmitoyllecithin above the thermal transition, it has been found that log S. shows an approximately linear decrease with increasing n up to about n = 8. Beyond n = 8 the decrease in log S. becomes increasingly more marked (Hubbell and McConnell, 1971). These results suggest that the region of the chain up to n = 8 is effectively all trans, but that the probability of a nontrans conformation increases rapidly with increasing n beyond this point. We see then that these nmr and esr results are broadly in agreement that the first half of the lipid fatty acid chain is relatively immobile, but that nearer the centre of the bilayer, disorder is considerably greater. The validity of the esr results is, however, not yet clear because of the possibility of distortions caused by the presence of the nitroxide group. Thus, unlike in dipalmitoyllecithin bilayers, the order parameters for spin labelled fatty acids incorporated into lemella of the sodium decanoate-n-decyl alcohol-water system decrease exponentially with distance between the nitroxide group and the carboxyl group (Seelig, 1970). In marked contrast, order parameters determined by deuterium nmr for lamella of the deuterated potassium laurate-water system are fairly constant for the first half of the chain, and then rapidly decrease (Charvolin et al., 1973). Clearly, a check of the order parameters for bilayers of dipalmitoyllecithin using the less perturbed deuterated chains is necessary. The gel to liquid crystalline phase transition has sometimes been described as a "melting" but this is somewhat misleading. In fact it has been shown that the change in volume for the lipid at the transition is about a factor of six less than the corresponding melting dilation ofpalmitic acid or tripalmitin, implying comiderably more order in the phospholipid system (Trauble and Haynes, 1971; Melchior and Morowitz, 1972). Similarly, whereas the change in configurational entropy AS per CH2 group for the melting of an


13

n-alkane or triglyceride is AS = 1.9 cal mole-1 deg-1, that for a lecithin is only AS = 1.25 cal mole- 1 deg- 1 (Phillips et al., 1969). Although these results have'generally been taken to imply that the lipid fatty acid chains are more ordered throughout their length than are free fatty acids or alkanes, it is more likely that they reflect a relatively high degree of order for the first part' of the chain in the liquid crystalline state, but with the ends of the chain having a fluidity similar to that in an n-alkane. The rapidly increasing disorder towards the centre, of the bilayers creates interesting packing problems. Because of its greater disorder, the volume occupied by a--CH2--group towards the centre of the bilayer will be greater than that for a--CH2--group near the glycerol backbone region. The increase in volume towards the centre of the bilayer can be accommodated in two ways: (i) by a bend in the fatty acid chains, with the upper portion of the chain tilted with respect to the bilayer plane (McFarland and McConnell, 1971); (ii) by a decreased packing density in the glycerol backbone region, perhaps with the extra space being taken up by water. As already described, in fully hydrated bilayers of dipalmitoyllecithin in the gel phase, the fatty acid chains are fully extended (all-trans) and tilted with respect to the bilayer surface. This is readily shown by measurement of the area per molecule in the bilayer. Thus the cross-sectional area of two fatty acid chains in the all-trans conformation is 40,6 A 2, and the area per molecule in bilayers of dipalmitoyllecithin in the gel phase at low hydration is 40.3 A 2. At maximum hydration, however, the area per molecule increases to 44.5 A 2 (Levine, 1973). This increase in surface area is accomplished by tilting the whole length of the fatty acid chain with respect to the surface of the bilayer. As the chains tilt, the cross-sectional area in a plane perpendicular to their axes remains constant, but the area per molecule projected at the surface of the bilayer increases. On the basis of esr studies of oriented multilayers of egg lecithin containing spin-labelled phospholipids, it has been suggested that in the liquid crystalline phase, the upper portion of the labelled fatty acid chain is tilted ca. 30° from the normal to the bilayer, whereas the lower portion is nearly perpendicular to the bilayer (McFarland and McConnell, 1971). On average, the bend occurs at about bond 8, and persists for a time longer than 10-a sec. Such a bent chain means that the area occupied at the bilayer surface by a--CH2-group in the first half of the chain is greater than that occupied by a--CH2--group in the latter part of the chain. The volume occupied by a--CH2--group is therefore greater for the lower portion of the chain than for the upper portion, but at the same time the packing of the ordered, upper portion of the chain is still relatively tight. However, the packing cannot be as tight as in the gel phase, since the area per lipid molecule in the liquid crystalline phase is ca. 70 A 2, and the area per molecule in the bent chain model with surface packing as close as in the gel phase would be only 45 A 2. Further, as discussed by McFarland and McConnell (1971), it is not clear whether or not the bent fatty acid chain is an artifact caused by the presence of the nitroxide group. Monolayer studies (Tinoco et al., 1972; Cadenhead and Muller-Landau, 1973) of spin-labelled fatty acids have shown that the nitroxide group has a tendency to orient at the air-water interface. It is therefore possible that in the lipid bilayer, when a nitroxide group is close to the beginning of the chain, it might tend to localize at the lipid-water interface, thus creating the observed bent fatty acid chain. No such bent chain could be detected by deuterium nmr in lamella of the deuterated potassium laurate-H20 system (Charvolin et al., 1973). Although this result only implies that there is no net bend in the chains with a lifetime of longer than about 10 - 6 sec, it clearly shows that in this system at least, it would be misleading to talk of any "longlived" net bend in the chains. Whether or not the lipid fatty acid chains can be described by a bent-chain model, it is clear that the packing in the glycerol backbone region of the lipid bilayer must be "looser" in the liquid crystalline phase than in the gel phase. It seems likely that this extra space will be taken up by water molecules, but the evidence for this is, as yet, rather scanty. Studies using differential scanning calorimetry and proton nmr indicate that ca. 6-10 molecules of water are "strongly bound" per lecithin molecule in a bilayer (Ladbrooke and Chapman, 1969; Gottlieb et al., 1973). This presumably includes water associated with the

14

A.G. LEE

15"0

~ I00 ~. 50

-I (Dimensionless)

OO

FIG. 6. A possible shape for the barrier to ion penetration presented by the lipid bilayer. The parameter ~ = 2x/d where d is the membrane thickness and x is the distance from the middle of the bilayer (from Hall et al., 1973).

lipid polar head groups as well as any water within the non-polar regions of the lipid bilayer. Deuteron nmr studies show that changes in water binding occur at temperatures near Tt (Salsbury et al., 1972) but again these changes could be associated with alterations in the structure of the polar head group (see below) rather than being due to any changes in the glycerol backbone region of the bilayer. Spin-label data can potentially provide information about water penetration into the bilayer, since spin label coupling constants are sensitive to the polarity of the environment. Studies with spin-labelled fatty acids and lipids incorporated into phospholipid bilayers have in fact been interpreted as showing that water molecules penetrate into the fatty acid chain region, to at least the C-2 position (Griffith et al., 1974). However, the same result would be obtained if the spin label moved up into the lipid-water interface, with no water penetration into the bilayer. To summarize, therefore, there is a marked gradient of motion within a lipid molecule in a bilayer when above the temperature of the gel to liquid crystalline phase transition. Motion increases from the glycerol backbone of the lipid both towards the polar head group and towards the terminal methyl group of the fatty acid chain. The "microviscosity" at the centre of the bilayer is similar to that for a short-chain alkane. The increasing disorder and increasing motion towards the centre of the bilayer imply that the volume occupied by a --CH2--group at the centre of the bilayer is greater than the volume occupied nearer to the glycerol backbone. In view of the "loose" packing in the glycerol backbone region of the bilayer, it seems likely that water molecules will be able to penetrate some way into the fatty acid chain region of the bilayer. Such penetration would produce an approximately trapezoidal polarity profile for the membrane. Recent studies of the nonactin mediated K + transport through black lipid (phosphatidylethanolamine) films have in fact been interpreted in terms of just such an energy barrier (Fig. 6) (Hall et al., 1973).

C. The Conformation of the Lipid Head Group The conformation of the zwitterionic polar group of lecithin is not yet agreed. There are two schools of thought: one believes that above the thermal transition temperature, the zwitterion is oriented with its axis normal to that of the chains whereas the other TABLE 3. TRANSITIONDATA FOR AQUEOUSLECITHINDISPERSIONS(Hinz and Sturtevant, 1972b)

Upper transition

Lower transition Lipid

T~

T~

AH (kcal mole- ~)

Co-operative unit

Dimyristoyllecithin Dipalmitoyllecithin DistearoyUecithin

13.5 34.0 49.1

23.7 41.7 58.2

6.26 + 0.18 9.69 + 0.21 10.84 + 0.17

200 + 40 70 +__ 10 80 + I0


15

believes that it is oriented parallel to that of the chains. Part of the confusion probably arises because of an apparent change in conformation for the polar group at a temperature just below that of the gel to liquid crystalline phase transition (Table 3; Ladbrooke and Chapman, 1969). Some studies suggest that above this transition, the polar group is oriented with its axis parallel to the chains. Thus proton relaxation studies in lecithin monohydrates have been interpreted in terms of a change in conformation from a bent structure in the gel phase to a straightened form in the liquid crystalline phase (Salsbury et al., 1970). Similarly, proton and 13C nmr studies of aqueous phospholipid dispersions suggest a change in conformation at the temperature of the pre-transition (Lee et al., 1972; Levine et al., 1972a). Below this transition, there is a very marked increase in the light scattering by sonicated phospholipid dispersions suggesting vesicle aggregation. This aggregation is prevented by the presence of multivalent ions. The simplest explanation is that above the transition, the zwitterionic choline phosphate group is in a conformation with t h e - NMe~ group extended so that it dominates the surface charge of the vesicle. Below the transition the choline phosphate dipole lies parallel to the surface of the bilayer, so that the surface is a mosaic of dipoles with no net charge. Aggregation of the vesicles could then occur by lateral dipole-dipole interactions in the surfaces of the vesicles. These interactions would be sensitive to cations binding strongly to the phosphate group, leaving the surface with net positive change, and hence preventing aggregation (Levine et al., 1972a). Also consistent with this type of change in headgroup conformation is the observation that the number of binding sites for 1-anilino-8-naphthaline sulphonate (ANS) and for metal ions increases markedly above the transition (Trauble, 1971a). Finally, studies of dipalmitoyllecithin using far infrared absorption have been interpreted in terms of a model with the choline-phosphate dipoles lying parallel to the plane of the bilayer below the transition (Colbow and Clayman, 1973). In contrast, however, Hanai et al. (1965) conclude from electrophoretic studies on egg lecithin dispersions at 20°C that the zwitterion is oriented with its axis normal to the chain, although they note that "this test of the polar group orientation is not an exceptionally good one owing to the approximations made". Phillips et al. (1972) have attempted to obtain the headgroup conformation by extrapolation of X-ray long spacings obtained for lecithins of chain length 14, 16 and 18 back to zero chain length, for which the long spacing would correspond to the long spacing of the polar groups. They conclude that in the gel phase, the polar groups of the lecithins are predominantly paralle.l to the long axis of the molecule (Phillips et al., 1972). Unfortunately, only three data points were available, of which one had to be discarded: complications could also arise from differences in the angles of chain packing between the various lecithins. The weight of evidence seems to favour an arrangement for aqueous lecithin dispersions with the headgroup folded down in the gel phase (Fig. 7a) and extended in the liquid crystalline phase (Fig. 7b). The arrangement of the headgroup in lecithins above the transition obviously provides much more scope for the binding of ions, water and other small molecules, as is observed. The lecithin-NMe~ groups, although relatively more restricted spacially below the transition than above it, still have appreciable motional freedom, as is shown by 13C T1 and line width studies (Levine et al., 1972a). Perhaps not surprisingly, the temperature of the conformational change for the lecithin headgroups is increased by the binding of multivalent ions. Thus in the presence of 50 rnM Eu 3 +, the temperature of the transition for t h e - - N M e ~ groups is increased by ca. 5°C. More interestingly, the temperature of the gel to liquid crystalline phase transition for the fatty acid chains is also increased by about 5°C. (Levine et al., 1973). This is discussed further on p. 34. The conformation of the glycerol backbone region of the phospholipids is not known in detail. Proton nmr data for solutions of dipalmitoyllecithin in organic solvents can be analysed to give the relative conformations about the C--C bonds of the glycerol group, but, unfortunately, line widths in the proton nmr spectra of aqueous lipid dispersions are too broad to allow the same sort of analysis (Birdsall et al., 1972). Crystal structures have been determined for glyceryl phosphoryl choline (Abrahamsson and Pascher, 1966) and

16

A.G. LEE

(a)

(b) FIG. 7. Possible conformations for the headgroups in dipalmitoyllecithin bilayers, (a) in the gel phase and (b) in the liquid crystalline phase. its c a d m i u m salt ( S u n d a r a l i n g a m a n d Jensen, 1965) a n d for a sulpholipid (Watts et al., 1972) but the relevance of these solid-state structures to the structure of lipids in a q u e o u s dispersion is n o t clear.


17

TABLE 4. LIPID SELF-DIFFUSIONCOEFFICIENTS Method Egg lecithin at 20~C Spin-labelled lipid in didihydrosterculoyl lecithin Spin-labelled steroid-lipid monolayer system Spin-labelled lipid in egg lecithin and cholesterol ( 4 : l ) a t 25'C Egg lecithin in egg lecithin and cholesterol (4: l) at 28°C

D (cm 2 sec- t)

Refs.

nmr

2 x 10-"

'

esr

1.8 x 10- 8

b

esr

1 x I0 -a

c

esr

7 x 10 - s

e

nmr

ca. 2 x 10 -9

"

( ' ) Lee et al. (1973). (h) Devaux and McConnell (1972). ~c) Sackmann and Trauble (1972). ca) Devaux, et al. (1973).

I11. L A T E R A L

DIFFUSION

OF

LIPIDS

IN

BILAYERS

As well as the fast internal motions within lipid molecules in the liquid crystalline phase, there is strong evidence for a fast diffusion of the lipid molecules in the plane of the bilayer. This evidence comes from esr studies of lipid bilayers and monolayers containing spinlabelled lipids and steroids (Devaux and McConnell, 1972; Trauble and Sackmann, 1972) and from proton nmr studies of unperturbed lipid bilayers (Lee et al., 1973). The self-diffusion coefficients obtained using these techniques are compared in Table 4. Diffusion has also been shown to occur in lipid systems in the gel phase, although the rates are unknown. Spin-labelled hydrocarbons incorporated into tristearin congregate into separate pools below the optical melting point. This occurs even when the samples are rapidly cooled, so that the spin labels are initially trapped in a solid triglyceride matrix. Although rates of diffusion were not measured, the effects of the diffusion could be detected ca. 5 sec after the sample had been cooled to more than 50°C below the bulk melting point (Cohn et al., 1974). It is not yet clear which is the best model to use to describe this diffusion: the choice of model depends very much on the extent of any long-range order postulated to be present in the gel and liquid crystalline states. In a crystal with a high degree of long-range order, four possible mechanisms for diffusion are usually considered (Shewmon, 1963; Nabarro, 1967). The three simplest possibilities are illustrated in Fig. 8. The first is diffusion by interchange by rotation about a midpoint. The second mechanism is by migration through interstitial sites. The third mechanism is by a molecule exchanging position with a vacant lattice site. The first two are likely to be high energy processes, since they involve considerable steric repulsion. The third mechanism accounts best for self-diffusion in plastic organic crystals such as cyclohexane (for which D ~ 10- 7 cm 2 sec- 1; Hood and Sherwood, 1966: Chadwick and Sherwood, 1971). For such a vacancy diffusion process, the self-diffusion coefficient can be expressed as O = va~'~, exp ( - A G * / R T)

(4)

where 3, is a correlation factor dependent upon the lattice geometry and the diffusion mechanism, ao is the nearest-neighbour distance in the lattice, v is the jump frequency of ~'-'

~7"q

x

X

x

, ~ ~'< ~-~

A

B

C

FIG. 8. Three basic mechanisms for diffusion. A. Interchange by rotation about a midpoint. B. Migration through interstitial sites. C. Exchange of positions with a vacant lattice site. J.P,a. 29.1

18

A . G . LE~:

the moving defect and AG* is the free energy of activation of the diffusion process (Shewmon, 1963). The free energy AG* can be expanded to give D = vao~'exp (ASs + AS")/R 'exp - ( A H I + AH,,)/RT

(5)

where AS/and AS,. are, respectively, the entropies of formation and migration of the defect and A H / a n d AH,. are the corresponding enthalpies. The fraction of the lattice present as vacancies can often be expressed as N,. = exp - ( A G / / R T )

(6)

where AG/is the free energy of vacancy formation. In general N,, -~ 10 - a mol. fraction at the melting point (Shewmon, 1963). From the value of the correlation factor it is possible to derive information about the mechanism of diffusion. Studies of this type with cyclohexane suggest that the vacancy is relaxed, that is, that a number of the surrounding molecules have collapsed to yield a disordered region around the vacancy (Chadwick and Sherwood, 1972). This is confirmed by measurement of the activation volume for the diffusion process in cyclohexane, which shows that the activation volume is some 70~o of the volume of a cyclohexane molecule (Foil and Strange, 1972). Vacancies are, however, not the only types of defect which might be important in diffusion. A more extensive type of defect often observed in crystals is the dislocation (Fig. 9). This can be regarded as though produced by the removal of a line of molecules terminating in the dislocation line. Around the dislocation line, the crystal is strongly disturbed, but further away the original coordination is restored. Rows of such dislocations often occur in crystals and are called mosaic or sub-grain boundaries. Figure 10 shows a raft of hexagonally packed soap bubbles which contains a row of dislocations forming a mosaic boundary, and three grain boundaries. A grain boundary is a boundary between two crystalline grains at which there is a large change oforientation, whereas a mosaic boundary is one where there is only a small change. Particularly at grain boundaries, molecular packing will be highly disordered. In such regions. therefore, the energy needed to form a vacancy or to move a molecule into a vacancy will be lower than for the rest of the lattice, so that these regions can act as "pipes" along which diffusion is greatly enhanced (Shewmon, 1963; Nabarro, 1967). However, since only a small proportion of molecules will be present in grain boundaries, grain-boundary diffusion has to be very much faster than vacancy diffusion before its effects can be seen. This

QOOQOOOt Q eO OOQOOO (3ocJ qt3 ODOD GOCD©O3OO O O3 OD (30 OO 0C30:3 13 ODODOO,3OCOC3OC)If3 FIG. 9. A dislocation in a square lattice. The open circles represent the position of the molecules in the perfect lattice. The filled circles represent the equilibrium positions adopted after a dislocation is formed by the removal of the central column of molecules beyond row tour (from Sworakowski, 1973).


19

FIG. 10. A raft of soap bubbles showing mosaic and grain boundaries where zones of differing orientation meet (from Smith, 1964).

is the case in non-plastic organic crystals such as napthalene, where two diffusion coetticients are detected, one slow (D --- 10- 11 cm 2 sec- 1) and attributable to vacancy diffusion, and the other fast (D -~ 10- 5 cm 2 sec- 1) and attributable to diffusion along grain boundaries (Sherwood and White, 1967). By analogy with these results it seems possible that diffusion in lipid bilayers in the gel phase will be by a defect mechanism since a relatively high degree of long-raflge order exists in this phase. There is in fact a little experimental evidence which suggests the presence of dislocations in the gel phase oflipids. Thus although the resonances in proton nmr spectra of lipids in the gel phase are typically very broad, a few weak but sharp components also appear (Chapman and Salsbury, 1966; Salsbury and Chapman, 1968). Similar sharp components have been observed in the spectra of long-chain paraffins and have been attributed to segmental motion of the chains ingrain boundaries and disordered regions (Odajimo et al., 1962; Van Putte and van den Enden, 1971). Further, X-ray diffraction studies of the membranes of Acholeplasma laidlawii (Engelman, 1971) and of bilayers of dipalmitoyllecithin (Levine, 1974) in the gel phase indicate the presence of grains of size ca. 400

A. The presence of some sort of defect is also highly likely on topological grounds. For a membrane to form a closed surface, it must have two-dimensional curvature, and this

20

A . G . LEE

/ ! /

•x

I

- --(~)

/

"-, \ \

-@

FIG. 11. The construction ofa Voroni polygon. Molecules 2, 3, 4 and 5 are direct geometric neighbours of 1, whereas molecule 6 is an indirect geometric neighbour.

is not in general possible for a perfect two-dimensional lattice. The easiest ways of allowing curvature are by the introduction of lattice dislocations or by bends at grain boundaries. In the liquid crystalline phase, there is clearly much less long-range order than in the gel phase. A description of diffusion in terms of a vacancy diffusion process may, therefore, tend to overestimate the degree of long-range order (Lissowski, 1974). A more valid description of the structure of the liquid crystalline phase might then be in terms of a homogeneous, coherent and essentially irregular assemblage of molecules, containing no crystalline regions or holes large er~ough to admit another molecule, as has been proposed for the structure of liquids (Bernal, 1959, 1964). A description of the structure of such a liquid is made in terms of the geometric neighbours of each molecule (Fig. 11). These neighbours can be used to define the boundaries of the region "occupied" by any molecule. Thus for any two molecules, A; and A> the boundary between Ai and Aj can be defined as the line which is the perpendicular bisector of the line joining Ai and Aj (Fig. 11). These boundaries then define the area "occupied" by the molecule, and this area is usually called the Voronoi polygon associated with that molecule. The surface of a two-dimensional liquid can be considered to be made up of the Voroni polygons associated with each molecule in that surface (Fig. 12). The importance of such a

FIG. 12. A schematic representation of the structure of a lipid bilayer in the liquid crystalline phase (adapted from Lissowski, 1974).


21

description lies in the fact that topological arguments show that for an irregular assembly of molecules, the average number of neighbours of any given molecule is six, and thus that the average Voroni polygon has six sides. For every five-sided Voroni polygon the surface must then also contain one seven-sided polygon (this can be seen in Fig. 12). The fundamental characteristics of a liquid are the irregular arrangement of molecules and the continual changes in the neighbourhood relations of the molecules. This continual change of neighbours gives rise to the high fluidity of liquids and to self-diffusion in the liquid (Bernal, 1959, 1964; see also Gleiter and Lissowski, 1971). Figure 12 probably presents a fair picture of the surface structure of a lipid bilayer in the liquid crystalline state. However, it is not yet possible to proceed beyond this purely geometric description to any sort of quantitative expression for rates of diffusion. IV. T H E

EFFECTS

OF TEMPERATURE

ON LIPIDS

A. The Phase Transition in Phosphatidylcholines The transition.between gel and liquid crystalline phases in lipids is generally very sharp and has been likened to melting. Thus in dimyristoyllecithin, the transition from 10~ lipid in the liquid crystalline phase to 90~ in the liquid crystalline phase can occur within about 0.2°C, although the width of the transition depends markedly on the prehistory of the sample (Hinz and Sturtevant, 1972b). The sharpness of the transition shows that the change of phase is a co-operative phenomenon. In this it is unlike a chemical transition (i.e. a chemical reaction) in which the equilibrium composition varies in a smooth and continuous way with change in a variable such as temperature. In a chemical case such as a dilute gas reaction ,4 -~ B, the probability that any given molecule is in state ,4 (or B) is independent of the state of the other molecules in the gas. In a phase transition, however, where A might represent a molecule in a dense phase and B a molecule in a dilute phase, there is a tendency for a large number of molecules to change as a group from state B to state ,4, because the molecules in state A can stabilize each other through intermolecular attractions. That is, the tendency for a particular molecule to change from one phase to another is not independent of the state of the other molecules around it. The sharpness of a phase transition depends on the number of molecules forced to cooperate in the transition: this number of molecules is called the co-operative unit of the transition. In an exact thermodynamic treatment, a first-order phase transition is defined as a transition for which there is a discontinuity in some thermodynamic function. Such "sharp" transitions only occur for an infinite co-operative unit and could correspond, for example, to the melting of a perfect infinite crystal (Fig. 13). Real crystals contain many crystal imperfections which can serve as nuclei for melting in local regions, so that the cooperative unit will be much smaller than the total solid. These small systems will exhibit

State A

4-N finite

~

infinite

State B

Vm Fro. 13. The variation of molar volume V. with temperature for finite and infinite size co-operative units.

A.G. LEI~

22

State B

la

State A

To

T

Fit;. 14. The definition of the transition temperature To in a system of small co-operative unit.

more or less gradual changes which approach discontinuities more closely the larger the system (Fig. 13). Although there is no exact thermodynamic treatment for transitions in small systems, an approximate treatment is possible (Hill, 1963). Most importantly, the temperature T, of the "phase transition" can be defined by the equal-distance theorem (Fig. 14), by making the distances ab = be. Only one attempt has so far been made to provide a model for the phase transitions in lipid bilayers and that is due to Nagle (1973). The transition is treated as an order-disorder transition, with all-trans fatty acid chains below 7], and yauche conformations appearing above 7],.The model provides a qualitative fit to the experimental data, but, unfortunately, it was necessary to assume that the diffusion coefficients for rotation about C - - C bonds were very small, so that contributions due to kinetic energy could be ignored: in fact, 13C nmr results show that these diffusion coefficients are large (Levine et al., 1972b). Useful information about the transition can, however, be obtained by the application of the simple model developed for transitions in linear polymers (Engel and Schwarz, 1970). It it assumed that the phase transition in lipid bilayers can be described as an all-ornone, two-state transition of the type A~-;B

where any given molecule can only be in state A (gel) or B (liquid crystalline). The degree of the transition 0 is defined as (7)

0 = CB/Ca

where Cs and CA are respectively the fraction of molecules in states B and A. Equilibrium in the system is not described in terms of single molecules but in terms of the co-operative unit for the transition. An "'apparent equilibrium constant" K can then be defined by K = 0/'1 - 0

(8)

The variation of K with temperature is given by the van't Hoff equation: d In K _ AHva,.t Hoff

1

dO R T 2 - = O(I - Oj d-T"

dT

19)

In particular, at the mid point of the transition 0 = ½ (and. by definition, T = T,) we have ~T~j.r,

=-

RTt

= 4 -tit r,

and so AH,.an,, Hoff

=

4 R T~ d-T" dO

(11)


23

The van't Hoffenthalpy will be greater than the enthalpy AH t determined calorimetrically, since the former corresponds to the transition of a co-operative unit and the latter corresponds to a single molecule. A co-operativity parameter ~r can now be defined by

O"

kAHvan.t Ha'f/ "

The smaller, the value of tr, the greater the co-operativity. For linear polymers of infinite length it has been shown that N = 1/tr I

(13)

where N is the size of the co-oper~/tive unit (Engel and Schwarz, 1970). The co-operative unit has been estimated in this way from calorimetric data for a number of aqueous dispersions of lecithins and values are given in Table 3: the size of the co-operative unit was found to vary with sample prehistory. From variations of 13C linewidth through the transition, the co-operative unit is estimated to be at least 70 molecules in dimyristoyllecithin (Lee et al., 1974a). The width of the transition can also be measured from the temperature dependence of the partition coefficient of the small spin label Tempo (IV) in the lipid-water system.

clxt3 When the lipid is in the liquid crystalline phase, the partition coefficient is large, but in the gel phase the solubility of Tempo is much reduced (Hubbell and McConnell, 1971). Tempo partitioning is therefore a sensitive measure of the phase transition. The partition coefficient is very conveniently measured using esr since separate signals are observed for Tempo partitioned into the bilayer and in solution in the aqueous phase. From studies of the temperature variation of Tempo binding the width of the transition in dimyristoyllecithin is estimated to be ca. 2°C, so that the co-operative unit is ca. I00 (Lee et al., 1974a). On sonication of these aqueous lipid dispersions to give small, single shell vesicles (diameter ca. 250 A) the width of the transition increases, corresponding to a drop in the size of the co-operative unit to ca. 8. There is also a decrease in the transition temperature, so that 13C nmr studies of sonicated dimyristoyllecithin indicate a transition centred at 22°C (Lee et al., 1974a). An alternative explanation of the Tempo partitioning data for sonicated dipalmitoyllecithin has been suggested by Sackmann et al. (1973). They suggest that Tempo is partitioning into the polar region of the bilayer, and is detecting the pre-transition at 24°C rather than the main fatty acid chain transition at 42°C. However, in unsonicated dipalmitoyllecithin, Tempo is certainly detecting the main transition at 42°C, and there is no obvious reason why the Tempo environment in sonicated and unsonicated lipid dispersions should be radically different. Indeed, it has been established (Hamilton and McConnell, 1968) that the ratio of the hyperfine splitting a, for the esr spectrum of Tempo in two different environments is related to the ratio of the environmental polarities, a n d a , ( H 2 0 ) / a , ( D P L ) --1.05 in both sonicated and unsonicated lipids (Lee et al., 1974a). The phase transition can also be detected using spin-labelled fatty acids or lipide incorporated into the lipid bilayer, but the data is rather confusing. A spin-labelled lipid incorporated into dipalmitoyllecithin shows a transition broader and to lower temperature in sonicated than in unsonicated dispersions (Taupin and McConnell, 1973). However, a number of spin-labelled fatty acids incorporated into sonicated dipalmitoyllecithin exhibit changes at different temperatures depending on the position of the spin label in the fatty acid chain (Sackmann et al., 1973), and, indeed, at low concentrations do not detect any transition at all (Hubbell and McConnell, 1971). Some, at least, of these differences can

24

A.G. LEE

probably be attributed to impurity effects caused by the added spin labels, which will prevent gel formation around the nitroxide group until after the bulk matrix has solidified. The small lowering of T, on sonication is a typical "small system effect" (Hill, 1963). T, is related to the enthalpy and entropy change of the transition by AH -~ TtAS. There are many possible reasons for differences in AH between single shell sonicated vesicles and multishell liposomes, including differences in interfacial tension and differences in electrostatic forces between the sheets of bilayer. There are also possible differences in AS. Thus it has been found that the apparent molal volume of dipalmitoyllecithin in the gel phase is ca. 1% greater in sonicated dispersions than in unsonicated dispersions, whilst the apparent molal volumes are equal for the two systems in the liquid crystalline phase (Sheetz and Chan, 1972). Although the interpretation of this result is complicated by possible differences in water binding to the two preparations (sonicated dispersions have been shown to aggregate below Tt; Lee et al., 1972), it could indicate a more disordered structure for the gel phase of the sonicated dispersions. To summarize, therefore, the range of temperatures over which the gel and liquid crystalline phases of a particular lipid coexist depends on the co-operativity of the transition. If the co-operativity is high, then the range of coexistence temperatures is small, but if the co-operativity is low, then the temperature range is large. Although the co-operativity is high for unsonicated dispersions of a single pure lipid, it may not be for lipid mixtures or for lipid-protein mixtures: this will be discussed later. It has been pointed out by Linden et al. (1973) that over the temperature range for which gel and liquid crystalline phases coexist, the system will exhibit a number of interesting thermodynamic properties. In particular, fluctuations in density within the bilayer will be large. In a single, pure phase, fluctuations in any thermodynamic property about its mean value will be very small. In a two-phase equilibrium, however, although density fluctuations within each pure phase will be normal, in any given element of volume, density fluctuations will be large, because the density can range between the two extreme values appropriate to the separate phases (Hill, 1956, 1960). Thus at any particular moment in time, the density will depend on the proportions of the two phases in the element of volume. Associated with these large density fluctuations the system will possess a large isothermal compressibility. The isothermal compressibility is defined as ~=

-V\~)r,~'

(14)

It can be shown (Hill, 1956) that the compressibility is related to fluctuations in density in the system by the equation (p - / 0 2

k T ~"

v

-

(p)2

(151

where (p - ~-~2/(fi)2is the mean square relative deviation in density from the mean density. In a one-phase system, the fluctuations in density are very small, so that the compressibility is very small. In a two-phase system, however, the large fluctuations in density mean that the compressibility is large. This effect on compressibility depends on the number of molecules N involved in the transition. For a first-order transition as defined mathematically, N ~ ~ and Jg ~ ~ (Hill, 1956). Thus the smaller the co-operative unit in a lipid phase transition, the broader the range of temperatures over which gel and liquid crystalline phases coexist, and so the greater the temperature range over which the system has a compressibility higher than in the pure liquid crystalline phase. However, the smaller the co-operative unit the smaller will be the increase in compressibility over that in the pure liquid crystalline phase. Evidence suggesting the importance of these effects will be discussed later. B. Pre-meltin# and Pre-freezin# Phenomena The process of melting in even simple organic molecules is generally very complex (Ubbelohde, 1965). A variety of pre-melting and pre-freezing phenomena are common, the


25

pre-melting phenomena making the solid more "liquid-like" and the pre-freezing phenomena making the liquid more "solid-like". Similar effects appear to occur in lipid bilayers at temperatures around that of the gel to liquid crystalline (order to disorder) transition. As already mentioned, as well as the main thermal transition associated with the fatty acid chains, aqueous lecithin dispersions exhibit a smaller pre-transition, which has been attributed to a transition in the polar head group region of the lipid (Ladbrooke and Chapman, 1969; Lee et al., 1974a). Interestingly, however, a similar pre-transition occurs in a number of solid n-alkanes just before the melting point. This has been attributed to the onset of co-operative rotation of the n-alkanes about their long axes in the hexagonal lattice (Muller, 1932; McClure, 1968). It seems possible, therefore, that for the lipids similar rotation about the molecular long axis will occur at the pre-transition since, at this temperature, the lipid head group has been postulated to change from a folded conformation (which would inhibit any such rotation) to an extended conformation. In the liquid crystalline phase of dimyristoyllecithin, 13C nmr data are suggestive of fast axial rotation (Lee et al., 1974b). Interestingly, esr studies of spin-labelled stearic acid incorporated into dipalmitoyllecithin bilayers show that the spin label has considerable freedom of motion just below the transition temperature (41°C) (Oldfield et al., 1972). This has previously been interpreted in terms of "local impurity pools" created by the nitroxide group (Raison et al., 1971) but could in fact correspond to rapid axial rotation in the phospholipids. Schindler and Seelig (1973) have estimated an effective correlation time for spin-labelled fatty acids in dimyristoyllecithin bilayers at 21.5 ° as Zc = 2.5 x 10- 9 sec. There is also evidence for pre-freezing phenomena in lipid bilayers. Such phenomena arise in organic liquids from a gradual extension of short-range order in the liquid as the temperature of the melt is lowered towards the freezing point. Studies of the heat capacities and viscosities of many organic liquids have led to the concept of small quasi-crystalline clusters of molecules present in the liquid phase together with freely dispersed molecules (Ubbelohde, 1965; Davies and Matheson, 1967; Ertl and Dullien, 1973). These clusters are pictured as short-lived dynamic arrangements of adjacent molecules having co-ordinated movements. The mean molecular density within the cluster will be higher than for freely dispersed molecules, and internal rotational freedom is probably inhibited for molecules within the cluster. With increasing temperature, these clusters gradually break up to give monomeric, freely dispersed molecules. Analogous groups of molecules, referred to as cybotactic groups, have been detected by X-ray diffraction in various liquid crystals (Gray, 1962; de Vries, 1970). The presence of such clusters has been suggested for bilayers of dioleoyllecithin in the liquid crystalline phase, based on a reduced partitioning of the small molecule Tempo (IV) into the quasi-crystalline clusters (Lee et al., 1974c). It is estimated that the fraction of lipid present in clusters is zero at 30°C but about 50~ at 2°C (the temperature of the gel to liquid crystalline phase transition is - 22°C). The presence of such clusters could account for breaks in activity for a number of membrane bound enzymes in the temperature range 20-30°C (see below). V. L I P I D M I X T U R E S

Since a biological membrane typically contains a wide variety of different lipids, the study of lipid mixtures is of considerable importance. The behaviour of mixtures of lipids is expected to depend on the extent of the similarity or dissimilarity between the component lipids. Mixtures of two very similar lipids might be expected to be miscible with each other in all proportions in both the gel and liquid crystalline phases, whereas two dissimilar lipids might be immiscible in the gel phase. The effects of temperature on mixtures of compounds are conveniently described in terms of a phase diagram. Such diagrams are, however, usually constructed for macroscopic systems where any phase transitions are sharp. There are a number of complications in any attempt to construct similar phase diagrams for microscopic systems, where phase transitions are more or less gradual. Before considering these complications, we will consider the two simplest types of phase diagram.

26

A.G. LEE

X

L

T

T t1 t2 t3 I

/

I

l

I-

M i x t u r e of solids

I

0

x1

x2

10

0

XB

10

XB

FIG. 15. Phase diagrams for (a) a system showing complete miscibility and (b) a system showing almost complete immiscibility in the solid phase.

In Fig. 15a is shown a typical phase diagram for two very similar compounds, completely miscible both in the liquid and solid phases. If we consider a mixture of composition x, then at any temperatures down to tl there will be a single liquid phase present of composition x. Similarly at temperatures below t3, there will be a single solid phase of composition x. At some intermediate temperature t2, however, both liquid and solid phases will be present: the liquid phase will have a composition xl, and thus be richer in the lower melting component, whereas the solid phase will have composition x2 and be richer in the higher melting component. If the compounds are very different there may be no cocrystallization, so that on cooling, the molecules will separate into regions corresponding to the two separate components. The phase diagram will now have the form of Fig. 15b. Phase diagrams can be constructed by studying the variation in some suitable parameter with temperature, for a number of mixtures of differing composition. One parameter that has been used to determine "phase diagrams" for lipid mixtures is Tempo partitioning (Shimshick and McConnell, 1973a, b). Two breaks are located in plots of Tempo partitioning against temperature for a particular lipid mixture, one corresponding to a point on the fluidus curve and one to a point on the solidus curve. It is here that complications could arise because of the small size of the lipid systems. Thus consider a lipid mixture which, if it were a macroscopic sample with infinite co-operativity, would give a phase diagram such as that in Fig. 15a. Above the critical temperature ta, a lipid mixture of composition x will be in a "fluid" state and the Tempo binding will be high (Fig. 16). Below this critical temperature, which defines the beginning of a broad course of phase separations, patches of lipids in a "solid" phase form in and coexist with the fluid lipid. Because the Tempo binding in the "solid" phase is low, the temperature t~ will be indicated by a drop in Tempo partitioning. As the temperature is decreased still further, the proportion of lipids existing in "solid" patches will increase, so that the Tempo partitioning will drop still further. When a second critical temperature t 3 is reached, all the lipid will be in a solid phase, and there will be a second break in Tempo partitioning. In the construction of a phase diagram such as that of Fig. 15a, it is assumed that for a mixture of composition x, the width of the phase transition (tl to G) is due to the formation of liquid and solid phases of different composition. In systems of low co-operativity, however, the phase transition can be broad even when the composition of the liquid and solid phases must be equal as in the example of a pure compound (Figs. 13, 14). To construct a phase diagram in the way suggested by Shimshick and McConnell (1973a, b) the phase transition must be very sharp, that is, it must be highly co-operative. Only then is there just a single temperature t2 at which liquid of composition xl and solid of composition x2 are in equilibrium. For unsonicated single lipid dispersions, the thermal transition is reasonably sharp, but, as we have seen, the transition is broad in sonicated lipid dispersions.

Functionalpropertiesof biologicalmembranes

27

The co-operativity of the transition for lipid mixtures is unknown, but it could well be less than for pure single lipids, and might well vary with the composition of the lipid mixture. To establish whether or not the lipid and solid phases formed at intermediate temperatures are of differing composition it would then be necessary to monitor some parameter characteristic ofeach component of the mixture rather than some single parameter for the whole mixture. This has not yet been done. Despite all these complications, it is at least possible to show that lipids of similar composition are miscible in both liquid crystalline and gel phases, whereas very dissimilar lipids are not completely miscible in the gel phase. Thus, mixtures of dioleoyllecithin with saturated lipids show limited solid solution formation (Phillips et al., 1970). Here differential scanning calorimetric data very clearly shows two separate peaks, one corresponding to crystallization of the bulk of the saturated lipid and one to crystallization of dioleoyllccithin. In the gel phase below - 22°C, the bilayers contain segregated areas of crystallized saturated lipids and of crystallized dioleoyllecithin. At - 22°C, the dioleoyllecithin in the bilayer melts. As the temperature is raised further, the saturated lipid starts to melt and the system contains regions of crystallized saturated lipids and regions of mixed liquid crystalline saturated and unsaturated lipids. In such a system the melting of the higher melting component will commence immediately after the lower melting component has transformed. Physically this can be pictured as the gradual melting of the saturated lipids which are in contact with pools of molten unsaturated lipids: these saturated lipids will have greater configurational freedom than those which are surrounded only by other saturated lipids. For mixtures of two very similar lipids such as distearoyllecithin and dipalmitoyUecithin, a continuous series of solid solutions is formed in the gel phase (Fig. 15a) (Phillips et al., 1970). For the distearoyllecithin-dimyristoyllecithin system, however, the "phase" diagram is probably of the type Fig. 15b, with very limited solid solution formation. During the process of crystallization in mixtures showing the behaviour of Fig. 15b it is necessary for lipid molecules to migrate within the bilayer to form segregated pools of the two components. Rapid lateral diffusion of lipids in the liquid crystalline phase has been established (see above). Equilibration of the gel phase could be achieved by diffusion within the gel phase or by rapid movement of the boundaries between the gel and liquid crystalline phases (Shimshick and McConnelk 1973a). These experiments give no indication of the size of the segregated pools in the gel phase but freeze-fracture experiments suggest that they might be quite large. It has been shown

40 I

I

30 I

20

I

i

I 3"30

A

T "C

I

0"7 0"6

0"5 f

0'4 0'3 02 0"1 0

I

I 3"20

i

I 3 ' 4 0 X 10 "~

¢'(°K")

16. T e m p o b i n d i n g to an aqueous dig~crsion of a mixture of dimyrismyllecithin and dJpalmitoyllecithin(DPL). The parameterfig approximately eqtufl to the fractionof Tempo dissolvedin the lipid.Triangles,76 tool % DPL; circles,51 tool % DPL, squares, 26 tool ~o DPL (from Shimshick and McConnell,1973a). FIG.

28

A.G. LEE

FIG. 17. A model of the fatty acid chai~cholesterol mixed phase viewed perpendicular to the plane of the lamellar phase. The circles represent the fatty acid chains and the irregular shapes are cholesterol molecules as projected down their long axes (from Engelman and Rothman, 1972).

that slow freezing of lipids in the gel phase gives rise to a band pattern in electron micrographs, whereas freezing of lipids in the liquid crystalline phase does not (Ververgaert et al., 1973a). Although this effect has been shown to be an artifact due to the rate of cooling (Ververgaert et al., 1973b), it may still nevertheless be useable to distinguish lipids in the gel and liquid crystalline phases. In liposomes containing distearoyllecithin and dioleoyllecithin, distinct banded and smooth areas are observed, possibly corresponding to segregated pools of distearoyllecithin and of distearoyllecithin-dioleoyllecithin, respectively. The dimensions of the banded area are in the 0.2-#m range (Ververgaert et al., 1973a). There has as yet been only one study on the motional parameters of phospholipid-phospholipid mixtures. In equimolar mixtures of dimyristoyllecithin and dipalmitoyllecithin it has been observed that motion about the terminal C--C bonds of the lipid fatty acid chains are faster than in the pure single lipids (Lee et al., 1974b). This is presumably a reflection of poor packing at the centre of the bilayer in mixtures of lipid with different chain length. Another much studied type of mixture is that of phospholipids with cholesterol. The details of this system are, however, still very far from clear. It is generally agreed that the addition of cholesterol to lipids in the gel phase "fluidizes" the lipid fatty acid chains, whereas addition of cholesterol to lipids in the liquid crystalline phase decreases the motional freedom of the chains (Phillips, 1972). But beyond such qualitative statements it is difficult to go. Phospholipid bilayers will take up cholesterol to a molar ratio of 1: 1, beyond which pure cholesterol crystallizes out (Lecuyer and Dervichian, 1969). On the basis of X-ray diffraction (Engelman and Rothman, 1972) and calorimetric studies (Hinz and SturtevanL 1972a), a structure for the 2:1 dipalmitoyllecithin-cholesterol bilayer has been suggested (Fig. 17). In this structure, the cholesterol molecules are assumed to be randomly oriented about their long axes, with each molecule completely surrounded by fatty acid chains. The minimum number of fatty acid chains necessary to completely surround a cholesterol molecule and prevent cholesterol-cholesterol contact is seven--this corresponds to a lipid-cholesterol molar ratio of 1.9: 1. In lipid bilayers containing less than 33 mole ~ cholesterol, the X-ray data are consistent with the presence of two phases: a pure lecithin phase and a mixed phase of lecithin and cholesterol of molar ratio 2: 1.


29

In the cholesterol-lipid phase, no strong configurational coupling is possible between the lipid fatty acid chains and the thermal phase transition is no longer detectable by differential scanning calorimetry (Hinz and Sturtevant, 1972a). With increasing cholesterol content from zero to 339/~ the transition enthalpy decreases (corresponding to a reduced number of iipids taking part in the transition) and the transition broadens (corresponding to a reduced co-operativity for the transition). Similar phase separation probably occurs in lipid-cholesterol bilayers where the lipid is in a liquid crystalline state (Engelman and Rothman, 1972). Measurements of the 13C relaxation times in 2:1 dioleoyllecithin-cholesterol mixtures show that the presence of cholesterol reduces mobility about all the C--C bonds in the fatty acid chains, but has no effect on motion in the lipid head group region (Lee et al., 1974b). Motion is probably more restricted at the beginning of the fatty acid chain than at the end. This is consistent with simple steric requirements. If the cholesterol molecule is arranged with its 3//-OH group at the lipid-water interface, then the first seven or so --CH2--groups of the lipid fatty acid chains will be in contact with the bulky ring region of the cholesterol molecule, and thus will be restricted in motion. Esr studies of spin labelled fatty acids incorporated into egg lecithin bilayers also show an increase in order parameter on addition of cholesterol (Schreier-iuccillo et al., 1973). Preliminary results suggest that addition of cholesterol to dioleoyllecithin or egg lecithin causes a significant reduction in the rates of lateral diffusion, but with no significant change in the activation energy for diffusion (Lee et al., 1973). This suggests that the reduction in diffusion coefficients is an entropy effect consistent with the maintenance of increased order during the diffusion, necessary to avoid unfavourable cholesterol-cholesterol contact. In contrast, the results obtained by Scandella et al. (1972) for the rates of diffusion of spin-labelled lipids in egg lecithin--cholesterol 4:1 mixtures would seem to imply that the rate of lipid diffusion is increased by the addition of cholesterol (see Table 4). However, their value for the self-diffusion coefficient in the absence of cholesterol was obtained by a method in which time dependence was observed directly (Devaux and McConnell, 1972), whereas the cholesterol results were obtained by a line shape analysis, which, it has been suggested, will overestimate the real rate of diffusion (Keith et al., 1973). The model for lipid-cholesterol mixtures that has been presented above is, however, far from universally accepted. For example, Darke et al. (1972) have come to a very different set of conclusions, largely based on their broad-line proton nmr studies. They interpret their data as showing that motion at the terminal methyl end of the lipid fatty acid chains is not restricted by the presence of cholesterol. However, the interpretation of broad-line proton nmr data is always difficult because individual resonances are not resolved, as they are in t3C nmr spectra. Darke et al. (1972) also believe that specific 1:1 lipid:cholesterol complexes are formed with a life time of > 30 msec, so that in non-equimolar mixtures, free lipid and lipid:cholesterol 1:1 complexes are present. However, these conclusions are again based on proton line width and area measurements, some of which are in doubt (Lee et al., 1974a). Phillips (1972) has also concluded from these studies that when cholesterol is added to mixed saturated and unsaturated lipids, "the first cholesterol molecules added probably complex preferentially with the more saturated (higher melting) molecules in the mixture". However, calorimetric studies of the effect of cholesterol on codispersions of dioleoyllecithin and distearoyllecithin show that cholesterol preferentially interacts with the dioleoyllecithin (de Kruyff et al., 1973). In biological membranes containing low concentrations ofcholesteroi, therefore, the cholesterol will not be randomly distributed when both gel and liquid crystalline lipid phases are present in the membrane, but will be preferentially associated with those lipids which are in the liquid crystalline phase. Shimshick and McConnell (1973b) have come to yet different conclusions based on their studies of Tempo partitioning into dipalmitoyllecithin-cholesterol mixtures. They obtain a phase diagram which exhibits solid phase immiscibility, but gives no indication of a phase change at a lipid:cholesterol 2:1 molar ratio. However, the agreement between their calculated and observed Tempo partitioning data is noticeably worse than for the phospholipid-phospholipid mixtures that they have also studied: effects of changing co-operativity and the possibility of cluster formation should perhaps be considered.

30

A.G. LEE

Before leaving the topic of lipid mixtures, the possibility of an asymmetric distribution of lipids between the two sides of the bilayer should be considered. Asymmetric lipid bilayers have been made by the apposition of the fatty acid chains of two lipid monolayers at air-water interfaces, when an aperture in a teflon septum separating the two aqueous phases is lowered through the interface. Such bilayer membranes are stable for several hours, during which the bilayer asymmetry is maintained, so that lipid flip-flop must be slow (Montal and Mueller, 1972; Montal, 1973). Asymmetric lipid bilayer vesicles can also be made by cosonication of equimolar quantities of phosphatidylglycerol and phosphatidylcholine: the resulting bilayer vesicles contain about 80~o of the phosphatidylglycerol and 40~o of the phosphatidylcholine in their outer layers, whereas the inner layers contain about 20~o of the phosphatidylglycerol and 60~o of the phosphatidylcholine (Michaelson et al., 1973). This asymmetry is due to the effects of radius of curvature on the electrostatic energy of the electric double layer* associated with the charged lipids. The electrostatic energy is minimized by concentrating the charged lipid in the outer surface of the bilayer. The asymmetry will be reduced by increasing radius of curvature or increasing ionic strength (Israelachvili, 1973). It is possible that the presence of metal ions in the bulk aqueous phase could induce an asymmetry between the two sides of the bilayer if the metal ions bound most strongly to one component of the lipid mixture. Such an effect has not yet been demonstrated. Similarly, it is possible that the presence of proteins will produce an asymmetry across the membrane; this possibility is discussed further in Section VIII.

VI. T H E

PERMEABILITY

OF

LIPID

BILAYERS

A number of theories have been proposed to account for the diffusion of small molecules across membranes. The possibility of large water-filled pores through the membrane has been considered, where the pores are of sufficient size that the water inside them has essentially bulk properties. For such a model, the activation energy for diffusion across the membrane should be closely similar to the activation energy for diffusion in water, and experimentally in a variety of membrane systems that is not found to be so (Price and Thompson, 1969; House, 1974). A second possibility is that the membrane contains small pores, where the pores are sufficiently small that the water inside them does not exhibit bulk properties. However, calculations suggest that pores of the requisite narrow dimensions ought to contain no water molecules at all for 98~o of the time and only one molecule for approximately 2~o of the time (Hirsch, 1967), in which case they can hardly be called water-filled pores. The most useful model appears to be that in which the small diffusing molecule .is assumed to dissolve in the bilayer and move across by diffusion. Zwolinski et al. (1949) have used this approach to describe the permeability of a homogeneous membrane. They regard the flow of molecules through the membrane as a series of successive jumps from one equilibrium position to another. Figure 18 shows an energy profile for a homogeneous membrane. The molecular jumps, of length 2, occur from one potential energy minimum to the next, with the rate of jumping given by a constant k. The permeability constant of the membrane is then given by 1

2

P

k~2

+

m

kmZ(k~/k~)

(16)

Here ks,, is the rate constant for diffusion from the solution into the membrane across the interface, kms is the rate constant for diffusion from the membrane out into the solution, and k,, is the rate constant for diffusion in the membrane. The number of jumps required to move across the membrane is m, so that m2 = ~ where fi is the membrane thickness. * The electric double layer is discussed on p. 35.

(17)


I

Aqueousl solution I I

Membrane

31

~ u $

solution

ksm

AA FXG. 18. A simple model for the potential energy profile of a small molecule diffusing across the bilayer.

The ratio k,~/k=s is the partition coefficient B for the diffusing molecule between the solution and the membrane. The first term on the right-hand side of eq. (16) represents the resistance encountered at the solution-membrane interface, and the second term depends on both the solubility and the rate of diffusion of the diffusing molecule in the membrane. For most non-electrolytes, we expect the partition coefficient to be small, and k~ ~ k ~ In the limit where kin, >> k~ >> kin, the rate determining step is diffusion in the membrane, and for reasonably small values of 2 (4-5 A); eq. (16) reduces to (Zwolinski et al., 1949; Price and Thompson, 1969) p ,.,_ k.B,~ _ DraB

m

(18)

2

Here, D , is the diffusion coefficient of the diffusing molecule in the membrane, and is given by k ~ 2. In the limit km >> k,~ >> k~, the slow step is diffusion through the solution-membrane interface, and the permeability becomes independent of the partition coefficient: p _ k~m2 = D ~

2

22"

(19)

The model of Zwolinski et al. (1949) is, of course, an oversimplification for the lipid bilayer. The microviscosity of the fatty acid chain regiofi decreases by a factor of ca. 15 from the glycerol backbone region to the centre of the bilayer. Thus we would expect km to vary through the membrane, depending on the distance of the diffusing molecule from the glycerol backbone region of the bilayer. The permeability coefficient for the membrane could be written as 1

P

2R +

dx Om(x)B(x)

(20)

where R represents the resistance of the membrane-water interface to solute flow, and Din(x) and B(x) give the diffusion coefficient and partition coefficient at a distance x from

the interface. Data for diffusion across the membrane is in fact usually discussed in terms of the simplified eq. (18). Thus the bilayer is considered to be a "slab" of hydrocarbon, and it is assumed that transient "holes" or pockets of free volume are opened up by thermal fluctuations which serve to carry the diffusing molecules across the membrane (Lieb and Stein, 1969; Trauble, 1971b)~ A temperature study of non-electrolyte permeation across red cell membranes has, however, shown that there are indeed two barriers to membrane permeation: one provided by the water-membrane interface and one by the membrane interior (Galey et al., 1973).

32

A.G. LEE

An intuitively attractive model for the diffusion of a small molecule across the membrane would involve the small molecule first entering the more "fluid" part of the hydrocarbon centre of the bilayer through a lattice vacancy or "transient pore" and then diffusing through the more "fluid" part of the hydrocarbon region in a pocket of free volume (Lee et al., 1974a). Addition of high molar ratios of cholesterol to egg lecithin bilayers reduces the permeability of the bilayers to water (Bittman and Blau, 1972). This is probably largely a reflection of reduced freedom of motion for the fatty acid chains, but could also be due to changes in packing density in the glycerol backbone region of the bilayer. Studies of the effect of low cholesterol levels, however, show that the situation is rather complex. The permeability of water increases to a maximum as cholesterol is added to a molar fraction of ca. 0.1. With more cholesterol, the permeability falls, and at a molar fraction of ca. 0.3 is back to the level in the absence of cholesterol: beyond this point there is a further steep reduction in permeability (Jain et al., 1973). X-ray diffraction studies, however, have shown a smooth increase in bilayer thickness on addition of cholesterol to egg lecithin up to a molar ratio of 2:1 (Lecuyer and Dervichian, 1969). A tentative explanation can be given in terms of the fluctuations expected to be present in two-phase systems. At low cholesterol levels, both pure lecithin and lecithin-cholesterol 2:1 phases will be present. As shown by X-ray studies, the densities of these two phases will be different, so that large fluctuations in density are possible in the membrane, associated with a high compressibility. Increased permeability to small molecules is therefore expected. At the same time, however, the permeability of the lecithin-cholesterol 2:1 phase is less than that of the pure lecithin phase, so that the permeability will pass through a maximum with increasing cholesterol content. The effect of lowering the temperature of a lipid bilayer will be both to reduce the mobility of the fatty acid chains and to increase the packing density in the glycerol backbone region. As expected, permeabilities of lipid liposomes to glycol, glycerol and erythritol decrease markedly with decreasing temperatures. However, in liposomes of dimyristoyllecithin the permeability to glycerol falls almost to zero below 22°C whereas the permeability to glycol is still appreciable at temperatures down to 15°C (de Gier et al., 1968, 1971). The process of permeation for organic molecules is clearly complex. At temperatures much below the transition temperature, the permeability increases dramatically {de Gier et al., 1968), perhaps because of the formation of extensive fractures in the vesicles when the lipids are in a tightly packed gel phase. The permeabilities of Acholeplasma laidlawii ceil'membranes were found to have a temperature dependence very similar to that of the extracted lipids (de Kruyff et al., 1973a; McElhaney et al., 1973). At temperatures above the phase transition temperature range for the membrane lipids, the permeability decreased with decreasing temperature. However, after the temperature had dropped to the lower end of the phase transition temperature range, an increase in permeability was noted. Since many broken cells and cell "ghosts" were then observed by phase microscopy, it seems that fracture of the cell membrane occurs because of the problems in packing when large proportions of gel phase lipids are present (McElhaney et al., 1973). The permeability of pure lipid bilayers to ions is generally very low. For egg lecithin vesicles at 4°C, the ion fluxes (mol cm- 2 sec- 1) for Na + and CI- are 8 x 10- 15 and 1.7 x 10- 18, respectively (Hauser et al., 1973). For comparison, the CI- flux through the red blood cell is 1.4 x 10-8 tool cm -1 sec-1 (Whittam, 1964). These low permeabilities presumably reflect the large electrostatic free energy needed 4o bring an ion from aqueous solution (dielectic constant E ca. 80) into a lipid phase of low dielectic constant (E ca. 2-3): this energy can be estimated to be ca. 46 kcal/mole. The permeability of liposomes of dipalmitoyllecithin to Na + increases markedly from the gel to the liquid crystalline phase (Papahadjopoulos et al., 1973). Interestingly, however, a maximum in Na + permeability is observed at the mid-point of the phase transition. The observation of a maximum in permeability at the point where equal amounts of gel and liquid crystalline phases are present is consistent with the thermodynamic arguments that the compressibility of the bilayer will be maximal at this temperature. Papahadjopoulos et al. (1973) explain their observations by suggesting an increased permeability


33

through the regions of disorder at the boundaries between discrete solid and liquid domains co-existing within the plane of the membrane. If permeability is greater in these boundary regions between domains than in either liquid or solid domains, then a maximum in diffusion will occur at the mid-point of the phase transition, where the fractional area of boundary regions within the membrane is greatest. These two explanations are saying very much the same thing, although in different languages. Vesicles of phosphatidylserine and phosphatidylglycerol exhibit a significant discrimination between the alkali metal ions, the maximum K+/Na + diffusion rate ratio being c a . 10 for phosphatidylserine vesicles at pH 7.4 (Papahadjopoulos, 1971). Addition of cholesterol to the phosphatidylserine vesicles reduces the Na +-K + discrimination, and in vesicles containing 15~o phosphatidylserine and 85% phosphatidylcholine at pH 7.4 the K*/ Na + diffusion rate ratio is only c a . 2. Vesicles of phosphatidylethanolamine exhibit no Na +-K + discrimination, and vesicles of phosphatidic acid exhibit slight discrimination, but only at low pH. The reasons for this discrimination are not yet clear. The permeability of lipid vesicle to Ca 2+ is less than to the alkali metal ions; for egg lecithin vesicles the permeability coefficient is a factor of I0-100 smaller (Vanderkooi and Martonosi, 1971). The permeability depends markedly on the lipid fatty acid chains. In dipalmitoyllecithin there is a marked increase in permeability at 40-50°C, and the permeability of dioleoyllecithin vesicles is considerably greater than those of dipalmitoyllecithin, particularly at higher temperatures. The permeabilities of lipid bilayers to metal ions can be increased dramatically by the addition of antibiotics such as valin.omycin, nonactin and enniatin B. These are all macrocyclic molecules, which form complexes with alkali ions in organic solvents with a high degree of specificity (Haydon and Hladky, 1972; Lauger, 1972). For example, the stability constant of the K + complex of valinomycin (V) in methanol is c a . 104 times greater than that for the Na + complex.

H3CX,/H3 H3~,,/H3 H3C, N,,/CH3 F_

CH3

CH

CH

i

I

J

CH ,

O-- CH-- .C.-- NH--CH-- C--O--OH-- ,C,- - NH-- CH-- , C - - ~ II

II

II

II

0

O

O

OI

-

I

|

(Y)

The structure of the K ÷-valinomycin complex (Fig. 19) is such that the K ÷ is contained within a polar environment within the complex, whilst the outer surface of the complex is hydrophobic (Shemyakin et aL, 1969). It is postulated that these antibiotics increase the Ck,,,

I

OC O 0

,/

.o

CS

(~)N O K

"~

~H-bond

FIG. 19. The structure of the valinomycin-K + complex (from Shemyakin et al., 1969). J P B 29 I .

("

34

A.G. LEE

permeability of the membrane to metal ions by acting as mobile carriers, enabling the metal ion to cross the membrane hydrocarbon region without being forced to leave its preferred polar environment. The effect of valinomycin is highly specific for certain alkali metal ions, and is considerably greater for K +, Rb ÷ and Cs + than for Li + or Na + (Lauger, 1972). Consistent with their role as mobile carriers, the effects of these molecules on ion permeability depends markedly on the state of the lipid fatty acid chains. The addition of cholesterol markedly reduces the valinomycin induced permeability of dioleoyllecithin vesicles to Rb + (de Gier et al., 1970). Reducing the temperature of black lipid films containing valinomycin through the lipid "phase transition" markedly increases their electrical resistance (Krasne et al., 1971 ; Stark et al., 1972). A similar effect is observed in the valinomycin mediated potassium conductivity of dipalmitoyllecithin bilayers. There is a marked drop in conductivity below the thermal transition temperature and a pronounced maximum at the transition attributable to the presence of two phases of different density (Wu and McConnell, 1973). An alternative to the carrier mechanism for ion transport is one where the transferring agent remains stationary in the membrane while the ions move relative to the binding sites of the transferring agent. The antibiotics gramicidin A and alamethicin are thought to act in this way, forming pores across a lipid membrane. The evidence for this view has been discussed by Haydon and Hladky (1972). These pores appear to be less sensitive to the state of the lipid fatty acid chains than are the mobile carriers. Thus temperature changes which led to a large decrease in the valinomycin induced conductivity of black lipid films have little effect on the gramicidin A induced conductivity (Krasne et al., 1971). VII. THE EFFECTS OF METAL IONS ON LIPIDS The arrangement of phospholipid bilayers with polar head groups oriented at the lipidwater interface means than an electrostatic field will extend out into the surrounding water. This field will be relatively small for the zwitterionic phospholipids, but will be appreciable for a lipid carrying a net charge. If the bilayer contains negatively charged phospholipids for example, then the effect of the electrostatic field will be to attract positively charged ions into the vicinity of the interface between the lipid bilayer and the aqueous solution. This will produce an electrical double layer, in which the charge in the membrane surface is screened by a layer of counter ions of opposite charge. However, counter ion adsorption onto the charged surface is also possible, resulting in a charge neutralization. In general these two effects might be expected to go together, with some ions binding to the surface and thus partially neutralizing the surface charge, the remaining surface charge giving rise to an electrical double layer. The simplest treatment of the electrical double layer is that of the Gouy-Chapman theory (Overbeek, 1952; Davies and Rideal, 1961). In this approach, the charges on the membrane are treated as if smeared out uniformly, and the charged ions in the aqueous medium are treated as point-like ions. There is as yet no detailed treatment for counterion adsorption, since the structures of the possible binding sites on the membrane surface are not known in any detail. The relative importance of these two effects has not yet been settled. McLaughlin et al. (1971) have tried to distinguish between them by studying the effects of divalent ions on the conductance of black lipid films of phosphatidylserine or phosphatidylglycerol containing the K ÷ carrier nonactin. They concluded that the alkaline earth ions act predominantly by a screening mechanism. However, Haydon and Hladky (1972) have pointed out that the treatment of McLaughlin et al. (1971) ignores other possibly important effects, so that the conclusions of McLaughlin et al. do not follow from the results. MacDonald and Bangham (1972) have studied the effects of monovalent ions on monolayers containing mixtures of phosphatidylcholine and phosphatidic acid, and show that even for monovalent ions the agreement with the Gouy-Chapman theory is only to within about 20~o. From monolayer and electrophoretic measurements, it is possible to measure an "apparent" association constant for Ca 2 ÷ which is low for zwitterionic phospholipids, but high (103-104) for negatively charged phospholipids such as pho sphatidic acid or phosphatidylserine (Barton, 1968; Dawson and Hauser, 1970; Seimiya and Ohki, 1973). The association


35

4'4

4.2

o/ -o-

4.0

o i l

/:///

3-8 ol _o 3-6

3'4

3"2

/ 3.0 0

I 20

I 40

I 60

-3 X 10

I 80 -2 (esu c m )

FIG. 20, The effect of surface Charge density on the apparent association constant K' for magnesium (open circles), calcium (closed circles), strontium (triangles) and barium (squares) (from Barton, 1968).

constant is only "apparent", however, since it will contain contributions both from specific adsorption onto the surface and from electrical double layer effects; because of the electrical double layer, the concentration of Ca 2+ at the surface of the charged membrane will be much higher than the bulk concentration, and would give rise to a high "apparent" association constant even in the absence of any specific adsorption. Measurements of "apparent" association constants have been reported for lipid vesicles formed from 1-stearoyl-2-oleoyl-phosphatidylcholinecontaining various proportions of 1stearoyl-2-oleoyl-phosphatidic acid (Barton, 1968). This study shows that surface charge density has a differential effect on the alkaline earth cations (Fig. 20). The larger cations show a rapidly increasing "apparent" association constant with increasing charge d~nsity. At low charge densities, the order of increasing "apparent" association constant is Mg 2+ > Ca 2+ > Sr 2+ > Ba 2+, whereas at high charge densities the order changes to i g 2 + > Ba 2+ > Sr 2+ > Ca 2+. Apparent association constants have not been measured for the alkali metal ions, but it is clear that they will be considerably smaller than for divalent metal ions. Thus Ca 2+ can be displaced from monolayers of phosphatidylserine by the addition of alkali metal ions, but a large excess is required: there is no detectable discrimination between the alkali metals (Rojas et al., 1966). These interactions of metal ions with lipid bilayers have been shown to have important effects on the packing of the lipid molecules in the bilayer. They strongly suggest that divalent ions are adsorbed onto specific sites on the membranes, whereas the alkali metals probably are not. Perhaps the most convenient parameter to study is the temperature of the lipid gel to liquid crystalline phase transition. The effect of the electrical double layer on the thermal transition temperature is seen in studies of pH dependence (Trauble and Eibl, 1974). If the membrane charges are assumed to be uniformly distributed over the membrane surface, then the electrical double layer energy can be calculated from the Guoy-Chapman theory. For monovalent ions, the free energy of the interface per cm 2 is (Trauble and Eibl, 1974) c~ = 2(k T / e X a ' / f )

(21)

36

A.G. LEE

where e is the electronic charge, J i s the surface area of one lipid molecule and a' is the charge per lipid polar group. The free energy of the interface is independent of the electrolyte concentration. If the aqueous solution contains divalent rather than monovalent ions, then ~b becomes ~bdivalent =

(kT/eXcr'/f).

(22)

For phosphatidic acid, the charge per polar group ~r' is connected to the degree of dissociation ~ of the phosphate groups by o' = e:~.

(23)

The effect of ionization on the transition temperature arises because the surface free energy will change at the transition. For phosphatidic acid there is a small change in a' between the gel and liquid crystalline states (Trauble and Eibl, 1974) presumably due to some rearrangement in the polar head group region. However, the major effect is the change in the surface area of the lipid at the transition. Iff~ and f2 are the molecular areas for T < T, and T > T,, respectively, then from eq. (21) the change in electrostatic free energy per tool. Aft: at the transition is AE = E~ - El

= -2Rr

f

f2) e

(24)

where

af=f2 - A . For a reversible phase transition at constant pressure, the molar free energies of the two states are equal at the transition temperature, so that AH = T, AS.

(25)

The enthalpy change AH can be considered to consist of the non-electrostatic term kM ° and the electrostatic term AE. If we ignore any effects of pH on AS (that is, if we assume that pH changes have no effect on the ordering of the lipids) then the change in transition temperature caused by electrostatic effects is A T , = T, -

T O = AE/AS.

(26)

From eq. (24), AT, is given by (Trauble and Eibl, 1974) am, - AE _

2Rr (Af

(af)']

(27)

Thus increasing ionization (increasing ~) will cause a decrease in transition temperature. The reason is simply that the electrostatic free energy of the bilayer in the liquid crystalline phase is smaller than in the gel phase, so that an increase in surface charge will favour the liquid crystalline phase. As an order of magnitude calculation, one can put AS = 22 cal deg- 1 mole- 1 as for dimyristoyllecithin and f2 = 70 A 2 and f, = 48 A 2 as for dipalmitoyllecithin. A change in o' from e to 2e at T ~- 50 ° will then give AT, = - 18°C (Trauble and Eibl, 1974). The effect of pH on the transition temperature of dimyristoylphosphatidic acid agrees well with these theoretical predictions. Phosphatidic acid has two ionizable protons, the first ionizing between pH 1 and 3 and the second between pH 7.5 and 10. The ionization leads to a stepwise decrease in T, as expected. This means that it is possible to trigger a transition from gel to liquid crystalline phase, or vice versa, at a fixed temperature by varying pH. Thus at 40°C, a change in pH from 7.5 to 8.5 will cause a phase change for dimyristoylphosphatidie acid from liquid crystalline to gel. Similar effects have been observed for phosphatidylserine and cephalin at high pH's, but the behaviour of these lipids at lower pH is unexpected. Phosphatidylcholine shows a marked decrease in transition temperature as the pH is increased from ca. 1 to 3, and then


37

no change on increasing the pH up to 12 (Trauble and Eibl, 1974). The results for the phosphatidylcholines could only be interpreted if the arrangement of the zwitterionic head° group in the surface were known, and as we have seen (p. 14) this is not the case. The addition of monovalent cations might have been expected to decrease the surface change a' and thus increase Tt (eq. (27)). Addition of Li ÷, Na ÷ or K ÷ to dimyristoylphosphatidic acid at rather high concentrations ( - 1 M) in fact lowers the transition temperature. This has been interpreted as an ionic strength effect which will tend to increase the dissociation of the dimyristoylphosphatidic acid (Trauble and Eibl, 1974). The effect of an ion such as T1 ÷ would be of interest, since this is more likely to bind to the surface of the bilayer (Levine et al., 1973; see also Lee, 1971). The addition of Ca 2 ÷ or Mg 2÷ to dimyristoylphosphatidic acid causes a marked increase in T,. The effects occur at too low concentrations to be explained by screening within the electrical double layer and indicate that specific adsorption to the surface is probably important. The effects of divalent ions increase with increasing pH, suggesting that the divalent cations interact preferentially with doubly ionized phosphatidic acid, although some interaction occurs with singly ionized (Trauble and Eibl, 1974). An increase in Tt has also been observed for dipalmitoyllecithin in the presence of E u a +, Nd a ÷ or UO 2 ÷ which was also attributed to metal ion binding to the surface (Levine et al., 1973). As already noted from monolayer studies, the addition of monovalent ions will displace divalent ions from the charged lipids (Trauble and Eibl, 1974). The effects of divalent metal ions on phosphatidylserine have not yet been studied in much detail. Nevertheless, it is clear that effects are similar to those with phosphatidic acid. Thus addition of alkaline earth ions to monolayers of phosphatidylserin~ cause a condensation, which has been attributed to complex formation between the metal ions and the phosphatidylserine. The effects of the metal ions increase in the order Mg 2÷ < Ba 2 + < Ca 2÷ (Papahadjopoulos, 1968). Further, the addition of Ca 2÷ to black lipid films of phosphatidylserine increases their electrical resistance (Okhi, 1969). More detailed information on the effects of Ca 2 ÷ comes from spin label studies. Butler et al. (1970) found that the ordering of the long axis of spin-labelled cholestane incorporated into films of lipid from beef brain increased on addition of metal ions. The effectivenessofthe metal ions increased in the order Na + = K + = Li + < M g 2 + = Ca 2+ < L a 3+ < Th 4+. The effects of Ca 2+ have also been studied on bilayers of phosphatidylserine (from beef brain) containing a spin-labelled phosphatidylcholine (Ohnishi and Ito, 1973). In the absence of Ca 2+, the spin-labelled phosphatidylcholine molecules are uniformly distributed throughout the membranes, but addition of Ca 2 + causes the separation of the phosphatidylcholine into segregated pools containing relatively few phosphatidylserine molecules. The segregation could be caused by an aggregation of the phosphatidylserine molecules largely involving the lipid headgroups, or it could be due to a transition to the gel phase for phosphatidylserine, caused by the Ca 2 + (and as described above for phosphatidic acid). The effects of Mg 2 + in this system are different to those of Ca 2 +, but no detailed conclusions could be drawn. Particularly interesting effects can be expected when Ca 2 + is added to just one side of a bilayer of phosphatidylserine. Adsorption of Ca 2 + to one side of a phosphatidylserine bilayer will tend to neutralize the charge on that side of the bilayer thus creating a difference in the surface energies between the two sides of the bilayer. Under these conditions groups of molecules will invert from one side of the bilayer to the other. In so doine, the permeability of the bilayer will increase and, under extreme conditions, the bilayer will break (Ohki, 1972). Such effects have been observed experimentally. Addition of Ca 2+ to both sides of a black lipid film of phosphatidylserine increases its electrical resistance. If Ca 2 + is added to just one side however, the electrical resistance decreases, and above a certain Ca 2+ concentration, the membrane breaks (Papahadjopoulos and Ohki, 1969; Ohki, 1970, 1972). Similar effects are seen in asymmetric lipid bilayers containing the neutral lipid glycerol dioleate on one side and the negatively charged cardiolipin on the other (Montal, 1973). Addition of Ca 2 + to the neutral side of the bilayer has no effect on electrical resistance, whereas addition of Ca 2 + to the negatively charged side of the bilayer causes a drop in resistance. As expected if the effect is caused by Ca 2 + binding, increasing ionic

38

A.G. LEE

strength reduces the effect of C a 2 + (Montal, 1973). Similarly, addition of local anaesthetics such as procaine, which apparently bind to the membrane displacing Ca 2+, also inhibits the effect of C a 2 + (Ohki, 1970). In the same way, addition of Ca 2 + to an aqueous suspension of phosphatidylserine liposomes preloaded with K + causes an increased rate of K + release. The effect is again inhibited by local anaesthetics (Papahadjopoulos and Bangham, 1966; Papahadjopoulos, 1970). Unexpectedly, however, no effects of C a 2+ a r e seen for liposomes containing mixtures of phosphatidylserine and phosphatidylcholine when the molar fraction of phosphatisylserine is less than 50~ (Papahadjopoulos and Bangham, 1966).

VIII.

BIOLOGICAL

MEMBRANES

A. The Structural Organization of Biological Membranes X-ray diffraction studies are consistent with the presence of substantial regions of lipid bilayer in a number of biological membranes, including myelin (Blaurock, 1971; Casper and Kirschner, 1971 ; Wilkins et al., 1971), retinal rods (Blaurock and Wilkins, 1972), erythrocytes (Wilkins et al., 1971), synaptosomes (Wilkins et al., 1971), sarcoplasmic reticulum (Wilkins et al., 1971 ; Dupont et al., 1973), purple membrane of Halobacterium halobium (Blaurock and Stoeckenius, 1971), PM2 .virus membrane (Harrison et al., 1971), Acholeplasma laidlawii (Engelman, •97•), and in lipid cytochrome C complexes (Blaurock, 1972, 1973). Such studies do not allow a quantitative assessment of the bilayer content of the membrane. It has, however, been shown that a large majority of the lipid in Acholeplasma laidlawii is free to undergo a thermal transition from a gel to a liquid crystalline state, analogous to that in a lipid bilayer (Engelman, 1971). Comparison of the heat of transition of the membrane with that of an equivalent amount of extracted membrane lipid also suggests that the majority of the lipids in the membrane are involved in the transition (Reinert and Steim, 1970), although this argument is by no means watertight (Chapman and Urbina, 1971). As a working model for the membrane, therefore, the bulk of the lipid is assumed to be in a bilayer form, with a fraction of the lipid possibly immobilized around the membrane proteins. It is further assumed that some proteins will span the membrane, whilst others will not. This constitutes the fluid-mosaic model of Singer and Nicolson (1972). Although intuitively it seems likely that, amongst others, the various transport proteins will span the membrane, the actual evidence for this is rather sketchy. Most experiments designed to test for the presence of proteins spanning the membrane have involved labelling of erythrocytes with a small "impermeable" label, and comparing the peptide labelling patterns of intact erythrocytes and leaky ghosts (see Bretscher, 1972). Any observed differences in labelling have then been interpreted as showing that when the erythrocyte membrane is made leaky on lysis, protein binding sites on the inner face of the membrane become accessible to the label, and thus that the protein spans the membrane. However, there are also several other possible explanations for differences in labelling between ghosts and intact cells. Labelling of membrane components by a permeating derivativizing compound can occur externally close to the bulk aqueous phase, in transit through the membrane, or on the internal face of the membrane. The extent of reaction will then depend on the reactivity of the groups involved, on the accessibility of the reactive site, and on the effective concentration of the label at the active site. Thus, for example, the presence of haemoglobin could be important in intact erythrocytes since it is present in such large concentrations: in its absence many "less reactive" proteins could become labelled because of label concentrations not possible in the presence of the protein (Schmidt-Ullrich et al., 1973). Rearrangement in the membrane following lysis could also explain differences in labelling. Experiments involving the lipophilic label 1-dimethylaminonaphthalene-5-sulphonyl chloride have, in fact, shown that the membrane peptides are differently reactive in intact cells and ghosts (Schmidt-Ullrich et al., 1973). A further problem is that erythrocytes are coated with a "fur" of carbohydrate obstructing much of the surface protein (Phillips and


39

Morrison, 1973): any changes in this surface coat on ghost formation could result in considerable changes in labelling pattern. It is difficult to see how any of the present type of labelling experiments can overcome these objections. In the absence of any firm evidence either way, it seems most likely that some protein will span the membrane, whilst some will penetrate only part way into the membrane, as in the Singer-Nicolson model (1972). The effects ofprotein on the lipid bilayer are as yet largely undetermined. Little is known about the nature of the lipid-protein interaction in membranes, although there is quite strong evidence for a shell of "immobilized" lipid around the membrane-bound proteins. This "immobilized" lipid is assumed to have properties different from the bulk lipid. Thus it is possible to remove lipid from the [Ca 2÷ + Mg 2 +] ATPase of sarcoplasmic reticulum by treatment with detergent, and the activity remains constant down to a lipid:protein molar ratio of 30: 1. If it is assumed that the part of the protein which penetrates the membrane is cylindrical, then its diameter would have to he ca. 40 A to accommodate thirty lipid molecules in a single bilayer shell. The molecular weight of the ATPase is 115,000, and if the membrane is 45 A thick, then these dimensions imply that approximately 50~ of the protein would protrude from the membrane surface (Warren et al., 1974c). Further evidence for an immobilized lipid shell comes from studies of the binding of Tempo to the sarcoplasmic reticulum membrane. The amount of Tempo bound is ca. 80~ of the amount bound by the equivalent amount of extracted lipid (Robinson et al., 1972). Since the lipid: protein molar ratio in sarcoplasmic reticulum is 90:1, there is enough lipid to form approximately three lipid shells around each protein molecule, of which only one is immobilized. Similarly, for the cytochrome oxidase-lipid complex, there is evidence from studies with spin-labelled fatty acids for a single bilayer shell of immobilized lipid surrounding the protein, the remainder of the lipid forming a normally fluid bilayer (Jost et al., 1973a, b). A similar kind of mosaic structure has been suggested for the cytochrome P450-cytochrome P450 reductase system of liver microsomes. Spin-labelled fatty acids incorporated into the membrane show a rapid rate of diffusion at 30°C. The activation energy for the reduction of the nitroxide-labelled fatty acids by the reductase, however, decreases abruptly above 32°C, suggesting that the enzyme system is enclosed by an immobilized phospholipid shell below 32°C, which undergoes a transition to a fluid state at 32°C (Stier and Sackmann, 1973). Again, the incorporation of rhodopsin into lipid bilayers containing spin-labelled lipids causes an increase in order parameter for the lipids, consistent .with a decrease in motional freedom for the lipid fatty acid chains (Hong and Hubbell, 1972). Studies of A. laidlawii and E. coil membranes using Tempo (Metcalfe et al., 1972; McConnell et al., 1972) or ANS (Trauble and Overath, 1973) partitioning suggest that ca. 80% of the membrane lipids are in a fluid state, and can take part in a lipid phase transition. There is considerable interest in the possibility that membrane bound proteins need particular lipids for activity, and thus that these lipids are in some way selected to form an immobilized ring of lipids around the protein~ However, experiments in which particular lipids are removed from membranes by the action of enzymes, detergents or organic solvents tend to be contradictory. Sometimes a particular lipid is reported to be necessary for activity of some particular enzyme, and sometimes it is not. The confusion lies in deciding whether the inactivation of the enzyme is due to removal of one specific lipid or to disruption of the membrane structure as a result of lipid removal. One of the most studied systems is the [Na÷-K+] ATPase of brain microsomes which has been lipid depleted using a variety of techniques, and reactivated by addition of lipids. Since it was often found that phosphatidylserine was the most effective of the reactivating phospholipids, it was suggested that this phospholipid was essential for the functioning of the native enzyme (see discussion in de Pont et al., 1973). However, it has now been shown that phosphatidylserine can be converted quantitatively into phosphatic[ylethanolamine by the action of phosphatidylserine decarboxylase, with no loss of [Na + + K ÷] ATPase activity (de Pont et al., 1973). Phosphatidylserine does not therefore have any specific role in maintaining [Na * + K ÷] ATPase activity. Similarly, despite earlier reports, it has now been shown that the regenerability of rhodopsin incorporated into lipid bilayers does not depend on

40

A.G. LEE

the presence of any particular phospholipid headgroup or fatty acid chain (Hong and Hubbell, 1973). On the other hand, the ternary protein-lipid complex isolated by Rothfield and Romeo (1971) from the outer membrane of S. typhimurium does show a specific lipid requirement. The complex consists of a galactosyl transferase which catalyses the addition of galactose to a lipopolysaccharide and has a specific requirement for a phospholipid which can be phosphatidylethanolamine but cannot be phosphatidylcholine. The mitochondrial enzyme D-fl-hydroxybutyrate dehydrogenase has also been purified and shown to require phosphatidylcholine for activity (Nielson and Fleischer, 1973). Apart from any possible lipid segregation attributable to the presence of proteins, it is likely that the lipid distribution in the bilayer will be non-random, simply because of the wide variety oflipids present. It is highly probable that some segregated pools of lipid will be in the gel phase whereas others will be in the liquid crystalline phase. Such an inhomogeneous distribution has been demonstrated in two bacterial systems. Pulse-labelled radioautographic studies of exponentially growing cells of Bacillus megaterium using radioactive palmitic acid show a highly non-uniform distribution of radioactively labelled phospholipids (Morrison and Morowitz, 1970). Similarly, studies of the induction of flgalactoside and fl-glucoside transport systems in mutants of E. coli that cannot synthesize unsaturated fatty acids suggest a lipid segregation in the gel phase. When these mutants are grown at 37°C in a medium supplemented with either elaidic acid or oleic acid, Arrhenius plots of the rate of/]-galactoside transport show discontinuities at 30°C and 13~'C respectively (Tsukagoshi and Fox, 1973). When cells are first grown on a medium supplemented in elaidic acid, and are then shifted to a medium supplemented in oleic acid before transport is induced at 37c'C, a single transition temperature for fl-galactoside transport is detected. This temperature lies between 30°C and 13°C, and shows that transport is primarily influenced by the average fatty acid composition of the membrane. If the transport is induced at 25°C after the shift in medium, however, two discontinuities are observed in Arrhenius plots of transport. This has been interpreted as showing that newly formed transport protein is incorporated into the membrane together with newly synthesized lipids. At 25°C, the lipid formed whilst in the elaidic acid medium will be largely in the gel phase, so that randomization of the lipid in the membrane will be slow: the membrane will then contain patches containing transport protein surrounded by the lipid synthesized in the elaidic acid medium and other patches in which the protein will be surrounded by the lipid synthesized in the oleic acid medium. At 37°C, however, all the lipid is in the liquid crystalline state, so that rapid lateral diffusion can occur in the membrane, and all the protein will experience the same, averaged, lipid environment. Lastly, discontinuities in the plots of enzymatic activity against temperature for a number of membrane systems occur at temperatures below the major lipid transition temperature, suggesting a heterogeneity in the lipid distribution, with particular enzymes in especially fluid environments. Such data are, however, not always completely clear: they are discussed further on p. 43. An asymmetry in lipid distribution has also been suggested between the two sides of the membrane. Thus Bretscher (1972) has postulated an asymmetrical distribution of lipids in the erythrocyte membrane, with phosphatidylcholine and sphingomyelin concentrated in the exterior half of the bilayer and phosphatidylethanolamine and phosphatidylserine concentrated on the interior. These results are based on experiments with a "membraneimpermeable" labelling reagent, and are open to the same criticisms as have been levelled at the protein labelling experiments. B. The State of Lipids in Biological Membranes Most membranes contain a complex collection of lipids (Ansell et al., 1973), and the correct lipid composition is presumably necessary for optimal functioning of the membrane, although the evidence for this is largely indirect. Many reports have appeared concerning the relationship between the growth temperature of bacteria and the fatty acid composition of the membrane lipids. In general, as the growth temperature increases, it is observed that the fatty acid chain length increases and the degree of unsaturation decreases. Since the longer and more saturated the fatty acid, the higher the transition temperature of the

Functional properties ofbiological membranes

41

lipid, it appears that the microorganism is preserving the proper "fluidity" of its membranes. Generally, the temperature range of the gel to liquid crystalline phase transition encompasses the growth temperature, so that lipids in both gel and liquid crystalline phases are probably present (see, for example, Steim et al., 1969; Melchior et al., 1970; Esfahani et al., 1971; Engelman, 1971). It is possible to speculate on the possible importance of this observation (p. 24) but any firm conclusions would require much more information from highly simplified systems. Studies of the physical properties of lipids in intact membranes are extremely difficult because of the complex, heterogeneous nature of the membrane. NMR studies of intact membranes have been reported, but can only give information about the averaged properties of some, undefined, group of the membrane lipids. Thus ca. 75~ of the lipids in sarcoplasmic reticulum contribute to the 13C nmr spectrum of this membrane (a fraction very similar to that of the fluid lipid component, as detected by Tempo binding) (Robinson et al., 1972). For these lipids, the 13C nmr studies suggest that C--C bond rotations are comparable to those in an aqueous dispersion of the extracted lipid, so that the bulk of the lipid is not restricted in motion by the presence of protein (Robinson et al., 1972). Similarly, in the retinal rod membrane, ~3C nmr studies show that a major proportion of the lipid fatty acid chains are free to undergo considerable conformational motion (Millet et al., 1973). Studies in which spin-labelled probes are incorporated into the membrane suffer from similar problems to the nmr studies since it is not known what proportion of the lipid environment of the membrane is being sampled by the probe. Indeed, it is known that in mixed lipid bilayers, spin-labelled fatty acids are excluded from any pools of lipid in the gel phase in preference for pools of lipid in the liquid crystalline phase (Oldfield et al., 1972). Spin-labelled lipids and fatty acids will therefore probably only report on the properties of the more "fluid" of the lipids in the membrane. Using fatty acids spin labelled at different positions in the chain, it has been found that the "fluidity" increases towards the centre of the membrane in A. laidlawii (Rottem et al., 1970), yeast (Eletr and Keith, 1972), viruses (Landsberger et al., 1971, 1972, 1973), mitochondria (Williams et al., 1972) and sarcoplasmic reticulum (Seelig and Hasselbach, 1971). As already described, the lipid directly in contact with membrane proteins is probably restricted in motion, but no details on the extent of this restriction are yet available. Estimates have also been made of the rate of lateral diffusion of'lipid molecules in biological membranes. The rate of lateral diffusion of spin-labelled lipids incorporated by fusion into sarcoplasmic reticulum membranes was estimated by extrapolation from experiments performed at 50-70°C to be ca. 7 x 10-a cm 2 sec-~ at 40°C (Scandella et al., 1972). From proton nmr experiments, a lower limit on the rate of lateral diffusion was estimated as 6 x 10-9 cm 2 sec-1 at 8°C (Lee et al., 1973). The rates of diffusion of spinlabelled fatty acids in liver microsomal membranes was estimated to be ca. 11 x i0- 8 cm 2 sec-~ at 30°C (Stier and Sackmann, 1973) and in E. coli membranes ca. 3 x 10-a cm 2 sec- 1 at 40°C (Sackmann et al., 1973). Although none of the numbers should be taken too seriously, they probably give a fair representation of the rates of lateral diffusion for the bulk of the lipids. The average distance travelled per second ~/x-2 for a two-dimensional lipid lattice is related to the diffusion coefficient by the equation O = ~2/2.

(28)

Thus if the lipids in the l-It-long bacterium can be characterized by a self-diffusion coefficient o f ca. 10-s cm 2 sec-1, then a lipid molecule can move from one end of the bacterium to the otherin the order of seconds. The rate of lipid flip-flop has been measured for spin-labelled phosphatidylcholines incorporated into excitable membrane vesicles prepared from the electroplax of Electrophorus electrieus (McNamee and MeConnell, 1973). The rate of transfer of the spinlabelled lipids from the inner to the outer surface of the membrane is characterized by a half-time of ca. 5 min at 15°C. This rate is an order of magnitude faster than the corresponding rate in pure phospholipid vesicles (Kornberg and McConnell, 1971). Whether

42

A.G. LEE

it is a true reflection of the rate of lipid transfer in an intact membrane, however, remains to be seen. It is quite possible that during the preparation of the membrane vesicles, discontinuities were formed, perhaps at the lipid-protein interfaces, which could account for the high rates of flip-flop. Alternatively, any fat droplets incorporated into the vesicles during their preparation could provide a site with a high flip-flop rate. If the rates of transfer for natural phosphatidylcholines in membranes are of this order of magnitude, however, it is difficult to see how an asymmetric lipid distribution of the type suggested by Bretscher (1972) could be maintained. C. The M o b i l i t y o f Macromolecules in Biological M e m b r a n e s

The observed translational freedom of motion for at least some of the lipids in biological membranes suggests that some membrane proteins will also be freely diffusing. According to the theory of translational Brownian motion for spherical particles, the translational diffusion coefficient is given by D -

kT 6na~l

(29)

where a is the radius of the particle and t/is the viscosity of the medium. The translational diffusion coefficient therefore depends rather weakly on molecular size. We can estimate that a cylindrical protein of molecular weight ca. 100,000 (a = 25 A) will diffuse about a factor of 10 more slowly than a lipid of molecular weight 800. Sackmann et al. (1973) have pointed out that eq. (29) may, however, be inapplicable in this case, since the protein and lipid molecules are of comparable sizes, and the derivation of eq. (29) assumes that the medium is effectively continuous. Using a "free-volume" model for the diffusion, it is estimated that a cylindrical protein of molecular weight 100,000 will diffuse a factor of 100 more slowly than the lipid molecules (Sackmann et al., 1973). The only accurate measurement of a protein diffusion rate in the surface of a membrane that has so far been reported is that for rhodopsin in the retinal rod membrane (Poo and Cone, 1974). The diffusion coefficient obtained was ca. 4 × 10- 9 c m 2 SeC- 1 at 20°C. Cone (1972) has also shown that rhodopsin molecules are rotating freely in the plane of the menlbrane, with a relaxation time of ca. 20 #sec at 20°C. The rhodopsin molecule apparently spans the membrane (Poo and Cone, 1974) and has an effective radius of ca. 1%30 A. The lipid environment of rhodopsin must therefore be highly fluid, and can be characterized by a viscosity of ca. 1 P. Trauble and Sackmann (1973) have shown that the rotational motion of the rhodopsin molecule can be explained very simply in terms of the appearance and disappearance of vacancies in the lipid matrix of the membrane, in positions next to local protrusions in the protein. X-ray evidence indicates that the average separation between nearest-neighbour rhodopsin molecules in the disc membrane is ca. 70 A (Blaurock and Wilkins, 1972). The average time r between collisions with another rhodopsin molecule can then be estimated as Z = sZ/4D

(30)

where s is the distance travelled between collisions. Assuming a diameter for rhodopsin of ca. 45 A, s = 25 A. The time between collisions is then ca. 4 #sec at 20°C (Poo and Cone, 1974). Put another way, the collision frequency between rhodopsin molecules is in the range l0 s 106 Collisions per second. It is too early to assess the possible significance of such high collision frequencies, but clearly it implies that rates of reaction between membrane proteins can be very high. Thus it has been suggested that the cytochrome b5 and NADH-cytochrome bs reductase in endoplasmic reticulum are randomly distributed, rather than bound in a "complex" or fixed array. Translational diffusion of reductase and cytochrome within the membrane are then required prior to their interaction (Rogers and Strittmatter, 1974). A high rate of lateral diffusion has also been estimated for antigens in the cell surface. Frye and Edidin (1970) fused cell lines of mouse and human origin and treated the hybrids with fluorescent antibodies, red for human antigens and green for mouse. After fusion, most hybrids initially showed distinct red and green halves, but within an hour the colours


43

were completely mixed. Since metabolic inhibitors did not prevent the mixing, whereas lowering the temperature did, the mixing was attributed to diffusion of the antigens in the cell structure. A diffusion coefficient of ca. 0.02 x 10- 8 c m 2 sec- 1 has been estimated (M. Edidin, in Scandella et al., i972). It is also probable that membrane immunoglobulins are free to diffuse in the surface of the membrane, since antiglobulin reagents cause aggregation into spots, which then collect into a large cap at one pole of the cell (Taylor et al., 1971; Loor et al., 1972; Unanue et al., 1972; Yahara and Edelman, 1972; Raft and de Petris, 1973). The movement of particles of carbon, gold, etc., on the cell surface might also be evidence for diffusion in the membrane. The movement of gold particles on the surface of 3T3 mouse fibroblasts can be characterized by two diffusion coefficients: a slow motion with D - 2 x 10-11 c m 2 SeC-1 and a faster motion with D - 2 x 10-,o c m 2 sec-1 (Albrecht-Buhler, 1973). Abercrombie et al. (1970) have proposed that such movement is caused by a rapid flow of the membrane from the leading ruffled edge of fibroblasts back to some "sink" in the central area; at the "sink", the membrane material is disassembled and is reassembled at the leading edge of the cell. A more likely explanation could be in terms of the circulation of membrane components in the surface, perhaps involving microfilaments. Evidence for translation of membrane proteins also comes from experiments showing reversible particle aggregation. The particles appearing in freeze-fractured erythrocyte ghosts undergo a reversible pH-dependent aggregation (da Silva, 1972) and a temperaturedependent aggregation has been observed for the alveolar membrane of Tetrahymena (Speth and Wunderlich, 1973). D. The Effects o f Temperature on Biological Membranes In intact biological membranes, the interpretation of the effects of changes in temperature is likely to be very difficult. Four factors have to be considered: (i) lipid segregation, (ii) lipid liquid crystalline to gel phase transitions, (iii) lipid cluster formation, (iv) reversible (or irreversible) conformational changes in proteins. In membranes containing a wide variety of lipids it is difficult to distinguish between these various possibilities. Such distinction is really only possible in highly simplified systems, containing a single protein and a single species of phospholipid, Such systems can now be prepared by a lipid-substitution technique, in which the lipid associated with a membrane protein is directly exchanged with defined exogenous lipid in the presence of detergent. Using this technique, the calcium transport protein from sarcoplasmic reticulum has been prepared, with more than 99~ of the original lipids replaced by dioleoyllecithin, dimyristoyllecithin or dipalmitoyllecithin (Warren et al., 1974a, b, c). The ATPase activity of the calcium transport protein has been found to be very sensitive to the liquid crystalline to gel phase transition of the substituted lipid (Fig. 21). The phase transition in dioleoyllecithin (DOL) is at about -22°C, and the DOL-substituted ATPase is active throughout the temperature range (but see below). On the other hand, dimyristoyllecithin (DML) has a phase transition at 24°C, and DML-substituted ATPase is inactive below this temperature. Although the phase transition temperature for dipalmitoyllecithin is at 42°C, the DPL-substituted ATPase has appreciable activity down to about 30°C. The reasons for these differences have not yet been established. Nevertheless, it is clear that the calcium transport protein exhibits no ATPase activity when its surrounding lipids are in a gel phase: this is not unexpected for a protein which is believed to span the lipid bilayer. The study of thermal effects in intact membranes has made much use of bacterial mutants in which the fatty acid composition of the membrane phospholipids depends on the fatty acid content of the growth medium. It has been found that the temperature dependence of the activity of several membrane bound proteins exhibit characteristic breaks at temperatures that are dependent on the fatty acid composition of the membrane (Lyons, 1972; Linden et al., 1973; Overath and Trauble, 1973). However, the interpretation of such results in molecular terms is far from easy. The first problem is that the incorporation of

44

A.G. Llv

DOL-ATPase

/

IU/mg4[

lo

-'2-o

-

-3o

Assay temperature

40

FK;.21. The temperaturedependenceof the ATPasc activityof the calcium transport protein associated with dioleoyllecithin(DOL), dimyristoyllecithin(DML) and dipalmitoytlecithin(DPL) (from Warrcn et al.. 1973cl. the fatty acid added to the growth medium is less than 100'~o: for example in E. coil mutants grown on a medium containing, respectively, elaidic acid and oleic acid, the percentage incorporations into the membrane phospholipids are ca. 80~'o and ca. 50% (Linden et al., 1973; Overath and Trauble, 19731. The membranes therefore contain a range of fatty acid chains, together with a variety of different lipid classes, so that temperature effects on the membrane are likely to be complex because of segregation of lipids in the plane of the membrane. The detection of phase transitions within the lipid portion of the bilayer using esr or fluorescent probe molecules is complicated by the fact that the probe molecule will probably be largely confined to the more fluid lipid pools of the membrane. There is also a more fundamental problem associated with the interpretation of the "phase transitions" observed in these membranes. If the co-operativity of the transition is low, then the transition will be intrinsically broad, and it would be very misleading to consider it as a more usual, sharp all-or-none type of phase transition (see p. 21). In spite of all these problems, it is clear that there is a correlation between thermal effects as detected for the lipid component of the membrane and variations of enzymatic activity with temperature. Thus E. coli grown on elaidic acid show lipid "transitions" at ca. 3040°C, and breaks in glucoside transport occur in the same temperature range (Linden et al., 1973; Overath and Trauble, 1973). A protein like the/~-galactoside carrier is presumably embedded in the lipid matrix and requires a suitably fluid environment to enable it to transport the galactose across the membrane. Conversion of the surrounding lipid from a liquid crystalline to a gel phase would increase the viscosity of the lipid phase and reduce its compressibility, so reducing the rate of sugar transport across the membrane. An exception to this generalization will occur at a temperature close to the liquid crystalline to gel phase transition. Here, as noted before, there will be an increase in the lateral compressibility of the membrane lipids, and so an increase in the rate of sugar transport might be expected as the temperature is lowered to a point corresponding to the onset of the phase transition (Linden et al., 1973). Such an effect has in fact been observed for/~-galactoside transport in E. coil cells grown on a medium supplemented with elaidic acid (Linden et al., 1973). Confirmation of the effect on enzyme activity in a simpler system is clearly important. Thermal effects in E. coli grown on a medium supplemented with oleic acid are less clear than those discussed above. Overath and Trauble (1973) interpret their spectroscopic data on both membranes and extracted lipids in terms of a single broad transition centred at between 11° and 17°C. depending on the conditions used, with a width of ca. 12°C. Linden et al. (1973) interpret their spectroscopic data in terms of two transitions, at 31.1 '~ and 15.8°C for the membrane, and at 27.3 " and 8.9°C for the extracted lipid. Overath and Trauble (1973) report a single break in sugar transport at 15- 16°C, whereas Linden et al. (1973) report three, at 26 '~, 21.8" and 14.4°C. Many of these temperatures are relatively high for


45

a membrane whose fatty acids are ca. 50% oleic acid, and the possibility that they are due to lipid cluster formation should be considered. Stronger evidence for an effect of lipid cluster formation on enzyme activity comes from studies of the [Ca 2÷ + Mg 2÷ ] ATPase from sarcoplasmic reticulum, after substitution of the exogeneous lipids with dioleoyllecithin. An Arrhenius plot of the ATPase activity of this system shows a discontinuity at ca. 30°C, at which temperature there is also a break observed in the binding of the spin label Tempo. This has been attributed to the effects of the formation of quasi-crystalline clusters of lipid in otherwise freely dispersed lipid (Lee e t al., 1974c). There is also some evidence for the importance of cluster formation in intact sarcoplasmic reticulum. Spin-label studies suggest a change in the properties of the membrane at ca. 25°C (Inesi e t al., 1973; Lee e t al., 1974c), which correlates with an abrupt decrease in the rate of release of accumulated Ca 2÷ as induced by the antibiotic X-537A at the same temperature (Scarpa e t al., 1972). Arrhenius plots of the ATPase activity of intact sarcoplasmic reticulum also show discontinuities at 26°C (Lee e t al., 1974c). Since the lipids of sarcoplasmic reticulum are highly unsaturated (Warren et al., 19"14b), this transition is unlikely to correspond to a liquid crystalline to gel phase transition. Similar effects in many other membrane systems at ca. 20°C could be due to cluster formation. In these relatively complex systems, no definitive assignment will generally be possible. A lipid liquid crystalline to crystalline transition has been observed in rat liver mitochondria at ca. 0°C by differential scanning calorimetry (Blazyk and Steim, 1972). Spinlabel studies, however, detect a change at ca. 25°C, at which temperature there are also discontinuities in Arrhenius plots of the damped volume oscillations of rat mitochondria (Williams et al., 1972; Tinberg et al., 1972) and in Arrhenius plots of succinate oxidase activity (Raison e t al., 1971). Further, Arrhenius plots of several enzyme activities in beef and rabbit heart mitochondria show breaks at ca. 20°C, which correlate with breaks observed in spin label studies (Lenaz et al., 1972; McMurchie e t al., 1973). Differential scanning calorimetric studies of rat liver microsomes also show a liquid crystalline to crystalline transition at ca. 0°C (Blazyk and Steim, 1972), consistent with the high content of unsaturated fatty acid chains (Dallner, 1968). Spin-label studies in guineapig liver microsomes detect a transition at ca. 19°C, which correlates with changes in the activities of microsomal glucose 6-phosphatase and UDP-gluceronyl transferase also at ca. 19°C (Eletr e t al., 1973). Spin-label studies of lamb kidney microsomes and Arrhenius plots of(Na ÷ + K ÷) ATPase activity both show discontinuities at ca. 20°C (Grisham and Barnett, 1973). Both rabbit kidney (Charnock e t al., 1973) and rat brain (Gruener and AviDor, 1956) (Na ÷ + K ÷) ATPases show discontinuities at 20°C. The mixing of surface antigens during the formation of man-mouse heterokaryons show a discontinuous increase with temperature at ca. 20°C. Again, the "time to fusion" during the chemically induced fusion of hen erythrocytes (Ahkong et al., 1973) and the rate of leak from mouse ascites cell during Sendai virus-induced fusion (Pasternak and Micklem, 1973) all show discontinuous increases between 20 ° and 30°C. IX. T H E B I O L O G I C A L

MEMBRANE:

CONCLUSIONS

The lipid and protein composition of most biological membranes is complex. This complexity is presumably connected with the wide variety of functions that the membrane is required to perform. Evena membrane such as the sarcoplasmic reticulurn, which contains predominantly one protein, a [Ca 2÷ - Mg 2÷] ATPase, contains many classes of lipid (Warren e t al., 1974c). The wide variety of lipids present in this membrane is not necessary for enzyme activity, since it is possible to prepare a system which can accumulate Ca 2+, and which contains just the [Ca 2÷ - Mg 2÷ ] ATPase and a single lipid, dioleoyllecithin. The net uptake of Ca 2÷ by this system is, however, very low, because of passive leak of Ca 2 + : dioleoyllecithin vesicles are known to be leaky to Ca 2 ÷. Some of the lipid complexity may be to prevent this passive leak. On the other hand, after depolarization of the sarcoplasmic reticulum membrane, C a 2 ÷ must be rapidly released from store to initiate

4¢,

A . G . LEE

muscle contraction. The lipids of sarcoplasmic reticulum might be involved in this process. Finally, the membrane lipids could be involved in maintaining and controlling the activity of the C a 2 + pump. Although it is known that some membrane-transport proteins are very sensitive to the nature of their lipid environment, the mechanism of action of these proteins is not yet clear. It is possible that after some suitable stimulus, a "pore" opens through the protein, and the diffusing molecule simply diffuses passively through this pore. Perhaps a more likely mechanism is the translocation of a substrate binding site from one side of the membrane to the other: this could either involve a change in configuration of the protein, or the rotation of the whole protein about an axis parallel to the surface of the bilayer. For the Ca 2 + pump of sarcoplasmic reticulum and for the Na + pump of nerve, either of the translocation mechanisms seems possible. The rate of turnover of these pumps (ca. 25 per sec for the Na + pump, and slower for the Ca 2+ pump) (see Ritchie, 1973) is not impossibly high for a mechanism involving protein rotation through a fluid membrane environment. On the other hand, the rate of passage ofNa + ions through a nerve membrane pore during the passage of an action potential is ca. 108 sec- 1. The maximum rate at which ions could arrive by diffusion at the mouth of the pore is only about ten times faster (Keynes, 1972). The Na + pore is therefore more likely to be a conducting channel than to correspond to a mobile carrier shuttling to and fro across the membrane: the chemical nature of this pore has yet to be established. There is also very little known about the mode of action of other membrane bound proteins. For a number of the reactions catalysed by such proteins, it has been suggested that there could be advantages in moving the active site into the lipid bilayer during the course of the reaction. This could be true, for example, for membrane dehydrogenases which accept hydrogen from water-soluble substrates and then transfer it to some lipophilic intermediate. Preliminary evidence in favour of such a movement has been obtained from spin-labelling studies (Kaprelyanz et al., 1974). In cell membranes containing a wide variety of proteins, both short-range ordering and some gross long-range ordering is likely. Thus there is now considerable evidence for the formation of a bilayer shell of relatively immobilized lipid around membrane proteins. It is reasonable to suppose that membrane proteins possess "binding sites" for lipid molecules, and that these sites have different affinities for different species of lipid. A membrane protein might then be able to segregate specific lipids in the membrane (either by selective binding or repulsion) and thereby acquire a local lipid environment which is compatible both with the protein's function and the cell's requirements. Experiments with fatty acid mutants of E. coli support this view. For example, the transition temperature occurring in Arrhenius plots of succinate dehydrogenase activity is higher in membranes derived from cells grown with linoleic acid as the medium supplement than in membranes derived from cells grown with elaidic acid as medium supplement: this is precisely the opposite of the result observed for proline transport (Esfahani, 1971). Such experiments strongly suggest the presence of local domains of membrane phospholipids differing in fatty acid composition from the average fatty acid composition of the membrane lipids. In bacteria at the optimal growth temperature, it is likely that lipids in both the gel and liquid crystalline phases will be present. In mammalian cells, on the other hand, the great majority of the glycerophospholipids contain one saturated (C 1) and one unsaturated (C2) fatty acid (White, 1973), so that most of the lipids would be expected to be in the liquid crystalline phase at body temperature. However, if cholesterol present in the membrane associated preferentially with particular classes of lipid, then this could cause a separation into domains of lipid of differing "fluidity". The presence of lipid phases of differing density in the membrane would lead to a high lateral compressibility, which could be favourable for the operation of a transport protein. The disordered regions present at the boundaries between the lipid phases could also be important for growth and repair of the membrane. Both lipids and proteins in membranes turn over many times during the lifetime of a cell (Siekevitz, 1972). Somehow the new material has to be inserted into the structure, presumably with the minimum perturbation of the membrane during the process. This can be achieved most easily at lattice disloca-


47

tions or grain boundaries (Harris and Striven, 1970), where the packing density will be lowest. Once within the membrane, proteins would have the choice of operating in a region of high or low fluidity, or in the region of lipid disorder between two phases. There is, as yet, very little information bearing on this point. It has, however, been found that whereas rhodopsin exists in a very fluid environment in the retinal rod membrane (Cone, 1972); a rhodopsin-like protein in the purple membrane fragment of Halobacterium halobium exists in a very rigid lipid environment, with crystalline, hexagonally packed lipid fatty acid chains (Blaurock and Stoeckenius, 1971). For those lipids not tightly bound to proteins, lateral diffusion in the plane of the membrane is probably fast and is appreciable even in the gel phase. Diffusion of small organic molecules within a lipid bilayer in a direction parallel to the plane of the bilayer is also fast. The diffusion coefficient of benzene in egg lecithin bilayers has been measured as 2 x 10 - 6 c m 2 see-1 (Rigaud et al., 1972). The possible biological importance of diffusion in two dimensions rather than in three has been discussed by Adam and Delbruck (1968). There is now considerable evidence for fast lateral diffusion of proteins in the plane of the bilayer. This implies a high rate of collision between proteins; the collision frequency between rhodopsin molecules is in the range 105-106 collisions per second (Poo and Cone, 1974). Such high collision rates mean that membrane proteins involved in a sequence o f reactions need not be located together in the membrane as some sort of complex. Nevertheless, many cells show a marked polarity of structure, and an explanation of this longrange order is necessary. In the hepatic cell, there are morphologically well-defined secretive and adsorptive surfaces with the former identified with microvilli adjacent to bile caniculi and the latter identified with junctional complexes connecting adjacent cells. It is also probable that different enzymes reside at these two locations: a (Mg 2+) ATPase at one and a (Mg 2+ - Na + K +) ATPase at the other (Siekevitz, 1972). Again, in adult skeletal muscle fibres, sensitivity to acetylcholine is normally restricted to the small end-plate regions where contact is established with nerve endings: a high population of acetylcholine receptors is postulated in this region (Diamond and Miledi, 1962). Localization of proteins in one region of the membrane could well produce a corresponding localization of particular lipid classes. Thus the enzyme 5'-nucleotidase has been found to be differentially localized in the liver plasma membrane, and can be isolated as a lipid-protein complex containing essentially only one class of lipid, sphingomyelin (Widnell and Unkeless, 1968). Microtubules within the cell may well have an important role in maintaining such longrange membrane structure by anchoring to suitable proteins in the membrane. Evidence for this comes from studies of phagocytosis by polymorphonuclear leucocytes (PMN). Normally such phagocytosis is not accompanied by internalization of membrane sites involved in specific transport ofaminoacids (Tsan and Berlin, 1971). After treatment of P M N with the drug colchicine which disrupts microtubules, however, phagocytosis leads to the internalization of large areas of membrane involved in active transport (Ukena and Berlin, 1972). This indicates a loss of a mosaic-like structure normally existing in PMN, with separation of phagocytic and specific transport sites, mediated by intact microtubules. The sensitivity of P M N to agglutination with concanavalin A is thought to be due to the presence of Con-A receptors in patches in the cell membrane. After treatment with colchicine or vinblastine, however, the agglutination is much reduced, again suggesting that microtubules maintain some fixed distribution of membrane constituents (Berlin and Ukena, 1972). Another type of ordering that might be important in membranes is an asymmetric distribution of lipid between the two sides of the bilayer. If a protein spanning the membrane were asymmetric, or if a protein was only situated in one half of the membrane, then this couldcause a non-equal distribution oflipids. Similarly, glycolipid synthesized on one surface of the membrane could interact preferentially with particular classes of lipid and induce an asymmetry. In membranes containing both negatively charged and neutral lipids, a potential difference across the membrane could also induce an asymmetric lipid distribution due to differences in the electrical double layer energy (McLaughlin and Harary, -

48

A.G. LEE

1974). Such effects might be important in the nerve membrane (Ohki, 1973; McLaughlin and Harary, 1974). An asymmetric distribution of lipids could also be important in determining the effects of divalent ions on biological membranes. It has been shown in a lipid bilayer containing cardiolipin on one side and glycerol dioleate on the other, that addition of Ca 2 + to the cardiolipin side causes a drop in electrical resistance, but addition to the other side has no effect (Montal, 1973). Addition of Ca 2 + to both sides of a bilayer of phosphatidylserine causes an increase in electrical resistance (Papahadjopoulos and Ohki, 1969; Ohki, 1970, 1972). In membranes containing relatively small fractions of phosphatidylserine or other negatively charged lipids, the addition of divalent ions could also have important effects since local changes in the state of compression of the lipid in the membrane could be transmitted appreciable distances away. Alternatively, if these lipids were closely associated with certain membrane proteins, then the activity of such proteins would be very responsive to local changes in pH or cation concentration. Control of this type might be important, for example, for the (Ca 2+ + Mg 2+) ATPase of sarcoplasmic reticulum, since this membrane contains phosphatidylserine (Marai and Kuksis, 1973). The sarcoplasmic reticulum is involved in the regulation of muscle function by accumulating Ca 2+ and reducing the concentration of Ca 2+ in the cytoplasm below 10 -7 M, thereby causing muscle relaxation. The reduction of Ca 2+ in the cytoplasm is achieved by transport of Ca 2+ across the sarcoplasmic reticulum membrane, by a process tightly coupled to ATP hydrolysis and catalysed by the (Ca 2+ + Mg 2+) ATPase in the membrane. The initial rate of this reaction is high, but falls markedly after a few seconds to a lower steady-state level (Tonomura, 1973). This change in activity fits it well for its physiological role: immediately after Ca 2 + has been released from store, the external Ca 2 + level rises. This Ca 2 + is then rapidly pumped back into store in the sarcoplasmic reticulure, and subsequently the ATPase has only to operate at a rate sufficient to overcome any passive leak of Ca e ÷ from store. Fragmented sarcoplasmic reticulum has three binding sites for C a 2+, with binding constants of 4 × 10 - 6 (~-site), 4 x 10 4 (/3-site) and 1 × 10 3 (7-site) (Ikemoto, 1974). The or-site is very probably the one involved in the activation of the ATPase, and thus in C a 2 + transport across the membrane. The second (/3) site appears to have little effect on the enzyme activity, whereas binding of Ca 2 + to the 7-site is involved in the inhibition of the ATPase (Ikemoto, 1974). It is then possible that the 7-site could correspond to phosphatidylserine on the inside surface of the membrane, and that C a 2 + binding to this lipid reduces the "fluidity" of the lipids surrounding the protein, causing an inhibition much as is observed by lowering the temperature. Although this is, at present, pure speculation, it is clear that serious consideration has to be given to the possibility of control of enzyme function by lipid-cation interaction. The importance of phosphatidylserine in neural excitation has been demonstrated by incubating lobster nerve with the enzyme phosphatidylserine decarboxylase. This converts phosphatidylserine into phosphatidylethanolamine, and the size of the nerve action potential decreases by about 40~o. If L-serine is now added to the incubation medium, phosphatidylserine is regenerated, and the height of the action potential returns to its former level (Cook et al., 1972). A role for Ca 2+ in neural excitation is consistent with the ideas of Tasaki (1968) who postulates that abrupt conformational changes occur in the axonal membrane during excitation, and that these changes can be triggered by variations in the molar ratios of monovalent to divalent cations. The effects of calcium are also important in the function of retinal rods. As a consequence of light absorption and visual pigment bleaching, the disc membranes release Ca 2 + from store. These Ca 2 + ions in turn block the flow of sodium ions across the plasma membrane of the cell, leading to hyperpolarization of the receptor. This polarization of the receptor cell membrane finally leads to the generation of a neural impulse (Mason et al., 1974). It is clear that the possibilities for interaction between membrane components are very considerable. Alterations in ionic balance across the membrane can have considerable effects on the state of the lipids in the membrane. In turn, this can affect the activities of


49

single proteins and affect the rate of interaction between proteins by altering rates Of diffusion within the membrane, and alterations in the physical state of lipids in one part of the membrane could have significant effects on the properties of distant parts of the membrane. Information on the functioning of biological membranes at a molecular level has, so far, come largel3/from the study of highly simplified systems: lipid bilayers and, at a preliminary level, single protein-lipid systems. The limitations of this type of approach should always be borne in mind: "The farther and more deeply we penetrate into matter, by means of increasingly powerful methods, the more we are confounded by the interdependence of its parts .... It is impossible to cut into this network, to isolate a portion without it becoming frayed and unravelled at all its edges" (Teilhard de Chardin, 1958). REFERENCES "A man will turn over halfa library to make one book" Samuel Johnson. as reported in Boswell's Life, 6 April 1775. ABERCROMaIE,M., HEAYSMAN,J. E. M. and PEGRUM, S. M. 0970) The locomotion of fibroblasts in culture. III. Movements of particles on the dorsal surface of the leading lamella. Exptl Cell. Res. 62, 389. AaRAnAMSSON, S. and PASCriER, I. (1966) Crystal and molecular structure of L-Ct-glycerylphosphorylcholine. Acta Cryst. 21, 79. ADA~ G. and DELaRUCK,M. (1968) Reduction of dimensionality in biological diffusion processes. In Structural Chemistry and Molecular Biology (ed. A. RICH and N. DAVlDSON),W. H. Freeman & Co., San Fransisco. ArIKONG, Q. F., CRAMP, F. C, FIStlER, D., HOWELL, J. 1., TAMPION,W., VERRINOER,M. and LucY, J. A. (1973) Chemically-induced and thermally-induced cell fusion:lipid-lipid interactions. Nature, New Biol. 242, 215. ALBRECrlT-BUHLER,G. (1973) A quantitative difference in the movement of marker particles in the plasma membrane of 3T3 mouse fibroblasts and their polyoma transformants. Exptl Cell. Res. 78, 67. ANSELL, G. B., HAWTHORNE,J. N. and DAWSON,R. M. C. (1973) Form and Function of Phospholipids, Elsevier, Amsterdam. BANGRAM,A. D. and HORNE,R. W. (1964) Negative staining of phospholipids and their structural modification by surface-active agents as observed in the electron microscope. J. Molec. Biol. 8, 660. BARTON, P. G. (1968) The influence of surface charge density of phosphatides on the binding of some cations. J. Biol. Chem. 243, 3884. BERLIN, R. D. and Ur~NA, T. E. (1972) Effect of colchicine and vinblastine on the agglutination of polymorphonuclear leucocytes by concanavalin A. Nature, New Biol. 238, 120. BERNAL,J. D. (1959) A geometrical approach to the structure of liquids. Nature, 183, 141. BERNAL,J. D. (1964) The structure of liquids. Proc. Roy. Soc. A, 280, 299. BIRDSALL,N. J. M., FEENEY,J., LEE,A. G., LEVINE,Y. K. and METCALFE,J. C. (1972) Dipalmitoyl-lecithin: assignment of the ~H and ~sC nuclear magnetic resonance spectra, and conformational studies. J. Chem. Soc., Perkins Trans. 11, 1441. BmDSALL, N. J. M., LEE, A. G., LEVINE,Y. K., METCALFE,J. C., PAR'rINGrOS, P. and ROaER'rs, G. C. K. (1973) Analysis of the ~sC T~ relaxation times in n-alkanes. J. Chem. Soc., Chem. Commun., p. 757. BITTMAN, R. and BLAU, U (1972) The phospholipid--cholesterol interaction. Kinetics of water permeability in liposomes. Biochem. !1, 4831. BLAUROCK,A. E. (1971) Structure of the nerve myelin membrane: proof of the low-resolution profile. J. Molec. Biol. 56, 35. BLAUROCK,A. E. (1972) Locating protein in membranes. Chem. Phys. Lipids, 8, 285. BLAUROCK,A. E. (1973) The structure ofa lipid--cytochrome C membrane. Biophys. J. 13, 290. BLAUROCK,A. E. and STOECKENIUS,W. (1971) Structure of the purple membrane. Nature, New Biol. 233, 152. BLAUROCK,A. E. and WILKINS,M. H. F. (1972) Structure of retinal photoreceptor membranes. Nature, 236, 313. BLAZYK,J. F. and S ~ I ~ J. M. (1972) Phase transitions in mammalian membranes. Biochim. Biophys. Acta, 266, 737. BRETSCm'R,M. S. (1972) Membrane structure: some general principles. Science, 181,622. BUTLER, K. W., DUGAS,H., SMITH, 1. C. P. and SCHNEIDER.H. (1970) Cation-induced organisation changes in a lipid bilayer model membrane. Biochem. Biophys. Res. Commun. 40, 770. CADENHEAD,D. A. and MULLER-LANDAU,F. (1973) Pure and mixed monomolecular films of 12-nitroxide stearate. Biochim. Biophys. Acta, 307, 279. CASPAR, D. L. D. and KIRSCHNER,D. A. (1971) Myelin membrane structure at 10 ]~ resolution. Nature, New Biol. 231, 46. CHADWICK, A. V. and SHERWOOD,J. N. (1971) Self-diffusion in molecular solids, in Diffusion Processes (ed. J. N. SHERWOOD,A. V. CHADWICK,W. M. MUIR and F. L. Swn~Tor~), Gordon & Breach, New York. CHADW~C~ A. V. and SHERWOOD,J. N. (1972) Lattice defects in plastic organic crystals. Part 7. Isotope mass effects in self-diffusion in cyclohexane, J'Chem. Soc. Faraday Trans. 1, 47. CHAPMA.~,D. (1968) Recent physical studies of phospholipids and natural membranes. In Biological Membranes (ed. D. CHAPMAS),Academic Press, London. CHAPMAN,D. and SALSaURY,N. J. (1966) Proton magnetic resonance studies of molecular motion in some 2,3diacyl-DL-phosphatidyl--ethanolamines. Trans. Faraday Soc. 62, 2607. CHAPMAN, D. and URa~A, J. (1971) Phase transitions and bilayer structure of Mycoplasma Laidlawii B. FEBS Lett. 12, 169. CHAPMAN,D, WmL~MS, R. M. and LADBROOKE,B. D. (1967) Thermotropic and lyotropic mesomorphism of some 1,2-diacyl-phosphatidyl cholines (lecithins). Chem. Phys. Lipid, I, 445. J.e.~. 29,'1

l~

50

A.G. LEE

CnARNOCK,J. S., COOK,D. A., ALMEIDA,A. F. and To, R. (1973) Activation energy and phospholipid requirements of membrane-bound adenosine triphosphatases. Arch. Biochem. Biophys. 159, 393. CHARVOLIN,J., MANNEVILLE,P. and DELOCHE,B. (1973) Magnetic resonance of perdeuterated potassium laurate in oriented soap-water multilayers. Chem. Phys. Lett. 23, 345. COGAN, U., SHINITZKY,M., WEBER, G. and NISHIDA,T. (1973) Microviscosity and order in the hydrocarbon region of phospholipid and phospholipid-cholesterol dispersions determined with fluorescent probes. Biochem. 12, 521. COHN, G. E., KEITH, A. D. and SNIPES,W. (1974) Spin label translational diffusion in solid tristearin. Biophys. J. 14, 178. COLaROW, K. and CLAYMAN,B. P. (1973) Far infrared absorption of the water-lecithin system, Bioehim. Biophys. Acta, 323, I. CONE,R. A. (1972) Rotational diffusion of rhodopsin in the visual receptor membrane. Nature, New Biol. 236, 39. COOK, A. M., LON, E. and ISmJIM1, M. (1972) Effect of phosphatidyl serine decarboxylase on neural excitation. Nature, New Biol. 239, 150. DALLNER, G. and ERNSTER,L. 0968) Subfractionation and composition of microsomal membranes: a review. J. Histochem. Cytochem. 16, 611. DARKE,A., FINER, E. G., FLOOK,A. G. and PHILLIPS,M. C. (1972) Nuclear magnetic resonance study of lecithin cholesterol interactions. J. Molec. Biol. 63, 265. DA SILVA,P. R. (1972) Translational mobility of the membrane intercalated particles of human erythrocyte ghosts. J. Cell Biol. 53, 777. DAVIES,D. B. and MATHESON,A. J. (1967) Influence of molecular rotation on some physical properties of liquids. Disc. Faraday Soc. 43, 216. DAVIES,J. T. and RIDEAL,E. K. (1961) Interfacial Phenomena, Academic Press, New York. DAWSON,R. M. C. and HAUSER,H. (1970) in Symposium on Calcium and Cellular Function (ed. A. W. CUTHBERT), Macmillan, London. DE GIER, J., MANDERSLOOT,J. G. and VAN DEENAN,L. M. M. (1968) Lipid composition and permeability of liposomes. Bioehim. Biophys. Acta, 150, 666. DE GIER, J., HAEST,C. W. M., MANDERSLOOT,J. G. and VAN DEENEN,L. U M. (1970) Valinomycin-induced permeation of S6Rb+ of liposomes with varying composition through the bilayers. Biochim. Biophys. Acta, 211, 373. DE GIER, J., MANDERSLOOT,J. G., HUPKES, J. V., MCELHANEY, R. N. and VAN BEEK, W. P. (1971) On the mechanism of non-electrolyte permeation through lipid bilayers and through biomembranes. Biochim. Biophys. Acta, 233, 610. DE KRUYFF, B., DE GREEF, W. J., VAN EYK, R. V. W., DEMEL, R. A. and VAN DEENEN,L. L. M. (1973a) The effect of different fatty acid and sterol composition on the erythritol flux through the cell membrane of Acholeplasma laidlawii. Biochim. Biophys. Acta, 298, 479. DE KRUVFF,B., DEMEL,R. A., SLOTBOOM,A. J., VAN DEENEN,L. L. M. and ROSENTHAL,A. F. (1973b) The effect of the polar headgroup on the lipid-cholesterol interaction: a monolayer and differential scanning calorimetry study. Biochim. Biophys. Acta, 307, 1. DE PONT, J. J. H. H. M., VAN PROOIJEN-VANEEDEN,A. and BONTING, S. L. (1973) Studies on (Na ÷ - K ÷)-activated ATPase. Phosphatidylserine not essential for (Na + - K÷)-ATPase activity. Biochim. Biophys. Acta, 323, 487. DEVAUX, P. and MCCONNELL, H. M. (197'2) Lateral diffusion in spin-labeled phosphatidylcholine multilayers, J. Am. Chem. Soc. 94, 4475. DEVAUX,P., SCANDELLA,C. J. and MCCON~ELL,H. M. (1973) Spin-spin interactions between spin-labeled phospholipids incorporated into membranes. J. MR#. Res. 9, 474. DE VRIES,A. (1970) X-ray photographic studies of liquid crystals. 1. A cybotactic nematic phase. Molec. Cryst. Liq. Cryst. 10, 219. DIAMOND,J. and MILEDI, R. (1962) A study of foetal and new-born rat muscle fibres. ,/. Physiol. 162, 393. DUPONT, Y., GABRIEL,A., CHABRE,M., GULIK-KRZYWlCKI,T. and SCHECHTER,E. (1972) Use of a new detector for X-ray diffraction and kinetics of the ordering of the lipids in E. coil membranes and model systems. Nature, 238, 331. DUPONT, Y., HARRISON,S. C. and HASSELBACH,W. (1973) Molecular organisation in the sarcoplasmic reticulum membrane studied by X-ray diffraction, Nature, 244, 555. EHRENSTEIN,G. and LECAR,H. (1972) The mechanism of signal transmission in nerve axons. Ann. Ret,. Biophys. Bioen~t. I, 347. ELETR, S. and KEITH, A. D. (1972) Spin-label studies of dynamics of lipid alkyl chains in biological membranes: role of unsaturated sites. Proc. Natl Acad. Sci. U.S.A. 69, 1353. ELETR, S., ZAKIM, D. and VESSEY,D. A. (1973) A spin-label study of the role of phospholipids in the regulation of membrane-bound microsomal enzymes. J. Molec. Biol. 78, 351. ENGEL,J. and SCHWARZ,G. (1970) Co-operative conformational transitions of linear biopolymers. An qew. Chem. Int. Ed. 9, 389. ENGELMAN,D. M. (1971) Lipid bilayer structure in the membrane of Mycplasma laidlawii. J. Molec. Biol. 58, 153. ENGELMAN,D. M. and ROTIqMAN,J. E. (1972) The planar organisation of lecithin-cholesterol bilayers. J. Biol. Chem. 2417, 3694. ERTL, H. and DULLIEN,F. A. L. (1973) Cluster sizes in the pre-freezing region from momentum transfer considerations and the relation of viscosity anomalies to the reduced freezing temperature. Proc. Roy. Soc. A, 335, 235. ESEAHANI,M., LIMBRICK,A. R., KNUTTON,S., OKA, T. and WAKIL, S. J. (1971) The molecular organization of lipids in the membrane of Escherichia coil: phase transitions. Proc. Natl. Acad. Sci. U.S.A. 68, 3180. FOLEAND,R. and STRANGE,J. H. (1972) Pressure dependence of self-diffusion in the plastic crystals hexamethylethane, norbornylene and cyclohexane. J. Phys. C. 5, L50. FRYE, L. D. and EDIDIN,M. (1970) The rapid intermixing of cell surface antigens after formation of mouse-human heterokaryons. J. Cell Sci. 7, 319.


51

GALEY,W. R., OWEN, J. D. and SOLOMON,A. K. (1973) Temperature dependence of non-electrolyte permeation across red cell membranes. 2. Gen. Physiol. 61, 727. GLEI~S, H. and LISSOWSKI,A. (1971) The rearrangement of atoms in high angle grain boundaries during grain boundary migration. Z. Metallkde, 61, 237. GOTTLIEB,A. M., INGLEFIELD,P. T. and LANGE,Y. (1973) Water-lecithin binding in lecithin-water lamellar phases at 20°C. Biochim. Biophys. Acta, 307, 444. GRAY, G. W. (1962) Molecular Structure and the Properties of Liquid Crystals, Academic Press, London. GRIFFITH,O. H., DEHLINGER,P. J. and VAN, S. P. (1974) Shape of the hydrophobic barrier of phospholipid bilayers (Evidence for water penetration in biological membranes). J. Membrane Biol. 15, 159. GRISHAM,C. M. and BARNETT,R, E. (1973) The role of lipid-phase transitions in the regulation of the (Sodium + Potassium) adenosine triphosphatase. Biochem.12, 2635. GRUENER,N. and AvI-DOR,Y. (1966) Temperature-dependence of activation and inhibition of rat-brain adenosine triphosphatase activated by sodium and potassium ions. Biochem. J. 100, 762. HAGELE,P. C. and PECHOLD,W. (1970) Struktur- und konformationsberechnung in polymeren, Kolloid-Z. u. Polymere, 241,977. HALL, J. E., MEADE, C. A. and SZABO,G. (1973) A barrier model for current flow in lipid bilayer membranes. J. Membrane Biol. 11, 75. HAMILTON,C. L. and McCONNELL,H. M. (1968) Spin labels. In Structural Chemistry and Molecular Biology (eds. A. R~CHand N. DAVlDSON),W. H. Freeman, San Fransisco. HANAI,T., HAYDON,D. A. and TAYLOR,J. (1965) Polar group orientation and the electrical properties of lecithin bimolecular leaflets. J. Theoret. Biol. 9, 278. HARRIS,W. F. and SCRIVEN,L. E. (1970) Function of dislocations in cell walls and membranes. Nature, 228, 827. HARRISON, S. C., CASPAR, D. L. D., CAMERINI-OTERO,R. D. and FRANKLIN, R. M. (1971) Lipid and protein arrangement in bacteriophage PM2. Nature New Biol. 229, 197. HAUSER,H., OLDANI,D. and PHILLIPS,M. C. (1973) Mechanism of ion escape from phosphatidylcholine and phosphatidylserine single bilayer vesicles. Biochem. 12, 4507. HAYDON,D. A. and HLADKY,S. B. (1972) Ion transport across thin lipid membranes: a critical discussion of mechanisms in selected systems. Quart. Re v. Biophys. 5, 187.. HEATLEY,F. (1974) Spin-lattice relaxation times of t3C satellites in proton nuclear magnetic resonance spectra and their use in determining molecular rotational diffusion constants. J. Chem. Sac. Faraday Trans. II, 70, 148. HILL, T. L. (1956) Statistical Mechanics, McGraw Hill, New York. HILL, T. L. (1960) Introduction to Statistical Thermodynamics, Addison-Wesley, Mass. HILL, T. L. (1963) Thermodynamics of Small Systems Part 1, W. A. Benjamin. New York. HINz, H.-J. and STURTEVANT,J. M. (1972a) Calorimetric investigation of the influence of cholesterol on the transition properties of bilayers formed from synthetic L-ct-lecithins in aqueous suspension. J. Biol. Chem. 247, 3697. HINZ, H.-J. and STURTEVANT,J. M. (1972b)Calorimetric studies of dilute aqueous suspensions of bilayers formed from synthetic L-g-lecithins. £. Biol. Chem. 247, 6071. Hmsc8, H. R. (1967) Relevance of the single-file model to water flow through porous cell membranes. Currents in Mad. Biol. 1, 139. HONG,K. and HUBBELL,W. L. (1972) Preparation and properties of phospholipid bilayers containing rhodopsin. Proc. Natl Acad. Sci. U.S.A. 69, 2617. HONG, K. and HUBBELL,W. L. (1973) Lipid requirements for rhodopsin regenerability. Bioehem. 12, 4517. HOOD, G. M. and SHERWOOD,J. N. (1966) Self-diffusion in cyclohexane single crystals. Molec. Cryst. l, 97. HORWlTZ, A. F. (1972) Nuclear magnetic resonance studies on phospholipids and membranes. In Membrane Molecular Biology (eds. C. F. Fox and A. D. KEITEI),Sinauer Assoc., Stamford. HOUSE,C. R. (1974) Water Transport in Cells and Tissues, Edward Arnold, London. HUANC;,C.-H. (1969) Studies on phosphatidylcholine vesicles. Formation and physical characteristics. Biochem. 8, 344. HUBBELL,W. L. and M CCONNELL,H. ~'~.(1971)Molecular motion in spin-labeled phospholipids and membranes. J. Am. Chem. Soc. 93, 314. IKEMOTO,N, (1974) The calcium binding sites involved in the regulation of the purified adenosine triphosphatase of the sarcoplasmic reticulum. J. Biol. Chem. 249, 649. INESI,G., MILLMAN,E. and ELETR,S. (1973) Temperature-induced transitions of function and structure in sarcoplasmic reticulum membranes. J. Molec. Biol. 81, 483. ISRAELACHVlLI,J. N. (1973) Theoretical considerations on the asymmetric distribution of charged phospholipid molecules on the inner and outer layers of curved bilayer membranes. Biochim. Biophys. Acta, 323, 659. JAIN, M. K., TOUISSAINT,D. G. and CORDES,E. H. (1973) Kinetics of water penetration into unsonicated liposomes, Effects of n-alkanols and cholesterol. J. Mere. Biol. 14, 1. JOST, P., WAGGONER,A. S. and GRIFFITH,O, H. (1971) Spin labeling and membrane structure, in Structure and Function of Biolooical Membranes (ed. L. I. ROTHFn~LD),Academic Press, New York. JOST, P. C., GRIFFITH,O. H., CAPALDI,R. A. and VANDERKOOI,G. (1973a) Evidence for boundary lipid membranes. Proc. Natl Acad. Sci. U.S.A. 70, 480. JOST.P., GR1FFITH,O. H., CAPALDI,R. A. and VANDERKOOI,G. (1973b) Identification and extent of fluid bilayer regions in membranous cytochrome oxidase. Biochim. Biophys. Acta, 311, 141. KAPRELYANZ,A. S., BINYUKOV,V. I., OSTROVSKII,.D.N. and GRIGORYAN,G. L. (1974) Mobility of protein molecules in Micrococcus Lysodeikticus membranes. FEBS Lett. 40, 33. KEITH, A. D., SHARNOFF,M. and COHN, G. E. (1973) A summary and evaluation of spin labels used as probes for biological membrane structure. Biochim. Biophys. Acta, 300, 379. KEYNES,R. D. (1972) Excitable membranes. Nature, 239, 29. KORNBERG, R. D. and McCONNELL, H. M. (1971) Inside-outside transitions of phospholipids in vesicle membranes. Biochem. 10, 1111. KRASNE, S., EISENMAN,G. and SzAao, G. (1971) Freezing and melting of lipid bilayers and the mode of action of nonactin, valinomycin and gramicidin. Science, 174, 412.

52

A.G. LEE

LADBROOKE, B. D. and CHAPMAN, D. (1969) Thermal analysis of lipids, proteins and biological membranes. A review and summary of recent studies. Chem. Phys. Lipid& 3, 304. LAGALY,G. and WEISS, A. (1971) Experimental evidence for kink formation. Angew. Chem. internal. Edit. 10, 558. LANDSBERGER, F. R., LENARD, J., PAXTON, J. and COMPANS, R. W. (1971) Spin-label electron spin resonance study of the lipid-containing membrane of influenza virus. Proe, Natl Acad. Sci. U.S.A. 68, 2579. LANDSBERGER, F. R., COMPANS, R. W., PAXTON, J. and LENARD, J. {1972) Structure of the lipid phase of Rauscher marine leukemia virus. J. Supramol. Struct. I, 5(1. LANDSBERGER, F. R,, COMPANS, R. W., CHOPPIN, P. W. and LENARD, J. (1973) Organization of the lipid phase in viral membranes. Effects of independent variation of the lipid and the protein composition. Biochem. 12, 4498. LAUGER, P. (1972) Carrier-mediated ion transport, Science, 178, 24. LECUYER, H. and DERVICHIAN, D. G. (1969) Structure of aqueous mixtures of lecithin and cholesterol. J. Molec. Biol. 45, 39. LEE, A. G. (1971) The Chemistry of ThaUium, Elsevier, London. LEE, A. G,, BIRDSALL, N. J. M., LEVINE, Y. K. and METCALFE, J. C. (19721 High resolution proton relaxation studies of lecithins. Biochim. Biophys. Acta, 255, 43. LEE, A. G., BIRDSALL,N. J. M, and METCALEE,J. C. (1973) Measurement of fast lateral diffusion oflipids in vesicles and in biological membranes by ~H nuclear magnetic resonance. Biochem. 12, 1650. LEE, A. G., BIRDSALL,N. J. M. and METCAt,FE, J. C. (1974a) Nuclear magnetic relaxation and the biological membrane. In Methods in Membrane Biology (ed. E. KORN), Plenum Press, New York. LEE, A. G., BIRDSALL, N, J. M., METCALFE, J. C., ROBERTS, G. C. K. and WARREN, G. B. (1974b) A determination of the mobility gradient in lipid bilayers by ~3C nuclear magnetic resonance. Submitted for publication. LEE, A. G., BIRDSALL,N. J. M., METCALFE,J. C., TOON, P. A. and WARREN, G. B. (1974c) Clusters in lipid bilayers and the interpretation of thermal effects in biological membranes. Biochem. 13, 3699. LENAZ, G., SECHI, A. M., PARENTI-CASTELLI,G., LANDI, L. and BERTOLI, E. (1972) Activation energies of different mitochondrial enzymes: breaks in Arrhenius plots of membrane bound enzymes occur at different temperatures. Biochem. Biophys. Res. Commun. 49, 536. LEVINE, Y. K. (1972) Physical studies of membrane structure. In Progress in Biophysics and Molecular Biology, Vol. 24 (eds. J. A. V. BUTLERand D. NOBLE), Pergamon Press, Oxford. LEVlNE, Y. K. (1973) X-ray diffraction studies of membranes. In Prooress in Surfiwe Science, Vol. 3 (ed. S. G. DAVISON), Pergamon Press. Oxford. LEVINE, Y. K. (1974) Personal communication. LEVINE, Y. K., BIRDSALL,N. J. M., LEE, A. G. and ME"fCALFE,J. C. (1972a) ~3C nuclear magnetic resonance relaxation measurements of synthetic lecithins and the effect of spin-labelled lipids. Biochem. 11, 1416. LEVINE, Y. K., PARTINGTON, P., ROBERTS, G. C. K., BIRDSALL, N. J. M., LEE, A. G. and METCALFE, J. C. (1972b) ~3C nuclear magnetic relaxation times and models for chain motion in lecithin vesicles. FEBS Lett. 23, 203. LEVINE, Y. K., LEE, A. G., BIRDSALL, N. J. M., METCALEE, J. C. and ROBINSON, J. D. (1973) The interaction of paramagnetic ions and spin labels with lecithin bilayers. Biochim. Biophys. Acta, 291,592. LEVINE, Y. K., BIRDSALL, N. J. M., LEE. A. G., METCALFE, J. C., PARTINGTON, P. and ROBERTS, G. C. K. (1974) Calculation of dipolar nuclear magnetic relaxation times in molecules with multiple internal rotations. J. Chem. Phys. 60, 2890. L1EB, W. R. and SXEIN, W. D. (1969) Biological membranes behave as non-porous polymeric sheets with respect to the diffusion of non-electrolytes. Nature, 224, 240. LINDEN, C. D., WRIGHT, K. L., McCONNELL, H. M. and Fox, C. F. (1973) Lateral phase separations in membrane lipids and the mechanism of sugar transport in Escherichia coll. Proc. Natl. Acad. Sci. U.S.A. 70, 2271. LIPPERT, J. L. and PETICOLAS,W, L. (1971) Laser Raman investigation of the effect of cholesterol on conformational changes in dipalmitoyl lecithin multilayers. Proc. Natl Acad. Sci. U.S.A. 68, 1572. LmPERT, J. L. and PETICOLAS,W. L. (1972) Raman active vibrations in long-chain fatty acids and phospholipid sonicates. Biochim. Biophys. Acta, 282, 8. LISSOWSKI, A. (1974) Personal communication. LOOR, R., FORNI, L. and PERNIS, B. (19721 The dynamic state of the lymphocyte membrane. Factors affecting the distribution and turnover of surface immunoglobulins. Eur. J. Immunol. 2, 203. LUZZATI, V. (1968)X-ray diffraction studies of lipid water systems. In Biological Membranes (ed. D. CHAPMAN), Academic Press, London. LYONS, J, M. (1972) Phase transitions and control of cellular metabolism at low temperatures. Cryobiol. 9, 341. MACDONALD, R. C. and BAYGHAM,A. D. (1972) Comparison of double layer potentials in lipid monolayers and lipid bilayer membranes. J. Membrane Biol. 7, 29. MACHTIGER, N. A. and Fox, C. F. (1973) Biochemistry of bacterial membranes. Ann. Rev. Biochem. 42, 575. MARAI, L. and Kt!KSIS, A. ( 19731 Molecular species of glycerolipids of adenosine triphosphatase and sarcotubular membranes of rabbit skeletal muscle, Can. J. Biochem. 51, 1248. MASON, W. T., FAGER, R. S. and ABRAHAMSON,E. W. (1974) Ion fluxes in disk membranes of retinal rod outer segments. Nature, 247, 562. MCCLURE, D. W. (1968) Nature of the rotational phase transition in paraffin crystals. J. Chem. Phys. 49, 1830. MCCONNELL, H. M., WRIGHT, K. L. and MCFARLAND, B. G. (1972) The fraction of the lipid in a biological membrane that is in a fluid state: a spin label assay. Biochem. Biophys. Res. Commun. 47, 273. McELHANEY, R. N., DE GIER, J. and VAN DER NEUT-KOK, E. C. M. (1973) The effect of alterations in fatty acid composition and cholesterol content on the nonelectrolyte permeability of Acholeplasma laidlawii B cells and derived liposomes. Biochim. Biophys. Acta, 298, 500. MCFARLAND, B. G. and McCO~NELL, H. M. (1971) Bent fatty acid chains in lecithin bilayers. Proc. Natl Acad. Sci. U.S.A. 68, 1274. MCLAUGHLIN, S. and HARARY, H. (1974) Phospholipid flip-flop and the distribution of surface charges in excitable membranes. Biophys. J. 14, 200. MCLAUGHL1N, S. G. A., SZABO, G. and EtSENMAN, G. (1971) Divalent ions and the surface potential of charged phospholipid membranes. J. Gen. Physiol. 58, 667.


53

McMtn~cr~m, E. J., RAISON,J. K. and CAUtNCROSS,K. D. (1973) Temperature induced phase changes in membranes of heart: a contrast between the thermal response of poikilotherms and homeotherms. Comp. Biochem. Physiol. ,biB, 1017. McNAt,~E, M. G. and McCONr~LL, H. M. (1973) Transmembrane potentials and phospholipid flip-flop in excitable membrane vesicles. Biochem. 12, 2951. MELCHIOR,D. L. and MOROWlTZ,H. J. (1972) Dilatometry of dilute suspensions of synthetic lecithin aggregates. Biochem. 11, 4558. MELCHIOR,D. L., MOROWITZ,H. J., STUR~VANT,J. M. and TSOr~G,T. Y. (1970) Characterization of the plasma membrane ofMycoplasma Laidlawii. Phase transitions of membrane lipids. Biochim. Biophys. Acta, 219, 114. METCALFE,J. C., BUtDSALL,N. J. M. and LEE,A. G. (1972) 13C NMR spectra of Acholeplasma membrane containing 13C labelled phospholipids. FEBS Lett. 21,335. MICrIAELSON,D. M., HORWla'-Z,A. F. and KLEIN,M. P. (1973) Transbilayer asymmetry and surface homogeneity of mixed phospholipids in cosonicated vesicles. Biochem. 12, 2637. MILLETT,F., HARGRAVE,P. A. and RAFTERY,M. A. (1973) Natural abundance laC nuclear magnetic resonance spectra of the lipid in intact bovine retinal rod outer segment membranes. Biochem~ 12, 3591. MONTAL,M. (1973) Asymmetric lipid hilayers. Response to multivalent ion. Biochint Biophys. Acta, 29~, 750, MONTAL,M. and MUELLER,P. (1972) Formation of bimolecular membranes from lipid monolayers and a study of their electrical properties. Proc. Natl Acad. Sci. U.S.A. 69, 3561. MORRISON, D. G. and MOROWlTZ,H. J. (1970) Studies on membrane synthesis in Bacillus meoaterium KM. J, Molec. Biol. 49, 441. MULLER,A. (1932) An X-ray investigation of normal paraffins near their melting points. Proc. Roy. Soc. A, 138, 514. NABARRO,F. R. N. (1967) Theory of Crystal Dislocations, Clarendon Press, Oxford. NAGLE,J. F. (1973a) Theory of biomembrane phase transitions, d. Chem. Phys. 58, 252. NAGLE, J. F. (1973b) Lipid bilayer phase transition: density measurements and theory. Proc. Natl Acad. Sci. U.S.A. 70, 3443, NIELSEN,N. C. and FLEISCI~R,S. (1973) Mitochondrial D-fl-hydroxybutyrate dehydrogenase. J. Biol. Chem. 248, 2549. ODAJIUA,A., SAUER,J. A. and WOODWARD,A. E. (1962) Proton magnetic resonance of some normal paraffins and polyethylene. J. Phys. Chem. 66, 718. OHKI, S. (1969) The electrical capacitance of phospholipid membranes. Biophys. J. 9, 1195. OrIKl, S. (1970) Effect of local anesthetics on phospholipid bilayers. Biochim. Biophys. Acta, 219, 18. OrlKl, S. (1972) Stability of asymmetrical phospholipid bilayers. Biochim. Biophys. Acta, 255, 57. OnKI, S. (1973) Evidence for a new concept of membrane potential. J. Theor. Biol, 42, 593. OHNISHI, S.-I. and Iro, T. (1973) Clustering of lecithin molecules in phosphatidylserine membranes induced by calcium ion binding to phosphatidylserine. Biochem. Biophys. Res. Commun. 51, 132. OLDFIELD,E., KEOUGH,K. M. and CHAPMAN,D. (1972) The study of hydrocarbon chain mobility in membrane systems using spin-label probes. FEBS Lett. 20, 344. OVERATn,P. and Ta_~UBLE,H. (1973) Phase transitions in cells, membranes, and lipids of Escherichia coll. Detection by fluorescent probes, light scattering and dilatometry. Biochem. 12, 2625. OVERBEEK,J. T. G. (1952) Electrochemistry of the double layer. In Colloid Science, Vol. 1, (ed. H, R. KRUVT), Elsevier, Amsterdam. PAPAHADJOPOULOS,D. (1968) Surface properties of acidic phospholipids: Interaction of monolayers and hydrated liquid crystals with uni- and bi-valent metal ions. Biochim. Biophys. Acta, 163, 240. PAeAHADJOPOULOS,D. (1970) Phospholipid model membranes. III. Antagonistic effects of Ca 2÷ and local anesthetics on the permeability of phosphatidylserine vesicles. Biochim. Biophys. Acta, 211,467. PAPAHADJOPOULOS,D. (1971) Na + - K + discrimination by "pure" phospholipid membranes. Biochim. Biophys. Acta, 241,254. PAPArlADJOPOULOS,D. and BANG,AM,A. D. (1966) Permeability of phosphatidylserine liquid crystals to univalent ions. Biochim. Biophys. Acta, 126, 185. PAPAHADJOPOULOS,D. and OnKl, S. (1969) Stability of asymmetric phospholipid membranes. Science, 164, 1075. PAPAHADJOPOULOS,D. and WATKrNS,J. C. (1967) Phospholipid model membranes. II. Permeability properties of hydrated liquid crystals. Biochim. Biophys. Acta, 135, 639. PAeArt~DJOI~OULOS,D., JACOBSON,K., Nut, S. and ISAC, T. (1973) Phase transitions in phospholipid vesicles. Fluorescence polarization and permeability measurements concerning the effect of temperature and cholesterol. Biochim. Biophys. Acta, 311, 330. PASrERNAK, C. A. and MICKLEM,K. J. (1973) Permeability changes during cell fusion. J. Memb. Biol. 14, 293. PrtmHes, M. C. (1972) The physical state of phospholipids and cholesterol in monolayers, bilayers and membranes. In Progress in Surface and Membrane Science, Vol. 5 (eds. J. F. DANrELLI,M. D. ROSENBERGand D. A. CADE~rrmAO),Academic Press, New York. PrtILLIPS, D. R. and MORRISON,M. (1973) Changes in accessibility of plasma membrane protein as the result of tryptic hydrolysis. Nature, New Biol. 242, 213. PHmLIPS, M. C., WILLIAMS,R. M. and CrtAeraAN,D. (1969) On the nature of hydrocarbon chain motions in lipid liquid crystals. Chem. Phys. Lipids, 3, 234. Pr,LHPS, M. C., LADaROOKE,B. D. and CnAMPMAN,D. (1970) Molecular interactions in mixed lecithin systems. Biochim. Biophys, Acta, 196, 35. PntLLn'S, M. C., FIr~R, E. G. and HAUSER,H. (1972) Differences between conformations of lecithin and phosphatidylethanolamine polar groups and their effects on interactions of phospholipid bilayer membranes. Biochim. Biophys. Acta, 290, 397. POO, M.-M. and CoI,,~, R. A. (1974) Lateral diffusion of rhodopsin in the photoreceptor membrane. Nature 247, 438. POWt~ALL,H. J. and SMrr., L. C. (1973) Viscosity of the hydrocarbon region of micelles. Measurement by excimer fluorescence. J. Am. Chem. Soc. 95, 3136. PRICE, H, D. and Tnor,wsoN, T. E. (1969) Properties of liquid bilayer membranes separating two aqueous phases: temperature dependence of water permeability. J. Molec. Biol. 41,443.

54

A.G. LEE

RADDA,(.J.K. and VANDERKOOI,J. (1972) Can fluorescent probes tell us anything about membranes? Biochim. Biophys. Acta, 265, 509. RAEE, M. C. and DE PETRlS, S. (1973) Movement of lymphocyte surface antigens and receptors: the fluid nature of the lymphocyte plasma membrane and its immunological significance. Fed. Proc. 32, 48. RAlSON,J. K., LYONS,J, M., MEHLHORN,R. J. and KEITH, A. D. (1971a) Temperature-induced phase changes in mitochondrial membranes detected by spin labeling. J. Biol. Chem. 246, 4036. RAISON, J. K., LYONS, J. M. and THOMSON,W. W. (1971b) The influence of membranes on the temperatureinduced changes in the kinetics of some respiratory enzymes of mitochondria. Arch. Biochem. Biophys. 142, 83. RAND, R. P. and SENGUPTA,S. (1972) Cardiolipin forms hexagonal structures with divalent cations. Biochim. BiDphys. Acta, 255, 484. REINERT,J. C. and STEIM,J. M. (1970) Calorimetric detection of a membrane-lipid phase transition in living cells. Science, 168, 1580. RIGAUD,J.-L., GARv-BoBo, C. M. and LANGE,Y. (1972) Diffusion processes in lipid-water lamellar phases. Biochim. Biophys. Acta, 266, 72. RITCHIE,J. M. (1973) Energetic aspects of nerve conduction: the relationships between heat production, electrical activity and metabolism. In Progress in Biophysics and Molecular Biology, Vol. 26 (eds. J. A. V. BUTLERand D. NOBLE),Pergamon Press, Oxford. ROBINSON, J. D., BIRDSALL,N. J. M., LEE, A. G. and METCALFE,J. C. (1972) 13C and 'H nuclear magnetic resonance relaxation measurements of the lipids of sarcoplasmic reticulum membranes. Biochem. 11, 2903. ROGERS, M. J. and STRITTUATTER,P. (1974) Evidence for random distribution and translational movement of cytochrome b 5 in endoplasmic reticulum. J. Biol. Chem. 249, 895. ROJAS,E., LETTVlN,J. Y. and PICKARD,N. F. (1966) A demonstration of ion-exchange phenomena in phospholipid mono-molecular films, Nature, 209, 886. ROTHFIELD, L. and ROMEO,D. (1971) Role of lipids in the biosynthesis of the bacterial cell envelope. Bacteriol. Rev. 35, 14. ROTHMAN,J. E. (1973) The molecular basis of mesomorphic phase transitions in phospholipid systems. J. Theor. Biol, 38, 1. ROTTEM, S.. HUBBELL,W. L., HAVEEICK,L. and McCONNELL, H. M. (1970) Motion of fatty acid spin labels in the plasma membrane of mycoplasma. Biochim. Biophys. Acta, 219, 104. SACKMANN,E., TRAUBLE,H., GALLA,H.-J. and OVERATH,P. (1973) Lateral diffusion, protein mobility, and phase transitions in Escherichia coil membranes. A spin label study. Biochem. 12, 5360. SALSBURV,N. J. and CHAPMAN,D. (1968) Nuclear magnetic resonance studies of diacyl-L-phosphatidylcholines (lecithins). Biochim. Biophys. Acta, 163, 314. SALSBURV,N. J., CHAt'MAN,D. and JONES, G. P. (1970) Hindered molecular rotation in 1,2-dipalmitoyl-L-phosphatidylcholine monodrate by nuclear magnetic resonance spin-lattice relaxation in the rotating frame (Tip). Trans. Faraday Soc. 66, 1554. SALSBURY,N. J., DARKE, A. and CHAPMAN,D. (1972) Deuteron magnetic resonance studies of water associated with phospholipids. Chem. Phys. Lipids, 8, 142. SCANDELLA,C. J., DEVAUX,P. and McCONNELL, H. M. (1972) Rapid lateral diffusion of phospholipids in rabbit sarcoplasmic reticulum. Proc. Natl Acad. Sci. U.S.A. 69, 2056. SCARPA, A., BALDASSARE,J. and INESl, G. (1972) The effect of calcium lonophores on li'agmented sarcoplasmic reticulum. J. Gen. Physiol. 60, 735. SCmNDLER,H. and SEELIG,J. (1973) EPR Spectra of spin labels in lipid bilayers. J. Chem. Phys. 59, 1841. SCHMIDT-ULLRICH,R., KNUFERMANN,H. and WALLACH,D. F. H. (1973) The reaction of l-dimethylamino-naphthalene-S-sulfonyl chloride (DANSC1) with erythrocyte membranes. A new look at "vectorial" membrane probes. Biochim. Biophys. Acta, 509, 353. SCHREIER-MucCILLO,S., MARSH, D., DUGAS, H., SCHNEIDER,H. and SMITH, I. C. P. (1973) A spin probe study of the influence of cholesterol on motion and orientation of phospholipids in oriented multibilayers and vesicles. Chem. Phys. Lipids, 10, I I. SEEEIG, J. (1970) Spin label studies of oriented smectic liquid crystals (A model system for bilayer membranes). J. Am. Chem. Soe. 92, 3881. SEELIG, J. and HASSELBACH,W. (1971) A spin label study of sarcoptasmic vesicles. Eur. J. Biochem. 21, 17. SEIMIYA,T. and OHKL S. (1973) Ionic structure of phospholipid membranes, and binding of calcium ions. Biochim. Biophys. Acta, 298, 546. SHEETZ,M. P. and CHAN,S. I. (1972) Effect of sonication on the structure of lecithin bilayers. Biochem. l 1, 4573. SHEMYAK1N,M. M., OVCHINNIKOV,YU. A., IVANOV,V. T., ANTONOV,V, K., VINOGRADOVA,E. I., SHKROB,A. M., MALENKOV,G. G., EVSTRATOV,A. V., LAINE,I. A., MELNIK,E. I. and RYABOVA,I. D. (1969) Cyclodepepsipeptides as chemical tools for studying ionic transport through membranes. J. Mere. Biol. I, 402. SHERWOOD,J. N. and WHITE, D. J. (1967) Self-diffusion in naphthalene single crystals. Phil. Mag. 16, 975. SHEWMON,P. (1963)Diffusion in Solids, McGraw-Hill, New York. SHIMSHICK,E. J. and McCONNELL,H. M. (1973a) Lateral phase separation in phospholipid membranes. Biochem. 12, 2351. SH1MSHICK,E. J. and McCONNELL, H. M. (1973b) Lateral phase separations in binary mixtures of cholesterol and phospholipids. Biochem. Biophys. Res. Commun. 53, 446. SIEKEWTZ,P. (1972) Biological membranes: the dynamics of their organization. Ann. Rev. Physiol. 34, I 17. SINGER, S. J. and NICOLSON,G. L. (1972) The fluid mosaic model of the structure of cell membranes. Science, 175, 720. SMITH, C. S. (1964) Structure, substructure, and superstructure. Rev. Mod. Physics, 36, 524. SPETn, V. and WUNDERLICH,F. (1973) Membranes of T etrahymena. II. Direct visualization of reversible transitions in biomembrane structure induced by temperature. Biochim. Biophys. Acta, 291,621. STARK, G., BENZ, R., PONE, G. W. and JANKO, K. (1972) Valinomycin as a probe for the study of structural changes of black lipid membranes. Biochim. Biophys. Acta, 266, 603. STEIM,J. M., TOURTELLOTE,M. E., REINERT,J. C., McEEHAHEV,R. N, and RADER, R. L. 0969) Calorimetric evidence for the liquid-crystalline state oflipids in a biomembrane. Proc. Natl Acad. Sci. U.S.A. 63, 104,


55

STIng, A. and SACKSmNN,E. (1973) Spin labels as enzyme substrates. Heterogeneous lipid distribution in liver microsomal membranes. Biochim. Biophys. Acta, 311,400. SUNDARALINGAM,M. and JENSEN,L. H. (1965) Crystal and molecular structure of a phospholipid component: L-ct-glycerophosphorylcholine cadmium chloride trihydrate. Science, 150, 1035. SWORAKOWSKI,J. (1973) Trapping states formed by structural defects in molecular crystals. Mol. Cryst. Liq. Cryst. 19, 259. TASAKI,I. (1968) Nerve Excitation, C. C. Thomas, Springfield, HI. TAUPIN,C. and McCONNELL, H. M. (1973) Membrane fusion. In Proceedings of the 8th FEBS Meeting, Amsterdam: Symposium on Biomembrancs. North Holland/American Elsevier, Vol, 28. TAYLOR,R. B., DUTTUS,P. H., RAFF,M. C. and DE PETRIS,S. (1971) Redistribution and pinocytosis of lymphocyte surface immunoglobulin molecules induced by anti-immunoglobulinantibody. Nature, New Biol. 233, 225. TEILHARDDE CHARDIN,P. (1959) The Phenomenon of Man, Harper Row, New York. THUDIC,UM,J. L. W. (1884) A Treatise on the Chemical Constitution of the Brain. Bailliere, TindaU & Cox, London. TINBERG,H. M., PACKER,L. and KEITH, B. D. (1972) Role of lipids in mitochondrial energy coupling: evidence from spin labelling and freeze-fracture electron microscopy. Biochim. Biophys. Acta, 283, 193. TINOCO, J., GHOSH,D. and KEITH,A. D. (1972) Interactions of spin-labelled lipid molecules with natural lipids in monolayers at the air-water interface. Biochim. Biophys. Acta, 274, 279. TONOMURA,Y. (1973) Muscle Proteins, Muscle Contraction and Cation Transport, University Park Press, Baltimore. TRAUBLE, H. (1971a) Phasenumwandlungen in lipiden mogliche sehaltprozesse in biologischen membranes. Naturwissenschaften, 58, 277. TRAUBLE,H. (197 lb)The movement of molecules across lipid membranes: a molecular theory. J. Membrane Biol. 4, 193. TRAOBLE,H. and EIaL, H. (1974) Electrostatic effects on lipid phase transitions: membrane structure and ionic environment. Proc. Natl Acad. Sci. U.S.A. 71,214. TRAUBLE,H. and HAVNES,D. H. (1971)The volume change in lipid bilayer lamellae at the crystalline-liquid crystalline phase transition. Chem. Phys. Lipid& 7, 324. TRAUBLE, H. and OVERATH,P. (1973) The structure of Escherichia Coli membranes studied by fluorescence measurements of lipid phase transitions. Biochim. Biophys. Acta, 307, 491. TRAUBLE,H. and SACKMANN,E. (1972) Studies of the crystalline-liquid crystalline phase transition of lipid model membranes. III. Structure of a steroid-lecithin system below and above the lipid phase transition. J. Am. Chem. Soc. 94, 4499. TRAUBLE,H. and SACKMANN,E. (1973) Lipid motion and rhodopsin rotation. Nature, 245, 210. TSAN,M.-F. and BERLIN,R. D. (1971) Effect of phagocytosis on membrane transport of nonelectrolytes. J. Exptl Med. 134, 1016. TSUKAGOSHI,N. and FOX,C. F. (1973) Transport system assembly and the mobility of membrane lipids in Escherichia coll. Biochem. 12, 2822. UBBELOHDE,A. R. (1965) Melting and Crystal Structure, Clarendon Press, Oxford. UKENA,T. E. and BERLIN,R. D. (1971) Effect of colchicine and vinblastine on the topographical separation of membrane functions. J. Exptl Med. 136, 1. UNANUE,E. R., PERKINS,W. D. and KARNOVSKY,M. J. (1972) Ligand-induced movement of lymphocyte membrane macromolecules. 1. Analysis by immunofluoreseence and ultra-structural radioautography. J. Exptl Med. 136, 885. VANDERKOOI,J. M. and MARTONOSI,A. (1971) Sarcoplasmic reticulum XVI. The permeability of phosphatidyl choline vesicles for calcium. Arch. Biochem. Biophys. 147, 632. VAN PUTTE,K. and VAN DEN ENDEN, J. (1971) On the nuclear magnetic resonance line shape of solid heptane. J. Phys. Chem. 75, 3901. VERVERGAERT,P. H. J. TH., VERKLELI~A. J., ELBERs, P. F. and VAN DEENEN, L. L. M. (1973a) Analysis of the crystallization process in lecithin liposomes: a freeze-etch study. Biochim. Biophys. Acta, 311, 320. VERVERGAERT,P. H~ J. TH., VERKLEIJ,A. J., VERHOEVEN,J. J. and ELBERS,P. F. (1973b) Spray-freezing of liposomes. Biochim. Biophys. Acta, 311,651. WAHL,P., KASAI,M. and CHANGEUX,J.-P. (1971) A study on the motion of proteins in excitable membrane fragments by nano-second fluorescence polarization spectroscopy. Eur. J. Biochem. 18, 332. WARREN,G. B., TOON,P. A., BIRDSALL,N. J. M., LEE, A. G. and METCALFE,J. C. (1974a) Reconstitution of a calcium pump using defined membrane components. Proc. Natl Acad. Sci. U.S.A. 71,622. WARREN,G. B., TOON,P. A., BIRDSALL,N. J. M., LEE,A, G. and METCALFE,J. C. (1974b) Complete control of the lipid composition of membrane bound proteins: application to a calcium transport system. FEBS Lett., 41, 122. WARREN,G. B., BIRDSALL,N. J. M., LEE,A. G. and METCALFE,J. C. (1974c) Lipid substitution: the investigation of functional complexes of single species of phospholipid and a purified calcium transport protein. In Membrane proteins in transport and phosphorylation (eds. G. S. AZZONE,M. E. KLINGENBERG,E. QUAGLIARIELLO and N. SILrPRANDI),Elsevier, p. 1. WATTS,P. H., PANGBORN,W. A. and HYBL,A. (1972) X-ray structure of racemic glycerol 1,2-(di-11-bromoundecanoate)-3-1-p-toluene sulphonate). Science, 175, 60. WHITE,D. A. (1973) The phospholipid composition of mammalian tissues. In Form and Function of Phospholipids (eds. G. B. ANSELL,J. N. HAWTHORNEand R. M. C. DAWSON),2nd ed., Elsevier, Amsterdam. WHITTAM,R. (1964) Transport and Diffusion in Red Blood Cells, Edward Arnold, London. WIDNELL,C. C. and UNKELESS,J. C. (1968) Partial purification of a lipoprotein with 5'-nucleotidase activity from membranes of rat liver cell. Proc. Natl Acad. Sci. U.S.A. 61, 1050. WILKINS,i . H. F., BLAOROCK,A. E. and ENGELMAN,n. M. (1971) Bilayer structure in membranes. Nature, New Biol. 230, 72. WILLIAMS,i . A., STANCLIFF,R. C., PACKER,L. and KEITH,A:D. (1972) Relation of unsaturated fatty acid composition of rat liver mitochondria to oscillation period, spin label motion, permeability and oxidative phosphorylation. Biochim. Biophys. Acta, 267, 444.

56

A.G. LEE

WE, S. H.-W. and McCONNELL, H. M. (1973) Lateral phase separations and perpendicular transport in membranes. Biochem. Biophys. Res. Commun. 55, 484. WUNDERLICH,B. (1973) Macromolecular Physics, Vol. 1, Academic Press, New York. YAHARA,I. and EDELMAN,G. M. (1972) Restriction of the mobility of lymphocyte immunoglobalin receptors by concanavalin A, Proc. Natl Acad. Sci. U.S.A. 69, 608. ZWOLINSKI,B. J., EYRING,H. and REESE, C. E. (1949) Diffusion and membrane permeability. J. Phys. Chem. 53, 1426.

Biological membranes.

Nanodomains in biological membranes.

Chloride channels of biological membranes.

Effects of sterilization methods on the physical, chemical, and biological properties of silk fibroin membranes.

Lipid polymorphism and the functional roles of lipids in biological membranes.

Biological Fuel Cells and Membranes.

Brownian motion in biological membranes.

Enzymatic oxidation of cholesterol: properties and functional effects of cholestenone in cell membranes.

Molecular interactions of anaesthetics with biological membranes.

Freeze-fracture morphology of biological membranes.

A microscopic view of phospholipid insertion into biological membranes.

Biological properties of a CD4 immunoadhesin.

Biological membranes. The semiotics of charge.

A computer simulation approach to quantify the true area and true area compressibility modulus of biological membranes.

Molecular Mechanics Approach for Linear Response Properties in Heterogeneous Environments.

LipidBuilder: A Framework To Build Realistic Models for Biological Membranes.

Monitoring peripheral protein oligomerization on biological membranes.

Lipid disequilibrium in biological membranes, a possible pathway to neurodegeneration.

Evidence for a large internal pressure in biological membranes.

Aluminium in biological environments: a computational approach.

Biological properties of interleukin 10.

Dynamic properties of polyelectrolyte calcium membranes.

Premature rupture of the membranes: a conservative approach.

Physical properties of Escherichia coli spheroplast membranes.