The evolution of membranes MYERBLOOM Department of Physics, University of British Columbia, 6224 Agriculture Road, Vc~rzcouver,B.C., Canada V6T 2A6 AND

C

OLE G. MOURITSEN

Can. J. Chem. Downloaded from www.nrcresearchpress.com by MICHIGAN STATE UNIV on 01/25/17 For personal use only.

Department of Structural Properties of Materials, The Technical UniversiQ of Derlmark, Buildir~g307, DK-2800 Lyngby, Derzmark Received September 23, 1987 This paper is dedicated to Professor J . A . Morrison

MYERBLOOM and OLEG. MOURITSEN. Can. J. Chem. 66, 706 (1988). Consideration of the influence of cholesterol on the physical properties of biological membranes leads to the conclusion that cholesterol increases the thickness of fluid membrane bilayers without appreciably increasing the microviscosity component of membrane fluidity. At sufficiently high cholesterol concentrations, the gel-liquid crystalline phase transition is completely eliminated in phospholipid-cholesterol mixtures and the system has the properties of a two-dimensional liquid over a wide range of temperatures. It is proposed that with the evolution of cholesterol and related sterols in an oxygen-rich atmosphere, the resulting modification of physical constraints on membrane properties made it possible for new evolutionary driving forces to manifest themselves leading to the peculiar properties of plasma membranes of eucaryotic cells. MYERBLOOM et OLEG. MOURITSEN. Can. J. Chem. 66, 706 (1988). Si on examine l'influence du cholestCro1 sur les propriCtCs physiques des membranes biologiques, on est amen6 B conclure que le cholesterol augmente 1'Cpaisseur des doubles couches des membranes fluides sans alfgmenter d'une facon appriciable la composante de la microviscosit6 de la fluidit6 de la membrane. A des concentrations de cholestCro1 suffisamment ClevCes, la transition de la phase cristalline gel-liquide est complktement CliminCe des mClanges phospholipide-cholestCro1 et, pour un grand intervalle de tempiratures, le systkme posskde les propriCtCs d'une liquide bi-dimensionnel. On est amen6 B proposer que, avec 1'Cvolution du cholestCro1 et des stCrols apparentCs dans un atmosphkre riche en oxygkne, la modification qui rCsulte de contraintes physiques sur les propriCtCs de la membrane fait qu'il est possible pour de nouvelles forces Cvolutionnaires de se manifester; ceci provoque l'apparition de propriitis particulikres pour les membranes plasmatiques des cellules eucaryotices. [Traduit par la revue]

Introduction The nature of the three-dimensional structure of integral membrane proteins (1-3), the insights on the molecular evolution of cholesterol provided by Bloch (4-6), and the interpretation of deuterium nuclear magnetic resonance ('H nmr) studies of the interaction between phospholipid molecules and integral membrane proteins (7- 1 1) lead us to propose that the presence of cholesterol, or related sterols where appropriate,' modified the physical properties of membranes in a manner which was crucial to the evolution of eucaryotic cells into their present form. At least two stages of membrane evolution have taken place. The jrst stage accounts for most natural membranes which do not contain cholesterol such as those of procaryotic cells and most organelles of eucaryotic cells. The second stage involves the evolution of eucaryotic cells or, at least, the plasma membranes of eucaryotic cells, upon the accumulation of oxygen in the atmosphere (12). The lipid composition of the plasma membranes of eucaryotic cells invariably includes a substantial fraction of cholesterol. Furthermore, Bloch (4, 5) has demonstrated that cholesterol evolved upon the appearance of oxygen in the atmosphere. He has speculated that the biological advantage associated with cholesterol may be due to the "reduced fluidity" or "increased microviscosity" which, it is claimed (13) the addition of cholesterol imparts to the liquid

h he word "cholesterol" as used in this paper should sometimes be interpreted as "cholesterol or related sterols". The "related sterols" may be ergosterol in plants or precursors to cholesterol and ergosterol such as lanasterol or cycloartenol (5).

crystallirie state of phospholipid bilayer membranes.* The discovery of hopanoid, triterpene derivatives in some procaryotic cells and in the form of "molecular fossils" of ancient times has led to the suggestion (14, 15) that these relatively rigid, anaerobically evolved, amphiphilic molecules play a similar "membrane reinforcement" role in some procaryotes to that played by aerobically evolved sterols such as cholesterol in eucarvotes. In this paper, we propose that the biosynthesis of cholesterol in an aerobic atmosphere removed a bottleneck in the evolution of eucaryotic cells. Our proposal, which concerns a crucial evolutionary role played by the physical properties of membranes, can be related to the broad, phenomenological characterization of the evolution of eucaryotic cells given by CavalierSmith (16, 17). He identifies "twenty-two characters universally present in eucaryotes and universally absent from procaryotes" and presents detailed arguments that, of these, the advent of exocytosis (and endocytosis) is the most likely to have provided the driving force for the evolution of eucaryotic cells into their modem form. As described below, we have been led via an unconventional interpretation of measurements of physical properties of membranes to identify a possible physical bottleneck to cytosis which was removed by the appearance of cholesterol. We hypothesize that, in addition to influencing the cohesive strength of membranes, the main role of cholesterol in this evolutionary step was to relax an important constraint on ' ~ l t h o u ~we h are concerned here with the influence of cholesterol on the physical properties of membranes, it s h ~ u l dbe' noted that cholesterol may play a role in metabolic functions as well (5).

BLOOM AIU'D

Can. J. Chem. Downloaded from www.nrcresearchpress.com by MICHIGAN STATE UNIV on 01/25/17 For personal use only.

membrane thickness imposed by the biological necessity of membrane fluidity. In attempting to establish the validity of our hypothesis, we have been confronted with a remarkable lack of relevant information on basic physical properties of membranes. The filling of this information vacuum will lead to a better understanding of membrane properties in addition to checking our hypothesis.

Relationship between membrane evolution and cellular evolution It is generally agreed that procaryotic cells came into existence almost 4 x lo9 years = 4 GA ago (1 GA = lo9 years). Procaryotic cells have only a single membrane and no nucleus. By contrast, most eucaryotic cells which exist today contain many organelles, each encapsulated within its own membrane. Our knowledge of the evolutionary relationship between procaryotic and eucaryotic cells is very fuzzy but it has advanced considerably during the past decade (18). Ten years ago, it was believed (19) that a procaryotic host cell spawned the first eucaryotic cell about 1.5 GA ago, approximately the time at which the earth's hydrosphere is thought to have become oxidizing (20). Cogent arguments have been made that the procaryotic to eucaryotic step occurred via a sub-line of wall-less anerobic procaryotes that gained the capacity for endocytosis (16, 17, 19). It is now known that there exist two distinct kingdoms of procaryotes called "eubacteria" and "archaebacteria" (18). These bacteria have quite different cytological and physiological properties from each other, operate optimally under different environmental conditions and have only distant and uncertain evolutionary linkages (18). If they do have a common ancestor, they parted in the early stages of cellular evolution. Furthermore, from an evolutionary standpoint, eucaryotic cells are as distant from eubacteria and archaebacteria as each of them is to the other. The impetus to the more quantitative and searching examination of evolutionary connectivity at the cellular level that has taken place in recent years has been supplied by the development of new methods of quantitative analysis of sequences of nucleic acids. In particular, the nucleic acid sequencing of ribosomal RNA (rRNA) has proven to be an effective molecular chronometer over the time scales, of order GA, of importance in bacterial and cellular evolution. In a recent comprehensive review of the subject, Woese (18) predicts that nucleic acid sequencing will stimulate greatly expanded knowledge of cellular evolution in the near future. A real understanding of the evolution of any biological system requires not only knowledge of changes in genetic characteristics but also insight into the interaction between the biological system and its surroundings. Whether or not a random change in the genes gives rise to a viable biological modification or not is determined by the environmental interactions. Contact between a cell and its surrounding environment is mediated by the plasma membrane, i.e. the outer membrane of the cell. Systematic differences exist between the composition of the plasma membranes of eubacteria, archaebacteria, and eucaryote cells (18, 21). We believe that these differences have an important evolutionary significance. We are now able to attempt an explanation of the relationship between membrane evolution and cellular evolution for eubacteria and eucaryote cells because of improvements in our understanding of the physical properties of biological membranes which have taken place during the past decade. We shall not discuss archaebacter-

ia, partly because the characterization of the physical properties of their membranes has not yet been adequately developed (21). More important, the phospholipid composition of the plasma membranes of eucaryotic cells resembles that of eubacteria but not that of archaebacteria. Thus, if our proposals concerning the importance in the evolution of eucaryotic cells of the modification of the physical properties of their plasma membranes by sterols are confirmed, this would imply a closer evolutionary relationship between eucaryotic cells and eubacteria than archaebacteria.

Relationship between biological activity and membrane fluidity, microviscosity, segmental orientational order, rigidity, and thickness It is now widely recognized and accepted that biological activity requires that membranes be in their fluid (liquid crystalline) state. When characterizing membrane fluidity, various authors sometimes use terms such as rigidity and microviscosity without adequately defining them, and often they are used in different senses. The confusion in describing the mechanical properties of membranes is caused by the conspicuous anisotropic nature of bimolecular fluid membrane sheet. To avoid ambiguities we shall here clarify the use of these terms and then review the influence of cholesterol on membrane fluidity. Pure phospholipid bilayer membranes undergo a phase transition upon heating from a solid "gel" phase to a fluid "liquid crystalline" phase. The phase transition is accompanied by two major dramatic changes (7, 22): (1) increased conformational freedom and flexibility of the acyl chains leading to decreased segmental orientational order and decreased bilayer thickness, and (2) increased lateral diffusivity of the lipid molecules parallel to the plane of the membrane and onset of a more rapid rotational diffusion about the long molecular axis corresponding to decreased microviscosity. The higher fluidity of the high temperature phase is therefore caused by changes in two different components, namely orientational order and microviscosity . In the case of pure phospholipid bilayer model membranes, there is a strong correlation between orientational order and microviscosity but this is not always the case. A prominent example central to our theme is that the introduction of cholesterol into phospholipid bilayer membranes increases the orientational order3 but does not increase the microviscosity appreciably (24-26). The requirement of fluidity for biological activity should therefore be interpreted in terms of the microviscosity aspect of fluidity. This interpretation of the influence of cholesterol is different from that given by Dahl et al. (13). They found a correlation between the biological viability of various sterols in the cholesterol evolutionary chain and increased "microviscosity". However, this correlation was made on the basis of steady state polarized fluorescence measurements. It is now known that such measurements on an anisotropic system do not actually measure microviscosity but rather a combination of microviscosity and the orientational order of the fluorescent probe (27, 28). A fluorescence depolarization study of membranes containing cholesterol has been carried out recently and the results analysed 3 ~ hassociated e decrease in the average membrane surface area per polar head is often called "condensation" in the literature (see, e.g. ref. 23).

Can. J. Chem. Downloaded from www.nrcresearchpress.com by MICHIGAN STATE UNIV on 01/25/17 For personal use only.

708

CAN. 1. CHEM.

with the anisotropic orienting potential included (29). In the light of these results, we can now re-interpret the earlier fluorescence measurements as indicating relatively little influence of cholesterol on the microviscosity component of the fluidity while giving a substantial increase in thickness as indicated by the orientational order of the lipid chains. In some cases the words "increased rigidity" have been used to characterize increased "orientational order" or "microviscosity". It is clear that both cannot always be correct. We would prefer to reserve the word "rigidity" in membranes as a description of the resistance of the bilayer to shear. Since the hydrophobic interior of a fluid membrane is well approximated as an incompressible fluid, small values of "area compressibility" imply small fractional changes in membrane thickness under a given change in stress. Evans and Needham (30) have shown that the addition of cholesterol (CHOL) to a sample of dimyristoylphosphatidylcholine (DMPC) vesicles resulted in a greatly reduced membrane area compressibility. At the same time, the phase transition was eliminated for molar ratios CHOL:(CHOL + DMPC) > 0.125, while the vesicles behaved like liquids with zero surface shear restoring forces even at temperatures well below the phase transition for pure DMPC. The combination of decreased area compressibility and zero surface shear restoring forces is perfectly compatible with the two dimensional character of the membrane fluidity. A close relationship exists between the average orientational order parameter of the acyl chains and the membrane thickness (7, 10, 31) which makes it possible to estimate the membrane thickness using 'H nmr measurements. Such estimates should be reasonably good whenever the symmetry axis for orientationa1 averaging is normal to the bilayer surface, i.e. they would not be reliable for the gel phase of pure phospholipid bilayers where the chains are tilted with respect to the bilayer normal, but the estimates would be good for the liquid-crystalline phase or for the gel phase close to the melting point for membranes with impurities such as cholesterol present which remove the chain tilt. The indirect determination of membrane thickness using 'H nmr is potentially very useful since there seems to be no reliable direct means of measuring membrane thickness for either synthetic membranes with impurities or biological membranes. Direct measurements using diffraction methods require that the membrane samples be prepared in the form of good quality smectic liquid crystals having long range order so that several orders of Bragg reflection may be obtained. This has been achieved for pure phospholipid bilayer systems (32, 33) but seems to be difficult for impure systems judging from the absence of good experimental data. Even in the pure systems, some differences of opinion exist on the interpretation of the diffraction data. A lattice model has been used (3 1, 34) to derive a relationship between the separation, d , of the first carbon atoms in the acyl chains of the two leaflets, each acyl chain having n C-$ bonds, in terms of the maximum separation, dm,, = 2.5n A, and the magnitude of the average acyl chain orientational order parameter, , The quantity I I is easily obtained from the average 'H splitting of lipids with perdeuterated acyl chains, i.e. from the first moment of the distribution of quadmpolar splittings (35, 36). Although the lattice model leading to eq. [ I ] is a simple one,

its predictions are in agreement with neutron diffraction studies of 'H-labelled phospholipid acyl chains in model membranes (37). Furthermore, the value of I I = 0.164 at 50°C for dipalmitoylphosphatidylcholine (DPPC) combined with the value of d = 26 A at 44°C from the X-ray diffraction experiments ofJewis and Engelman (33) would require a value of dm,, = 39 A corresponding to n = 3912.5 = 15.6, a very reasonable value for an acyl chain of 16 carbons. Reasonable results are also obtained using the measurements of Cornell and Separovic (32) for DPPC. The most complete measurements of the influence of cholesterol on 'H nmr splittings in a model membrane have been made by Vist (38). His measurements confirm those reported previously by Stockton and Smith (23) and Oldfield (8). For example, the average 'H nmr splitting in DPPC increases by about 60% at 50°C when 25% cholesterol is added (38) so that I I changes from about 0.16 to about 0.26. +cording to eq. [I], this corresponds to a change in d from 26 A to about 30 A. Even more striking is the behaviour of erythrocyte membranes where the influence of the large cholesterol concentration was found to give rise to a fluid membrane down to -5°C (39). The average quadmpolar splitting of DPPC in the erythrocyte membranes increased monotonically between 50°Coand -5°C to an effective hydrophobic thickness of d = 35 A which is approximately 90% of the maximum value attainable with fully extended acyl chains. Yet, the lipid component of erythrocyte membranes is fluid as indicated by their 'H nmr spectra (39) and measurements of the lateral mobility of lipid probes (25). In summary, then, we list the effects of physiological concentrations of cholesterol on phospholipid bilayer model membranes. 1. The gel-liquid phase transition is eliminated. 2. No static surface rigidity is observed over a wide temperature range indicating fluid-like macroscopic properties. 3. Translational and rotational diffusion constants are not strongly affected in the high temperature region indicating a fluid-like microviscosity. 4. The area compressibility is greatly reduced leading to greater cohesion and reduced permeability. 5. The orientational order increases indicating greater membrane thickness and "condensation" of the membrane lipids.

Membranes in procaryotic cells - the first stage of membrane evolution Most biological activity takes place at moderate temperatures and, in particular, not at temperatures well above 37°C. A consequence of this is that biological membranes must operate at moderate temperatures. When combined with the requirement of fluidity, this implies that the thickness of bilayers composed of lipids extracted from membranes which do not contain cholesterol must fall within a narrow range of values. For example, lipid molecules with longer saturated acyl chains lead to thicker bilayers (33) but also higher gel-liquid crystalline phase transition temperatures. Thus, membranes composed mainly of longer lipids would be in the gel phase at moderate temperatures. Shorter lipids would not form stable bilayers. Longer lipids having unsaturated bonds give rise to thinner bilayers than saturated chains of the same length (33) so that the thickness of fluid bilayers of such lipids is not so different at moderate temperatures from that due to shorter saturated chains. The structure and amino acid sequence of integral membrane proteins indicate that they are predominantly made up of a number of a-helical segments which span the lipid bilayer. On

Can. J. Chem. Downloaded from www.nrcresearchpress.com by MICHIGAN STATE UNIV on 01/25/17 For personal use only.

BLOOM AND MOURITSEN

the basis of known primary sequences and diffraction studies, it has been demonstrated that the positions of the hydrophilic and hydrophobic parts of integral membrane proteins tend to be correlated with their counterparts in the lipid molecules comprising the membrane bilayer (1-3). In addition, 2H nrnr studies on a large number of bacterial and reconstituted membranes (see, e.g. Table 6 of ref. 9) all gave the unanticipated result that of the acyl chains in the fluid phase was unaffected by the presence of physiological concentrations of proteins. These results lead to the following hypothesis which, we suggest, governs some important aspects of the evolution of membranes in procaryotic cells. Hypothesis I . Integral membrane proteins operate in an optimum manner in membranes whose natural hydrophobic thickness matches the hydrophobic thickness of the protein. We have pointed out in a theoretical analysis of the significance of the matching of the hydrophobic regions of lipids and proteins (10) that the 2H nrnr experiments were all carried out on lipids and proteins whose hydrophobic thickness were approximately matched. Thus the addition of such proteins would not modify bilayer thickness, and hence ,providing that Hypothesis 1 was correct. Strong support of Hypothesis 1 also comes from studies of the activity of a variety of membrane enzymes (40, 41) and ion-channels (42) as a function of membrane thickness. By manipulating the bilayer hydropobic thickness either by changing the lipid chain length or by incorporating alkanes and anaesthetics (43,44) it is invariably found that the activity is optimized at a certain thickness. Perhaps the most striking confirmation of Hypothesis 1 comes from a systematic study of the influence of an integral membrane protein on the phase transition temperatures of phospholipid bilayer model membranes having different acyl chain lengths (45). The evolutionary driving force which leads to the validity of this hypothesis presumably comes from the superiority of a strain-free membrane, under normal operating conditions, for survival under changes in environmental conditions, e. g. changes in stress, and for the optimum function of protein dynamics in analogy with cytoplasmic proteins (46). Although we have emphasized that life at moderate temperatures would promote membranes having a relatively narrow range of thickness values, it is well known that some species of procaryotic cells are viable at abnormal temperatures and pressures (see, e.g. ref. 47). Hypothesis 1 also applies to such systems, of course. Lipids capable of being fluid under a certain range of external constraints can provide the basis of cellular membranes providing that appropriate integral membrane proteins can be evolved which match their thickness. Whatever the cellular system, the capacity of most of the integral membrane proteins to function properly will be severely limited if the stresses on the membrane are extended beyond the range within which the membrane retains its fluid property. We believe that a sensitive parameter in relation to membrane fluidity is membrane thickness. Pure lipids undergo large discontinuous changes in membrane thickness as a result of the gel-liquid crystalline phase transition. Thus, some physical processes which require large excursions in membrane thickness may tend to stimulate a phase transition. This would be true of mixtures of phospholipids with short and long chains because of phase separation effects. It is striking that, although procaryotic cells were present in an anaerobic atmosphere for more than lo9 years they did not evolve any cells with organelles and/or compartmentalized

709

nuclear material. Anaerobic molecular evolution did take place in procaryotic cells. We suggest that although the recently discovered triterpene derivatives evolved and are capable of replacing some aspects of cholesterol's influence on membrane properties (14, 15), they lacked the ability possessed by cholesterol to facilitate a crucial stage in the evolution of the type of internal structures associated with eucaryotic cells.

Plasma membranes in eucaryotic cells - the second stage of membrane evolution In his review of the distinction between the properties of procaryotic and eucaryotic cells, Cavalier-Smith (17) identifies twenty-two universal eucaryotic characteristics which are universally absent from procaryotic cells. It is likely that a small number of these characteristics, possibly a single one, represent the driving force for the evolutionary development of eucaryotic cells as we see them today. We propose here that, with the appearance of large amounts of molecular oxygen in the earth's atmosphere between 2.3 and 1.5 GA ago (20), a bottleneck in the evolution of eucaryotic cells was removed by the resultant incorporation of sterols in the plasma membrane. In an earlier paper, Cavalier-Smith (16) argued that the advent of exocytosis (and endocytosis) which makes it possible for a cell to export (or import) material at a much faster rate, could provide the "evolutionary driving force" to explain the development of the other universal eucaryotic characteristics. He suggested that cytosis would have been inhibited by the rigid peptidoglycan cell wall present in many procaryotic cells and universally absent in eucaryotic cells (17). This would require the replacement of cell walls by a still strong, but more flexible plasma membrane. We propose here that the development of a plasma membrane having suitable physical properties only became possible with the appearance of molecular oxygen, which gave rise to the evolution of sterols leading to cholesterol and related sterols (4, 5). We propose in a second hypothesis below that the crucial physical property imparted to membranes by cholesterol was its ability, described in an earlier part of this paper, to increase the thickness of fluid membrane bilayers by a large amount without appreciably increasing the microviscosity component of membrane fluidity. Hypothesis 2. The evolution of eucaryotic cells required the evolution of plasma membranes containing cholesterol or related sterols and thus capable of maintaining their fluidity and integrity as membranes over a wide range of external constraints. Implicit in this hypothesis is the idea that the process of cytosis requires membrane stretching and/or changes of local membrane curvature which would disrupt the integrity of the membrane in the absence of fluid-like microviscosity. Such fluid-like properties would allow large membrane cirvature without abnormal increases in permeability. Presumably, this desirable property would be facilitated by rapid redistribution of molecular species in response to local stresses as discussed recently by Sackmann et al. (48). Eucaryotic cells are on the average one order of magnitude larger in linear extension than procaryotic cells. The mechanical strength required for supporting larger cells could well be provided by cholesterol-containing plasma membranes. In fact, it has recently been demonstrated in model membranes that for a given fluidity, vesicle fusion is facilitated and larger vesicles are stabilized for increased acyl chain orientational order induced by cholesterol (49).

710

CAN. J. CHEM. VOL. 66, 1988

Can. J. Chem. Downloaded from www.nrcresearchpress.com by MICHIGAN STATE UNIV on 01/25/17 For personal use only.

Experimental considerations Further support or rejection of the conjectures concerning membrane evolution presented in this paper will require more information on the thermodynamic, mechanical, and dynamic properties of mixtures of lipids, sterols, and proteins as well as natural membranes. Here are a few useful questions and studies which could be approached in the laboratory. It would be of interest to complement the study of fluorescence depolarization measurements in membranes containing a series of precursors of cholesterol (1 3) with other measurements of the type discussed earlier in this paper such as 2H nmr calorimetry, and micromechanics, where possible. If the evolution of sterols in membranes to the biologically optimized forms of cholesterol and ergosterol took place for the type of physical reason we have postulated, it may be possible to identify the crucial physical parameters from a systematic study of model systems. In this respect, some triterpene derivatives such as the hopanoid molecules should be included in the study since these molecules would be expected to be further removed from cholesterol in their crucial physical properties. Some scattered measurements on such properties have been reported, but we make no attempt to review them here since none of them represent complete studies and they have not been focussed in relation to our hypotheses. It would be highly desirable to carry out a comparative study of the phase diagrams of a mixture of phospholipids with a selected hopanoid and with a selected primitive sterol such as lanosterol. This might explain the observation (50) that, with the exception of tetrahymanol in a ciliate, hopanoids, or similar cyclization products of squalene do not occur as components in membranes of eucaryotic cells. Experimental studies of the phase diagrams of lipidcholesterol mixtures have been carried out using a variety of techniques, including 2H nmr (38, 51), electron spin resonance (52), electron microscopy (53), and micromechanical measurements on individual vesicles (30). In the case of DPPCcholesterol mixtures, a new fluid phase, sometimes called the "B phase" (51), has been shown to occur at concentrations of cholesterol above about 22%. Between about 8% and 20% cholesterol, the system consists of a mixture of the B phase and either the gel phase for T < 41°C or the liquid crystalline phase for T > 41°C. A theoretical analysis of the cholesterol-DPPC system (54) has led to a phase diagram capable of fitting the available experimental observations on the DPPC-cholesterol system (38,5 1-53) very well (see Fig. 1). Ipsen et al. (54) have characterized the gel phase of the pure lipid as a "solid-ordered phase" (so) and its liquid crystalline phase a "liquid-disordered (Id) phase". The B phase is described as a "liquid ordered (lo) phase" in which the acyl chains are ordered due to the lipid-cholesterol interaction while two dimensional fluidity is retained. This ordering gives a fluid phase of higher cohesive strength (30), presumably because of the increased effectiveness of the van der Waals interactions between neighbouring lipid molecules. The B phase seems, therefore, to have desirable properties for the plasma membranes of eucaryotic cells. The B-phase membranes are fluid over a wide temperature range, corresponding to a large change in membrane thickness, and have great cohesive strength. They have no phase transition corresponding to the gel-liquid crystalline (so-ld) transition which occurs in the absence of sterols. It is possible that the evolution of sterols in membranes was driven by the advantage of the B phase appearing at the lowest concentration of sterol. This conjecture should be easy to test

0

10

20

30

MOLE % CHOLESTEROL FIG. 1. Phase diagram for DPPC-cholesterol mixtures proposed by Vist (38) and Davis (51) and interpreted theoretically by Ipsen et nl. (54). This diagram is based on a large number of experiments including several cited in this paper (30. 38, 51-53). The effect of the lipid-cholesterol interaction (54) is to stabilize, at high cholesterol concentrations, a "liquid-ordered" (lo) phase. At low cholesterol concentrations, the usually-discussed "solid-ordered" (so) (i.e. gel) and "liquid-disordered" (Id) (i.e, liquid crystalline) phases are obtained.

experimentally by carrying out careful measurements of the phase diagrams of mixtures of lipids and various sterols found on the evolutionary pathway to cholesterol and ergosterol (4,5). For example, experiments are underway in the laboratory of one of us (M.B.) to compare the lowest concentration at which the B phase occurs with that of cholesterol in sterol-DPPC mixtures (Fig. 1). An interesting system to study in this way would be a-cholesterol-phospholipid mixtures. Dufourc et al. (55) have shown that the a-cholesterol molecule adopts a tilted conformation in phospholipid bilayer model membranes while the P-cholesterol ring structure lies with its plane parallel to the bilayer normal. They have speculated that this is why the P-cholesterol molecule occurs in natural systems while the a-cholesterol molecule does not. We ask here whether the tilt of a-cholesterol is less favourable to the stability of B phase and hence less favourable to the establishment of thermomechanical properties appropriate for plasma membranes of eucaryotic cells. As described in various references cited in this paper, transient fluorescence depolarization and nmr relaxation measurements offer the possibility of determining the translational and rotational diffusion rates of lipid molecules and/or molecular probes in these same model systems. The relationship between such molecular dynamical properties and thermodynamic and mechanical properties is of great interest. The methods described above can be applied to study some biological membranes though in some cases it will require the use of molecular probes in membrane extracts. It would be interesting to know whether there are systematic differences between the hydrophobic thickness of plasma membranes of eucaryotic cells and each of the membranes of organelles of the same cells and those of procaryotic cells. In the case of plasma membranes of eucaryotic cells, it would be highly desirable to also study the thickness of membranes formed from their separated lipids. Since plasma membranes have a large amount of polyunsaturated lipids (17), it may well be that the role of

Can. J. Chem. Downloaded from www.nrcresearchpress.com by MICHIGAN STATE UNIV on 01/25/17 For personal use only.

BLOOM AND I

cholesterol is not to increase the membrane thickness. but rather to restore the thickness to values "normal" for saturated chains while imparting greater cohesive strength without changing the fluidity. Measurements of acyl chain orientational order carried out with perdeuterated fatty acid probes in erythrocytes (56) were in good agreement with those carried out later using perdeuterated phospholipid molecules. Such probes could be used in biological membranes in the experimental program outlined above. It would also be of value to study cells which are abnormal in their cholesterol content. The cells of cancer tumours are an example (57). Much insight into the structure of integral membrane proteins and how their geometry relates to membrane geometry is now obtained from the amino acid sequence, when available (1-3). It would be interesting to compare the amino acid sequences of membrane proteins of plasma membranes with those of the organelles of eucaryotic cells to check them as indicators of membrane hydrophobic thickness and/or variability of thickness.

Concluding remarks As discussed near the beginning of this paper, the three known kingdoms of cells parted company early in the evolutionary history, as perceived via the rRNA molecular chronometer (18). The approximate dates at which eubacteria, archaebacteria, and eucaryotes diverged from each other are not known, however. Up to now, changes in nucleic acid sequences of rRNA have produced reliable information on the time-ordering of evolutionary events, but not on their tempo. We can say, however, that procaryotic cells (eubacteria and archaebacteria) have existed for more than 3.5 GA, including the first GA in an anerobic atmosphere, without evolving any variants with internal compartments in the manner of what are now called eucaryotic cells. Furthermore, the plasma membranes in all eucaryotic cells characterized up to now contain sterols while essentially no procaryotic membranes contain sterols. Though the progenitor of eucaryotic cells may have evolved before the appearance of molecular oxygen in the atmosphere some 2.3 GA ago and the oxidation of the hydrosphere about 1.5 ago (20), the fact that the biosynthesis of sterols requires O2 (4-6) implies unequivocally that the plasma membranes of eucaryotic cells evolved between 2.3 and 1.5 GA ago. We have been led, following a detailed reconsideration of the striking effects of cholesterol on the physical properties of phospholipid bilayer model membranes, to postulate that these correlations and anti-correlations are no accident. We argue in this paper that prior to the appearance of cholesterol and related sterols, the available molecular building blocks for biological membranes were incapable of generating physical properties of a plasma membrane compatible with the evolution of modem eucaryotic cells. The available data on physical properties of membranes are insufficient to test our argument adequately. It is not obvious at this time that molecules such as the cholesterol-resembling hopanoids, which evolved anaerobically in procaryotes and which are capable of replacing cholesterol in living systems under certain experimental conditions, were incapable of providing the physical properties necessary for facilitating special functions such as cytosis which could have provided the evolutionary driving force for eucaryotic cells. The subtle interplay between thermodynamic and micromechanical prop-

erties and rates of molecular motion which may be required to promote a process such as cytosis, while still maintaining membrane strength and permeability control, has not yet been examined carefully and exhaustively. When this is done, and it will require a considerable effort, it should be possible to demonstrate whether our hypothesis concerning eucaryotic evolution is correct or not. Whatever the outcome, the required experimental studies will generate considerable understanding concerning the systematics of the physical properties of membranes in relation to their biological function. A few final words should be added about the relationship of our proposals to the well known and widely accepted symbiosis theory for the evolution of eucaryotic cells. We do not believe that our proposals are in disagreement with the symbiosis theory. On the contrary, it is unreasonable to expect cells to have evolved to the stage of being capable of swallowing an entire bacterium and allowing it to survive in its interior without having previously gone through a pre-bacterium-swallowing evolutionary phase, i.e. cells would first have had to develop and practice the swallowing facility on smallerobjects. We have only dealt with this earlier stage in our paper. Whether subsequent internal incorporation of entire, smaller cells would necessarily follow such a development, and/or did actually follow it, must be settled by arguments and criteria other than those discussed here.

Acknowledgements We thank Guy Cox, Ford Doolittle and Evan Evans for helpful discussions. We wish to dedicate this paper to Professor J.A. Morrison whose rigorous use of thermodynamics to characterize the physical and chemical properties of a variety of molecular systems is a model for the study of complex systems such as the biological membranes considered in this paper. This research was supported by NATO Research Grant No. 392183. The research of M. Bloom was also supported by the Natural Sciences and Engineering Research Council of Canada and that of O.G. Mouritsen by the Danish Natural Science Research Council under Grant J. nr. 5.21.99.72. 1. R. HENDERSON. 111 Membranes and intercellular communications. Edited by R. Balian, M. Chabre, and P. F. Devaux. Elsevier/North-Holland, Amsterdam. 1981. pp. 229-249. 2. P. N. T. UNWINand R. HENDERSON. Sci. Am. 250, No. 2, 78 (1984). 0 . EPP,K . MIKI,R. HUBER,and H. MICHEL. 3. J. DEISENHOFER, Nature, 318, 618 (1985). 4. K. BLOCH.Scielice, 150, 19 (1965). 5. K. BLOCH.C.R.C. Crit. Rev. Biochem. 14, 47 (1983). 6. K. BLOCH.I n Biochemistry of lipids and membranes. Edited by Dennis E. Vance and Jane E. Vance. Benjamin/Cummings Pub. Co. Inc. 1985. pp. 1-24. and A. SEELIG. Q. Rev. Biophys. 13, 19 (1980). 7. J. SEELIG I n Membranes and transport. Vol. 1. Edited b y E. 8. E. OLDFIELD. N. Martinosi. Plenum Press, New York 1982. pp. 115-123. 9. P. F. DEVAUX. I n Biological magnetic resonance. Vol. 5. Edited by L. J. Berliner and J. Reuben. Plenum Press, New York. 1983. pp. 183-299. 10. 0 . G. MOURITSEN and M. BLOOM. Biophys. J. 46, 141 (1984). 11. M. BLOOMand I. C. P. SMITH.I n Progress in protein-lipid interactions. Edited b y J . J . H. H. M. De Pont and A. Watts. Elsevier, Amsterdam. 1985. pp. 6 1-88. J. LEWIS,M. RAFF,K . ROBERTS and J . D. 12. A. ALBERTS, D. BRAY, WATSON. Molecular biology of the cell. Garland Publ. Co., New York and London. 1983. Chap. I . 13. J. DAHL,C. DAHL,and K . BLOCH.Biochemistry, 19, 1462 (1980).

712

CAN. J. CHEM. VOL. 66. 1988

14. M. ROHMER,P. BOUVIER,and G. OURISSON. Proc. Natl. Acad. Sci. U.S.A. 76, 847 (1979). ~ ,ALBRECHT,and M. ROHMER.Sci. Am. 250, 15. G. O u ~ n s o P. No. 8, 44 (1984). 16. T. CAVALIER-SMITH. Nature, 256, 463 (1975). In Molecular and cellular aspects of 17. T. CAVALIER-SMITH.

Can. J. Chem. Downloaded from www.nrcresearchpress.com by MICHIGAN STATE UNIV on 01/25/17 For personal use only.

18. 19. 20.

21. 22.

23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36.

macrobial evolution. Society for General Microbioligy Ltd., Symposium 32. Edited by M. J. Carlile, J. F. Collins, and B. E. B. Moseley. Cambridge University Press. 1981. pp. 33-84. C. R. WOESE.Microbiol. Rev. 51, 221 (1987). L. MARGULIS. Origin of eucaryotic cells. Yale University Press, New Haven, CT. 1975. J. C. G. WALKER,K. KLEIN,M. SCHIDLEWSKI, J. W. SCHOPF, D. J. STEVENSON, and M. R. WALKER.In Earth's earliest biosphere. Edited by J. W. Schopf. Princeton University Press, Princeton, NJ. 1983. M. DE ROSA,A. GAMBACORTA, and A. GLIOZZI.Microbiol. Rev. 50, 70 (1986). J. SEELIG.In Membranes and intercellular communications. Edited by R. Balian, M. Chabre, and P. F. Devaux. ElsevierINorth-Holland, Amsterdam. 1981. pp. 15-78. G. W. STOCKTON and I. C. P. SMITH.Chem. Phys. Lipids, 17,

37. G. ZACCAI,G. BULDT,A. SEELIG,and J. SEELIG.J. Mol. Biol. 134, 693 (1979). 38. M. R. VIST. M.Sc. Thesis. University of Guelph, Guelph, Ont., Canada, 1984. J. H. DAVIS.M. BLOOM,J. WESTERMAN, and 39. B. MARAVIGLIA, K. W. A. WIRTZ.Biochem. Biophys. Acta, 686, 137 (1982). C. A. KEIGHTLEY,G. A. SMITH,C. D. 40. (a) A. JOHANNSSON,

RICHARDS, T. R. HESKETH, and J. C. METCALFE. J. Biol. Chem. 256, 1643 (1981); (b) A. JOHANNSSON, G. A. SMITH,and J . C. METCALFE. Biochim. Biophys. Acta, 641,416 (1981).

41. C. MONTECUCCO, G. SMITH,DEBBENI-SALA, A. JOHANNSSON, and R. BISSON.FEBS Lett. 144, 145 (1982). Y. M. GALANTE, J. P. DILGER,and D. A. HAYDON. 42. J. R. ELLIOT,D. NEEDHAM, Biochim. Biophys. Acta, 735, 95 (1983). 43. D. A. HAYDON,B. M. HENDRY,S. R. LEVINSON,and J. REQUENA. Nature, 268, 356 (1977). 44. R. G. ASHCROFT,H. G. L. COSTER,and J. R. SMITH.Nature, 269, 819 (1977). Biophys. J. 49, 11 11 (1960). 45. J. RIEGLERand H. MOHWALD. and J. A. MCCAMMON. Sci. Am. 254, No. 4, 30 46. M. KARPLUS (1986). ~ YANNANOS. ~ Science, 228,1101 (1985). 47. E. F. D E L O N GA.~A. H.-P. DUWE, and H. ENGELHARDT. Faraday 48. E. SACKMAN, 251 (1976). G. LINDBLOM, L. B. A. JOHANSSON, and G. ARVIDSON. BioDiscuss. Chem. Soc. 81, 353 (1986). chemistry, 20,2204 (1981). 49. R. SAEZ,F. M. GONI, and A. ALONSO.FEBS Lett. 179, 311 N. L. THOMPSON and D. AXELROD.Biochem. Biophys. Acta, (1985). 597, 155 (1980). FEMS Microbiol. Lett. 13, 131 (1982). 50. K. PORALLA. J. L. R. RUBENSTEIN, B. A. SMITH,and H. M. MCCONNELL. 51. J. H. DAVIS.ROC. Int. School Phys. "Enrico Fermi" on The Roc. Natl. Acad. Sci. U.S.A. 76, 15 (1979). Physics of NMR in Biology and Medicine. 1986. Edited by B. K.KINOSITA,S. KAWATO,and A. IKEGAMI. Biophys. J. 20,289 Maraviglia. In press. 52. D. M. RECHTENWALD and H. M. MCCONNELL.Biochim. (1977). F. JAHNIG.Proc. Natl. Acad. Sci. U.S.A. 76, 6361 (1979). Biophys. Acta, 20, 4505 (1981). G. VAN GINKEL,L. J. KORSTANJE, H. V A N LANGEN, and Y. K. and H. M. MCCONNELL. Biochem. Biophys. 53. B. R. COPELAND LEVINE.Faraday Discuss. Chem. Soc. 81, 49 (1986). Acta, 95, 599 (1980). E. EVANSand D. NEEDHAM. Faraday Discuss. Chem. Soc. 81, 0 . G. MOURITSEN, H. WENNER54. J. H. IPSEN,G. KARLSTROM, STROM,and M. J. ZUCKERMANN. Biochim. Biophys. Acta. 905, 267 (1986). J. SEELIG.Q. Rev. Biophys. 10, 353 (1977). 162 (1987). B. A. CORNELLand F. SEPAROVIC. Biochem. Biophys. Acta, and I. C. P. 55. E. J. DUFOURC,E. J. PARISH,S. CHITRAKORN, 733, 189 (1983). SMITH.Biochemistry, 23, 6062 (1984) B. A. LEWISand D. M. ENGELMAN. J. Mol. Biol. 166, 211 G. WEEKS, and D. GODIN. 56. J. H. DAVIS, B. MARAVIGLIA, (1983). Biochim. Biophys. Acta, 550, 362 (1979). H. SCHINDLER and J. SEELIG.Biochemistry, 14, 2283 (1975). and B. B . LAVIETES. C.R.C. Crit. Rev. Biochem. 57. P. S. COLEMAN J. H. DAVIS.Biophys. J. 27, 339 (1979). 11, 341 (1981). J. H. DAVIS.Biochim. Biophys. Acta, 737, 117 (1983).