摘要:

GENERAL DISCUSSION Prof. R. P. Rand (Brock University, Ontario, Canada) turned to the paper by Carvell, Hall, Lyle and Tiddy. There are practically insurmountable technical problems in maintaining full hydra- tion of hydrophilic surfaces using water vapour. Fig. 1 shows the hydration of phos- phatidylcholine bilayers that results from their equilibration with water of chemical potential different from bulk water by the values of the right ordinate. From full hydration to ca. 10 cal mol-' large excess dextran solutions of directly measured osmotic pressure P were in direct contact with the lipid. Shown with those pressures as ordinates are both equivalent relative humidiiies and temperature increases that would create those relative humidities starting at 100%. A relative humidity of 99.9% induced by a temperature increase of 0.2 K is sufficient to remove one half of the total water.For this water, temperature control to *0.003 K is required to attain the stability of hydration achieved using osmotic pressure.. Serious misinterpretation of results can arise if it is assumed that 100% (nominal) relative humidity achieves full hydration. Practically, temperature fluctuations can result in water condensing on non-lipid surfaces or in the lipid seeing a time-averaged relative humidity less than 100%. We have seen consequent standing gradients of d spacing near air bubbles in closed sample holders containing an excess of water where bilayer separations never reach maximum values when equilibrated only with water vapour. This problem is particularly striking in the case of charged bilayers that never swell indefinitely unless in contact with bulk water.Dr V. A. Parsegian (National Institutes of Health, Bethesda MD) said: I have three questions. (1) Vapour sorption measurements at >go% relative humidity are risky unless one has very good temperature control and one is prepared to wait out tedious kinetics. One should have been forewarned of such problems by the work of Jendrasiak and Hasty,' who showed that toward 100% relative humidity, the charged lipid phosphatidyl- serine (PS) imbibed less water than either of the neutral species phosphatidy- lethanolamine (PE) or phosphatidylcholine (PC). In liquid water we know that PS swells indefinitely [e.g. ref. (2)] while zwitterionic PE and PC swell to finite limits.In vapour, PC was observed' to take up more water than PE or PS. This is consistent with the idea that vapour pressure sorption probes only strong hydration forces [e.g. ref. (2) and (3)]. I realize that the fact of a significant critical micelle concentration frustrates the application of osmotic stress on your surfactants, but how have you satisfied yourself that you would see similar forces in liquid water? Do your lamellae swell to the same extent in an excess of pure liquid water as in 100% relative humidity vapours? In this vein, your reported activity error Aa = 0.002 can be fairly significant near full hydration. For example, if the vapour pressure is 99.8% rather than loo%, then the actual stress will be P = -kT/ V, In 0.998 = -1.38 x In 0.998 = 2.93 x dyn cm-2 at 45 "C (k, T, V, being the Boltzmann constant, absolute tem- perature and water molecule volume, respectively.All points plotted as log,,P = 6.4 or less are indistinguishable. Is this a problem? (2) What is the source of scatter at low hydrations in fig. 2 ( a ) at high pressures where the water activity error should not matter? (3) I think it is worth noting explicitly that the averaging (of microscopic states of water) that goes into a mean quadrupolar splitting is qualitatively different from the x 3 18/30 x 257258 I General Discussion mole ratio (water/lipid) 5 10 15 20 25 0.5 0.1 9 --. h a 0.01 0.001 - 85.000 - 95.000 - 99.000 - 99.900 -99.990 j5t 0 0 0 0 0 0 -2 0 1 I 1 I a 5 10 15 20 25 I vapour pressure interbilayer separation/A Fig.1. See text. thermodynamic average that leads to a free energy of water. Strictly speaking a model with one layer of water ‘bound’ to a surface and with the remaining water completely ‘free’ should not lead to any force between parallel planar layers. Fitting your force data with two constants, as you have done, does not really establish a fundamental equivalence between quadrupolar splitting and water activity. It may be though, that you are seeing steric undulatory forces of the type proposed by Helfrich and recently elaborated to include long-range elastic force^.^ These forces might give the apparently greater decay constraints A that you see by increasing tem- perature. 1 G. L. Jendrasiak and J. H. Hasty, Biochirn. Biophys. Acra, 1974, 79, 337.2 M. E. Loosley-Millman, R. P. Rand and V. A. Parsegian, Biophys. J., 1982,40, 221. 3 V. A. Parsegian, N. L. Fuller and R. P. Rand, Roc. Natl Acad. Sci. USA, 1979, 76, 2750. 4 E. A. Evans and V. A. Parsegian, Roc. Natl Acad. Sci. USA, 1986,83, 7132. Dr G. Cevc (Uniuersity of Essen, West Germany) added: Your measurements indicate (table 1) that the decay length of the interfacial repulsive force strongly changes with the surfactant type and temperature, even if bilayers do not change their phase state. Do you think that these variations in your data are significant and, if so, could you present us your interpretation based on the combined results of your X-ray diffraction and nuclear magnetic resonance measurements? Dr G. J. T. Tiddy and Dr I. G. Lyle ( Unilever Research, Port Sunlight) replied to the three questions in turn.There are, of course, problems and limitations associated with vapour-pressure measurements as with other techniques for probing the chemical potential of water. Since the solubilities of the non-ionic surfactants we investigated are several orders of magnitude higher than those typical of phospholipids, we envisaged problems in applying the osmotic stress technique in our systems. Our experimental procedure involves distilling a known volume of water, from a reservoir in an evacuated apparatus, directly into the cooled sample vessel. The sampleGeneral Discussion 259 vessel and Baratron head are then closed off from the vacuum line and the sample is stirred to mix the surfactant and water thoroughly.After equilibration the vapour pressure is measured. Thus we know with certainty the sample composition; we do not rely on sorption of water from the vapour phase, which might indeed be risky. Our quoted accuracy in water activity of k0.002 depends on the measurement of sample temperature to within *0.02 "C. As Prof. Rand and Dr Parsegian rightly point out, this implies that we cannot reliably measure a,> 0.998, which corresponds to log,, P < 6.44 at 25 "C. While we do not regard this as a problem, it certainly does prevent us measuring weak repulsive forces. These weak, longer-range interactions do exist in the non-ionic surfactant systems and result in their swelling to very large separations, as indicated in fig. 1 of our paper. Unfortunately the X-ray peaks become so broad at high water contents that it has proved impossible to determine a maximum or equilibrium d-spacing in excess water.We have not attempted to measure the swelling of non-ionic surfactants equilibrated in 100% relative humidity vapour (or at any other relative humidity for that matter, for the reasons given by Prof. Rand and Dr Parsegian). In a separate study' we have measured vapour pressures for a mixed phospholipid lamellar phase at low water content, using the Baratron apparatus, and obtained results in good agreement with those of Cowley et aL3 from osmotic stress. Now to Dr Parsegian's second question: At low surfactant hydrations, mixing of the viscous lamellar phase samples becomes difficult. Although we compensate for this by extending equilibration periods, it is possible that errors might arise from incomplete mixing.Measurements are routinely made during water removal as well as during addition to check that significant hysteresis does not occur. In fig. 2(a) we have drawn 'best-fit' straight lines through the experimental points to examine how closely these correspond to an exponential 'hydration force'. There is no a priori reason to suppose that the data should be fitted by simple exponential decay functions of this type. And his third question: We agree that the averaging of water states inherent in mean n.m.r. A values and water activities are quite different. Hence the fact that the two sets of data can be reconciled lends powerful support for our water binding hypothesis.We regard the aqueous region between the bilayers as a mixture of polyoxyethylene groups and water (or 'mash', borrowing Dr Parsegian's terminology). Hence a bound/free equilibrium gives repulsive forces, even with only one layer of bound water, as long as none of the free water separates EO groups on opposing layers. We measure forces only when these opposing headgroups overlap, as expected for this mechanism. Beyond contact, where undulation forces could predominate, the repulsions are too weak to be detected by our technique. Also, when the water activities approach close to 1.0, changes in the amount of bound water per surfactant must be very small. Finally we turn to Dr Cevc. Yes, the effect is real, but we do not consider the exponential treatment to be relevant for these systems: we rationalise our data in terms of the water binding model.1 K. Hammond, I. G. Lyle and M. N. Jones, Colloids Surf.', in press. 2 A. C. Cowley, N. L. Fuller, R. P. Rand and V. A. Parsegian, Biochemistry, 1978, 17, 3163. Dr J. Yarwood ( University of Durham) said: I should like to commend to this meeting the use of infrared spectroscopy to study the perturbation of water molecules near polar organic interfaces. As an example of what may be done I present some very recent results from our laboratory (obtained by Dr W. F. Pacynko in collaboration with Dr G. J. T. Tiddy) on the spectra of water in the various lyotropic liquid-crystalline phases of dodecyltetramethyl ammonium chloride ( C12TACl). The material forms a well known' series of mesophases as the amount of water is reduced.Fig. 2 shows the way in which the water bands at 2520cm-' [corresponding to vs(OD) of HDO; 5% D20 in H20] and at 2110 cm-' [corresponding to the association transition, v,+ vL- v,; described as vA(H20)] change in frequency, shape and relative intensity on going from 'pure' water260 General Discussion 1 I I I I I 1 1 I I I I I 2800 2600 2400 2200 2000 1900 1800 wavenumber/ cm-’ Fig. 2. Infrared red spectra of water in different mesophases of CI2TACl: (A) pure water, 5% D20 in H20, (B) the HI hexagonal phase, (C) the V, cubic phase, (D) the L, lamellar phase. The temperature is 50°C. to the lamellar phase at 93.8% surfactant. The 21 10 cm-’ band clearly shifts and becomes relatively much more intense than the vs(OD) band (fig.3). Indeed, in the lamellar phase (La) (the phase most closely related to the membrane structures discussed at this meeting) the v,( H,O) band comprises two components, each presumably related to water bound at the interface. Since, at this concentration, the waterlsurfactant mole ratio is between 1 and 2, it is possible that the two bands represent cation and anion interactions with water. Both vA( H20) and v( OD) bands are considerably narrowed, probably pointing to a reduction in the breadth of distribution of perturbed water molecules as the water/ surfactant mole ratio decreases. The absence of discontinuities in the intensity ratios (fig. 3), band-broadening or frequency shifts at any of the phase boundaries indicates (a) that water molecules probe the short-range interactions at the head groups and not the long-range forces associated with the ‘packing’ of micelles into the various phases and (b) that the perturbation of water in these systems (although substantial) is a gradual process, without any sharp change from one ‘species’ to another.The spectra therefore provide a sensitive indicator of the way in which interfacial interactions develop with decreasing inter head-group separation. Quantitative intensity data would enable water ‘activity’ to be monitored in surfactant mesophases, providing further support for the general validity of the equilibrium model proposed by Carvell, Hall, Lyle and Tiddy at this Discussion. 1 R. F. Balmbra, J. S. Clunie, J. M. Corkill and J. F. Goodman, Trans. Furuduy SOC., 1962, 58, 1661. Mr F.A. M. Leermakers, Dr J. M. H. M. Scheutjens and Prof. J. Lyklema (Agricultural University of Wageningen, The Netherlands) addressed Prof. Bailey: Direct measurements of the interaction between DLPC bilayers adsorbed on mica surfaces show both attraction and a strong short-range repulsion (fig. 3 in the article by Afshar-Rad et aZ.) Preliminary calculations of membrane-membrane interaction based on an a6 initio statistical ther- modynamical theory’ show very similar behaviour. Fig. 4 is an example of the interactionGeneral Discussion .o 26 1 - 0 - 2 3*0i .o ‘ I II !I I 8 1 1 2 I I’ I I I I I l l 1 1 : I 1 1 I I L1 H1 0 0 0 0 0 C,,TACl (wto/o) Fig. 3. The vA/voD intensity ratios (determined from spectra shown in fig. 2) as a function of surfactant concentration across the phase diagram of CI2TAC1.The temperature is 50 “C. -0.02 -0.04 15 20 25 30 M Fig. 4. Calculated interaction curve for membranes composed of non-ionic lecithin-like molecules with two apolar tails of 16 segments, a glycerol backbone and three polar head-group segments. M is the distance between the two centres of the membranes in units of ca. 0.13 nm. F / n is the free energy of interaction per lipid molecule. The interaction parameters are: xAB = 1.4, xAS = 1.6, xBS = 0 for tail-head, tail-solvent arid head-solvent, respectively. The trans-gauche energy is (275/300) kT.262 Genera 1 Discussion curve found for interacting membranes composed of lecithin-like molecules above T, . In contrast to the mica experiments, in our calculations the membranes are able to adapt their thickness and surface area when they come close.The repulsion at short separation is therefore (especially above T,) much weaker than reported for mica-supported layers. We have not yet included the van der Waals interaction of the membranes. The attraction between the membranes is a result of incomplete shielding of the apolar phase by the head groups, so that some hydrophobic interaction between the bilayers occurs and the repulsion originates from the overlap of the two bilayers. Any repulsive force between head groups and tails reduces and may even prevent the attraction between the bilayers. 1 F. A. M. Leermakers, J. M. H. M. Scheutjens and J. Lyklema, Biophys. Chem., 1983, 18, 353. Prof. A. I. Bailey (Imperial College, London) replied: We find the predictions of Prof.Lyklema and coworkers interesting and are pleased that they agree qualitatively with our results. However, without any inclusion of a van der Waals term in the calculation it is difficult to draw any quantitative conclusions. Dr G. J. T. Tiddy (Unilever Research, Port Sunlight) asked: How do the surface separations where the 'jump' is observed compare to those reported by Prof. Ninham for hydrophobic surfaces? What is the mechanism of this attraction? Prof. A. I. Bailey and Dr P. F. Luckham (Imperial College, London) replied: The range and strength of the attraction between the two lipid surfaces plus myelin basic protein, and between the two mica surfaces bearing adsorbed MBP, are similar to those reported by Ninham and Pashley' in their study of hydrophobic surfaces.The attraction commences at lipid or mica surface separations of ca. 14 nm and the resultant adhesion is of the order of 10 mN m-', although there is variation between experiments. The second part of this question is more difficult to answer. The strong attraction between the protein covered surfaces is of great significance considering the proposed role of MBP in the stacking of the myelin membrane. We believe the attraction to be due a protein-protein interaction, as it is too strong and too long-range to be due to van der Waals forces between the underlying mica surfaces or the lipid surfaces them- selves. The origin of the attractive force may be of a hydrophobic nature, although the necessary disruption of the water structure over a range of 14 nm (ca.60 water molecules) is a subject of controversy. Nevertheless, the protein has hydrophobic regions, and when adsorbed directly on mica, causes the mica to become hydrophobic as evidenced by a high contact angle. Other explanations are less satisfactory. Polymer effects leading to attraction, such as interactions between polymers in a poor solvent2 (0.5 mm-') and polymer bridging3 (50 pN m-I) produce far smaller attractions than those observed here. The bridging possibility is further reduced since in an experiment where one mica surface was coated with an adsorbed layer of MBP and second surface was freshly cleaved mica the attraction was reduced (this residual attraction being due to van der Waals attraction between the mica surfaces which were now separated by 1.6 nm).We are continuing to investigate this problem. 1 R. M. Pashley, P. M. McGuiggan, B. W. Ninham, and F. Evans, Science, 1985, 229, 1088. 2 J. Klein, Nature (London), 1980, 288, 248. 3 J. Klein and P. F. Luckham, Nature (London), 1984,308, 836. Prof. E. A. Evans (Unviversity of British Columbia, Vancouver, Canada) said: This paper presents some very interesting and important results on forces between bilayer membranes fixed to mica substrates. However, consideration of the results brings out a major concern with the method as applied to the study of membrane bilayers: what is the water-gap distance between bilayer surfaces? The instrument is quite accurateGeneral Discussion 263 with regard to the distance between silvered surfaces; but when ‘invisible’ layers (such as surfactant bilayers) are present, the interbilayer separation can only be derived by indirect calculations.In this paper, for example, the authors show a stress-free (secon- dary minimum) position for two DLPC bilayers at ca. 7 nm (which is equivalent to twice one bilayer thickness). Thus there would be no water between the layers, in contradiction to X-ray diffraction studies (which show a ca. 2.5-3.0 nm water gap). Also, results of Horn’ and Marra and Israelachvili* give more reasonable values of ca. 9- 10 nm for the same stress-free position. Again, however, results from the latter study’ are also in question, since they arrive at a water gap of 1.3 nm between DPPE bilayers in the crystalline state, whereas X-ray studies show 0.4nm or less.Hence, although this approach gives accurate measurements of forces and reasonable measurements of distances when separations are greater than a few nm, there remains a significant uncertainty of the order of a fraction of a nanometre or more for determination of distances between ‘invisible’ bilayers. This uncertainty is of critical importance in the analysis of long-range interactions at close proximity, and especially to the form of the theoretical relations for these forces in this range. 1 R. Horn, Biochim. Biophys. Acta, 1984, 778, 224. 2 J. Mama and J. Israelachvili, Biochemistry, 1985, 24, 4608. Prof. A. I. Bailey and Dr P. F. Luckham (Imperial College, London) replied: In our work distances are measured with reference to the contact of the mica sheets in air.This may be different to the contact position in water by up to 0.7 nm. For reasons such as this the absolute error present in distance measurements in lipid experiments may be greater than is usually supposed, although the interferometric technique can give resolution down to 0.1 nm. The DLPC has been found partially to desorb from the surfaces if the bathing solution is not saturated with DLPC monomer. This results in the thickness being reduced by up to ca. 20% over a period of hours. In the preliminary results presented (fig. 3 ) thinning of this type has occurred. We have carried out further experiments where the solution has been saturated with DLPC. The stress-free secondary minimum was then located at 8.8f0.5 nm.This value is comparable to twice the repeat period obtained by X-ray diffraction for DLPC multilayers at high relative humidity.’ No desorption was ever observed in the cerebroside sulphate-cholesterolmyelin basic protein membranes where identical results were obtained in the absence and presence of lipid in the aqueous phase. 1 T. J. McIntosh, Biochim. Biophys. Acta, 1978, 513, 43. Dr L. R. Fisher (CSIRO, Sydney, Australia) asked Prof. Bailey: Could you tell us how the reference distance D=O was established in your experiments? In particular, was it measured in air or with the surfaces under water? I ask this question because it is known that there is a ca. 0.5 nm layer of adsorbed material on the mica surfaces in air, and that this layer apparently dissolves in water, allowing the surfaces to come into closer contact.Prof. Bailey replied as follows: The point raised by Dr Fisher has been covered in the reply to Prof. Evans. Prof. T. W. Healy (University of Melbourne, Australia) commented: The data of fig. 2 for interaction of two mica surfaces in water and those of fig. 3 for two DLPC bilayers interacting in water, illustrate the fact that the deposition process totally eliminates the double-layer repulsion interaction that is observed for the bare mica surfaces. Whatever264 General Discussion the structure of the deposited DLPC film, it is clear that the film precludes development of the mica surface charge and/or the penetration of counter ions to allow expression of the mica substrate double layer.It would follow that in the L-B deposition process there is insufficient water in the film to allow penetration of counter-ions. The earlier Israelachvili and Marra results also show the same effect as evidenced in fig. 2 and 3, indicating that their ‘adsorption’ process for forming the lipid bilayer also appears to exclude water. The total lack of e.d.1. repulsion from the lipid bilayer itself needs to be considered. It is possible that: (i) the zwitterionic head group lies flat in the plane and that the counter-ion population out from the wall is not discriminated sufficiently to yield a nett diffuse layer charge, and/or (ii) the process of pushing two surfaces together folds, as a regulation response, the positive end of the head group down into the plane, and/or (iii) the sharp variation in charge (i.e., zero, negative, positive, zero) is the nitrogen centre extended out, over the e.d.1.is itself an unfavourable cyclic variation in micro- potential that again, as a regulation response will ensure that the zwitterion lies flat. I believe such issues still need to be explored, in view of the finding from probe studies that local negative potentials are observed within the lipid bilayer structure. Prof. A. I. Bailey and Dr P. F. Luckham (Imperial College, London) replied to Prof. Healy as follows: Our results, together with those or Israelachvili and Marra, show that when the surfaces are coated with a bilayer of DLPC, the mica substrate double layer, due to the dissolution of potassium ions from the mica surface, is not present.The mica is immersed in the water of the Langmuir tro.ugh before deposition takes place; any surface charge would be developed at this stage, so that it is unlikely that the DLPC is inhibiting the formation of charge. It is possible, however, that penetration of counterions into the region where the polar headgroup interacts with the mica surface is inhibited. We have the following comments concerning the lack of a double layer from the lipid bilayer itself. In studies of stacked; oriented phosphatidylcholines,’ it has been shown that the headgroup lies flat in the plane of the bilayer. If this were the case in our experiments no double layer would be expected. However, if the headgroups are in a different configuration some form of regulatory response must occur as the two surfaces approach.In our experiments no probe is used and therefore questions of the influence of perturbations of the bilayer by the probe do not arise. 1 R. G. Griffin, L. Powers and P. S. Pershan, Biochemistry, 1978, 17, 2718. Dr D. S. Dimitrov (Bulgarian Academy of Sciences, SoJia, Bulgaria) said: These results presented by Dr Fisher and Prof. Haydon are undoubtedly the first experimental data for the profile of approaching lipid bilayers. There are several important observa- tions, which deserve special attention and can be compared with theoretical results.’ ( 1 ) It is observed that a dimple forms only for the case of fast approach. The dimple forms when the driving force F is larger than the force due to membrane tension T, which is of the order of RsT where Rs is the radius of spherical membranes).In these experiments the driving force can be calculated by using the Taylor formula. I have estimated it for the cases of fast and slow approach to be (at a separation h = 150 nm) 1.1 and 0.0031 dyn, respectively. Then the ‘deformation’ thickness f/T is equal to 0.26 cm and 7.2 x cm, respectively. For the case of fast approach it is larger than the membrane radius (0.1 cm), i.e. a dimple will form. For the case of slow approach it is smaller than the membrane radius, i. e. there is no dimple, as observed experimentally. For the later case the ‘deformation’ thickness is higher than the separation h (150 nm); therefore a plane-parallel film should form, as observed experimentally.In the latter case electrostatic intermembrane forces can significantly contribute to the formation of the flat liquid film. A flattened region has also been observed, however,General Discussion 265 when the bilayers were brought together in 0.01 mol dm-3 sodium chloride solutions. In this case the Debye length is 3 nm and the electrostatic forces are suppressed. Consequently, the electrostatic forces can be important but even when they are very small a flat film can be formed due only to hydrodynamic and membrane tension effects. The estimate of the driving force by using the Taylor formula may be incorrect because of the membrane deformation. Another way to estimate the driving force, especially for deformed membranes, is by using the formula for the radius of membrane contact.However, during approach, in order to keep the rate constant the force should be continuously increased. Therefore, at the final equilibrium state (or the state before fusion) the forces for the fast and slow approach may be of the same order of magnitude. Inertial forces may also contribute to dimple formation. Estimates have shown that for this case the Reynolds number is <0.1. It seems, however, that the major reason for dimple formation is membrane instability to bending. It occurs when the driving force F is larger than the force due to membrane tension which opposes bending. When the driving force is smaller, there are two cases: a small deformation (i-e. the membranes retain their spherical shapes) and large deformations (where the membranes are deformed and a flat region forms).(2) The speed of the osmotically driven approach of the bilayers is 0.35 pm s-’. We have shown that the parameter which determines whether the water will flow through the bilayers or within them is E~ = 12pLR2/ h3 where p is the viscosity of the liquid layer between the membranes, L is the hydraulic permeability coefficient and R is the radius of the contact area. For these experiments this parameter is of the order of smaller than unity. This means that the water pre- dominantly flows within the bilayers in the region of close opposition and through the bilayers outside this region. Indeed, if we suppose that the driving pressure is the capillary one, 2T/R,, and the rate of membrane approach obeys the Reynolds law, we get a value of the order of 0.1 pm s-l, which is very close to the one experimentally observed, having in mind the approximations made.However, this does not explain the observed independence of the rate of approach from the distance. This behaviour indicates that permeability is responsible for the membrane approach. Further work will help us to understand what is the underlying mechanism in this case. (3) Especially important is the demonstration and measurement of the critical thickness of rupture of the liquid film between the membranes. I have shown theoreti- cally2 that this should occur and calculated that the critical thickness of rupture for such systems should be in the range 10-100 nm. Based on the assumption that only van der Waals forces act, for the parameters of this experiment I have calculated a value of 56nm.This value is of the order of magnitude of the experimentally observed quantity. The higher value can be due to the existence of a small electrostatic positive disjoining pressure, nonhomogenities in the film thickness, etc. The simplest possible formula was used,2 in which very rough approximations were made: h,, = 0.8(2 R2Hwd2/ V T ) ’ ’ ~ where H, is the Hamaker constant and d is the membrane thickness. I have also estimated the time of growth of the critical fluctuation wave to be ca. 0.8 s, which seems to be a reasonable value. The important thing is that this is evidence that the fluctuation mechanism can be responsible for the instability of thin films between membranes and this can be of basic significance for the kinetics of membrane fusion.1 D. S. Dimitrov and R. K. Jain, Biochim. Biophys. Acta, 1984, 779, 437; D. S. Dimitrov, Progr. Surf: 2 D. S. Dimitrov, Colloid Polym. Sci., 1982, 260, 1137; D. S. Dimitrov and D. V. Zhelev, Stud. Biophys., Sci., 1983, 14, 295; D. S. Dimitrov, I. B. Ivanov and T. T. Trykov, J. Membr. Sci., 1984, 17, 79. 1985, 110, 105.266 General Discussion Dr L. R. Fisher (CSIRO, Sydney, Australia) replied to Dr Dimitrov as follows. Your recently published formula for the critical thickness of rupture provides a nice explana- tion for the fact that we find no significant difference between the critical thickness of rupture for an aqueous draining film between glycerol mono-oleate-n-decane or glycerol mono-oleate-n-hexadecane bilayers.The value of h,, is proportional to ( d2/ T ) ''6. For glycerol mono-oleate-n-decane, d = 5.77 nm and for glycerol mono-oleate-n-hexadecane, d = 4.03 nm.' The respective values for the membrane tension T are 7.7 and 4.2 mN m-'.2 Thus the value of ( d 2 / T)lj6 is 1.28 in the first case and 1.25 in the second, a difference which is within our limits of error. Regarding your comment on the flow paths for water when the bilayers are brought together osmotically, the approach speed of 0.35 p m s-' is quite compatible with the known permeability coefficient of 52.7 f 0.4 p m s-' for glycerol mono-oleate-n- hexadecane bilayers? and this fact, together with the lack of dependence of approach speed on bilayer separation, gives us confidence that the main pathway for water flow is through the bilayers in this case.1 J. P. Dilger, Biochim. Biophys. Acta, 1981, 645, 357. 2 D. Needham and D. A. Haydon, Biophys. J., 1983, 41, 251. 3 R. Fettiplace and D. A. Haydon, Physiol. Rev., 1980, 60, 510. Dr V. A. Parsegian (National Institutes of Health, Bethesda, MD) said: the uninten- ded action of alkane solvent has been an enduring worry to students of artificial bilayer systems. Sensitivity to this often hidden variable has become more pertinent to me since seeing the results presented in our own paper at this meeting. Alkanes can actually effect lamellar-to-hexagonal phase transitions. What might they be doing to bilayers that are pushed togethtr as in your apparatus? Are contact and bending influenced by the decane and hexadecane in your preparation? Is there any way to maintain a reservoir of alkane to control its chemical potential? How will one relate the bilayer response in your system to the behavior of cell membranes or of solvent-free artificial systems? Dr L. R.Fisher (CSIRO, Sydney, Australia) replied: Our measured values for aqueous film drainage rates and critical thicknesses of rupture are unlikely to have been influenced by the details of bilayer composition and structure (except insofar as these affect macroscopic variables such as interfacial tension), since the bilayers were separated by more than 28 nm during these measurements. Where an effect of the hydrocarbon solvent may arise, though, is when the bilayers approach closely once the aqueous draining film has collapsed. It is by no means certain, for example, that solvent-free glycerol monoleate bilayers would fuse in the manner described. If properties which could be affected by hydrocarbon solvent (e.g. bending moments, or the tendency to HII phase formation) do in fact affect fusion, then of course the presence of hydrocarbon may have an influence. This may not be entirely bad, since changes in hydrocarbon content may then be useful as a tool to investigate the mechanism of fusion. The chemical potential of the hydrocarbon reservoir can be controlled over a narrow range by changing the lipid concentration, and for alkanes a wide range of volume fractions in the bilayer may be achieved by varying the alkane chain length. A relationship between lipid bilayer fusion and cell membrane fusion, while likely, has not yet been established, and in neither case is the mechanism of fusion known. We can only hope, albeit with some justification, that studies on simple membrane analogues will provide sufficient clues to resolve the puzzle eventually. Dr Fisher then added a more general comment. I should perhaps point out that the accuracy with which we can perform our distance measurements depends very much on the distance being measured. We are limited by the accuracy (ca. *O.l YO) with which we can measure the intensity of the reflected laser beam. This *0.lo/~ translates to k0. 1 nm: for surfaces 20-50 nm apart. The accuracy rapidly worsens at smaller distances, though and, for example, a separation of 2 nm can at best be measured to k0.4 mm.

ISSN:0301-7249    

DOI:10.1039/DC9868100257

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

Faraday Discuss. Chem. SOC., 1986, 81, 267-280 Giant Vesicle Bilayers composed of Mixtures of Lipids, Cholesterol and Polypeptides Thermomechanical and (Mutual) Adherence Properties Evan Evans* Pathology and Physics Departments, University of British Columbia, Vancouver, Canada V6T 1 W5 David Needham Pathology Department, University of British Columbia, Vancouver, Canada V6T 1 W5 Micromechanical tests of giant vesicle bilayer elasticity and bilayer-bilayer adhesivity have been carried out on vesicles made from mixtures of lipids, cholesterol and polypeptides. Mixtures of phosphatidylcholine (PC) and phosphatidylethanolamine (PE) exhibited ideal solution behaviour over a temperature range that covered both liquid-to-gel phase transitions. Addition of cholesterol (CHOL) to a saturated chain lecithin (DMPC) reduced, broadened and shifted to higher temperature the main crystalline-to-liquid acyl chain transition.Cholesterol greatly reduced the membrane area com- pressibility and increased membrane cohesion to levels exhibited by frozen acyl chain bilayers, but maintained the bilayer in a liquid state. Addition of amphiphilic polypeptides to PC and PC-CHOL mixtures slightly increased bilayer compressibility at high temperatures; when the temperature was lowered, bilayer compressibility for DMPC-CHOL-peptide mixtures was greatly reduced to below that of the single component lipid. Cholesterol appeared to change from an association with the lipid at low temperatures to an association with the protein at high temperatures. Membrane cohesion correlated with a simple fracture energy model where the level of tension required to lyse vesicles is proportional to the square root of the elastic area compressibility modulus.Free-energy potentials for adhesion of mixed lipid (PC and PE) bilayers showed a dramatic increase as the mole fraction of PE approached unity. Based on published values of separation distance between lamellae at full hydration for each pure component, the free energy potentials for adhesion of mixed-lipid bilayers were calculated from theoreti- cal models of the van der Waals attraction and hydration repulsion then compared with the measured values of adhesion energy. The correlation provides strong evidence that the hydration repulsion can be represented by a surface-potential theory.The lamellar configuration of biological membrane structures is peculiar to the preferen- tial assembly of amphiphilic molecules (e.g. lipids, proteins etc.) into membrane double layers. The strong preference for the lamellar configuration is evidenced by the very slow rate of exchange of membrane molecules between an encapsulating membrane surface and its adjacent aqueous phases. Consequently, the encapsulating membrane of a biological cell or artificial lipid vesicle behaves as a closed system for periods of time on the order of hours. In this condensed state, membranes exhibit solid or liquid-like material behaviour with the common feature of limited surface compressibility (i.e. great resistance to changes in surface density). Micromechanical experiments are used to establish the nature of the material structure of the membrane.Combined with tem- perature changes, mechanical experiments provide data for decomposition of membrane deformation energies into internal energy and configurational entropy contributions. 267268 Giant Vesicle Bilayers With the use of well defined chemical compositions (i.e. specific ratios of lipids, proteins etc.), mechanical experiments provide direct insight into the chemical state and origin of the material properties observed for natural biological membranes. As condensed states of matter, membranes can interact non-specifically via long-range electrostatic, electrodynamic and solvation forces. The non-specific or colloidal forces that act between these thin membrane materials are commonly recognized as the superposition of van der Waals attraction, electrostatic repulsion and hydration repulsion.Even though the magnitude of each of these interactions can be quite large, the free-energy minimum at stable contact is extremely low, which makes measurement difficult. Here again, micro- mechanical experiments provide a direct method for measurement of the chemical affinity (cumulated free-energy of interaction) between membrane surfaces. In this paper, we will report results which show how mixtures of lipids, cholesterol and polypeptides affect the thermo-mechanical and mutual interaction properties of bilayer membranes. In addition, we will discuss prominent physical features of bilayer mem- brane structure and interactions which have been derived from these results: e.g.the mixing behaviour of phosphatidylcholine and phosphatidylethanolamine; the effects of cholesterol and polypeptides on membrane cohesion and physical state (liquid or solid); evaluation of the theoretical prescription for van der Waals attraction between bilayer membranes and evidence in support of a surface potential law for the hydration repulsion. Lipid mixtures exhibit the feature that they spontaneously form closed vesicular capsules when rehydrated from an anhydrous state. Although small in number, a few of these capsules are of sufficient size [( 10-20) x cm] that they can be subjected to micromechanical deformation and adhesion tests. Because of the liquid interior and exterior, micropipette aspiration and manipulation of the giant vesicles provide simple techniques to measure vesicle membrane mechanical and adhesion properties (plate 1).First, we will outline the theoretical methods for analysis of membrane deformation and adhesion experiments, then describe results for single- and multi-component mem- brane bilayers. Membrane Thermomechanics Membrane thermoelasticity is simply the inter-relation of the equations of state for the membrane material to the combined first and second laws of thermodynamics for reversible Membrane materials are characterized by two surface equations of state for the assumption that the membrane structure is isotropic in the surface plane. The first equation of state relates the mean or isotropic tension to changes in surface density or fractional change in area and temperature, dT,=K(da+& dT) where T, is the isotropic membrane tension, (Y is the fractional change in local area and T is temperature.Hence, the surface isotropic equation of state is characterized by two differential parameters, elastic area compressibility modulus (K) and thermal area expansivity (&), which can be easily measured by micropipette aspiration of bilayer vesicles. The procedure is to aspirate a vesicle with small micropipette [ca. 5 x lod3 cm diameter as shown in plate 1( a ) ] and then perform two experiments: (1) increase the aspiration pressure at fixed temperature and observe the increase in vesicle membrane projection length inside the pipette; (2) increase the ambient temperature at fixed aspiration pressure and again observe the increase in vesicle-membrane projection inside the pipette.Since the volume of aqueous contents inside the vesicle is held constant because of internal solutes, changes in vesicle-membrane projection length inside thePlate 1. Video micrographs of micromechanical tests of ( a ) bilayer elasticity and (6) bilayer- bilayer adhesivity. Pipette dimensions are c a 5 x cm and vesicles are ca. 2 x cm in diameter. (To face p. 268)E. Evans and D. Needham 269 pipette are a direct measure of the fractional change in total vesicle-membrane AA z [2nRP( 1 - R,/RO)]AL. Furthermore, the tension is uniform over the entire vesicle surface and is given by the product of pipette suction pressure and the pipette radius divided by a simple function of the vesicle diameter: Consequently, the fractional change in surface area or density can be directly related to change in membrane tension or change in temperature; the slopes of these relations provide direct measures of the area compressibility and thermal area expansivity respec- tively (the elastic area compressibility modulus is given by the reciprocal of the compressi- bility).A major uncertainty in this type of experiment is the number of bilayers which make up the total vesicle membrane thickness. Fortunately, by selection of the most optically transparent vesicles (with interference contrast microscopy) and evaluation of the membrane elastic modulus, it is possible to discriminate between one, two and more layers since the elastic modulus groups around discrete values where the lowest value is characteristic of a single b i l a ~ e r .~ Thus, we are confident that these measurements truly represent a single bilayer structure. Typical values for the elastic area compressibil- ity modulus ( K ) of lecithins in the La state (above the acyl chain crystallization temperature) are in the range of (1-2)x 102dyncm-' and values for the thermal expansivity (ci) range from (2-6) x K-'. In the Lpf phase below the acyl chain crystallization temperature, the area compressibility modulus increases by cu. an order of magnitude (cu. lo3 dyn cm-'), whereas the thermal area expansivity is reduced to The thermoelastic coefficients ( K , ci) define the internal energy and enthalpy of expansion for fractional changes in surface density at constant temperature based on simple thermodynamic relations.'.2 The internal energy of expansion is equal in magni- tude and opposite in sign to the enthalpy of expansion.The enthalpy of expansion is the reversible heat per unit area exchanged with the environment in proportion to area K-' or l e ~ s . ~ - ~ expansion and is given by (2) E T( g) T = TKk. For ideal surface equations of state, the enthalpy of expansion is equal to the surface pressure inside the bilayer. Lecithin bilayers above the acyl chain crystallization tem- perature exhibit enthalpies of expansion on the order of lo2 erg cm-2 or ca. 5 kcal mol-' of lipid, comparable to twice the interfacial free energy density (surface tension) of a hydrocarbon-water interface. The surface isotropic equation of state (just described) represents material properties that characterize both solid and liquid condensed membrane structures.However, for solid membrane materials (e.g. frozen lipid bilayers below the acyl chain crystallization temperature), there exists an additional equation of state that relates the in-plane membrane shear forces to surface shear deformations at constant surface density and temperature. Shear rigidity properties are significant determinants of cell and frozen vesicle deformability ; however, these properties are not crucial to the understanding of the solution chemistry of the surface and will not be considered in this paper. (See references that deal with membrane shear and bending rigidities for frozen lipid b i l a y e r ~ ~ .~ and red blood cell m e m b r a n e ~ . ~ ~ ~ ~ ~ ) Thermomechanical Properties of Bilayer Mixtures The only previous mechanical measurements on membrane mixtures were for the natural red blood cell membrane, which is a composite of lipids, integral and peripheral proteins,270 Giant Vesicle Bilayers I .o V. I 0 10 20 30 40 50 T/ "C Fig. 1. Single vesicle cooling experiments for SOPC-POPE mixtures. Vesicle areas were normal- ized by the values at a common temperature of 25 "C. Note that for PE contents below 33% some of the SOPC appears to freeze separately. The transition temperatures observed for these mixtures are consistent with an ideal mixture model with no enthalpy of mixing. POPE/(POPE+SOPC): 0, 0.0; 0, 0.33; A, 0.60; 0, 0.80; 0, 0.90. cholesterol etc.At high temperature (> 48 "C), the peripheral protein cytoskeleton is disrupted; the resultant membrane structure exhibits an elastic area modulus of ca. 300 dyn cm-' and thermal area expansivity of ca. 1 x K-*. Thus, the enthalpy of expansion is ca. 100ergcm-2, comparable to a lecithin bilayer above the acyl chain crystallization temperature, but less compressible. In addition, the tension for lysis of the red cell at high temperature is comparable to values for lecithin bilayers, i.e. ca. 3-6 dyn cm-'. Hence, what are the effects of lipid mixture, cholesterol and integral proteins on membrane compliance and cohesion? Lipid Mixtures The phosphatidylcholine and phosphatidylethanolamine classes of lipids, especially those with asymmetric chain composition, are major components of cell membranes.Thus, we have chosen to study two of these lipids with similar chain length and composition and with gel to liquid-crystalline phase-transition temperatures in the range accessible to our technique. These were 1 -stearoyl-2-oleoylphosphatidylcholine (SOPC) and 1 -palmitoyl-2-oleoyl phosphatidylethanolamine (POPE). To ensure that both lipids were electrically neutral, the measurements were made at pH 6.0. As described before, observation of the aspirated length of a vesicle held at constant pressure in a micropipette provides a sensitive measure of the vesicle area change as the temperature is changed. Results of single vesicle cooling experiments for SOPC- POPE mixtures are plotted in fig. 1 over a temperature range that spans both gel-liquid- crystalline phase transitions." With the onset and final temperatures observed for each mixture transition, we have constructed a phase diagram and found that it is in close agreement with an ideal mixture model (i.e.no enthalpy of mixing and an ideal entropy * For heating and cooling rates of 0.5 K min-', some hysteresis in the area uersus temperature curves was observed. At the (liquid)-(liquid-solid) boundary, Tmelt was cu. 1 K above TfreeZe and increased to cu. 2 K for higher rates of temperature change. At the (solid-liquid)-( solid) boundary, very little hysteresis was seen: Tmelt - Tcreeze < 0.5E. Evans and D. Needham 27 1 Table 1. Thermoelastic properties of SOPC-POPE bilayers POPE (SOPC+ POPE) K/dyn cm-' ci/ K-' T / "C 0 199.6 f 12.7 3.28 f 0.68 15.0 0.33 209.9 * 14.1 3.79 f 0.24 22.0 0.60 221.4* 11.2 3.51 f 0.57 25.0 0.80 233.7 f 26.0 3.53 f 0.49 26.0 of mixing)." Hence, the two lipids are essentially miscible in both liquid and solid phases.The form of the cooling curves supports this conclusion except for PE contents below 33 YO, where some of the SOPC appears to freeze separately. Such immiscibility effects are commonly observed for saturated PC-PE mixtures''-'* although the phase diagram is more complicated owing to the presence of Pp, phases.13 Values for elastic area compressibility moduli, measured at temperatures just above the upper transition. temperature ( T / T, = 1.02-1.03) for the L, phase of both com- ponents, show a slight linear compositional dependence (table 1).The linearity is consistent with ideal mixing. Thermal area expansivities for the bilayer mixtures were fairly constant and similar to values for other lecithins in the L, phase well above the acyl chain crystallization temperature. The enthalpy of expansion is elevated at tem- peratures close to (within 5-10 K above) the liquid-gel transition and diminishes at higher relative temperatures in the L, phase. Based on observation of the tensions required to lyse (rupture) vesicles, these neutral lipid mixtures exhibited bilayer cohesion levels similar to single component membranes (although the SOPC and SOPC-POPE mixtures were slightly more cohesive than DMPC). Lipid-Cholesterol Mixtures Mixtures of dimyristoylphosphatidylcholine (DMPC) and cholesterol (CHOL) were used to examine the effects of cholesterol on the main and pre-gel to liquid-crystalline phase transitions.Vesicle area versus temperature plots (fig. 2) show that the transition broadens and shifts to higher temperatures with increasing cholesterol concentration, in agreement with regular solution theory." Similar results have been obtained by scanning calorimetry and n.m.r. s t ~ d i e s . ' ~ , ' ~ For pure DMPC, the magnitude of the area change represents both acyl chain condensation and chain tilt with respect to the bilayer plane. In the Pp phase, the chains line up parallel to the normal to the projected plane because of a surface 'ripple' or superlattice; when these ripples are removed (either by mechanical stress' or at low temperatures) the chains become tilted with respect to the projected plane and the projected area slightly increases.Upon the addition of cholesterol (even small amounts), the Ppv phase disappears, which indicates a tilted geometry for any frozen lipid. The total area change over the transition is reduced by cholesterol and disappears at 50 mol YO where the thermal area expansivity is 1.33 x K-'. By estimating the area fraction for each component (liquid and gel-phase lipid and cholesterol-lipid complex), the results are consistent with formation of a 1 : 1 DMPC-CHOL complex;16 the area change over the broad transition is simply attributed to uncomplexed lipid. Both above and below the observable transition the elastic area compressibility modulus ( K ) is greatly increased as the cholesterol content is increased (table 2).The value for the 1 : 1 lipid-cholesterol complex was found to be at least 800 dyn cm-', which is comparable to that for DMPC in the Lp crystalline phase. However, for all concentra- tions above 12.5 mol YO (which was only weakly solid), the vesicle bilayers behaved as272 Giant Vesicle Bilayers ' / -0- .o . I I I I T/"C Fig. 2. Vesicle area vs. temperature plots for several mixtures of cholesterol and DMPC. Here, vesicle areas were normalized by values at a common temperature of 20 "C. Also, pure DMPC exhibits two separate area us. temperature relations for the low-temperature phase below 23 "C: the dotted line is for the Ppr phase where the surface is in the form of a superlattice of ripples with the acyl chains oriented normal to the projected plane of the superlattice; the dashed line is the area relation for the same vesicle after the superlattice (ripples) have been eliminated by mechanical stress to form a plane surface with the acyl chains tilted at ca.25-33 O, i.e. Lp, geometry.' Addition of cholesterol eliminates the Pp, phase superlattice and progressively reduces and broadens the transition, but maintains the bilayer in a fluid state. CHOL/(CHOL+DMPC): 0, 0.0; A, 0.125; U, 0.25; 0, 0.33; a, 0.40; A, 0.50. Table 2. Thermoelastic properties of lipid-cholesterol mixtures composition K/dyn cm-' K-' T / "C DMPC ( L a ) 144.9 f 10.5 6.81 f 1.0 29 4.17*0.2 35 DMPC ( Lp,) 855.3 f 140.1 1 .o 8 CHOL (DMPC+CHOL) 0.125 396.9 2.83 15.5 0.33 646.8 1.97 15 559.0 3.1 25 0.40 600 2.3 35 0.50 685 1.33 22 SOPC (La) 199.6 f 12.7 3.28 f 0.68 15 CHOL (SOPC + CHOL) 0.5 865 f 166.6 1.62zk 0.16 15E.Evans and D. Needham 273 16 12 I 3 h m 2 I k- 4 0 0 10 20 30 40 50 fi/dyn‘/2 crn-’I2 Fig. 3. Cumulative plot of tension levels required to ruputure vesicles us. the square root of the elastic area compressibility modulus. The line represents a simple ‘fracture energy’ model based on a critical surface energy beyond which the surface expands without limit. *, DMPC (La phase); 0, EYPC; 0, SOPC; 0, SOPC-CHOL ( l : l ) , DMPC (L,, phase); A, SOPC-CHOL (3 2); 0, SOPC-CHOL-P,6; A, soPc-p,6- liquids with no surface shear rigidity even at temperatures well below the DMPC phase transition. Thus, compared to lipid bilay-ers in the liquid state, introduction of cholesterol forms a tight complex with the lipid which greatly reduces the bilayer compressibility (and permeability as evidenced by the extremely slow response of vesicles to osmotic dehydration), but maintains the liquid-like (fluid) state of the bilayer well below the acyl chain crystallization temperature of the lipid. The addition of cholesterol to lipid bilayers not only reduces the area compressibility but also greatly enhances bilayer cohesion based on observation of the tension levels required to lyse vesicles.This general feature has been seen in all of our vesicle studies, i.e. the tension (mechanical stress) levels at lysis increase as the elastic area compressibil- ity is reduced. Even though the membrane failure must surely be a stochastic process and depend on the time of exposure to the stress, these results are consistent with a simple ‘fracture energy’ model which is based on a critical surface energy, 2, beyond which the surface yields, e.g.1/2Ka2< i? i.e. the lysis tension should be inversely proportional to the square root of the compressi- bility, This simple naive model correlates reasonably well with lysis tensions observed for many vesicle membrane compositions: SOPC-CHOL mixtures, single component lipids in both the solid and liquid phases, and lipid-protein mixtures to be discussed next (fig. 3). Lipid-Protein Mixtures In model^'^-'^ of lipid-protein interaction, the mismatch between hydrophobic lengths of lipid chains and the transmembrane peptide sequence plays a major role in the274 Giant Vesicle Bilayers Table 3.Elastic area compressibility modulus of lipid, lipid- cholesterol and lipid-cholesterol-peptide bilayer mixtures composition K/dyn cm-* T/"C SOPC ( L a ) SOPC-CHOL (3:2) SOPC-peptide,, (5 : 0.05) SOPC-CHOL-peptidel, (3 : 2 : 0.05) DMPC (La) DMPC-CHOL DMPC-CHOL-peptide,, (3 : 2) (3 : 2 : 0.05) 199.6 + 12.7 362.7 f 20.6 161.1 *23.6 168.9 f 24.6 144.9 f 10.5 600 945 f 218 245 1933~31 15 15 15 15 29 35 12.5 f 0.5 29.5 34.4* 1.3 solution-phase behaviour. We present here some results from preliminary experiments on a simple lipid-peptide system where two synthetic amphiphilic peptides were used: 16 or 24 hydrophobic leucine residues bounded by two lysines at the N and C terminals.* Thus, the peptide forms a long hydrophobic a-helix with hydrophilic ends.Both peptides have been shown to broaden and lower the phase transition of DMPC2' The concentra- tion dependence of the transition enthalpies suggests a thermodynamic model which does not involve complexed boundary lipid.20. We have examined the effect of these peptides on area compressibility modulus of both lipid and lipid-cholesterol bilayers; results are listed in table 3. The introduction of peptide,, into SOPC bilayers at a soPc-P16 concentration of ( 5 : 0.05) clearly increases the bilayer compressibility. Even though SOPC-CHOL (3 : 2) is much less compressible than pure lipid, the introduction of peptide at the same total lipid-CHOL-peptide ratio (3 : 2 : 0.05) produces an unexpectedly large increase in bilayer compressibility. The values indicate that the presence of peptide interferes with formation of the lipid- cholesterol complex.Additional evidence that incorporation of the peptide alters the lipid-cholesterol interaction ( i e . 'frees-up' the lipid component at high temperatures) was obtained from measurements of area compressibility on DMPC-CHOL-peptide,, mixtures. At temperatures 15-20 "C above the main DMPC crystallization temperature (24 "C), the lipid-cholesterol-protein vesicles exhibited the onset of a broad but sig- nificant area transition (not present for the lipid-cholesterol mixture); here, the area compressibility was much greater with the peptide present than for the lipid-cholesterol mixture (table 3). As the temperature was reduced, there was a striking reduction in area compressibility where the modulus increased to values even greater than for the lipid-cholesterol mixture.These results indicate that cholesterol appears to change 'partners' as the temperature is increased. In the low-temperature region, cholesterol forms a close cohesive complex with the lipid; at high temperatures, cholesterol associates with the peptide to form islands in the compressible liquid phase of the lipid. Reduction in area compressibility with incorporation of a long peptide and an increase in area compressibility with incorporation of a short peptide is consistent with the lipid wetting model where the lipids adjacent to the 'particle' are either extended or shortened, respectively, with concomitant changes in lipid average area compressibility.* The peptides were gracious gifts from Dr R. Hodges, University of Alberta, and Dr M. Bloom, University of British Columbia.E. Evans and D. Needham 275 Non-specific Interactions between Neutral Lipid Bilayers Based on X-ray diffraction studies of hydration experiments with lamellar lipid-water phases,21,22 it has been concluded that the non-specific colloidal interaction between neutral lipid bilayers is composed of weak van der Waals attraction opposed by strong hydration repulsion. Because of the steep exponential decay of the repulsive hydration forces, the energetics of assembly or adhesion of lipid bilayers can be viewed conceptually as an approach along a soft van der Waals attraction to a specific limit or barrier determined by the magnitude and decay of the strong hydration repulsion.As-such, the free-energy potential for assembly of the bilayers to stable contact ( i e . at the free-energy minimum) essentially represents the van der Waals attractive potential at the equilibrium contact distance. Although the functional form and basic physics of the van der Waals attraction are well establi~hed,23*~~-~~ the strong repulsive hydration force is founded on empirical Dispersive polarization based theories have been proposed which represent hydration forces by a potential field analogous to electrostatic double-layer However, there has been no direct verification of this potential law although recent calorimetric data for lamellar phosphatidy- lethanolamine-water phases at various states of hydration strongly support the theoreti- cal In addition to the primary action of van der Waals attraction and hydration repulsion, repulsion may be enhanced by secondary effects due to thermally excited undulations (bending fluctuations) of the bilayer~.*~.~' These mechanical excitations do not appear to alter greatly the close range stresses dominated by the hydration repulsion; however, there is predicted to be a significant effect on the value of free-energy potential for assembly of bilayers because of the outward displacement of the equilibrium contact distance caused by the slow decay of the 'fluctuation-enhanced' rep~lsion.~' Stable adhesion of bilayer vesicles is promoted by the free-energy potential for bilayer adhesion, but is opposed by mechanical rigidity of the vesicle capsules.31 The variational expression for mechanical equilibrium translates into the familiar Young equation which relates the free-energy potential for membrane assembly ( y ) to the membrane tension, where 8, is the included angle between the membranes.Based on this simple relation, micromechanical experiments are used to directly measure the free-energy potential for assembly of the vesicle membranes to adhesive contact. The procedure is outlined as follows: vesicles are first slightly dehydrated to produce an initial flaccid state in which there is an excess of membrane area (over that for a sphere of equivalent volume). One vesicle is aspirated with a suction micropipette at high pressure such that the membrane tension is large and the vesicle forms a rigid surface.A second vesicle is aspirated with another pipette, but at low, variable suction pressure. The second vesicle is then manoeuvred into the proximity of the rigid vesicle surface where adhesion is allowed to occur in discrete steps controlled by the low pipette suction pressure [shown in plate 1 ( b ) ] . The vesicle membrane tension is determined directly from the pipette suction pressure, pipette radius, and the geometry of the adherent vesicle. From observation of fractional extent of coverage of the rigid vesicle surface, the relation between adherent vesicle tension and adhesion area is directly determined. Based on a simple numerical algorithm for the equilibrium geometry of the vesicle adherent to a rigid sphere, correlation with experimental data provides the free energy of adhesion per unit area of contact formation (i.e. the free-energy potential for assembly of membrane b i l a y e r ~ ) .~ ~ Reversibility of the adhesion process can be directly evaluated by observation of the decrease in contact area as the suction pressure applied to the adherent vesicle is increased. This approach has been used successfully to measure the free-energy potential for adhesion between lecithin bilayers in aqueous salt buffers and in salt buffers containing high concentrations of surface adsorbent macromo1ecules.33i34 Neutral276 Giant Vesicle Bilayers lecithin bilayers in 0.1 moldm-’ NaCl solutions yield values of the order of erg cm-2.33 Correlation of results, both for vesicle adhesion and lamellar lipid-water phase dehydration experiments, with theoretical prescriptions for bilayer-bilayer interactions is based on the free-energy potential given by where Phyd is the coefficient for the hydration repulsion and hhyd is the characteristic decay length; AH is the Hamaker coefficient for the van der Waals attraction and f( z,/ z,) is a weak function of the ratio of the distance between bilayers, z,, to the bilayer thickness, zl, with a value close to unity.22 The last term in the free-energy expression is the free-energy excess due to thermally excited mechanical undulations of the bilayers.” The interlamellar stress is derived from the potential by differentiation with respect to the interlamellar separation, Reversible, equilibrium contact is specified by the stress-free condition (Le. the secondary minimum): (T = 0.The three parameters ( Phyd, Ahy& A H ) are determined by the optimum fit of the interlamellar stress function to dehydration (coupled with X-ray diffraction observations) experiments and by comparison of the free-energy potential at equilibrium separation to the adhesion energy measured in vesicle adhesion tests. It is expected that the van der Waals attraction coefficient and the decay length for the hydration repulsion will be essentially the same for lipids witka similar acyl chain composition. Hence, variations in equilibrium contact distance (and thus adhesion energies) for these lipids will be determined solely by the magnitude of the coefficient for the hydration repulsion, Fig. 4 shows the increase in free-energy potential predicted for adhesion of bilayers (with fixed values for van der Waals coefficient and decay length for the hydration repulsion) as the magnitude of the hydration repulsion is decreased from values representative of neutral lecithin^.^',^^ The solid curve in fig.4 shows the free-energy potential that includes ‘fluctuation-enhanced’ repulsion whereas the dashed line was calculated without fluctuation effects. For both predictions, the stress-free state becomes unstable at low values of Phyd. which indicates collapse to direct molecular contact (elimination of the secondary minimum). Such instabilities may explain the problems associated with hydration of some lipids (e.g. PE). It should be noted that the parameters obtained in the optimum fits to data derived from dehydration of lamellar lipid-water phases and vesicle adhesion tests differ between correlations with and without fluctuation-enhanced repulsion.30 Most significant is the difference in van der Waals attraction coefficient which is ca.two-fold larger for the situation where there is augmented repulsion; also, the decay length for the hydration repulsion is slightly less than that calculated for the situation without fluctuation effects. As a consequence of the increased value of van der Waals coefficient, the free-energy potential for adhesion of bilayers with thermally excited undulations exceeds that of bilayers without undulations when the hydration repulsion is low. Lipid Bilayer Mixtures: Mutual Adherence Properties In order to evaluate the functional form of the van der Waals attraction as well as more subtle aspects of other interactions, adhesion experiments have been carried out on vesicles made from mixtures of phosphatidylcholine and phosphatidylethanolamine, which is known to form lamellar phases at much lower separation than the lecithins.36 By choosing similar chain composition for both lipids (SOPC and POPE), it was hypothesized that the results given in fig.4 could be correlated with adhesion tests for vesicles made from SOPC-POPE mixtures. The basic premise is that the hydrationE. Evans and D. Needham 277 0.15 0.10 N \ E 2 . + 0.05 0.05 4 unstable : i I I A t i i l I I ) 0.5 I .o =/(lo-’’ dyn cm-2)1/2 Fig. 4. Predicted values of the free-energy potential for adhesion of bilayers (with fixed values for van der Waals attraction and decay length for the hydration repulsion) as a function of the magnitude &yd of the hydration repulsion.Values of the parameters (&d, Ahyd and AH) were determined by the optimum fit of a theoretical prescription for the interlamellar stress to data derived from X-ray diffraction studies of lecithin-water phase dehydration and by comparison of the theoretical relation for free-energy potential at equilibrium separation to the adhesion energy measured in lecithin vesicle adhesion tests.22,33,35 The solid curve (hhyd = 2.4 A; AH = 1.4 X erg) shows the predicted free-energy potential that includes ‘fluctuation-enhanced repulsion’, whereas the dashed line (Ahy, = 2.8 A; AH = 0.7 X erg) is the prediction without fluctuation effects.30 Note: the secondary minimum becomes unstable at low values of Phyd, which indicates collapse to molecular contact.Also shown are the separation distances appropriate to this range of hydration repulsion. repulsion is characterized by a surface potential proportional to28 and that the source for the potential resides with each lipid molecule, hence, where it is assumed that the area per molecule is the same for both lipid components. Furthermore, the free energy of interaction is the product of ‘hydration’ surface potentials for the interacting layers,28 i.e. phydx i’278 Giant Vesicle Bilayers 000 I I I I I I I 0 0 0 2 04 06 08 10 12 X,, = POPE/(SOPC + POPE) Fig. 5. Free-energy potential for adhesion of mixed SOPC and POPE bilayers as a function of PE content.Theoretical predictions for free-energy potentials of adhesion both with and without fluctuation-enhanced repulsion were derived from the results in fig. 4 with the prescription that the hydration repulsion is characterized by surface potential proportional to the square root of Phyd and the assumption that the source for the potential resides with each lipid molecule. 0, Data obtained in symmetric vesicle adhesion tests where both bilayers had the same composition; A, data for asymmetric vesicle adhesion tests where the bilayers had separate POPE-SOPC compositions of either 4: 1 or 9: 1 mole ratios (these data are plotted at ‘mean’ concentrations determined by the geometric average of hydration potentials for each surface); *, were derived from experiments where it was observed that pockets of multilamellar lipid adjacent to the vesicle surface could be drawn into the outer vesicle membrane by micropipette aspiration of the vesicle capsule (it is likely that these lipid pockets were separate regions of pure PE).Brackets about average values for the data represent the standard deviations derived from 5 to 10 adhesion tests; data points which do not have brackets imply that the standard deviation was within the size of the symbol used to mark the average value. for symmetric membrane adhesion and Phyd OC $1 $2 for asymmetric membrane adhesion (where the membranes have different compositions). Thus, if the hydration potentials characteristic of each pure component (SOPC and POPE) are known, we can use the above relations with theoretical results like those presented in fig.4 to predict the free-energy potential for adhesion of bilayers with any mixture composition. Based on established values for separation distances between PE bilayers in the La state (and the self-consistent observation that PEs have limited lamellar phase stability), we can use the lower limits of Phyd illustrated in fig. 4 to characterize the ‘hydration’ surface potential for POPE. With this value and the value of Phyd determined previously for lecithins, we predict the free energy potential for adhesion as a function of POPE/(SOPC+POPE) mole ratio (fig. 5). Also shown in fig. 5 are data obtained from vesicle-vesicle adhesion tests with both symmetric and asymmetric vesicle compositions.The free-energy potentials for adhesion predicted both with and without fluctuation-enhanced repulsion clearly exhibit the same characteristic depen- dence on PE mole fraction as results from the symmetric vesicle adhesion tests (fig. 5 , open circles). Comparison of the theoretical predictions with observed adhesion energies indicates that the van der Waals attraction and hydration re ulsion must be slightly modified for the close-range interactions (between 10 and 20 w , which is the resultantE. Evans and D. Needham 279 effect from the fluctuation-enhanced repulsion model). The results strongly indicate that, for situations where both vesicles have the same composition, the free-energy potential for adhesion of the bilayers can be predicted from a surface potential function which is given by a simple summation of the products of area fractions times ‘hydration’ potentials for each pure component (analogous to surface charge density dependence in electrostatic double layer theory).Additional support for a potential-based theory was obtained from asymmetric adhesion tests which are also shown in fig. 5 as open triangles plotted at ‘mean’ concentrations determined by the geometric average of ‘hydration’ potentials calculated for each surface. (The asymmetric tests involved pure SOPC vesicles and POPE-SOPC vesicles with either 4 : 1 or 9: 1 mole ratios.) Thus, for asymmetric situations where the vesicles have different compositions, the free-energy potential for adhesion can be predicted from the geometric mean of ‘hydration’ surface potentials.For PE contents >go%, there was a large variance in the measured adhesion energies which indicated compositional variation from vesicle to vesicle (note: * 5 YO variation in vesicle composition accounts for the spread in adhesion energies). Also, we frequently observed vesicle fusion events with compositions above 90% PE. This behaviour is consistent with the steep rise in adhesion energy at low Phyd with the subsequent collapse to molecular contact predicted by the theoretical calculation. This work was supported by the Medical Research Council of Canada through grant MT 7477. D.N. was supported in part by a NATO/S.E.R.C. post-doctoral fellowship. References 1 E. Evans and R. Waugh, J. Colloid Interface Sci., 1977, 60, 286.2 E. Evans and R. Skalak, Mechanics and Thermodynamics of Biomembranes (C.R.C. Press, Boca Raton, 3 R. Kwok and E. Evans, Biophys. J., 1981,35, 637. 4 E. Evans and R. Kwok, Biochemistry, 1982, 21, 4874. 5 D. Needham and E. Evans, Biophys. J., (to be submitted); Biophys. J., 1985,47a. 6 E. Evans and D. Needham, Biophys. J., (to be submitted); Biophys. J., 1985, 47a; E. Evans, in Adu. 7 E. Evans, in The Rheology of Blood, Blood Vessels and Associated Tissues, ed. Hwang and Gross (Sitjhoff 8 E. Evans, Biophys. J., 1982, 43, 27. 9 R. Waugh and E. Evans, Biophys. J., 1979, 26, 115. Florida, 1980), p. 272. Solid State Phys., 1985, xxv, 735. and Noordhoff Int, 1981), pp. 1-23. 10 G. N. Lewis and M. Randall, Thermodynamics (McGraw-Hill, New York, 1961), p. 723. 11 E. J. Shimshick and H. M. McConnell, Biochemistry, 1973, 12, 2351. 12 D. Chapman, J. Urbina and K. M. Keough, J. Biol. Chem., 1974, 249, 2512. 13 A. Blume, R. J. Wittebort, S. K. Das Gupta and R. G. Griffin, Biochemistry, 1982, 24, 6243. 14 S. Mabrey, P. L. Mateo and J. M. Sturtevant, Biochemistry, 1978, 17, 2464. 15 M. R. Vist and J. H. Davis, Can. Biophys. SOC. Abs. (Banff, 1984). 16 F. T. Presti, R. J. Pace and S., I. Chan, Biochemistry, 1982, 21, 3831. 17 S. Marcelja, Biochem. Biophys. Acta, 1976, 455, 1. 18 J. C. Owicki, M. W. Springgate and H. M. McConnell, Roc. Natl. Acad. Sci. USA, 1978, 75, 1616. 19 0. G. Mouritsen and M. Bloom, Biophys. J., 1984, 46, 141. 20 M. R. Morrow, J. C. Huschilt and J. H. Davis, Biochemistry, in press. 21 D. Le Neveu, R. P. Rand, D. Singell and V. A. Parsegian, Biophys. J., 1977, 18, 209. 22 V. A. Parsegian, N. Fuller and R. P. Rand, Proc. Natl. Acad. Sci. USA, 1979, 76, 2750. 23 E. J. W. Verwey and J. Th. G. Overbeek, Theory of the Stability of Lyophobic Colloids (Elsevier, 24 V. A. Parsegian, Annu. Rev. Biophys. Bioeng., 1973, 2, 221. 25 J. Marra and J. Israelachvili, Biochem., 1985, 24, 4608. 26 D. W. R. Gruen and S. Marcelja, J. Chem. SOC., Faraday Trans. 2, 1983, 79, 225. 27 B. Jonsson and H. Wennerstrom, Chem. Scr., 1983, 22, 221. 28 G. Cevc and D. Marsh, Biophys. J., 1985,47, 21. 29 W. Helfrich, 2. Naturforsch., Teil a, 1978, 33, 305. 30 E. Evans and V. A. Parsegian, Proc. Natl. Acad. Sci. USA, in press. 31 E. Evans and V. A. Parsegian, Ann. N. Y. Acad. Sci., 1983, 416, 13. Amsterdam, 1948), p. 205.280 Giant Vesicle Bilayers 32 E. Evans, Biophys. J., 1980, 31, 425. 33 E. Evans and M. Metcalfe, Biophys. J., 1984, 46, 423. 34 E. Evans and M. Metcalfe, Biophys. J., 1984, 45, 715. 35. L. J. Lis, M. McAIister, N. Fuller, R. P. Rand and V. A. Parsegian, Biophys. J., 1982, 37, 657. 36. J. M. Seddon, K. Harlos and D. Marsh, J. Biol. Chem., 1983, 258, 3850. 37 V. A. Parsegian and D. Gingell, J. Adhesion, 1972, 4, 283. Received 19th December, 1985

ISSN:0301-7249    

DOI:10.1039/DC9868100267

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

Faraday Discuss. Chem. Soc., 1986, 81, 281-290 Membrane Bending Elasticity and its Role for Shape Fluctuations and Shape Transformations of Cells and Vesicles Erich Sackmann,* Hans-Peter Duwe and Hans Engelhardt Physics Department (Biophysics Group), Technical University of Munich, 0-8046 Garching, Federal Republic of Germany Experimental evidence for the essential role of the curvature elasticity and the membrane-asymmetry-controlled spontaneous curvature for shape fluctuations and shape transitions of liposomes and erythrocytes is presented. Non-spherical flaccid vesicles can form a limited number of stable shapes such as discocytes (disc-shape), stomatocytes (cup-shape) two-vesicle states (small vesicle inside or outside large one), or beaded chains of vesicles. Transformations between these shapes are triggered by changes in the area difference between the inner and the outer monolayer in agreement with the bilayer-coupling model of Svetina and Zeks.Improved mechanical models of the red-blood-cell membrane have been prepared by incorporation of linearly polymerized macrolipids into vesicles, which then exhibit similar elastic constants as the RBC. The formation of two-vesicle states by changes in the global spontaneous curvature suggests that important cellular transport processes such as endocytosis and exocytosis may be triggered by biochemically induced expansions or compressions of one monolayer of the plasma membrane. The essential role of lateral redistribution processes for the final triggering of the vesicle detachment is stressed.Shape fluctuations, shape transformations and membrane instabilities associated with the migration of cells or the transmembrane transport by vesicles are both fascinating and of upmost biological importance. An example of a membrane instability is the formation and expulsion of a vesicle from the cell envelope for the purpose of transport- ing products out of the cell, called exocytosis. Another is the reverse process, the endocytosis (or phagocytosis) by which material (or foreign particles) attached to the membrane surface is taken up by cells. Both types of instabilities are initiated by a local change in curvature leading to an invagination (in the case of endocytosis) or a protrusion as is illustrated in fig. l ( b ) (below). It is an intriguing and important task of modern membrane biophysics to explain the above cellular events in terms of the phenomenological physical properties of the plasma membrane and its biochemically triggered structural changes.It is the main purpose of this contribution to provide experimental evidence from model membrane studies that the shape changes and membrane instabilities can be explained in terms of very simple principles, namely the bending elasticity and intrinsic curvature of the cell plasma membrane as suggested by several theoretical studies.lP3 Plasma membranes are composed of three coupled layers: (1) the lipid-protein double layer, (2) a quasi-two-dimensional network of protein filaments at the cytoplas- matic side and (3) a gel-like sheet at the outer side called the glycocalix. The glycocalix is formed mainly of the large head groups (peptide/saccharide-copolymers) of the glycoproteins and the oligosaccharides of the glycolipids and controls the cell-cell contact.The membrane-coupled cytoplasmatic meshwork is only two-dimensional for erythrocytes as shown in fig. 1( a), whereas it spans the whole cytoplasm in the case of other cells. Nevertheless one can find a layer coupled to the bilayer also in these cases. 28 1282 Bending Elasticity and Cell Shape Fig. 1. ( a ) Cartoon-like representation of cell plasma membrane of erythrocytes composed of the lipid-protein bilayer at the centre, the glycocalix at the outer side and the cytoskeleton at the cytoplasmatic side of the cell envelope. The latter is composed of spectrin filaments (cu.1000 8, long) and actin oligomers which are interconnected to form a network which is coupled to membrane-bound proteins (band 111 and glycophorin) by coupling proteins (such as ankyrin).' Recent model membrane studies provided strong evidence that the rather flexible and positively charged spectrin filaments bind also electrostatically to the phosphatidylserin which is mainly distributed in the inner monolayer.20 Most probably, the spectrum binding plays an important role in the stabilization of the asymmetric lipid distribution.8 (b) Illustration of local protrusion of cell plasma membrane caused by decoupling of cytoskeleton from the lipid-protein bilayer. The cytoskeleton is responsible for the high global stability of cells which allows them to squeeze through very narrow channels without damage.Simultaneously, it allows for instabilities on a local scale such as membrane-vesicle fusion or the process of exocytosis which is illustrated in fig. l(b). The latter process is initiated by a local change in curvature which could be caused by a local (for instance biochemically induced) detachment of the cytoskeleton from the lipid-protein bilayer. In the model membrane experiments presented below we provide evidence for this view.E. Sackmann, H-P. Duwe and H. Engelhardt + 7 . 8 P 4 discocyte 283 s t o m ato cy te echinocyte sp h erocyte Fig. 2. Shape transformation of the red blood cell. The dimensions of the discocyte correspond to a solution of isotonic osmolarity (300 m o ~ m ) .~ The relative volume (with respect to the spherical shape) is V/ V, = 0.6. The discocyte-stomatocyte transition is induced by a pH decrease or addition of positive amphiphatics:8 the discocyte-echinocyte transformation by a pH increase or addition of negative amphiphatics and cholesterol depletion.4p8 The spherocyte is formed by osmotic swelling. Shape Transformations of Cells Fig. 2 presents a second fascinating property of cell membranes: the shape transforma- tions. These are most clearly observed and studied for the case of erythrocytes. Under normal physiological conditions the cell is shrunk, i.e. its volume is smaller than that of a sphere of the same surface area. Although the cell could thus assume any shape it always forms a discocyte unless it is sick or the cellular environment is changed.However, even then only a limited number of shapes is observed. These are: the cup-shape (stomatocyte), the sea urchin or crenated shape (echinocyte) and the spherical cell (spherocyte): which are summarized in fig. 2. Since the cell interior contains only haemoglobin the ability of forming a limited number of cell shapes must be a membrane property. It is an often debated question whether the stabilization of a certain cellular shape is achieved by the cytoskeleton or by the b i l a ~ e r . ~ . ~ One piece of experimental evidence for the first view is that removal of the lipid-protein bilayer by Triton XlOO retains the discoid shape. On the other hand, a number of theoretical papers appeared which explained the cell shape in terms of the curvature elastic properties of the membrane. First Canham3 showed that the discoid shape corresponds to a minimum value of the membrane-bending energy.Later Deuling and Helfrich2 introduced the concept of spontaneous curvature and showed that each shape corresponds to a minimum of the bending energy, but a different value of the spontaneous curvature C,: which is for instance negative ( C , = -0.6 pm) for discocytes and positive for echinocytes. Recently, Svetina and Zeks' combined the concept of the spontaneous curvature with the so-called bilayer coupling hyp~thesis,~ which states that the two leaflets of the bilayer can change their area in a different way in response to an external perturbation (such as a pH change). These authors could explain the different cell shapes in terms284 Bending Elasticity and Cell Shape fi AAlAA, Fig.3. Phase diagram of cell shapes according to bilayer-coupling model of Svetina and Zeks.' The lines give the pair of V / V, and AAIAA, values for which the shapes indicated by small letters correspond to a bending energy minimum. of two phenomenological parameters: ( 1 ) the reduced volume of the cell, V/ V,, where V, is the (maximum) volume of a sphere of given membrane area A; and (2) the relative difference in area AA/AA,= (Ao-Ai)/AA, between the outer and the inner leaflet of the membrane, where AA, is the corresponding value for the spherocyte. By minimizing the total bending elastic energy (Ryl+ R;') dA (where R,' and R;' are the principal radii of curvature) under the assumption that the reduced volume V / V, and area difference AA/AA, are held constant these authors could establish a phase diagram of the cell shapes which is given in fig.3. The interesting results are: (1) A non-spherical shell can divide up into two, three or more spheres, but only two different radii can coexist. The smaller spheres can protrude from or invaginate into the larger one. (2) For the relative volume of the erythrocyte V/ V, = 0.6, the discoid shape corresponds indeed to a bending energy minimum. Below we present experimental evidence for the validity of the Helfrich-Svetina-Zeks model. Model of Erythrocyte Membrane The above concept' is based on the idea that the two halves of a plasma membrane can expand or contract independently, but that the shape of a cell changes if the area of one monolayer is changed owing to the coupling of the two leaflets at the centre of the bilayer.In this picture the cytoskeleton is part of the inner leaflet and a change of its structure or its coupling to the inner monolayer will lead to a shape change of the cell. The recent introduction of polymerizable lipid opens the possibility of constructing mechanically equivalent models of plasma membranes.'o711 One class of such lipids carries functional groups at one end of a spacer molecule which is attached to the polar head group of the lipid by the other end [cJ: plate 3 (later)]. Giant vesicles composedE. Sackmann, H-P. Duwe and H. Engelhardt 285 Fig. 4. Vesicle composed of mixture of a monomeric and a cross-linked lipid bilayer may be considered as two-dimensional macromolecular solution. Polymerizable lipids containing one functional group (x) (right side) form linear macrolipids.This may in addition be cross-linked by adding some percent of a polymerizable amphiphile with two functional groups. This two- dimensional solution of a cross-linked macrolipid may be considered as a mechanical model of erythrocyte. of two-dimensional solutions of linearly polymerized lipids in normal phospholipids ( cJ: fig. 4) can be prepared by swelling of appropriate mixtures of the two types of lipids and subsequent photopolymerization of the vesicles formed by the swelling. Such partially polymerized lipid layers are first of interest as model systems for two- dimensional macromolecular solutions and exhibit some typical mechanical features of plasma mernbranes.l0 More realistic models can be prepared by addition of a small amount of a lipid with two functional groups per head group.Then the linear macrolipids are also interconnected (cJ: fig. 4) forming a two-dimensional network, similar to the spectrin-actin network. Thermally Excited Shape Fluctuations and Measurement of Curvature Elastic Constant Non-spherical vesicles and cells exhibit pronounced shape fluctuations which can be well observed under a phase-contrast microscope and which for erythrocytes leads to the phenomenon which is known as 'the fli~kering'.'~.'~ These shape fluctuations are thermally excited and can be exploited for very accurate and undisturbed measurements of the curvature elastic constants of plasma membranes12 and pure lipid b i l a y e r ~ .' ~ ~ ' ~ Plate 1 shows as an example two images of a flaccid vesicle taken at a time interval of 1 s and the difference in the contours at the two times can be clearly seen. Recently, a fast computer-controlled image processing system has been developed, which enables a quantitative evaluation of the shape changes in terms of a Fourier expansion of the contour R ( 4 ) - R o w =c uq exP(-iq4) 4 where R ( 4 ) is the instantaneous radius vector of the contour as measured from the centre of the average contour Ro(+) as a function of the rotational angle 4 (cJ plate 1 for definition of parameters). A method has been established" which allows determina- tion of the curvature elastic constant by measurement of the mean-square amplitudes (u',) of the individual Fourier components of wavevector q ; although the shape fluctu- ation of the shell has to be expanded in terms of spherical harmonics Y1,,, ( 0 , 4 ) .The286 Bending Elasticity and Cell Shape 1 5 10 15 Fig. 5. Plot of curvature elastic constant K , of vesicle of a solution of 20 mol % of polymerized lipid [(x, )I)-POMECY cJ plate 3 0 1 in dimyristoylphosphotidylcholine (DMPC) as obtained by the Fourier analysis of fhe shape changes. Measurements performed at 32 "C. The second mode ( q = 2) zxhibits a considerably smaller value of K , , which has been attributed to surface tension effects. Above the eighth mode the relaxation times of the surface excitations become smaller than the response time of the camera and K , increases apparently.Above the tenth mode the amplitudes u, become smaller (u, < 10 nm) than the spatial resolution of the video system and have been truncated. elastic constant, K , , is related to the mean-square amplitude of each mode q according to where N ( q ) is a numerical factor which can be easily calculated from an equation derived in a previous paper [eqn (6) of ref. (15)]. At present the contour fluctuations can be measured up to the 10th order, corresponding to a resolution of q-' = 1.000 A. This means that K , is simultaneously measured 10 times by each experiment and can thus be determined very accurately. Fig. 5 shows the result for a solution of a linearly polymerized macrolipid (of the structure shown in plate 1) in DMPC at a concentration of 20% of polymerizable lipids.For pure DMPC the bending elastic constant is K , = 3.5 x erg and is thus ca. a factor of two smaller than that for the partially polymerized vesicle. It is remarkable to note that the value of the curvature elastic constant of the erythrocyte membrane is also a factor of two larger than for DMPC, which provides some evidence that the partially polymerized vesicle may be considered as mechanical model of red blood cells. Shape Transitions of Giant Bilayer Vesicles The Helfrich-Svetina-Zeks model is attractive because it explains the stability of a limited number of cellular shapes in terms of a very simple concept. In addition it strongly suggests that a change in the global intrinsic curvature may trigger localized processes such as exo- or endo-cytosis owing to the transitions to two-vesicle states (cJ: fig.3). In the following we report shape changes of simple bilayer vesicles which provide strong experimental evidence for the model of Zeks and Svetina. Plate 2 shows an example of shape transformations of a vesicle composed of one lipid component which is associated with a phase transition of the bilayer. The lipidPlate 1. Images of a flaccid vesicle taken at time intervals of 1 s. (To face p. 286)Plate 2. Shape transformation of vesicle of lipid with positively charged (ammonium) head group which was studied in ref. (10). The transitions are caused by variation of the temperature in the neighbourhood of the lipid phase transition (onset of transition at temperature T, = 41 "C).(a)-( 6)-( c): ( a ) Shape transformation from discoid shape to two-vesicle state at increasing the temperature. The process is reminiscent of exocytosis. (a)-( d)-( e): Shape transformation of the same vesicle from discoid to two-vesicle shape triggered by decreasing the temperature, reminiscent of process of endocytosis. Note the small piece of non-swollen lipid in (d) and (e), which is taken up by the vesicle. T/"C: ( a ) 42, (6) 42.5, (c) 43, ( d ) 41.5, (e) 41.Plate 3. Endocytosis-like shape change of vesicle composed of DMPC and a polymerizable lipid," 4.16-POMECY, of the structure shown in fig. 6(b) caused by polymerization of this lipid. ( a ) Mixture of 80 mol % of DMPC and 20% of 4,16-POMECY after swelling in 50 mmol dm-3 mannitol solution at 29 "C.(b), ( c ) : Transformation occurring after 2 min of U.V. irradiation with a low-pressure mercury lamp. The irradiation was terminated before (b) and ( c ) were taken. The polymerization process leads to a shrinking of the vesicle.Plate 4. Demonstration of metastability of shapes of vesicles. Transition of a vesicle from quasi-spherical to two-sphere state is accomplished by slight mechanical agitation of the vesicle kept between microslides. ( a ) Vesicle at resting state; (6) two-vesicle state obtained after slightly knocking the cover glass; ( c ) same vesicle after 1 min, when it has rotated by 90 "C; (d) transient shape after a new mechanical agitation; (e) re-stabilization of two-vesicle state.E. Sackmann, H-P.Duwe and H. Engelhardi 287 possesses a positively charged (ammonium) head group and unsaturated hydrocarbon chains. Bilayers of this lipid exhibit a broad fluid-to-solid (gel) transition at T, = 41 "C. The vesicle in plate 2(a) was formed by swelling the lipid at 50 "C. If it is heated (at a rate of 1 K min-') it splits into two adhering vesicles of different radius at ca. 43 "C. Decreasing the temperature again the shape transformation is first completely reversed. An invagination appears if the vesicle is cooled below the starting temperature, e.g. to T=41 "C. The shape changes are completely reversible if the temperature cycle is repeated. These remarkable shape transformations can be explained in terms of a difference in the charging of the surfaces of the inner and outer monolayers.First it should be noted that owing to a higher lipid packing density the surface potential and therefore the surface concentration of counterions at the lipid-water interface is higher in the crystalline than in the fluid state. Secondly, if ions dissociate from the surface, the ionic strength in the inner (confined) volume will increase much more than that of the (infinitely large) outer medium. If we start from a discocyte [plate 2(a)] and increase T more ions will dissociate from the outer surface than from the inner. Consequently, the outer monolayer will expand with respect to the inner one. This results in a positive spon- taneous curvature and according to the Svetina-Zeks model, in the splitting of an outer vesicle. Decreasing T below the starting point, fewer ions can adsorb to the inner monolayer than to the outer owing to the confined number of counterions in the inner volume.This will result in an relative lateral compression of the outer monolayer. The decrease in AAIAA, will lead to an invagination according to fig. 3. It is important to note that the vesicle of this lipid is still fluid at the temperature of plate 2( e), exhibiting pronounced thermally excited shape fluctuations. Another point to be emphasized is that the lipid is polymerizable owing to the unsaturated C-C bonds," and in order to observe the above effect the vesicles have to be annealed for some hours. Therefore it cannot be excludea that some of the lipid has polymerized, so that the bilayer may not be a pure one-component system.Another shape transformation which is triggered by polymerization in a mixed bilayer composed of 80 mol O/O of a conventional lipid (DMPC) and 20% of a polymerizable lipid is shown in plate 3 and fig. 6. The latter lipid [called 4,16-POMECY] is a phospholipid to the phosphate head group of which a spacer is attached, which carries a polymerizable (methacryl) group at the other end. The spacer has about the same length as the hydrocarbon chains and is therefore rather bulky. Plate 3 shows the shape transformation of a discoid vesicle caused by polymerization of the POMECY by U.V. irradiation. First the average area of the vesicle shell decreases, as indicated by the transition to a more swollen shape. Simultaneously, an invagination is formed, resulting in a smaller vesicle within a larger one.Note that the transition from ( b ) to ( d ) occurs within 2 min after termination of the photopolymerization. The shape transformation can again be explained in terms of the Svetina-Zeks model as follows: first it is known that POMECY forms linearly connected macrolipids [cJ: plate 2(f)] and that polymerization leads to a reduction in the molecular area." Secondly, owing to the large size of the head group, the concentration of POMECY is higher in the outer leaflet than in the inner. This is also suggested by small-angle neutron scattering experiments (W. Pfeiffer and W. Knoll, unpublished results). For the above reasons, polymerization will lead to a larger decrease in area of the outer monolayer, which, according to fig.3, is expected to cause an invagination. In several cases we did not succeed in producing shape transformations, although the spontaneous curvature was changed, for instance, by the addition of ions to one side of a charged vesicle. One reason is that we were too far from one of the phase boundaries of fig. 3. Another is that the transformations are hindered by a high energy barrier between two shapes. Evidence for the latter comes from the experiments presen- ted in plate 4, which show that shape transformations can be triggered by mechanical288 Bending Elasticity and Cell Shape b- Na+ (m,n)-POMECY m=4; n =16 Fig. 6. (a) Possible explanation of shape transition in terms of a larger reduction of the area per lipid molecule in outer monolayer as compared with the inner leaflet.This effect is due to asymmetric distribution of POMECY, which is preferentially distributed in outer monolayer due to large head group. (b) The structure of 4,16-POMECY. agitation. In this experiment a transition of a pure DMPC vesicle from a quasi-spherical shape ( a ) to one with a small vesicle inside a large one [(b) and ( c ) ] occurs after knocking lightly on the cover glass. After knocking a second time a further invagination appears for a short time [top of plate 4 ( d ) ] then the vesicle relaxes to the original two-vesicle state. Also in other cases we observed that shape transformations occurred only under non-equilibrium situations, e.g. in temperature gradients. Concluding Remarks and Biological Implications The present work provided experimental evidence that the bilayer bending elasticity in combination with the asymmetry of the two coupled monolayers can account for the stabilization of a limited number of cellular shapes, in agreement with the Helfrich- Svetina-Zeks hypothesis.It appears that the shape is determined solely by the ratio ( VAAJAAV,) of the relative volume to the relative area difference. In particular, flaccidE. Sackmann, H-P. Duwe and H. Engelhardt 289 bilayer vesicles may assume two of the well known erythrocyte shapes, the discocyte and the stomatocyte, as well as the two types of two-vesicle states predicted by the Svetina-Zeks model. It appears that the discoid shape is the state of lowest bending energy for small values of AA/ AA,, in agreement with the theory.Another often observed shape is the beaded chain of small vesicles predicted by Helfrich and Deuling.2 One particularly interesting aspect of the present model membrane study is that shape transformations may be triggered by slight changes in the area difference (or spontaneous curvature). The finding that a change in a global membrane property such as the spontaneous curvature may lead to the protrusion (or invagination) of a small vesicle is most attractive from the point of view of cellular processes such as exocytosis or endocytosis. It suggests that these processes may be triggered by a change in the environment. Of course, the second step in these processes, namely the detachment of the vesicles, requires an additional biochemical process.In biological membranes additional features come into play: (i) the coupling of part of the cytoskeleton to the inner monolayer of the plasma membrane most likely plays an essential role in the maintenance of the asymmetric distribution of the lipid com- ponents between the inner and the outer monolayer.' The cytoskeleton therefore helps to control the area difference AA and thus the stabilization of a certain cell shape. (ii) Any local curvature may be rapidly stabilized by a lateral reorganization of the membrane components; for instance by the accumulation of lipids or membrane proteins with large head groups at the outer leaflet of a protrusion.'6717 In fact, this may happen in the partially polymerized bilayer of plate 3. (iii) Local protrusions or invaginations may be stabilized by the adsorption of macromolecules to the membrane surface.An example is the coated pit, which is stabilized by the adsorption of clathrin to the inner monolayer. (iv) Owing to the coupling to the cytoskeleton, plasma membranes exhibit also shear elasticity which could also play a role in the stabilization of a certain cell shape.*8319 We have seen that the shape transitions of vesicles and cells can be triggered by slight changes in the environment or the composition of the membrane. This demon- strates that subtle biochemical events may be transformed into drastic structural changes. It is clearly a consequence of the two-dimensionality of the membrane. No such effects could be achieved in three-dimensional systems. This work is part of a research project on polymerized membranes supported by the Deutsche Forschungsgemeinschaft (Sa 246/ 17- 1).Additional support by the L.-Lorentz- Stiftung and the Fonds der Chemischen Industrie is also acknowledged. The authors are also most grateful for helpful discussions with Svetina and Zeks from Ljubljana. This paper was written during a visit to the Biophysics Institute of the Academia Sinica. One of the authors (E.S.) wishes to thank the Academia Sinica for its hospitality on this occasion. References 1 S. Svetina and B. Zeks, Biomed. Biochem. Acta, 1983, 42, 86. 2 H. J. Deuling and W. Helfrich, Biophys. J., 1976, 16, 861. 3 P. B. Canham, J. Theor. Biol., 1970, 26, 61. 4 B. Deuticke, Biochim. Biophys. Acta, 1968, 163, 494. 5 M. P. Sheetz and S. J. Singer, Proc. Natl. Acad. Sci. USA, 1974, 71, 4457. 6 M. Bessis in Living Red Blood Cells and their Ultrastructure (Springer-Verlag, Berlin, 1973). 7 B. W. Shen, R. Josephs and L. Steck, J. Cell Biol., 1984, 99, 810. 8 C. W. M. Haest, G. Plasa, D. Kamp and B. Deuticke, Biochim. Biophys. Acta, 1978, 509, 21. 9 A. Zachowski, P. Fellman and P. F. Devaux, Biochim. Biophys. Acta, 1985, 815, 510. 10 H. Gaub, R. Buschl, H. Ringsdorf and E. Sackmann, Biophys. J., 1984,45, 725. 11 E. Sackmann, P. Eggl, C. Fahn, H. Bader, H. Ringsdorf and M. Schollmeier, Ber. Bunsenges. Phys. 12 K. Fricke and E. Sackmann, Biochim. Biophys. Acta, 1984, 803, 145. 13 K. Fricke and E. Sackmann, Eur. Biophys. J., submitted for publication. Chem., 1985, 89, 1198.290 Bending Elasticity and Cell Shape 14 M. B. Schneider, J. T. Jenkins and W. W. Webb, J. Phys., 1984,45, 1457. 15 H. Engelhardt, H. P. Duwe and E. Sackmann, J. Phys. Lett., 1985, 46, 395. 16 A. Perov, S . A. Selezner and A. Derzhanski, Acta Phys. Polonica, 1979, ASS, 385. 17 E. Sackmann, in Biological Membranes, ed. D. Chapman (Academic Press, London, 1984), vol. 5, chap. 18 A. E. Evans and R. Skalak, in Mechanics and Thermodynamics of Biomembranes (CRC Press, Boca 19 V. S. Markin, Biophys. J., 181, 36, 1. 20 E. Sackmann, Sui Sen-fang, R. Maksymiw and T. Uromov, in Biomembrunes on Receptor Mechanisms 3, pp. 106-143. Raton, Florida, 1980). (Catania Conference), ed. E. Bertoli and D. Chapman, to be published. Received 16th December, 1985

ISSN:0301-7249    

DOI:10.1039/DC9868100281

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

Faraday Discuss. Chem. SOC., 1986, 85, 291-301 Adsorption of Phospholipid Vesicles on Solid Surfaces Simon Jackson, Miguel D. Reboiras,? Ian G. Lyle and Malcolm N. Jones* Department of Biochemistry, University of Manchester, Manchester M 13 9PL and Unilever Research Laboratories, Bebington, Wirral, Merseyside L63 3JW Sonicated vesicles have been prepared from dipalmitoylphosphatidylcholine (DPPC) and phosphatidylinositol (PI) and their mixtures covering a range of composition. The adsorption of lipid from the vesicle dispersions onto the surfaces of glass beads has been measured by a batch procedure using a radiochemical assay. Lipid deposition onto the glass surface occurs uia vesicle adsorption rather than monomeric lipid adsorption and a fluorescent assay has been used to demonstrate that the vesicles disrupt on contact with the glass.The adsorption isotherms are of the Langmuir type and for pure DPPC the limiting areas at the glass-aqueous interface are 0.39 nm2 molecule-' and 0.64 nm2 molecule-' at 25 and 50 "C, respectively. These figures suggest that a monolayer of DPPC is formed at the interface. Limiting adsorption could not be correlated with the electrokinetic properties of the vesicles and was not markedly dependent on the acyl chain length for a series of diacylphosphatidylcholines. The physical and chemical factors which determine the deposition of lipid vesicles from aqueous media onto solid surfaces have not been studied in depth apart from a few studies of liposome adsorption on clay,' asbestos2y3 and Biobeads4 and incidental references to liposome adsorption by gel filtration5 columns and membrane filters.6 From the known characteristics of vesicles viewed as colloidal particles and of solid-aqueous interfaces it might be envisaged that vesicle deposition would be deter- mined initially by the classical combination of a repulsive force arising from the interaction of the electrical double layers associated with the vesicle and the solid surface and the attractive dispersion force between the vesicle and the solid.Vesicles are not, however, permanent rigid structures, and depending on their size and chemical composi- tion and that of the aqueous medium they can distort, aggregate, disrupt and fuse with each other. Deposition of vesicles onto a solid surface could give rise to any particular one or a combination of these processes.Horn' reports that unilamellar phos- phatidylcholine vesicles break open and adhere to a mica surface to form a bilayer coating, although the evidence for this was indirectly obtained from the measured separation between two such surfaces when pushed together. Further compression of the closely apposed bilayers can result in fusion into a single bilayer. Neither the mechanism of the bilayer formation nor the fusion process is fully understood. In an attempt to understand the process of vesicle deposition on solid surfaces we have investigated adsorption of lipid from sonicated phospholipid vesicles prepared from mixtures of dipalmitoylphosphatidylcholine (DPPC) and phosphatidylinositol (PI) on glass beads.Experimental Materials L-a-Dipalmitoylphosphatidylcholine (DPPC), ca. 99% pure (product number P 0763), L-a-dimyristoylphosphatidylcholine (DMPC), ca. 98% pure (product t Present address: Departmento de Electroquimica, Facultad de Ciencias, Universidad Autonoma, Canto Blanco, Madrid 28034, Spain. 29 1292 Vesicle Adsorption number P 0888), and L-a-dilauroylphosphatidylcholine (DLPC), ca. 99% pure (product number P 1263), were obtained from the Sigma Chemical Co. Ltd, Poole, and were used as supplied. Phosphatidylinositol (PI) was obtained as the sodium salt, molecular weight 846; from Lipid Products, South Nutfield. L-a-Dipalmitoylphos- phatidyl( N-[ 3H]methyl)choline (code TRK 673) was obtained from Amersham Inter- national. Ballotini glass beads (high index, barium titanate) were obtained from Potters Ballotini Ltd, Barnsley, and were sieved using wire mesh sieves to obtain a sample of beads with diameters in the range 63-75 pm. The distribution of bead diameters was measured by image analysis and the beads were found to have a mean diameter of 67.44 f 7.66 pm.Dipicolinic acid was from the Sigma Chemical Co. Ltd (product number D-0795) hydrated terbium( 111) chloride, 99.9% pure (product number 6965-00), and sodium deoxycholate (product number 5039-00) were from Koch-Light, Colnbrook. All other reagents were of analytical grade. The buffer of pH 6.6 was prepared from 20 mmol dm-3 imidazole ( ~ K ~ 6 . 9 5 3 ) and hydrochloric acid and had an ionic strength of 0.0138 mol dm-3. Methods Vesicle Preparation Vesicles were prepared by adding the required amounts of phospholipid stock solutions in chloroform (total volume 10 cm3, total lipid concentration 3 mg ~ m - ~ ) plus 2.0 pCi [3H]DPPC together with 50 cm3 of chloroform-methanol mixture (4: 1 v/v) to a 1 dm3 round-bottomed flask.The solvent was removed by rotary evaporation at a temperature of 50-55°C. The resulting homogeneous lipid film was flushed with nitrogen and hydrated by addition of 9.5 cm3 of imidazole buffer, pH 6.6, previously saturated with nitrogen and heated to 60°C. The resulting suspension was hand-shaken to produce multilamellar liposomes, sealed under nitrogen and sonicated in a Decon FSlOO frequency-sweep sonicator for 1 h at 60°C. The small unilamellar vesicles were then centrifuged at 105g for 1 h using an MSE Prepspin 75 centrifuge at 25 "C to remove any remaining multimellar liposomes.The supernatant containing the unilamellar vesicles was used within 1 h of preparation. Treatment of Glass Beads The Ballotini glass beads were cleaned using the following protocol: 0.2 kg batches of beads were successively washed with 2 x 20 cm3 aliquots of methanol, 2 x 20 cm3 aliquots of chloroform followed by 1 x 20 cm3 of methanol. Following solvent evaporation they were again washed with 3 x 25 cm3 aliquots of 50% (v/v) nitric acid and finally with 10 x 50 cm3 aliquots of distilled water. The beads were then dried at 110 "C for 12 h allowed to cool to room temperature before use. Adsorption Measurements Vesicle suspensions prepared as described above were diluted with imidazole buffer pH 6.6 to give total lipid concentrations in the range 1 x mol dm-3.The suspensions were added to samples of glass beads to known weight (1-3 g) in screw- capped soda-glass vials of 2.5cm3 capacity. The suspensions were weighed into the vials, care being taken to eliminate air bubbles by stirring, and the vials were attached to a vertically rotating wheel of 11 cm radius and rotated at 30 r.p.m. in an air thermostat for the required time interval, after which the beads were allowed to settle to the bottom to 1 xS. Jackson, M. D. Reboiras, I. G, Lyle and M. N. Jones 293 of the vials and 0.1 cm2 aliquots of the supernatant removed for assay of the amount of unadsorbed lipid. The aliquots were added to 2 cm3 of scintillant (PPO/POPOP cocktail T, B.D.H.Chemicals) and counted on a Beckman LS 9800 scintillation counter. Vesicle Disruption Assay The extent of disruption of the vesicles on adsorption was determined by a modification of the terbium dipicolinic acid (DPA) fluorescence assay for monitoring vesicle Vesicles were prepared as described above encapsulating 0.15 mol dm-3 DPA, extra- vesicular DPA being removed by gel filtration on a Sephadex G75 column (10 x 2 cm) using imidazole (20 mmol dm-3), pH 6.6 buffer eluant. The vesicle peak in the elution profile was detected by scintillation counting of 0.05 cm3 aliquots from each 1 cm3 fraction. The fractions containing the vesicles were pooled and centrifuged at lo5 g for 1 h at 25 "C. The resultant supernatant was used in the adsorption measurement.Following equilibrium with glass beads the extent of release of encapsulated DPA was determined by measuring the fluorescence on addition of terbium chloride (final con- centration 0.25 mmol dm-3) using a Perkin-Elmer 204 fluorescence spectrophotometer with an excitation wavelength of 276 nm at an emission wavelength of 491 nm. The fluorescence signal was used to calculate the percentage of vesicles disruption together with the signal for 100% disruption determined by addition of terbium chloride (final concentration 0.25 mmol dm-3) and sodium deoxycholate (final concentration 1.7% w/v) to disrupt the vesicles totally. Controls were also made for fluorescence arising from leakage of DPA during gel filtration and from addition of sodium deoxycholate and used to correct the data where necessary.Glass Bead Surface Area The surface area (in m2kg-') of the glass beads was determined by the B.E.T. gas adsorption method' ' using a Quantasorb sorption system (Quantachrome Corp., Syosset, N.Y. 11791). The beads were found to have surface areas of 104k 1 m2 kg-' using krypton adsorption and 109 f 2 m2 kg-' using nitrogen adsorption. Micro electrophoresis Electrophoretic mobilities ( u ) of large multilamellar liposomes were measured using a Rank Bros. (Bottisham, Cambridge) microelectrophoresis apparatus with a flat cell and grey platinum electrodes as previously described.12 Zeta potentials ( 5) were calculated from the Smoluchowski equation: s = 77UIEOEr ( 1 ) where for dilute aqueous media we took the viscosity q = 8.904 x lop4 N s m-2, the relative permitivity of the media ~ , = 7 8 .5 and the permittivity of vacuum E ~ = 8.854 x C2 J-' m-' at 25 "C. The zeta potentials of the glass beads were determined from measurements of the mobilities of suspensions of 'fines' produced by shaking 0.01 kg of beads with ca. 25 cm3 of buffer or salt solution. This produced a small but sufficient concentration of suspended particles for detection with the microelec- trophoresis microscope. Results The times required for equilibrium adsorption of lipid from aqueous dispersions of sonicated vesicles onto glass beads were determined over the vesicle composition range from pure DPPC to pure PI. Fig. 1 shows a time course for adsorption from pure DPPC and DPPC-PI (75 : 25 wt O/O) vesicles which shows that equilibrium adsorption is attained294 Vesicle Adsorption Fig. 1.Time course for the adsorption of phospholipid from sonicated phospholipid vesicles onto barium titanate (Ballotini) glass beads in imidazole (20 mmol dmP3) buffer pH 6.6, ionic strength 0.0138 moi dmP3 at 25 "C. 0, Pure dipalmitoylphosphatidylcholine vesicles (initial lipid concentration 2.34 x lop4 mol drn-j); 0, dipalmitoylphosphatidylcholine-phosphatidylinositol (75 : 25 wt "/o) vesicles (initial lipid concentration 6.24 x mol dm-3). rapidly (ca. 1.5 h). Similar data were obtained for other vesicle compositions, and on the basis of these measurements equilibrium adsorption isotherms were measured using a 2 h equilibrium period. Fig. 2 shows a typical adsorption isotherm for lipid adsorption from DPPC-PI (75 : 25 wt YO) vesicles.We found considerable scatter of the data points, and to obtain reasonable precision on statistical analysis of the data a large number of samples were assayed (ca. 120 per adsorption isotherm). The isotherms were typical of Langmuir adsorption, and fig. 3 shows a plot based on the linearised Langmuir adsorption isotherm for the DPPC-PI (75 : 25 wt '/O ) vesicles: where (x/m) is the number of moles of lipid adsorbed to rn grams of beads, c is the free lipid concentration and K is the adsorption constant. To avoid the bias towards the least precise points at high free lipid concentrations inherent in least-squares fitting of the linear plots, the adsorption constants and limiting adsorptions (x/ rn)max were obtained by fitting the adsorption isotherms (as in fig.2) by non-linear regression using the method of Walmsley and Lowe.13 The parameters obtained from this analysis are given together with the electrophoretic mobilities and zeta potentials of the vesicles in table 1. Table 2 shows the electrophoretic properties of the glass beads in several aqueous systems. The effect of phospholipid acyl chain length on the adsorptionS. Jackson, M. D. Reboiras, I. G. Lyle and M. N. Jones 295 60 - 0 ' = ' 0 - 0 0 t F 1 I I 0 20 40 60 free lipid concentration/ lo-' mol dm-3 Fig. 2. Adsorption isotherm for the adsorption of lipid from sonicated dipalmitoylphos- phatidylcholine-phosphatidylinositol (75 : 25 wt % ) vesicles onto barium titanate (Ballotini) glass beads in imidazole (20 mmol dm-3) buffer pH 6.6, ionic strength 0.0138 mol dmP3 at 25 "C.free lipid concentration/ lov5 mol dm-3 Fig. 3. Langmuir plot for the adsorption of lipid from sonicated dipalmitoylphosphatidylcholine- phosphatidylinositol (75 : 25 wt %) vesicles onto barium titanate (Ballotini) glass beads in imi- dazole (20 mmol dmP3) buffer pH 6.6, ionic strength 0.0138 mol dm-3 at 25 "C.296 Vesicle Adsorption Table 1. Limiting adsorption and adsorption constants for the adsorption of lipid from sonicated dipalmitoylphosphatidylcholine-phosphatidylinositol (DPPC-PI) vesicles in aqueous solution (20 mmol dm-3 imidazole buffer, pH 6.6) onto barium titanate (Ballotini) glass beads vesicle composition 5" (25 "C) (~/rn),,,/lO-~ mol kg-' K / lo" dm3 mol-' DPPC : PI U /mv (wt%) /10-8m2s-'V-' [ref. (12)J 25 "C 50 "C 25 "C 50 "C 1oo:o 0.184 f 0.010 -2.36 45.7f1.6 27.9f0.5 3.52f0.63 12.1f1.2 75 : 25 2.94 f 0.18 -37.7 52.1*1.2 27.6k0.5 6.54k0.55 30.425-6 50: 50 5.09 f 0.16 -65.2 49.6 f 1.1 30.2 f 0.9 8.92 f 0.87 45.7 * 12.7 25 : 75 6.55 * 0.05 -83.9 38.3f1.6 34.4k0.9 15.1f3.1 39.8k8.0 0: 100 8.74f0.16 -111.9 53.7*2.5 40.3*1.6 9.55*2.3 22.0f4.9 Table 2.Electrophotetic properties of barium titanate (Ballotini) glass beads medium mobility, u {" 025 "C) / lo-' m' s-l V-' /mv 20 mmol dm-3 imidazole, pH 6.6, I = 0.0138 mol dmP3 0.60 f 0.02 -7.7 2.69 f 0.27 -34.4 10 mmol dmP3 Tris, pH 7.4 1.74 f 0.19 -22.3 20 mmol dm-3 Tris," pH 7.4 1.51 f0.32 -19.3 sodium chloride, I = 0.0138 mol dm-3 Tris( hydroxymethy1)aminomethane. Table 3. Limiting adsorption and adsorption constants for the adsorption of 1,2-diacylphos- phatidylcholines from sonicated vesicles in aqueous solution (20 mmol dm-3 imidazole buffer, pH 6.6) onto barium titanate (Ballotini) glass beads at 25 "C phospholipid (chain melting temperature, TF', area per molecule (x/ m)Inax area per K at 25 "C in bilayers'l) / mol kg-' molecule/nm' / lo4 dm3 mol-' DLPC (- 1.8 "C,I4 0.59 nm2) 37.0 f 0.6 0.48 17.3 f 3.4 DMPC (23.7 "C,15 0.58 nm2) 35.0* 0.8 0.5 1 18.4 * 3.4 DPPC (41.1 OC,I6 0.50 nm') 45.7 f 1.6 0.39 3.52 k 0.63 characteristics is shown by the data given in table 3.Also given are the chain-melting temperatures of the phospholipids and the areas per molecule in bilayers. It was observed that equilibration of the vesicles with the glass surface of the beads resulted in a change in the turbidity of the vesicle suspensions.Fig. 4 shows the absorbance change for DPPC and DPPC-PI (75 : 25 wt %) vesicles as a function of time. These results demonstrate a change in the nature of the vesicle dispersion due to interaction with the beads. To investigate this further the extent of surface-catalysed vesicle disruption was assessed by measuring the fluorescence from the terbium- dipicolinic acid complex formed on release of encapsulated dipicolinic acid into a solution containing terbium ions. Vesicle disruption is shown as a function of the amount of glass beads in fig. 5 and as a function of the percentage of adsorbed lipid in Fig. 6. It is seen in fig. 5 that between 80 and 90% of the encapsulated dipicolinic acid is released after exposure to the glass beads and there is an initial linear relationship between the extent of release and the amount of lipid adsorption (fig.6).S. Jackson, M. D. Reboiras, I. G. Lyle and M. N. Jones 297 Fig. 4. Time course for the change in absorbance at 700nm on exposure of vesicles to barium titanate (Ballotini) glass beads in imidazole (20 mmol dm-3) buffer pH 6.6, ionic strength 0.0138 mol dm-3 at 25 "C. a, Pure dipalmitoylphosphatidylcholine [initial lipid concentration 2.34 x mol dmP3 0.5 kg (beads) dm-3]; 0, dipalmitoylphosphatidylcholine-phosphatidy- linositol [initial lipid concentration 6.24 x lop4 mol dmP3, 1.3 kg (beads) dm-3]. Discussion Equilibrium adsorption is attained within 1-2 h for all the systems studied (fig.l ) , and it is pertinent to consider whether deposition of lipid occurs by interaction of intact vesicles with the glass surface or via adsorption of monomeric lipid in equilibrium with the vesicles. The number of particles ( q ) having diffusion coefficient D and concentration n reaching a unit area of surface in time t is given by17 q = n( Dt/2)'l2. (3) Considering pure DPPC vesicles having a diffusion coefficient" of 1.4 x lo-" m2 s-' and a mean concentration of ca. 4 x loP5 mol mP3 it follows that q(vesic1e) == 1 x lo-'' t''2. The concentration of monomeric DPPCI9 will be no greater than 4.7 x mol m-3, and we estimate that the diffusion coefficient will be ca. 2 0 ~ lo-" m2 s-l, hence q(monomer) = 5 x t'12. It follows that the relative rates of diffusion of vesicles to monomer reaching the surface in unit time will be of the order of 20: 1, although this ratio will decrease with increase in vesicle size on increasing the PI content of the vesicle^'^ to ca.3 : 1. Taking into consideration that the vesicles contain ca. 14 000-74 000 lipid molecules depending on their size," the delivery of lipid to the surface will be predominantly via vesicles rather than via monomer. mol (monomeric lipid) kg-', the surface area of the beads is 107 m2 kg-' hence the relative times taken Considering the data in fig. 1, the limiting adsorption is ca. 44 x298 amount of glass beads/ kg Fig. 5. Extent of disruption of dipalmitoylphosphatidylcholine vesicles following 2 h equilibrium with barium titanate (Ballotini) glass beads in imidazole (20 mmol dm-3) buffer pH 6.6, ionic strength 0.0138 moldmP3 at 25°C.Vesicle disruption was determined as the ratio of free to encapsulated dipicolinic acid. Initial lipid concentration (2.44 f 0.089) x mol dmP3. for 4.1 x loP6 mol (lipid) m-2 to arrive at the surface is ca. 9 s as vesicles and ca. 1.9 x 10' h as monomer. If adsorption was diffusion controlled the process would be almost instantaneous. However, the data in fig. 1 clearly demonstrate that other factors control the rate of lipid adsorption. It has been shown previously that incorporation of PI into DPPC multilamellar vesicles decreases both the chain melting temperature and the enthalpy of the gel-lamellar liquid-crystalline transition, and at a mole fraction of 0.5 the transition is eliminated.20 These observations have been interpreted in terms of the removal of one DPPC molecule from participation in chain-melting on addition of each PI molecule.Thus the compo- sition range in table 1 spans both a range of mesomorphic as well as electrokinetic properties of the vesicles. While there are variations in limiting adsorption across the composition range these do not correlate with the electrokinetic properties of the vesicles. There are, however, significant differences in limiting adsorption between 25 and 50 "C. For pure DPPC for example the limiting adsorptions are (45.7 * 1.6) x loP5 mol kg-' at 25 "C and (27.9 * 0.5) x loP5 mol kg-' at 50 "C. Taking the surface area of the beads as 107 mz kg-' gives values of 0.39 nmz molecule-' (25 "C) and 0.64 nm2 molecule-' (50 "C) for the limiting areas.The area of a DPPC molecule determined by X-ray diffraction measurements on the gel and lamellar liquid crystalline phases at 25 and 50 "C are 0.50 and 0.69 nm2, respectively.2' On this evidence there seems little doubt that at saturation the glass is covered with a monolayer of DPPC with a lower packing density at 50 "C than at 25 "C. The adsorption constants increase with temperature, from which it follows that adsorption is endother- mic. The packing density is not markedly affected by changing the acyl chain length of the lipid from DLPC to DPPC (table 3). For DPPC-PI vesicles limiting adsorptionS. Jackson, M. D. Reboiras, I. G. Lyle and M. N. Jones 299 x 100) initial concentration - final concentration initial concentration lipid adsorbed Fig.6. Extent of disruption of dipalmitoylphosphatidylcholine vesicles following 2 h equilibrium with barium titanate (Ballotini) glass beads [(0-3) x kg] as a function of lipid adsorption in imidazole (20 mmol dm-3) buffer, pH 6.6, ionic strength 0.0138 mol dmP3 at 25 "C. Initial lipid concentration (2.44* 0.08) x mol dm-3. is comparable to pure DPPC, but in the absence of detailed bilayer structure for these systems it is not possible to consider the adsorption data in terms of molecular dimensions. Finally we consider the mechanism of the adsorption process and the orientation of the lipid in the adsorbed monolayer. The vesicles used in the experiments have a negative surface charge density, which in the case of the DPPC-PI vesicles increases markedly with PI content as reflected by the change in their 6" potentials (table 1).The beads are also negatively charged (table 2). It is noteworthy that the 6" potential of barium titanate glass is very dependent on the ionic composition of the aqueous medium, even if the ionic strength is kept constant. Given that both the vesicles and the glass are negatively charged there will be a potential barrier to adsorption for vesicles incorporat- ing PI. Fig. 7 shows some potential-energy curves for pure DPPC, DPPC-PI (50: 50 wt YO) and pure PI spherical vesicles on interaction with a flat glass plate with 6" potential -7.7 mV. The curves were computed using an equation for the attractive dispersion energy given by Langbein22 in which the dielectric properties of water, glass (quartz) and hydrocarbon were determined as a function of frequency from data given by Par~egian,~~ Chan and Richmond24 and Ninham and Parsegian," respectively.The repulsive electrostatic energy was calculated from an equation given by Arminski et al.26 The curves show that PI imparts very large electrostatic repulsion energies to the system. For DPPC-PI (50: 50 wt "10) vesicles and pure PI vesicles the energy barriers are 15 x 1 O-20 J (separation 0.7 nm) and 147 x 1 O-*O J (separation 0.2 nm), respectively. There are corredsponding secondary minima of - 1 . 3 0 ~ (separation 8.5 nm) and300 160 140,- 120- 100- 2 2 80- 9 :: Vesicle Adsorption - 1 - 2 0 t / -401 - 1 . 1 4 ~ J (separation 11.4 nm). The thermal energy at 25 "C ( k T ) is 0 .4 ~ lop2' J. It follows that the vesicles could be trapped in the secondary minima, but only for pure DPPC vesicles will the attractive dispersion interaction be strong enough for adsorption in the primary minimum. While the initial interaction between vesicles and the surface will be controlled by a balance between electrostatic and dispersion forces, other factors must determine the fate of the vesicles once in the vicinity of the interface. The release of encapsulated dipicolinic acid clearly demonstrates that the vesicle structure is disrupted. It is relevant to note that the ability of solid surfaces to disrupt bilayer structure has been previously observed in the context of the rupture of l y s o ~ o m e s ~ ~ and the haemolysis of erythrocytes.28 In the experiments described here the glass beads tumble through the vesicle suspension which raises the possibility of vesicle disruption by a compression process between apposing glass surfaces.However, experiments carried out using columns of glass beads also confirmed monolayer adsorption of pure DPPC2' A monolayer at the solid-liquid interface would presumably be oriented to minimise contact between the acyl chains ;; i / I # I I IS. Jackson, M. D. Reboiras, I. G. Lyle and M. N. Jones 301 and water which suggests that the phospholipids are adsorbed with their head groups uppermost. This situation would be analogous to the formation of phospholipid monolayers at the air-water interface from vesicle dispersion^^^-^^ and must involve exchange of lipid molecules from the outer layer of the vesicle bilayer to the interface.We thank the S.E.R.C. for a CASE studentship for S.J., the British Council for financial assistance for M.D.R. and Dr J. Potter for assistance with the surface-area measurements. References 1 N. Murase and K. Gonda., J. Biochem. (Tokyo), 1982, 92, 271. 2 M. C. Jaurand, J. H. Thomassin, P. Baillif, L. Magne, J. C. Touray and J. Bignon, Brit. J. Znt. Med., 3 M. C. Jaurand, P. Baillif, J. H. Thomassin, L. Magne and J. C. Touray, J. Colloid Interface Sci., 1983, 4 J. Phillippot, S., Mutaftschiev and J. P. Liautard, Biochim. Biophys. Acta, 1983, 734, 137. 5 C. H. Huang, Biochem., 1969, 8, 344. 6 S. E. Schullery and J. P. Garzaniti, Chem. Phys. Lipids, 1973, 12, 75. 7 R. G. Horn, Biochim.Biophys. Acta, 1984, 778, 224. 8 L. Ter-Minassian-Saraga and G. Mandelmont, J. Colloid Interface Sci., 1982, 85, 375. 9 R. Sundler and D. Papahadjopoulos, Biochim. Biophys. Acta, 1981, 649, 743. 10 R. Sundler, N. Duzgunes and D. Papahadjopoulos, Biochim. Biophys. Acta, 1981, 649, 751. 11 S. Brunauer, P. H. Emmett and E. J. Teller, J. Am. Chem. SOC., 1938, 60, 309. 12 M. D. Reboiras and M.N. Jones, Colloids Surf.', 1985, 15, 239. 13 A. R. Walmsley and A. G. Lowe, Comput. Method Program. Biomed., 1985, 21, 113. 14 S. Mabrey and J. M. Sturtevant, Proc. Natl Acad. Sci. USA, 1976,73, 3862. 15 H. J. Hinz and J. M. Sturtevant, J. Biol. Chem., 1972, 247, 3697. 16 S. C. Chen, J. M. Sturtevant and J. Gaffney, Proc. Nut1 Acad. Sci. USA, 1980, 77, 5060. . 17 K. J. Mysels, Introduction to Colloid Chemistry (Interscience, New York, 1959), chap. 8, p. 175. 18 K. Hammond, M. D. Reboiras, I. G. Lyle and M. N. Jones, Biochim. Biophys. Acta, 1984, 774, 19. 19 R. Smith, and C. Tanford, J. Mol. Biol., 1972, 67, 75. 20 K. Hammond, I. G. Lyle and M. N. Jones, J. Colloid Interface Sci., 1983, 99, 294. 21 M. J. Janiak, D. M. Small and G. G. Shipley, J. Biol. Chem., 1979, 254, 6068. 22 D. Langbein, J. Adhesion, 1969, 1, 237. 23 V. A. Parsegian, in Physical Chemistry: Enriching Topics from Colloid and Surface Science, ed. H. van 24 D. Chan and P. Richmond, Proc. R. SOC. London, Ser. A , 1977, 353, 163. 25 B. W. Ninham and V. A. Parsegian, Biophys. J., 1970, 10, 646. 26 L. Arminski, S. Weinbaum and S. Chien, J. Colloid Interface Sci., 1982, 90, 390. 27 G. Weissmann and G. A. Rita, ,Nature (New Biol.), 1972, 240, 167. 28 J. Depasse and J. Warlus, J. Colloid Znterface Sci., 1976, 56, 618. 29 S. Jackson, M. N. Jones and I. G. Lyle, unpublished observations. 30 F. Pattus, P. Desnuelle and R. Verger, Biochim. Biophys. Acta, 1978, 507, 62 31 H. Schindler, Biochim. Biophys. Acta, 1979, 555, 316. 32 F. Jahnig, Biophys. J., 1984, 46, 687. 33 K. Tajima and N. L. Gershfeld, Biophys. J., 1985, 47, 203. 1980, 37, 169. 95, 1. Olphen and K. J. Mysels (IUPAC Commission, Theorex La Jolla, California, 1979, chap. 4. Received 10th December, 1985

ISSN:0301-7249    

DOI:10.1039/DC9868100291

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

Faraday Discuss. Chem. SOC., 1986, 81, 303-311 Liposome Electro formation Miglena I. Angelova" and Dimiter S. Dimitrov Central Laboratory of Biophysics, Bulgarian Academy of Sciences, B1.21, 11 13 Sojia, Bulgaria Liposome formation and lipid swelling on platinum electrodes in distilled water and water solutions in d.c. electrical fields have been investigated for different amounts of a negatively charged lipid (mixture from 71% PC, 21.5% PE and 7.5% PS), and a neutral lipid (DMPC). Negatively charged lipids do not form liposomes without field when the thickness of the dried lipid layer is of the order or less than that corresponding to 90 bilayers. The rate and extent of swelling of layers thicker than 90 bilayers is largest on the cathode, smaller without fields and smallest on the anode.The theory, based on the assumption that osmotic and electrostatic forces drive lipid swelling and liposome formation, is in semi-quantitative agreement with the experimental data; in particular, it gives the observed linear dependence of the rate of swelling on the inverse lipid layer thickness. To induce liposome formation for layers thinner than 90 bilayers it was necessary to apply a negative potential which is proportional to the logarithm of the inversed layer thickness. The characteristic critical potential is proportional to RTk/F; R being the gas constant, Tk the absolute temperature and F the Faraday constant. This indicates that redistribution of counterions may be the cause which increases the repulsive electrostatic intermembrane forces to overcome van der Waals attraction.For thicknesses below 10 bilayers, formation of very thin-walled liposomes of narrow size distribution and mean diameter of the order of 30 p m was observed. These liposomes grow in size before detachment and a formula for the kinetics of growth has been derived, which is in very good agreement with the experimental data. The effects of d.c. fields on DMPC swelling are smaller and lead to formation of liposome-like structures of different appearence. Bilayer separation and bending are pre- requisites for liposome formation from hydrating lipids. Therefore, a poss- ible molecular mechanism is that membranes should be destabilized to bend and fuse to form liposomes. This requires the right proportion between structured regions, in the form of bilayers, and defects and/or non-bilayer structures, and in many cases external constraints, in particular, electrical fields.Liposome formation from dried lipid in water solutions requires bilayer separation and bending. External electrical fields can facilitate both: they can decrease the inter- membrane attraction and can induce instability of bending. This work suggests some experimental evidence and theoretical estimates that electrical fields can induce lipid swelling and liposome formation. After the pioneering work of Bangham et aL' liposomes have been thoroughly investigated [see, e.g. ref. (2)] because of their great application and potential in membrane research, medicine and technology. Relatively little is known, however, of the mechanism of their Helfrich3 and later LasiC4 have used the concept for the curvature elastic energy and boundary interaction energy to explain the formation of liposomes in a bulk liquid, especially liposomes produced by the detergent-removal method. However, these authors have not considered effects of solid surfaces and external fields and the kinetics of liposome formation.In previous work9-" we have shown that liposome formation may depend on the type of the surface, where the lipid 303304 Liposome Electroformation A T I I 1 + 0 0.6 1.2 1.8 UCJ v Fig. 1. Dependence of the logarithm of the EggL layer thickness on the applied voltage above which thin-walled liposomes are observed. Temperature = 30 "C, ZCr,,, = 0.6 pm. is hydrated, the thickness of the lipid layer and the temperature. It was suggested that surface-lipid interactions can determine the possibility for and the kinetics of liposome formation, as well asstheir yield and physicochemical characteristics.By decreasing the thickness of the lipid layer we can obtain a higher yield of liposomes per unit lipid. In later work'* we have shown that negatively charged lipids swell faster and reach greater extents of swelling on the negative electrode than on the positive one or without fields. Theoretical estimates, based on the assumption that osmotic and electrostatic forces are dominant, have given predictions in semiquantitative agreement with the experimental results. This work considers the effects of external d.c. fields on lipid swelling and liposome formation for different thicknesses of the lipid layer.The basic concept is that the interplay between external fields and intermembrane forces may allow mechanisms of liposome formation to be studied. Experimental We have used two types of lipids: (1) egg lecithin from dried egg yolk (EggL), (Sigma P-5394), which contains 71 Yo phosphatidylcholine (PC), 21.5% phosphatidyl- ethanolamine (PE) and 7.5% phosphatidylserine (PS), Le. it behaves as a negatively charged lipid; and (2) dimyristoilphosphatidylcholine (DMPC), (Fluka 41803), syn- thetic, 99%, which is neutral. The lipids were dissolved in a chloroform-methanol 9: 1 mixture, and a lo3 cm-3 drop was deposited on each of the two parallel cylindrical platinum electrodes (diameter 0.5 mm, separation between their surfaces 0.5 mm).The solvent was then evaporated under nitrogen. An electric field was applied and distilled water or water solution added. All the experiments were performed at a d.c. voltage below 3 V because above this voltage we have observed formation of gas bubbles. The lipid swelling and liposome formation were observed under phase contrast. In some cases Ficoll was added as an aid to visualising the thin-walled liposomes. The tem- perature was kept above the phase-transition temperature of both lipids (in most cases 30 "C). Fig. 1 shows the dependence of the logarithm of the EggL layer thickness 1 = I,, on the applied voltage U = U,,, for which liposomes were observed at the negative electrode after 20 min. The critical thickness for the hydrated lipid to form thin-walled liposomes without electrical field (i.e.U,, = 0 ) is ZcriO = 555 nm which corresponds to 90 hydrated bilayers; the repeat distance for maximally hydrated EggL being 62.5 A.13 Each point is the mean of 10 experiments. It is seen that the applied voltage U,, to induce liposome formation increases very strongly with decreasing lipid layer thickness. The dependence U,,[ln ( lcr/ lc,;O)] follows a linear relationship, In (ZJ = - 1.96 U,,, with a correlationM. I. AngeZova and D. S. Dimitrov 305 t Fig. 2. The rates of swelling (corresponding to plate 2) on the cathode ( -), anode (+ ) and without fields (0), as functions of the inverse hydrating lipid layer thickness, I-'. coefficient of 0.975. In the region above 40 effective dried bilayers the EggL pre- dominantly swells to form myelin figures, tubes and multilamellar liposomes.From 40-10 effective dried bilayers the number of myelin figures, tubes and thick-walled liposomes decreases. Below 10 effective dried bilayers only very thin-walled liposomes of an average diameter of 30 pm are observed. On the positive electrode there is no liposome formation for any d.c. voltage if the effective dried bilayers are less than 90. For larger thicknesses the applied field decreases the rate of swelling. The effects of d.c. fields on lipid swelling and liposome formation from DMPC are much smaller and rather different from those for EggL. Plate 1 shows formation of liposome-like structures, which appear different on both electrodes ( U = 1.2 V). Lipo- somes from DMPC can form without electric field even for the smallest lipid layer thickness we have investigated (12- 15 effective dried bilayers), but the number and size of DMPC liposomes increases in presence of d.c.field, especially at the cathode for voltages > 1.2 V. Plate 2( a ) shows the time course of swelling of EggL (100 effective dried bilayers) in a 1 mol dm-3 sucrose solution without field ( a ) and with an applied voltage of 0.4 V ( b ) . Fig. 2 presents the respective rates of swelling. It is seen that in all cases (cathode, anode and without field) the experimental points lie well on straight lines in the plot of dZ/dt vs. Z-'; Z being the thickness of the hydrating lipid layer and t the time. The correlation coefficients are > 0.995. The corresponding equilibrium distances are: I,- = 253 pm, I,+ = 43 pm and Zeo = 50 pm.In 2.5 mol dm-3 sucrose no liposomes had formed306 Liposome Electroformation Table 1. Liposome diameter 0, height H and contact radius R, at different times of growth number time/min D / p m H / p m 2R,/pm 85 1 100 115 85 2 100 115 85 3 100 115 85 4 100 115 85 5 100 115 85 6 100 115 29 31 32 22 23 24 33 35 39 30 - 32 32 32 33 34 36 20 22 23 15 17 20 23 26 29 32 - 33 33 33 28 31 33 29 30 31 20 23 24 30 32 33 14 - 0 0 0 23 24 25 after 2 h. Plate 3 shows liposome electroformation from EggL (40 effective dried bilayers) under 0.6 V d.c. field in distilled water (b), while without field only tubes, clumps and other structures are observed (a). Plate 4 shows liposomes formed from EggL (5-6 effective dried bilayers) at 30 "C and 2 V in distilled water at the cathode.The liposomes attached to the electrode surface grow by ca. 10% in 115 min. Those liposomes which are separated do not change their dimensions. All the liposomes are a little elongated in the direction of the field lines. Table 1 gives the changes in the diameter 0, height H and contact radius R, for five liposomes shown in plate 4. Liposome 4 detaches from the electrode and goes into the bulk after 95 min. Liposome 5 is separated from the surface, but remains close to it and does not change its size. Owing to technical difficulties the contact radius R, is measured for the line of visual contact, which may not represent the 'true' contact, but gives one additional geometrical characteristic. Theoretical We consider a spherical liposome attached to a solid surface, as shown in fig.3. The liposome increases in size under the action of osmotic and electrostatic forces and the membrane tension acts to decrease the liposome radius. The rate of change of radius can be expressed as dR/dt = LA) ( 1 ) Ap = RTk( ci - c,) + UU/ L, - 2 T / R where L is the hydraulic permeability coefficient, p is the driving pressure, R is the gas constant, Tk is the absolute temperature, ci and c,, are the solute concentrations inside and outside the liposome, U is the applied voltage, (T is the membrane surface charge density, L, is the distance between electrode surfaces and T is the membrane tension. The internal concentration can be expressed as ci = ci,Ri/ R3 (2)Plate 1.Formation of liposome-like structures on the anode (+ ) and cathode ( - ) from DMPC in distilled water at 30 "C, and 1.2 V d.c. The effective dried lipid layer thickness corresponds to 12 bilayers. The bar represents 100pm. (To face p. 306)1 min 3 min 5 min 15 min 30 min Plate 2. The time course of swelling of EggL (effective dried lipid layer thickness 100 bilayers) in 1 mol dm-3 sucrose: (a) without field and ( b ) for applied voltage of 0.4 V. The bar represents 100 km.Plate 3. Effects of d.c. fields on liposome formation from EggL in distilled water. Effective dried lipid layer thickness 40 bilayers; temperature = 30 "C. ( a ) Without field and ( b ) with applied voltage 0.6 V. Ficoll was added after 20 min to improve the phase contrast.The bar represents 100 pm.Plate 4. Growth of EggL liposomes (effective dried lipid layer thickness 5 bilayers) in distilled water on the negative electrode at d.c. voltage 2 V after ( a ) 85, ( b ) 100 and (c) 115 min. The bar represents 50 pm.M. I. Angelova and D. S. Dimitrov 307 f Fig. 3. A liposome model as part of a spherical surface. where c,, is the concentration at R = R,; Ro being the radius of the tension-free liposome. For small changes of the radius AR =sR - Ro << R, eqn ( 1 ) combined with eqn (2) becomes dAR/ d t = L[ RTk( ci, - c,) + UCT/ Le - ( 3RTkCi0/ Ro 4- 4K / Ri) AR] (3) where T is represented as T = 2KAR/ R, and K is an elasticity constant. Integrating eqn (3) yields AR = R,[1 -exp ( - t / ~ ) ] (4) where A R , = [ RTk( Ci0 - C,) 4- UCT/ Le]/ (3 RTkci,/ Ro + 4K / Ri) 7 - l = L(3RTkciO/Ro+4K/Ri) and an initial condition AR = 0 at t = 0 was used.When AR < 0, eqn (4) is no longer valid. In this case T = 0 and the liposome inflation is much faster, so the approximation ARC< R cannot be used and the liposome is not spherical. In that case eqn ( 1 ) leads to a formula for the rate of swelling presented in our previous work: l 2 dZ/dt = C U ( I-' - l;') ( 5 ) where (Y = LRTkcehOn a/L(RTkcj- ua/Le) and I is the layer thickness, n is the number of bilayers, c is the solute concentration for an initial thickness between plane-parallel bilayers of ho [see fig. 4 ( a ) ] . The above formulae do not describe the kinetics of the initial stages of liposome formation because hydration forces, intermembrane electrostatic forces and van der Waals forces are not taken into account.If the driving pressure Ap = po exp ( - h / h ) , i.e. hydration forces dominate, then the time to increase intermembrane separation from 0 to h is t = A[exp ( h / X ) - l]/LpO. ( 6 ) This time is very short if h G 10 A. For larger separations, electrostatic intermembrane forces and van der Waals forces between the membranes and between the membranes and the electrode can be important up to membrane separations of the order of 0.1 pm. The external electric field can change the intermembrane electrostatic forces by changing the concentration of the counterions. This concentration is proportional to exp [ - U/(RTk/F), where F is the Faraday constant. Decreasing the counterion con- centration leads to an increase in the intermembrane repulsive electrostatic forces.For308 Liposome Electroformation Fig. 4. ( a ) A plane-parallel model of a hydrated lipid layer. ( b ) - ( d ) Possible mechanisms of liposome formation. smaller layer thicknesses the van der Waals forces are greater and larger fields should be applied to overcome them. Thus, we can suppose that to induce liposome formation or lipid swelling electrically: ( lcr/ L , o ) rn = exp [ - ucr/ ( R Tk/ F ) 1 (7) where rn is a constant of the order of unity, lcr is the critical thickness of lipid swelling and liposome formation and l C r . 0 = for U c r = 0. Discussion The exact mechanism of liposome formation on surfaces is still unknown. Our observa- tions and experimental data of others' have revealed several facts which need explana- tion: (1) liposomes form only when the temperature is above that of the main phase transition; (2) the rate of liposome formation depends on the temperature, e.g.at 4°C they form for 24-48 h, while at 70 "C they form for ca. 30 min; ( 3 ) during formation,M. I. Angelova and D. S. Dimitrov 309 part of the liposomes or all of them can internalize neutral external solutes, e.g. sucrose, and even large solid particles; (4) the internal solution has an osmolarity of the order of 2 mol m-3, conductivity 20 to 50 pS cm-’, and pH 6.4 when the liposomes are formed in distilled water (osmolarity 0.1 mol m-3, conductivity 0.2 pS cm-’, pH 6.8); (5) lipo- somes do not form in solutes of high ionic strength, e.g.in solutions of monovalent salts of concentration > mol dm-3; ( 6 ) additives, such as deoxycholic acid or tocopherol, improve liposome formation; (7) thin-walled liposomes, almost uniform in size, form when the lipid layer thickness is of the order of 5-10 bilayers (layers of larger thicknesses yield multilamellar liposomes and tubes, while layers of smaller thicknesses do not change); (8) thin-walled liposomes from negatively charged lipids form on the cathode and not on the anode or without field. These facts suggest that the structure of the dried lipid layer should have domains of bilayers, perhaps parallel to each other, separated by defects forming holes or lines. To form liposomes we have to have an optimum proportion between structured regions and defects.The formation and structure of the dried lipid layer is very im- portant for liposome formation. When water or a solution is added, it goes through the bilayers and/or through the defects, driven by the hydration forces and, perhaps to a lesser extent, by undulation forces. The time for this hydration stage is very short: it takes s to increase the interbilayer separation from 0 to 10 A, according to eqn (6) (A = 2.6 A, po = 7 x lo9 dyn cmP2, L = 3 x cm2 s g-’). For larger separations repul- sion can be determined from osmotic and electrostatic forces which must overcome van der Waals attraction. At this stage the bilayer should bend in order to form liposomes. There are several possibilities. Owing to the higher energy of the defects which expose hydrophobic parts of the bilayers to water, line tension will appear.This will tend to decrease the length of the defects and, together with the action of repulsive forces, will result in bending instability. The membrane can bend in several ways: (1) the line of defects is attached to the other bilayers (or solid surface) and the bilayer under consideration bends, forming a bulging which increases in size due mainly to the osmotic forces [fig. 4 ( b ) ] ; neutral solutes or solid paiticles can enter through defects or can be internalized additionally by forming pores or by ‘endocytosis’; (2) the line of defects closes, forming a bag which internalizes part of the external solution and the liposome continues to inflate [fig. 4( c)] (however, in this case the osmotic forces are very small, if not zero, and it is difficult to explain liposome formation in water); (3) other possibilities include fusion of bilayers, forming a ‘bag’, which inflates to form a liposome [fig.4(d)], fusion of small liposomes to yield a larger one etc. It seems that several bilayers attached to the solid surface are highly structured and attracted to it. The next bilayers are not influenced so much by the solid surface. Therefore, ‘good’ liposomes can be formed when the lipid bilayer has an optimum thickness. Electrical fields can overcome the lipid-solid surface interactions, and to induce liposome formation from very thin lipid layers. According to eqn (7) the slope of the curve In 1 DS. U,, is equal to FlrnRT‘. The result from the experimental data, shown in fig.1, is 1.96 V-’, which is equal to the value of this expression for m = 1.5. This agreement does not prove the statement that the interplay between van der Waals and induced electrostatic intermembrane forces determines the critical potential which can destabilize the bilayers exposed to bulk water. Other factors include direct electro- static interactions of the bilayers with the field [see eqn (l)], restrictions due to the finite thickness of the lipid bilayer etc. This problem indicates the necessity to calculate electrostatic intermembrane forces rigorously for a multilamellar system in external electric fields, and the respective contribution of van der Waals forces, dipole-dipole interactions etc. Presently we are making some efforts in this direction, especially in dynamic conditions, based on our previous work on the dynamics and stability of membrane system^.'^-'^ As we have pointed out in a previous paper,12 lipid swelling to macroscopic distances (of the order of p m per lipid bilayer) can be driven by osmotic310 Liposome Electroformation forces.Therefore, it should be expected that higher osmolarity of the solution will decrease the rate and the extent of swelling. Plate 2( 6) for a 1 mol dm-3 sucrose solution shows that this is the case. The final equilibrium thickness 1, calculated from the experimental data [see eqn ( 5 ) ] , is 50 pm, i.e. smaller than le = 130 p m measured in ref. (12) for d;stilled water. There is no liposome formation in 2.5 mol dmP3 sucrose solutions and PBS (0.3 mol drnp3).In these cases the osmotic forces, and the intermembrane electrostatic repulsive forces, respectively, are supressed. The last stages of liposome electroformation on surfaces involve increase in size and detachment. After detachment liposomes do not change their dimensions, as is seen from table 1, but they become a little elongated. This demonstrates the effect of the negative surface charge of the liposome membrane. Another confirmation for that statement is the observed electrophoretic motion of the liposomes. We have not observed indications for electrokinetic effects. The maximum relative area change (AA/ A = 2A0/D0) is of the order of 0.3 (liposome 3), commonly 0.2 (liposomes 1, 2 and 5). This corresponds to membrane tensions K = 100 dyn cm-' for one bilayer which are rather high (the critical membrane tension of rupture is of the order of 3-5 dyn cm-1.20 This result may indicate that during liposome growth material can come from the lipid layer or that liposomes are formed from several bilayers, which may rupture and slip over each other.However, numerical estimates based on eqn (4) show very good agreement with the experimental data shown in table 1. If we assume that cio corresponds to the osmolarity, measured by Mueller et d,* cio = 2 mmol dm3, c, < 0.1 mmol dm-3 and also that U = 2 V, (T = 0.1 C mW2, L, = 0.5 mm, Tk = 293 K, K = 100 dyn cm-', Ro = 16 p m and L = 3 x lo-'* cm2 s g-', we obtain ARm = 1.9 p m and T = 23 min. This shows that osmotic forces, and electrostatic forces (their contribution is of the order of 10%) drive liposomes to grow in size.The detachment of the liposomes is determined by the external electric field and Brownian motion, which must overcome the attraction to the electrode surface, lipid layer or other liposomes. The attraction can be due to interaction of the induced dipoles of the liposomes with other induced dipoles or with the electrode. If so, the attraction force will be proportional to the square of the field intensity, and increasing the applied voltage will not help much to detach liposomes. A better way is to switch off the electrical field after the liposomes have been formed, and to wait for a while and/or to shake the suspension if possible. Shaking, however, may damage the liposomes. The liposomes we have obtained by electroformation are rather sensitive to mechanical constraints; for example, extrusion through capillaries (diameter less than 0.5 mm, speed above 0.1 cm s-l) can lead to their destruction.One final point to make is that these experimental and theoretical results are the first on lipid swelling and liposome formation on metal surfaces and in electrical fields. Therefore, they are incomplete or even some of them might be wrong. Further work, now in progress in our laboratory, will show the significance and potential of liposome electroformation for science and technology. References 1 A. D. Bangham, M. M. Standish and J. C. Watkins, J. Mol. Biol., 1965, 13, 238. 2 Liposome Letters, ed. A. D. Bangham (Academic Press, London, 1983). 3 W. Halfrich, Phys. Lett., 1974, 50a, 115. 4 D. D. LasiC, BBA Report, 1982, 501. 5 D. D. LasiC, Periodicum Biologorum, 1983, 147. 6 N. Oku, J. F. Scheerer and R. C. MacDonald, BBA, 1982, 694, 384. 7 R. I. MacDonald, N. Oku, R. C. MacDonald, in Liposome Letters, ed. A. D. Bangham (Academic 8 P. Mueller, T. F. Chien and B. Rudy, Biophys. J., 1983, 44, 375. 9 D. S. Dimitrov, J. Li, M. I. Angelova and R. K. Jain, FEBS Lett., 1984, 176, 398. 10 M. I. Angelova and D. S. Dimitrov, Biophys. J., 1985, 47, 163a. 11 D. S. Dimitrov and M. I. Angelova, Proc. Biotech '85 Europe, 1985 (Geneva, May 1985), p. 655. Press, London, 1983), p. 63.M. I. Angelova and D. S. Dimitrov 311 12 D. S. Dimitrov and M. I. Angelova, Studia Biophysica, 1986, in press. 13 L. J . Lis, M. McAlister, N. Fuller, R. Rand and V. Parsegian, Biophys. J., 1982, 37, 657. 14 D. S. Dimitrov, Biophys. J., 1981, 36, 21. 15 D. S. Dimitrov, Colloid PoIym. Sci., 1982, 260, 1137. 16 D. S. Dimitrov, in Progress in Surface Science, ed. S. G. Davison (Pergamon Press, New York, 1983), 17 D. S. Dimitrov, R. K. Jain, BBA Reviews on Biomembranes, 1984, 779, 437. 18 D. S. Dimitrov, in Electrical Fields in Biological Material, ed. U. Zimmermann (Chemie-Verlag, Berlin, 19 D. S. Dimitrov and D. V. Zhelev, Studia Biophysica, 1985, in press. 20 R. Kwok & E. Evans, Biophys. J., 1981, 35, 627. V O ~ . 14, pp. 295-424. 1986), in press. Received 3rd December, 1985

ISSN:0301-7249    

DOI:10.1039/DC9868100303

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

Faraday Discuss. Chem. Soc., 1986, 81, 313-327 The Distribution of Substituted Phenols into Lipid Vesicles Stanley S. Davis and Michael J. James Department of Pharmacy, University of Nottingham, University Park, Nottingham NG72RD Nicholas H. Anderson? ICI Plant Protection Division, Jealotts Hill Research Station, Bracknell, Berkshire It is shown that solute partitioning into phospholipid liposomes is relevant to the study of structure-biological activity relationships. Data on the ther- modynamics of such partitioning are reviewed, from which it is clear that the magnitude of the enthalpy (AHtr) and entropy (AS,,) of transfer varies greatly with the solute. Results are reported on the partitioning of a series of phenol and anisole solutes into cyclohexane, octan-1-01 and dimyris- toylphosphatidylcholine (DMPC) liposomes below their phase transition temperature.For all three systems, AH,, and AStr were derived from van't Hoff plots. Whereas partitioning into octan-1-01 was always associated with a loss of enthalpy and was enthalpy driven for a number of solutes, transfer of these same compounds into cyclohexane led to an increase in enthalpy and was entropy driven. These differences are related to the hydrogen- bonding properties of the solutes and the fact that octan-1-01 but not cyclohexane can hydrogen-bond. Surprisingly, all the solutes partitioned to a greater extent into the DMPC liposomes at 22°C than into the organic solvents on a mole fraction basis. The free energy of transfer (AG,,) into liposomes was more closely related to that for octan-1-01 than for cyclo- hexane.Differential scanning calorimetry demonstrated that as little as 1 '/o mol 4-methylphenol per mole DMPC caused some broadening of the liposome phase transition and reduced the enthalpy of transition. This perturbation of the gel structure was probably also responsible for the very large AHtr and AStr observed for this and other solutes partitioning into the liposomes. Larger substituent groups than methyl did not always lead to increased values of enthalpy and entropy of transfer, as might have been expected. Inter alia substituent effects are discussed in terms of their interac- tion with phospholipid uia hydrogen bonding, their steric effects and their effects on the pK, and hence hydrogen-bonding ability of the phenolic hydroxy group.The results from the phenols and a number of substituted anisoles suggest that hydrogen bonding between solute and phospholipid is an important factor affecting the thermodynamics of partitioning. In free- energy terms both octan-1-01 and DMPC liposomes appear to be equally good models for the partitioning of a small number of phenols into red blood-cell membranes. It is concluded that in thermodynamic terms neither octan-1-01 nor cyclohexane is a good model for the gel phase of DMPC liposomes. In the past, many attempts have been made to correlate the activity of xenobiotics with their structures, either from the standpoint of the interaction of the solute with a receptor (pharmacodynamic response) or with the transport of the xenobiotic to its site of action (pharmacokinetics).' In the transport process, the permeation of solutes (e.g.drugs and t Present address: University of Bristol, Department of Agricultural Sciences, Long Ashton Research Station, Bristol BS18 9AF. 313314 Distribution of Phenols into Lipid Vesicles Table 1. Changes in partition at the transition temperature of DMPC liposomes (from Diamond and Katz") butyramide 0.507 0.409 11.9 98.0 34.2 321.5 ethyl acetate 2.52 2.39 7.38 113.4 32.4 387.8 pesticides) through biological membranes will be an essential step in determining biological response and permeability will be dictated by the size of the solute (diffusivity) and its affinity for the membrane (equilibrium partition coefficient) .2 Partition experi- ments conducted using immiscible isotropic liquids [ e.g.water (buffer) and octan-1-01, or olive have provided data that have been used to correlate permeability data in the hope of finding a universal 'solvent' that will mirror the biophases and the properties of membrane^.^.^ Not too surprisingly simple organic liquids such as octan-l- 01 and hydrocarbons have not been successful in this respect.6 The proposed model membrane systems have included not only isotropic liquids but also mi~elles,~~* cells (erythrocytes, plant and tumour cells799), vesicles ( l i p o s o m e ~ ) ~ ~ ~ ~ ~ ~ - ~ ~ and biological structures (toad urinary bladder,23 sea urchin eggs'). The phospholipid vesicle (liposome) has become a popular model system since in nature phospholipids are believed to provide the essential permeability barrier for cells.Liposomes are easy to prepare and characterise and they provide a system within which the solute can partition to different domains (e.g. non-polar hydrocarbon core, or to the more polar region of the head groups). Wright and B i n d ~ l e v ~ ~ have compared the thermodynamics of permeation of non- electrolytes in the toad urinary bladder with similar data obtained for erythrocytes, liposomes and bulk liquid phases. They concluded that the lipids in the biological membrane were in a highly ordered configuration in both the bladder and red cell as compared with bulk liquid and liposomes above their phase transition (in the liquid- crystalline phase). Overall, they believed that the partition of solutes into the plasma membrane resembled partitioning of solutes into frozen (gel state) liposomes where the enthalpy of transfer (AHtr) was high and the entropy (ASt,) of transfer was also high and positive.The change in thermodynamic quantities that can occur on passing through the phase-transition temperature (T,) is well illustrated by data taken from the work of Diamond and Katz" (table 1). Of interest is the fact that the free energy of transfer below and above T, is not greatly different. However, striking differences are seen in the values for AH,, and AS,, above and below the phase-transition point. The large changes in these values are not reflected in AGt, because of compensation effects. The large change in ASt, below phase transition is attributed to the disruption of the ordered bilayer structure by the penetrating solute molecule, such that it becomes more fluid.This effect has been well documented for cholesterol which can intercalate into phos- pholipid bilayer~.~' However, in a recent review Lee26 has stated that membrane lipids must be in the liquid-crystalline phase to support enzyme activity and hence this is the state of lipids in most cell membranes. The question of whether a liposome above or below its phase-transition temperature is a more appropriate model for a biological membrane is a matter of discussion. Studies on the partition of solutes into liposomes (and isotropic liquids) have usually reported data conducted at a single temperature, and consequently it is not possible to derive enthalpy and entropy contributions to gain insight into the mechanism of soluteS.S. Davis, M. J. James and N. H. Anderson 315 -- I I 1- I 1 - i\, I 01 I 1 1 1 1 I 1 3.1 3.2 3.3 3.L 3 . 5 3 . 6 3.7 lo3 K/ T Fig. 1. The partitioning of solutes between liposomes and aqueous phase, the effect of temperature. All data refer to DMPC liposomes except for hexane (DOPC liposomes). The partition coefficient data are expressed in molal concentration units: (1) cortisone, (2) progesterone,20 (3) hydrocor- tisone, (4) phenothiazine deri~ative,'~ ( 5 ) h e ~ a n e , ~ (6) n-butyramide" and (7) phenols.16 transfer. The data in table 1 demonstrate well that the partition (free-energy) term can perhaps hide more than it reveals, and therefore studies conducted over a range of temperatures to allow AHtr and AS,, to be estimated are recommended.It is realised that the derivation of thermodynamic values from the van't Hoff plot is beset with inaccuracies and microcalorimetry would be m refer able.^^ However, such methodology is far from simple, and even studies with simple liquids such as octan-1-01 have proved difficult. Data from some previous studies on the partition of solutes into liposomes at a range of temperatures are shown in fig. 1. The values have been plotted according to van't Hoff coordinates. The data all refer to dimyristoylphosphatidylcholine (DMPC), except data for h e ~ a n e , ~ where the phospholipid was dioleylphosphatidylcholine (DOPC). The gel phase of DMPC is believed to be highly ordered, with the alkyl chains close-packed predominately in an all trans conformation.26 Below the 'pre-transition' temperature of 13.5 "C the alkyl chains are tilted from perpendicular to the plane of the bilayer surface, but adopt a perpendicular orientation above this temperature.Interestingly, the data for hexane and DOPC liposomes demonstrated that for hexane (which would be expected to partition into the hydrocarbon regions of the liposome) the partition coefficient decreased with increase in temperature (Le. AH,, was negative). A similar conclusion was reached by Miller et aZ.,14 who studied the solubility of hydrocarbon gases in DMPC (and egg lecithin) liposomes. The other plots in fig. 1 show that partition coefficient increases with increase in temperature ( i e . AH,, is positive). In some instances the phase transition of DMPC is clearly demonstrated by a change in gradient and in some cases by a discontinuity.This discontinuity has been attributed to 'freezing out' of the solute and is associated with the gain in enthalpy that arises from stronger interaction between phospholipid chains.14 In other cases (e.g. hydrocortisone) there is little that distinguishes a phase transition. Above T,, AHtr is generally smaller than below, but progesterone demonstrates the opposite effect. From such data it is clear that the structure of the solute can have a critical bearing on the thermodynamics of partitioning into liposomes.3 16 Distribution of Phenols into Lipid Vesicles The current work was undertaken to follow the thermodynamics of partitioning of a range of structurally related solutes into DMPC liposomes.We have chosen substituted phenols and anisoles, since these compounds are readily available with a wide range substituents; they can be purified easily if required and they can be analysed by U.V. spectroscopy. Furthermore, they have been used previously by ourselves and others in extensive investigations on the thermodynamics of partitioning using isotropic sol- vents3’28-31 and l i p o ~ o m e s . ~ ~ ’ ~ ~ The studies reported here were all conducted below the phase transition of the liposomes. It is appreciated that the gel state might not be directly relevant to the structure of real membrane systems, but these conditions were selected because thermodynamic changes (especially AS,,) would be expected to be greater below T, (see fig.l), and thus the experimental uncertainties would be less and the quality of the derived data would be much better below T, than above. In future studies we intend to repeat our measurements at temperatures above T,. The actual solutes examined were chosen in order to study the effects of alkyl chain length, the size of the added group, substituent position and the effect of hydrogen bonding. Comparison has been made with the partitioning of the same solutes between water and bulk isotropic liquids (octan-1-01 and cyclohexane) as well as with limited data from other biological models (erythrocytes). Experimental Materials The solutes were obtained from commercial sources and purified if necessary by estab- lis hed techniques (redistillation, recrystallisat ion).Dimyristoyl-L- a -phosphatidylcho- line (98% ) was obtained from Sigma. Thin-layer chromatography produced a single spot. Water was distilled from a glass still. All other chemicals were of reagent grade. Preparation of Liposomes Stock solutions of the solutes were prepared at a concentration of ca. 2mg of solute per 100 cm3 of 0.15 mol dmP3 aqueous sodium chloride solution (2 x mol dm-3) at a pH from 5.5 to 6.0. Known weights of DMPC (ca. 40 mg) were dissolved in a small volume of chloroform in glass stoppered round-bottomed flasks (50 cm’). The solvent was removed by rotary evaporation to leave a uniform layer of lipid on the surface of the flask. The solute solution was added (10 cm’) and the lecithin hydrated at 35 “C (above T,) with shaking to give large multilamellar liposomes (MLV) of a size in the region of 8-10 p m as determined by Coulter-counter measurements.Attention to the hydration time and method of agitation was important in order to obtain MLV with a reproducible size range and total number. The resultant liposome suspension was divided into two aliquots and duplicates examined. Part it ion Measurements The method for following the change of partition coefficient with temperature was based on that described by Katz and D i a m ~ n d . ~ ~ - ~ ~ Flasks containing the suspended liposomal preparation were equilibrated in a shaking water bath for 48 h. The suspension was then transferred to a thermostatted centrifuge tube and centrifuged (MSE High Speed 25) at 25 000 r.p.m.for 1 h. The concentration of the solute in the supernatant was measured by U.V. spectroscopy. The supernatant was returned to the liposome pellet and the liposomes redispersed, and the contents returned to the equilibration bath. TheS. S. Davis, M. J. James and N. H. Anderson 3 17 procedure was carried out at 5,10,14,18 and 22 "C. The pH of the system was measured on each occasion. Preliminary experiments had shown that the choice of equilibration time was appropriate and that the selected combination of concentrations for the solute and lecithin enabled a wide range of partition coefficient values to be studied. The effect of the repeated processes of shaking, centrifugation, resuspension and pipetting on measured partition coefficient values was closely examined.The partition coefficients were found to remain constant within experimental error. A slight reduction in liposome size was observed during the course of the experiments. However, other preliminary studies as well as the work of Ahmed et all8 have failed to demonstrate any significant relation between liposome size and derived partition coefficient values. Since the overall time period for each temperature range could be a matter of days, attention was paid to the possible degradation of the solute. A number of substances selected originally for investigation were found to be unstable and consequently were not examined further. During each batch of experiments a standard liposome prepar- ation containing 4-methylphenol was used as a control system. Calculation of Partition Coefficient The mole-fraction concentration scale was used to aid comparison with data for partition- ing into the organic solvent.The concentration of solute in the liposames was determined from the reduction in concentration in the aqueous phase on partiti~ning.~' Organic Solvent-Water Distribution Oil-water distribution coefficients were obtained using a rapid-mix filter-probe The aqueous and organic phases were mutually saturated before use. Each experimental arrangement was thermostatted at a given temperature (hO.1 "C), and at equilibrium the concentration of the solute in the aqueous phase was determined from the U.V. absorbance measured in the linear region of the Beer-Lambert plot. Calculation of Enthalpy and Entropy of Transfer The enthalpy and entropy of transfer were derived from the temperature dependence of the partition coefficient by fitting the derived data to a straight line (ACp =r 0)" and using linear regression analysis.For the liposomes the mean standard errors ( 0 ) in AHtr and AStr were 15.8 and 11%, respectively, excepting two results giving high percentage errors because of low numerical values for AH,, and AS,,. For octan-1-01 the standard errors in AHtr and AStr were <10 and 15%, respectively, except when numerical values of AHtr and AStr were very small. For cyclohexane the standard errors in AH,, and AS,, were ca. 5% for all solutes. Differential Scanning Calorimetry The phase-transition temperature of the DMPC system and the effect on bilayer organisa- tion caused by partitioning solute (4-methylphenol) were examined by differential scanning calorimetry using a Perkin-Elmer DSC2 instrument calibrated using an indium standard.The measured pretransition temperature ( 13.5 "C), transition temperature (23.5 "C) and enthalpy of transition (23.9 kJ mol-') for DMPC liposomes compared well with literature The effect of increasing concentrations of 4-methylphenol on the phase transition was measured.318 Distribution of Phenols into Lipid Vesicles Table 2. Thermodynamic parameters for the transfer of 4-methylphenol into different batches of pure DMPC liposomes liposome AGtr(22 "C) AHtr A&" batch no. /kT mol-' / kJ mol-' /J mol K-' 1 -22.56 71.2 f 9.0 314f31 2 -21.98 88.6 f 25.0 374 f 87 3 -2 1.70 68.4 f 9.2 305 f 32 4 -2 1.96 97.3 f 18.9 404 * 66 5 -22.79 119.3 f 19.6 481 f68 6 -22.70 111.0* 12.6 453 * 44 7 -22.63 111.3* 15.9 454 f 55 8 -20.72 68.7 f 15.7 303 f 55 mean -22.13 f 0.4 92.0 f 16.5 386 f 58 " Data calculated from partition coefficient ( K ) expressed as ratio of mole fractions (standard error in AGtr is estimated as *0.4 kJ mol-'; equivalent to error in K of *20°/0).Results and Discussion Effect of Solute on Liposome Structure The perturbation of the liposome structure by 4-methylphenol was investigated by d.s.c. This showed that the enthalpy (AH,) and entropy (AS,) of transition of the liposomes decreased as the mole fraction of 4-methylphenol increased up to 0.10, the highest level measured. Even at a mole fraction of 0.01, the values of AH, and AS, were slightly lower than those for pure DMPC liposomes.The main transition temperature remained unchanged, even at the highest solute concentration, although the temperature for the onset of transition was slightly lowered. The enthalpy of the pretransition process was reduced with increasing solute concentration, until it was barely detectable at a solute mole fraction of 0.10. The solute concentration finally selected for the measurement of partition values was the lowest practicable using spectrophotometric analysis. At this concentration, which corresponded to a mole fraction of 0.011 4-methylphenol in phospholipid at 22 "C, the enthalpy of transition was 18.4 kJ mol-' as compared with 23.9 kJ mol-' in the absence of solute. The entropy of transition was reduced from 81 to ca.61 J mol-' K-'. Thus it appears that even this low solute concentration is modifying the liposome structure to a measurable extent, presumably by causing some disordering of the gel state and broadening the temperature range of the phase transition. The highest mole fraction found in the lipid phase was with 4-propylanisole, when a value of 0.020 was recorded at 22 "C. This mole fraction of 4-methylphenol 'would have reduced the enthalpy of transition to ca. 16.8 kJ mol-'. Distribution of Phenols and Anisoles General Observations Using a standard reference solute, 4-methylphenol, the free energy of transfer ( AGtr) was found to be highly reproducible between almost all liposome preparations (table 2) and most values of AHtr and AStr were the same at the 95% confidence level.However, since there did appear to be some real differences, comparisons of the partitioning behaviour of all solutes were made relative to 4-methylphenol as a reference compound.S. S. Davis, M, J. James and N. H. Anderson 319 Relative values are designated by a superscript r: A Gf;methylphenol , etc. In order to discuss the interaction of the phenolic solutes with the liposome bilayers, it is desirable to obtain data on the thermodynamics of solute hydration, so that the contribution of the loss of water of solvation to the overall transfer process can be considered. These data are not available, but partitioning from water into organic solvents such as cyclohexane and octan-1-01 is useful in considering partitioning into liposomes. Calorimetric data on a series of 3-alkoxyphenols show that removal of the solute from water accounts for some 5.2-5.8 kJmol-' and solvation by octan-1-01 for -3.6 to -2.67 kJ mol-' in the overall enthalpy of transfer of -8 to -9 kJ m01-l.~' The rapid-mix filter-probe system is particularly useful for determining the tem- perature dependence of solute partitioning into organic solvents and is believed to give more reliable data than those obtained by the 'shake flask' procedure.37 Agreement between the results reported here (tables 3 and 4) and that obtained for a number of phenols by the 'shake flask' method3* was variable.Partitioning into octan-1-01 of all the phenol solutes was associated with AHt,<O and AS,,> 0 except for three of the more polar compounds. For nine of the 18 solutes the overall free-energy change was dominated by the enthalpy term, and this is almost certainly related to the ability of the phenols to hydrogen-bond to water-saturated o~tanol.~' For these same solutes, partitioning into cyclohexane was less favoured than into octan-1-01 as typified by the results for 4-methylphenol (table 4).The enthalpy of partitioning was positive for all solutes and hence the driving force was the gain in entropy associated primarily with the removal of hydrogen-bonding solute from water. The perturbation caused to the liposome structure by the solute is likely to be particularly important for the highly structured gel phase used in this Any loss of structure would lead to a gain in entropy and a loss of enthalpy of interaction between phospholipid molecules.These changes could be larger than those caused by the interaction of solute with phospholipid, unlike the situation for partitioning into organic solvents where solute-solvent interactions predominate. The results obtained for the thermodynamics of partitioning of 4-alkyl- and 4-halogeno-phenols into DMPC lipo- somes16 using solute concentrations approximately five times greater than in this work differ from those reported here. The differences in AH,, and AS,, were particularly notable and support the view that relatively low solute concentrations can affect the liposome structure. Surprisingly, for all of the solutes used in this study, the free energy of transfer into the gel phase of DMPC liposomes was more favourable than for partitioning into octan-1-01, in spite of a large enthalpy of transfer (AHt,> 20 kJ mol-') for 13 solutes.Thus for these solutes the large entropy gain more than compensated for the large H,, term and this entropy gain must be mainly related to a loss of structural order in the gel phase caused by the introduction of the solute. Similar large entropies of transfer have been noted For three solutes AH,, was small (< 10 kJ mol-'), although still positive, taking the mean value for 4-methylphenol as a reference point, and in each case the AS,, term was also positive. Detailed discussion of the behaviour of different structural types of solute is given below. Bearing in mind the mechanistic and thermodynamic differences in the partitioning of solutes into octan-1-01, cyclohexane and liposomes, it was interesting to examine the situation in terms of free energy alone, as is normal practice in comparing partitioning into different organic phases.28 For the 16 solutes for which data were available for all three organic phases, the following relationships were found: AG~iposome = -0.50 + 0.43AG~,,,,01 r = 0.65w h) 0 Table 3.Thermodynamic parameters for the transfer of substituent groups (relative to methyl) into DMPC liposomes and into octanol phenol substituent liposomes octan-1-01 cyclo hexane liposome batch no. pK, AG:, (22 "C) AH:, AS;, AG:r (22 "C) /kJ mol-' /kJ mol-' / J mol-' K-' /kJ mol-' unsubstituted 4-methyl 4-ethyl 4-propyl 4-isopropyl 4-fluor0 4-chloro 4-bromo 2-methyl 3-methyl 4-methoxy 4-nitro 4-cyano 4-acetyl 4-methoxy- carbonyl 4-acetamido 3-hydroxy 3-methoxy 3-ethoxy 9.92 10.26 5 10.26 5 10.2 5 10.26 2 9.81 2 9.38 2 9.34 1 10.29 1 10.09 3 10.21 3 7.15 7 7.97 8 8.05 8 8.34 8 9.80 6 9.27 6 9.65 6 9.76 - 0.0 -1.42 -1.75 -3.21 -0.46 -2.11 -3.10 1.29 1.83 1.44 -0.98 -0.34 4.7 1 -2.52 5.27 -0.41 -0.36 -0.47 - 0.0 -39.2 -88.1 -83.6 -58.1 -70.4 -73.6 -39.6 -3.5 77.5 -58.8 8.2 -89.2 -71.3 -65.0 -43.6 27.6 -3.70 - 0 -125 -292 -272 -198 -23 1 -239 -139 -19 - 196 258 29 -318 -232 -238 - 146 95 -11 2.46 0.0 -2.46 -4.69 1.06 - -3.21 -4.09 -0.27 -0.21 3.38 0.16 1.89 3.25 0.30 - 6.6 1 2.29 -0.02 AS[r AG:r (22 "C) AH:, As:, /kJ mol-' / J mol-' K-' /kJ mol-' /kJ mol-' /Jmol-' K-' b 0.9 -5 4.22 0.7 -11 G -1.9 2 -3.25 -1.2 7 3 9 E: 0.0 0 0.0 0.0 0 g.8 % -1.6 11 -6.97 -5.0 - -3.8 -8.6 -8.7 0.4 -0.3 5 .O -10.2 -0.7 -4.3 -12.1 - -17 -18 -15 3 0 6 -35 -8 -25 - 42 - - - 3.55 -0.7 - 14 3 E 3.56 1.8 -4 $ s 0.12 -2.6 -9 0 -0.92 -3.9 -10 z -1.18 -3.5 -7 0.5 1 -0.6 -3 6.78 9.8 11 11.83 28.2 56 10.65 14.5 13 CI.E s z. 6.48 7.4 4 2 2 - - - - - - - - -8.0 -49 -1.4 -7 4.1 1 -2.0 -20 0.1 1 1.25 -2.2 -11S. S. Davis, M. J. James and N. H. Anderson Table 4. Thermodynamics of transfer of 4-methylphenol to octan-1-01 and cyclohexane 321 AG,,(22 "C)" AH,, AS,," solvent /kJ mol-' /kJ mol-' /J mol-' K-' octan- 1-01 -16.55 -7.3 31 cyclohexane -3.67 18.6 75 " Data calculated from partition coefficient expressed as ratio of mole fractions. lo3 K/ T 3 . L 3.5 3.6 6 Fig. 2. The effect of temperature on liposome-water partition coefficient (log,, K ) for some substituted phenols (corrected for differences between liposome preparations): ( 1) 4-nitro, (2) 3-hydroxy, (3) 4-acetyl, (4) 4-methyl and (5) 4-methoxy.Thus in free-energy terms, octan-1-01 is a reasonable model for the liposomes, particularly if the two compounds which deviate from +his relationship (fig. 3) are excluded, to give eqn (3): AG&,osome = -0.53 + 0.48AG~,,,,,1 r = 0.74 (3) (data for 3-hydroxy- and 4-acetyl-phenols excluded). With cyclohexane no improvement to the relationship given in eqn (2) (which is not significant at the 90% confidence level) could be found by similarly excluding one or two solutes. Thus we conclude that for solute with hydrogen-bonding properties octan-l- 01 is a much better model than cyclohexane for partitioning into gel-phase phospholipid bilayers, although in thermodynamic terms neither is satisfactory.A 1 ky lp h en o 1s The partitioning of 4-methylphenol into the DMPC liposomes is expected to be favoured by its structure, with a polar hydrogen-bonding group attached to an essentially planar322 Distribution of Phenols into Lipid Vesicles hydrocarbon group. Insertion of such a molecule into the gel structure is likely to cause some local disorder and hence entropy gain, but the phenolic hydroxy group should be able to hydrogen bond to the phosphate ester while the hydrocarbon part of the molecule could interact via van der Waals forces with the hydrocarbon chains of the phospholipid. The expected trends were observed in that the higher alkyl compounds partitioned more readily into the liposomes, with the partitioning being entropy driven.The enthalpy of removal from water is unlikely to vary greatly with the length of the alkyl and therefore the differences in AH:, relate largely to differences in interactions with the phospholipid bilayer. The insertion of the 4-methyl- and 4-ethyl-phenols is highly disfavoured in enthalpy terms, and this must be related to the loss of favourable interactions between the phospholipid molecules. However, a large increase in entropy results from this disturbance to the gel state of the phospholipid, and so AG:, is more favourable than for partitioning into octan-1-01. The smaller AH:, and AS:r terms for the higher alkyl phenols imply that these compounds cause less disturbance to the phospholipid structure or are more immobilised within it.It was notable that the isopropyl phenol partitioned into the liposomes to a greater extent than the n-propyl analogue, since the opposite is observed for partitioning into organic solvents, and chain branching can reduce permeation rates through membrane^.^ The lower AG:, for the isopropyl compound appears to be due to the entropy gain outweighing the less favourable AH:r. Substituent Position The effects of substituent position would be expected to be more important for partition- ing into a gel phase where molecular shape and spatial distribution of functional groups could be important, than into an isotropic organic solvent and the comparative data for liposomes and two organic solvents shows this to be the case.In cyclohexane the enthalpy of transfer of 2-methylphenol, in which the methyl group is in close proximity to the phenolic -OH, is lower than that for the 3- and 4-methyl analogues. A similar pattern was found for partitioning into liposomes, although here the difference in AH:, was much larger (table 3). In the liposome compensating change in AS:, resulted in a free energy of transfer close to those of the 3- and 4-methyl compounds. In the case of cyclohexane it could be argued that the shielding effect of the 2-methyl group is reducing the enthalpy of solvation by water and hence facilitates the transfer to a non-polar environment. However, this argument cannot explain the large difference in behaviour observed between the 2- and 4-methylphenols partitioning into liposomes.If the phenolic hydroxy group was in the polar head-group region, the adjacent methyl group would not penetrate far into the membrane structure and therefore could be less perturbing than the 3- or 4-methyl groups. Hence the entropy and enthalpy of transfer of the 2-methyl group would be expected to be lower than that of the 3- or 4-methyl groups. Further experiments with other compounds such as 2-, 3- and 4-methyl- acetanilide would be needed to test this hypothesis. Halogenop h enols The qualitative picture was found to be similar to that for the alkyl phenols in that AH:, and AS:, both decreased with increasing size and lipophilicity of the halogen. However, AH:, was smaller at a given AG:, value than for the alkylphenols, and this may be related to the lower pK, of halogenated species, particularly the 4-chloro- and 4-bromo- phenols (pK, 9.38 and 9.34, respectively) as compared with 4-methylphenol (pK, 10.26), since a lower pK, would lead to stronger hydrogen bonding between the phenolic OH and phosphate head group.38S.S. Davis, M. J. James and N. H. Anderson 323 L - 2 - 0 - . 2 - 0 0 0 3. 4. 0 5 0 8 O 6 0 70 2 0 - 6 -1 - 2 0 2 L 6 AG[,/ kJ mol-' (octan-1-01) Fig. 3. The relative free energy of transfer (AG:,) of substituted phenols into DMPC liposomes and octan-1-01. 0, Polar substituents; @, alkyl and halogeno substituents. Phenol substituents as follows: (1) 4-acetyl, (2) 3-hydroxy, (3) 3-methyl, (4) 2-methyl, ( 5 ) 4-methyl, (6) 4-nitro and (7) 4-methoxycarbonyl. Polar Phenols Results with phenols containing a number of more or less polar substituents are presented in table 3. Several different effects can be envisaged from the introduction of polar groups into the parent phenol.The acidity and hydrogen-bonding ability of the phenolic OH group may be altered. The polar group itself may be able to act as a hydrogen-bond donor or acceptor. The dipole moment of the molecule will be changed. As discussed above, the position (3- or 4-), size and shape of the substituent may also be important in determining interactions in a highly structured gel phase. Considering the relative free energies of transfer to the liposomes with those into octanol, where the latter can be regarded as the lipoidal phase with hydrogen-bonding capability, it can be seen from fig.3 that partitioning into liposomes broadly paralleled that into octan-1-01 for polar substituents, excepting the 3-hydroxy and 4-acetyl groups. The former partitioned more into liposomes than expected from its behaviour with octanol, while the converse was true for the latter. In the past AH vs. AS relationships have usually been used to determine if a common mechanism underlies a particular process, e.g. solute transfer from aqueous to organic phases.41 However, when AH and A S are both derived from the van't Hoff relationship, colinearity can arise because errors in these two terms are correlated. In this situation to test for enthalpy-entropy compensation AG should be plotted against AH at the harmonic mean temperature of the e~perirnents.~~ For octan-1-01 there was no significant linear relationship between AG,, (at the harmonic mean temperature of the experiments, 15 "C) and AH,, for the substituted phenols, but from a graphical plot it could be seen that the 3-hydroxyphenol did not follow the general pattern. This phenol is the only one with two hydrogen-bond donor groups, and thus it might form a ternary complex with two molecules of octanol.This could explain the relatively high loss in entropy and hence unfavourable AG,, for the observed AH,,. Even when data for this compound were excluded, the relationship between AG,, and AH,, was still not significant, showing that the relative importance of the different factors affecting partitioning into octan-1-01 varied between solutes. For cyclohexane, however, AG,, at the harmonic mean temperature of 26 "C was related to AH,, for those phenols where data were available (table 3): AGt, = - 11.4 + 0.47AHt, r = 0.84.(4)324 - 4 - -6 Distribution of Phenols into Lipid Vesicles 0 06 0 . - I I I 1 I I I 1 01 0 3 Fig. 4. The relative free energy and enthalpy of transfer (AG:,, AH:r) of substituted phenols into DMPC liposomes at 286.7 K. 0, Polar substituents; 0, alkyl and halogeno substituents. Phenol substituents as follows: ( 1 ) 4-methoxy, (2) 3-methoxy, (3) 3-ethoxy, (4) 3 hydroxy, ( 5 ) 4-nitro, (6) 4-methoxycarbonyl, (7) 4-acetamido and (8) 4-acetyl. Kinkel et al. have made a similar comparison of solute transfer into a hydrocarbon solvent, with similar results.37 The essential difference between the two systems is that partitioning into a pure hydrocarbon is dominated by the enthalpy and entropy of solute removal from water, whereas with octan-1-01 significant solute-octanol interactions can occur.Enthalpy and entropy of hydration tend to be inter-related,43 and hence AG,, and AH,, are related for transfer into cyclohexane. Given this background, no such relationship was expected for phenol transfer into DMPC liposomes, but examination of the data in graphical form (fig. 4) showed a marked trend excepting two solutes. Excluding the data for these compounds (4-acetyl and 4-acetamidophenols) the following relationship was found: AG,, = 0.03 +0.052AHt, r = 0.91 ( n = 16). For solute transfer to liposomes the variation in AH,, was some 20 times greater than that in AGt, [eqn ( 5 ) ] , in contrast to the results for cyclohexane. It is difficult to explain why A G,, was particularly unfavourable for the 4-acetyl and 4-acetamido groups, since other similar polar substituents such as nitro and methoxycarbonyl showed quite different behaviour.All these four polar substituents gave low enthalpies of transfer, but entropies of transfer were very unfavourable so that the AGtr represented the small difference between these opposing forces. The thermodynamic data for the 3-hydroxy group also followed this pattern. It is probable that this type of behaviour is related to the hydrogen-bonding ability of these polar substituents, which resulted in a strong interaction between the polar substituent and the phospholipid in the case of -OH and -NHCOMe, or interaction via the water associated with the pho~pholipid~~ in the case of the other substituents.Additionally, the increased acidity of the phenolic hydroxy group in three of these compounds (table 3) would also strengthen the hydrogen-bonding ability of the -OH. Thus a strong interaction with the phospholipid-water complex could occur with both the -OH and polar substituent, resulting in a negative substituent enthalpy of transfer counterbalanced by an associated loss of entropy, relative to the methyl substituent. The other three polar substituents, 4-cyano, 4-methoxy and 3-methoxy, gave a positive enthalpy and entropy of transfer relative to 4-methyl, resulting again in only small free-S. S. Davis, M. J. James and N. H. Anderson Table 5.Thermodynamic parameters for transfer of substituted anisoles to DMPC liposomes relative to 4-methylphenol and comparison with related phenols anisole liposome AG:,(22 "C) A K r substituent batch no. /kJ mol-' /kJ mol-' /J mol-' K-' 4-chloro 4 -5.79 4.8 36 4-bromo 4 -5.23 -7.7 -8 4-propyl 3 -8.39 -6.2 8 4-nitro 3 -1.92 -13.2 -5 1 4-nitro octan-1-01 -0.16 6.3 22 anisole liposome AGOH-OMen AHOH-OMea ASOH-OMea substituent batch no. /kJ mol-' /kJ mol-' /J mol-' K-' 4-choro 4 3.68 -75.2 -267 4-bromo 4 2.13 -65.9 -231 4-propyl 3 6.64 -81.9 -300 4-nitro 3 -2.80 -45.6 -145 4-nitro octan- 1-01 0.32 -16.5 -57 325 AGfr for phenol -AG:r for anisole, etc. energy differences relative to this reference substituent. Thus the driving force for transfer is an entropy gain, as for the 4-methyl compound.It seems likely that for these compounds the relatively polar substituent is introduced into the non-polar region of the phospholipid bilayer. The smallest polar substituent, -CN, is closest to -Me in its behaviour, whereas the 4-methoxy group has a large AH,, and AStr. The pK, of the 4-methoxy compound is slightly higher than that of the 4-methyl reference solute, and thus the hydrogen bonding of the former would be expected to be slightly less. However, it appears that the introduction of the -0Me into the lipid region of the liposome is energetically very unfavourable. The 3-methoxy compound is more acidic than the 4-methoxy analogue, and this factor, combined with the difference in geometry from the 4-methoxy analogue, may contribute to its lower AHt, and AStr.Extension of the 3-methoxy to 3-ethoxy leads to a reduction in AHtr and AStr (table 3), as would be expected from the behaviour of the alkylphenols. An isoles In order further to explore the importance of the interaction of the phenolic hydroxy group, the behaviour of a number of anisoles in which the hydroxy group was replaced by methoxy were examined. Results are given in table 5. For all five anisoles, both the enthalpy and entropy change on transfer were more positive than for the corresponding phenols and, except in the case of the 4-nitro analogue, resulted in a more favourable free energy of transfer. These results are consistent with the hypothesis that the hydrogen bonding of the phenolic OH group is important in determining the partitioning behaviour of the phenols and lowers the enthalpy of transfer.31 It was interesting to note that the free-energy change caused by replacement of the acidic hydroxy of nitrophenol by methoxy was positive, contrary to the situation observed for partitioning into octan-1-01 and cyclohexane.In the liposome it is possible that this compound is orientated with the nitro group in the polar outer region and the methoxy in the lipid interior, an arrangement quite different to that envisaged for 4-nitrophenol.326 Comparison of Solute Transfer into Liposomes and Erythrocyte Membrane The partition coefficients of a number of phenols into erythrocyte (red blood cell) membranes has been estimated (ratio of solute concentration in molal units).45 Linear relationships between these data and the corresponding partition coefficients (based on molal concentrations) into liposomes and octan-1-01 have been examined.Using data on 4-methyl-, 4-fluoro-, 4-chloro-, 4-bromo-, 4-methoxy- and 4-methoxycarbonyl-phenols the following equations were derived: Distribution of Phenols into Lipid Vesicles log Kmembrane = -2.O-k 1.35 log Kliposome log Kmembrane = -0.59 + 0.89 log Ko,tanol r = 0.97 r = 0.99. Considering the small number of compounds the relationship between membrane and liposome partitioning is satisfactory. However, eqn (7) shows that octan-1-01 is also an acceptable model in free-energy terms. (6) (7) Conclusions The results reported here demonstrate that the thermodynamics of partitioning of solutes into the liposome gel phase can be radically different from those into isotropic organic solvents and also different in quantitative terms for the pattern typically observed above the transition temperature. The steric and electronic properties of the solutes are more important in determining behaviour than is the case for solvents.Both the d.s.c. and partitioning data suggest that even small quantities of solute (on a mole fraction basis) can cause significant disruption to the phospholipid gel phase, and this disruption needs to be taken into account in interpreting the thermodynamic data on partitioning. Further interpretation of our results would involve the investigation of solute- phospholipid interactions and liposome structure using a technique such as n.m.r.spectroscopy with appropriate probe molecules as necessary. We thank the S.E.R.C. for a research grant and with ICI plc for a CASE award for M.J.J. References 1 Y. C. Martin, Quantitative Drug Design, A Critical Introduction (Marcel Dekker, New York, 1978). 2 B. E. Cohen, J. Membr. Biol., 1975, 20, 205. 3 S. S. Davis, T. Higuchi and J. H. Rytting, Adu. Pharm. Sci., 1974, 4, 73. 4 R. Collander, Physiol. Plantarum, 1954, 7, 420. 5 B. E. Cohen, J. Membr. Biol., 1975, 20, 235. 6 Y. C. Martin, J. Med. Chem., 1981, 24, 229. 7 S. A. Simon, W. L. Stone and P. Busto-Latorre, Biochim. Biophys. Acta, 1977, 468, 378. 8 C. Treiner, J. Colloid Interface Sci., 1983, 93, 33. 9 V. A. Levine, D. Dolginow, H. D. Landahl, C. Yorke and J. Csejtey, Pharm. Res., 1985, 259. 10 J. M. Diamond and Y.Katz, J. Membr. Biol., 1974, 17, 121. 11 J. A. Dix, J. M. Diamond and D. Kivelson, Proc. Natl Acad. Sci. USA, 1974, 71, 474. 12 W. L. Stone, J. Biol. Chem., 1975, 250,4368. 13 M. Tomkiewicz and G. A. Corker, Biochim. Biophys. Acta, 1975, 406, 197. 14 K. W. Miller, L. Hammond and E. G. Porter, Chem. Phys. Lipids, 1977, 20, 229. 15 M. K. Jain and L. V. Wray, Biochem. Pharmacol., 1978, 27, 1294. 16 J. A. Rogers and S. S. Davis, Biochim. Biophys. Acta, 1980, 598, 392. 17 M. Ahmed, J. S. Burton, J. Hadgraft and I. W. Kellaway, Biochem. Pharmacol., 1980, 29, 2361. 18 M. Ahmed, J. S. Burton, J. Hadgraft and I. W. Kellaway, J. Membr. BioL, 1981, 58, 181. 19 P. G. Ruijrok, Naunyn-Schmied Arch. PharmacoL, 1982, 319, 185. 20 M. Arrowsmith, J. Hadgraft and I. W. Kellaway, Biochim. Biophys. Acta, 1983, 750, 149. 21 M. M. Saket, K. C. James and I. W. Kellaway, Int. J. Pharm., 1984, 21, 155. 22 P. Chatelain and R. Larvel, J. Pharm. Sci., 1985, 74, 783. 23 W. R. Galey, J. D. Owen and A. K. Solomon, J. Gen. PhysioL, 1973, 61, 727. 24 E. M. Wright and N. Bindslev,-J. Membr. Biol., 1976, 29, 289.S. S. Davis, M. J. James and N. H. Anderson 327 25 H. Hauser and M. C. Phillips, Prog. Surf: Membr. Sci., 1979, 13, 297. 26 A. G. Lee, Proc. Nutr. Soc., 1985,44, 147. 27 W. Riebesehl and E. Tomlinson, J. Phys. Chem., 1984, 88, 4770. 28 A. Leo, C. Hansch and D. Elkins, Chem. Rev., 1971, 71, 525. 29 I. Kojima and S. S. Davis, Int. J. Pharm., 1984, 20, 247. 30 A. E. Beezer, W. H. Hunter and D. E. Storey, J. Pharm. Pharmacol., 1983,35, 350. 31 N. H. Anderson, S. S. Davis, M. James and I. Kojima, J. Pharm. Sci., 1983, 72, 443. 32 Y. Katz and J. M. Diamond, J. Membr. B i d , 1974, 17, 69. 33 Y. Katz and J. M. Diamond, J. Membr. B i d , 1974, 17, 87. 34 Y. Katz and J. M. Diamond, J. Membr. Biol., 1974, 17, 101. 35 D. A. Wilkinson and J. F. Nagle, Biochemistry, 1981, 20, 187. 36 S . Mabrey and J. M. Sturtevant, Methods Membr. Biology, 1978, 9, 237. 37 J. F. M. Kinkel and E. Tomlinson, Int. J. Pharm., 1981, 9, 121. 38 J. A. Rogers and A. Wong, lnt. J. Pharm., 1980,6, 339. 39 A. G. Lee, in Membrane Fluidity in Biology, ed. R.C. Aloia (Academic Press, New York, 1983), vol. 2, 40 J. Rubin, B. Z. Senkowski and G. S. Panson, J. Phys. Chem., 1964, 68, 1601. 41 J. E. Leffler and E. Grunwald, Rates and Equilibria of Organic Reactions (Wiley, New York, 1963), p. 128. 42 R. R. Krug, W. G. Hunter and R. A. Grieger, J. Phys. Chem., 1976,80, 2335. 43 J. A. V. Butler, Trans. Faraday Soc., 1937, 33, 229. 44 0. H. Griffith, P. J. Dehlinger and S. P. Van, J. Membr. B i d , 1974, 15, 159. 45 H. Machleidt, S. Roth and P. Seeman, Biochim. Biophys. Acta, 1972, 255, 178. p. 43. Received 13th December, 1985

ISSN:0301-7249    

DOI:10.1039/DC9868100313

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

Faraday Discuss. Chem. SOC., 1986, 81, 329-337 Membrane-spanning Symmetric and Asymmetric Diyne Amphiphiles Hubert Bader and Helmut Ringsdorf * Institut f u r Organische Chemie, Johannes Gutenberg- Uniuersitat, 0-6500 Mainz, Federal Republic of Germany Symmetric membrane-spanning diyne amphiphiles of different molecular designs, e.g. single-chain and identical or mixed double-chain compounds, have been synthesized, starting from 10,12-docosadiyne-l,22-diol (1) and 10,12-docosadiynylenedihydrogendirnaleate (2), respectively. Esterification of 2 and/or addition of functional thiols at the maleic acid double bond yield symmetric a,o -dipolar diyne amphiphiles which show structural resemblance to lipids found in the biomembranes of archaebacteria. Asym- metric membrane-spanning diyne amphiphiles are obtained by a straightfor- ward route from 22-hydroxy- 10,12-docosadiynoic acid ( 12) by acylation with functional aromatic dicarboxylic acid anhydrides.Vesicles are formed from a number of symmetric and asymmetric diyne lipid analogues on dispersion in aqueous media, as shown by light- and freeze-fracture electron microscopy and dye entrapment experiments. Stabilization of lecithin liposome mem- branes towards temperature-induced dye release is effected by incorporation of a macrocyclic membrane-spanning diyne amphiphile. Ultraviolet irradi- ation of dispersions from asymmetric diynes yields blue and red solutions, indicating that topotactical diacetylene polymerization takes place in these lipid assemblies. Membrane-spanning Lipids and their Relevance in Natural and Artificial Membranes Although known since the late 1970s,' the feasibility of a,w-dipolar lipids and amphi- philes, in which the hydrocarbon chains span the hydrophobic membrane core, in reconstituted and artificial model membranes has not yet been fully appreciated.Only recently have biologists started to investigate the compositions of the membrane- spanning lipid fractions isolated from archaebacteria, a class of micro-organisms whose membranes have been found to resist extreme environmental conditions, such as strongly acidic pH, high salt concentrations and abnormally high temperatures.* These bio- membranes feature unusual lipid structures such as dipolar isoprenoid glycerol tetra- ethers (fig. 1 ) and comprise a major fraction of the membrane lipids in these bacteria.They are assumed to play an active role in stabilization against physical stress (tem- peratures up to 110 "C) and chemical stress (pH 1). Upon dispersion in aqueous media, these lipids, like naturally occurring lecithins and synthetic amphiphile~,~ form liposomes (vesicles): spherically closed mono- or multi-compartment model membranes used as powerful tools in membrane research, compartimentation processes and drug delivery in uiuo. Okahata and Kunitake' were the first to synthesize dipolar quarternary ammonium amphiphiles, which give liposomes if the chain between the polar headgroup is of an appropriate length. The first reported polymerizable membrane-spanning surfactants were derived from 10,12-docosadiyne-l,22-diol ( l).4 More recently, Fuhrhop' and coworkers prepared dipolar bipyridinium compounds and established the reaction of thiols with maleic acid esters as a simple route to functional dipolar lipid analogues.329330 Membrane-spanning Diyne Amphiphiles r OR’ s 0 - P- PHOSPHATIDYLGLYCEROL I 0- Fig. 1. Isoprenoid ( C40) glycerol tetraethers from archaebacterial membranes. Fig. 2. Schematic representation of polymerizable bipolar amphiphiles. In this article we describe attempts to vary the structure of polymerizable dipolar amphiphiles systematically and investigations of the properties of these membrane- spanning molecules. Structural Variations of Membrane-spanning Amphiphiles In general, the schematic structures for the symmetric synthetic analogues of archaebac- terial lipids as given in fig.2 can be envisaged. Saturated single-chain and macrocyclic double-chain dipolar amphiphiles have been reported recently.6 The polymerizable symmetric lipid analogues corresponding to the structures given in fig. 2 are summarized in tables 1 and 2. Experimental The synthesis of the starting compound 10,12-docosadiyne-l,22-diol (1) has been described el~ewhere.~H. Bader and H. Ringsdorf Table 1. Symmetric esters from 10,12-docosadiynylenedihydrogendimaleate (4) C02[CH2]9-CGC-C-C-[CH2]902C COzR R02C 331 _ _ _ ~ ~~ compound R Table 2. Symmetric derivatives from 10,12-docosadiynylenedithiosuccinic acid and its esters A A X-S CO2H HO2C S-X compound X 7 -CH,C02H 8 -[CH212C02H 9 - [CH2I2SO3Na 10 4CH212NH2 C02[CH2]9-CfC-C=C-[CH2]902C C02[CH2]9-C-C-C-C-[CHz]gO2C SOH2COzH and (11) H02CCH2S 10,12-Docosadiynylenedihydrogendimaleate (2) 3.35 g (10 mmol) of 1 and 1.96 g (20 mmol) maleic acid anhydride in 60 cm3 toluene were refluxed for 4 h, the solvent removed and the solid residue recrystallized from an ethyl acetate-hexane mixture.Yield 4.5-5 g (85-95%), m.p. 81 "C. C30H4208 (530.66): calc. C 67.90, H 7.98; found C 67.48, H 8.00. The symmetric esters 3-5 were obtained by routine procedures7 as oily or waxy substances. Bis( 10,12-docosadiynylene)dimaleate (6) 2.12 g (4 mmol) of 2 and 0.5 g p-toluenesulphonic acid in 400 cm3 toluene were kept under reflux, as a solution of 1.34 g (4 mmol) of 1 in 50 cm3 toluene was added dropwise for 12 h. The reaction mixture was then refluxed for another 2 h, the solvent evaporated and the waxy residue subjected to column chromatography, using silica gel and a hexane-ethyl acetate 6 : 1 mixture as eluent.The product fractions were combined, the organic solvent removed and the solid residue recrystallized from ether. Yield 0.53-0.93 g332 Membrane-spanning Diyne Amphiphiles (16-28%), m.p. 49-50 "C. H 9.49. acid and its esters, a typical procedure is given for 8. C52H7608 (829.18): calc. C 75.32, H 9.24; found C 74.91, For the synthesis of symmetric derivatives from 10,12-docosadiynylenedithiosuccinic 2,2'-Bis(carboxyethylthio)-3,3'-( 10,12-docosadiynylenedioxydicarbonyl)dipropionic Acid (8) 531 mg (1 mmol) of 2 and 117 mg (1.1 mmol) 2-mercaptopropionic acid were dissolved in 10 cm3 dry THF. 2 cm3 of N,N,N',N'-tetramethylethylenediamine (TMEDA) were added and the mixture stirred overnight at room temperature.The solvent was evapor- ated, the oily residue dissolved in ethyl acetate and the solution was washed consecutively with 1 mol dm3 H2S04 and water. After drying and removal of the solvent, the compound was further purified on silica gel using a CH,CI,-MeOH 9 : 1 mixture as eluent. Yield 490 mg (66%), m.p. 143 "C. C,,H,,O,,S, (742.94): calc. C 58.20, H 7.33, S 8.63; found C 56.83, H 8.11, S 8.98. 22-Hydroxy-10,12-docosadiynoic Acid (12) 3.3 g (10 mmol) of 10,12-doc~sadiynolid~ were added to a solution of 1 g sodium methoxide in 20 cm3 dry methanol; the mixture was warmed until a clear solution was obtained and then stirred at room temperature for 1 h. Methanol was removed, cold 1 mol dm-3 H,SO, was added and the product was extracted into ethyl acetate.The organic phase was washed with water, dried over MgSO, and evaporated to dryness. The solid residue was recrystallized from an ether-hexane mixture. Yield 2.5 g (71%), m.p. 70-73 "c. C22H3603 (348.53): calc. C 75.82, H 10.41; found c 75.65, H 10.44. l-Carboxy-lO,l2-docosadiynyl-22-(22,4-dicarboxy)benzoate (13) 800 mg (2.3 mmol) of 12 and 460 mg (2.4 mmol) trimellitic acid anhydride were dissolved in 25cm3 CHCl,; 250mg (2.5mmol) triethylamine was added and the mixture was stirred at room temperature for 72 h. The solvent was removed, the solid residue triturated repeatedly with ice-cold 1 mol dmP3 HCl, filtered off , washed with water and dried. For further purification, the compound was recrystallized from an acetone-hexane mixture. Yield 1 g (8O%), m.p.121 "C. C3,H4008 (540.65): calc. C 68.87, H 7.46; found C 68.38, H 7.43. Synthesis of Symmetric Dipolar Diyne Amphiphiles Glaser's oxidative coupling of functional terminal acetylenes has proved useful for the synthesis of long-chain dipolar diynes such as 10,12-docosadiyne-l,22-diol (1) and 22-hydroxy-10,12-docosadiynoic acid (12). These are the starting materials for symmetric and asymmetric membranes-spanning diacetylene lipid analogues, potentially polymeriz- able by U.V. light., 10,12-Docosadiynylene-dihydrogendimaleate (2), obtained from the reaction of (1) with maleic anhydride, is readily converted into a variety of amphiphilic compounds by esterification (route A) and/or addition of functional thiols at the maleic acid ester group (route B), as shown in scheme 1.Thus the possible structures for symmetric polymerizable dipolar amphiphiles given above may easily be realized. Synthesis of Asymmetric Dipolar Amphiphiles Basically, all the lipid structures for symmetric amphiphiles can be adopted for their asymmetric counterparts, which have two headgroups differing in chemical structure,H. Bader and H. Ringsdorf 333 COOR X-S COOH,R Scheme 1 HC=C-(CH2)9OOC(CH2)8-CECH I 1 f O O C > (CH,), C2,-diynolid j + HO(CH2)9-CEC-CEC- (CH2)g-COOH (12) Scheme 2 charge and size. Chemical modification of the headgroups of symmetric compounds to yield asymmetric amphiphiles gave poor yields and difficult purification problems and was therefore abandoned.In contrast to this strategy, the a priori asymmetric 22-hydroxy-l0,12-docosadiynoic acid (12) was prepared from 10-undecanoic acid 10-undecanoate via 10,12-docosa- diynolid in high yields according to scheme 2. The o-hydroxy carboxylic acid was selectively converted into single-chain asymmetric anionic dipolar amphiphiles with headgroups of distinctively different degrees of bylkiness by reaction with cyclic aromatic dicarboxylic anhydrides. The structures of these membrane-spanning diacetylene surfac- tants are given in table 3 . Behaviour of the Dipolar Diyne Amphiphiles at the Air-Water Interphase The synthetic lipid analogues were spread from organic solutions at the air-water interphase of a Langmuir film balance to evaluate their ability to form stable lipid monolayers.Not surprisingly, all the compounds exhibit fluid phases at fairly large surface areas per molecule, indicating an intense interaction of both headgroups at the interphase. Under all conditions applied, e.g. variations in temperature, salt content and pH of the subphase, no crystalline or solid-analogue phases could be observed. This is exemplified by the surface pressure-area diagram in fig. 3 . The extreme fluidity334 Membrane-spanning Diyne Amphiphiles Table 3. Asymmetric diyne amphiphiles Y-(CH,),-CEC-CEC-(CH,),-COOH compound Y 13 14 15 C02H C02H -i 20- E z E t: 10- ,(4) 1 A/nmz molecule-' Fig. 3. Surface pressure-area diagram of 4 and 11 at 20 "C on water. found in archaebacterial lipids3 has an analogy in the synthetic amphiphiles' lack of bulky methyl side groups in the isoprenoid chains of their natural analogues.The assumption that dipolar amphiphiles become erect in order to build a palisade- like solid monolayer when the surface area per molecule is reduced could not be substantiated by these findings. Despite this, attempts to obtain solid-analogue monolayers from asymmetric dipolar amphiphiles with headgroups markedly different in polarity still appeared promising. Expectations that the less polar and hydrophilic hydroxy group in 12 would be pushed out of the air-water interphase at smaller surface areas per molecule, thus allowing pressure-area diagrams similar to those of mono- functional long-chain carboxylic acids, were frustrated by the results from spreading experiments (fig.4). A further increase in the predominant hydrophilicity of the carboxy- lic acid headgroup by addition of CdCl, to the subphase had no significant effect.H. Bader and H. Ringsdorf 335 E A/nm2 molecule-' Fig. 4. Surface pressure-area diagram of 12 on water (a) and CdC12 solution (b). Liposomes from Dipolar Amphiphiles Spherically closed single-wall ('monolayer') liposomes can be generated by either swel- ling lipid films in a humid atmosphere at elevated temperatures (80 "C) or by ultrasonica- tion. The swelling method affords large tubular structures similar to the myelin figures observed with conventional lipids from which large (ca. 1 pm) liposomes are tied off (plate 1). As can be seen by light microscopy from the fluctuations of the membranes, the fluidity of the dipolar lipids is not affected by cooling the samples to temperatures as low as 4°C.From preliminary liposome experiments it became obvious that only double-chain symmetric amphiphiles are useful candidates to build stable compartments by themselves. The moderately polar single-chain symmetric diynes form liposomes only in the presence of ch~lesterol,~ whereas the highly polar sulphonic acid compound dissolves in water. Dye (carboxyfluorescein) entrapment experiments show the existence of enclosed inner aqueous compartment in liposomes from symmetric diyne lipids. The efficacy of enclosure is remarkably higher than in liposomes from conventional lipids owing to the larger diameter of these vesicles. The high fluidity of the monolayer membranes is responsible for the fairly rapid release of entrapped marker from the inner space.Incorporation of cholesterol significantly reduces the permeability of these liposomal membranes. Mixed Liposomes from Dimyristoyl Phosphatidylcholine and a Macrocyclic Diyne Amphiphile (1 1) Although liposomes from a single lipid species are useful in studying basic membrane properties, those from mixtures of structurally different lipids are a closer match to the complexity of membrane constituents in natural systems. With membrane-spanning amphiphiles as components in mixtures with conventional bilayer-forming lipids one might assume that the former would act like integral proteins, which also span the membrane and strongly stabilize the bilayer lipid arrangement. The rationale for this effect comes from a thermodynamic estimation.To create membrane defects and remove a membrane-spanning amphiphile embedded in a dilayer assembly, one needs to strip the inner hydrophilic headgroup from its water shell and push it through the hydrophobic membrane core. However, this process is energetically very unfavourable and therefore improbable. Freeze-fracture electron microscopy confirms the uniform mixing of the membrane- spanning macrocyclic amphiphile 11 with the phospholipid dimyristoyl phos- phatidylcholine (DMPC) at a 1 : 1 molar ratio at 17 and 4 "C (plate 2). The differences in headgroup structure and chain length do not induce a phase separation, and the336 100 80 60- W - 2 8 10- 20 Membrane-spanning Diyne Amphiphiles - X A A 0 - A 0 0 0 8 0 8 O A 8 - 8 8 I 2; 3b 35 10 45 50 T / "C Fig.5. Dye release on heating from liposomes from x, DMPC; A, DPPC; 0, 11 and U, 1:l mixtures of 11 and DMPC. mixed vesicles are devoid of 'ripple structures' characteristic for pure phospholipid domains. At higher temperatures-the increased mobility of the lipids induces more defects and the liposome membrane will become more permeable toward entrapped molecules. The release of the highly water-soluble fluorescent marker carboxyfluorescein (CF) from mixed liposomes on heating is a valuable test for the above hypothesis. The results of this experiment are shown in fig. 5. For comparison, liposomes from pure DMPC were prepared with 50 mmol dm-3 CF at 25 "C. These liposomes completely lose the entrapped dye during passage over a g.p.c.column, whereas vesicles from pure samples of 11 leak the marker gradually as the temperature is raised, with 87% of the originally entrapped fluorescent molecules being released at 50°C. In contrast to the pure components, the membrane permeability of liposomes from a 1 : 1 mixture of DMPC and 11 increases slowly, and only 43% of the dye diffuses from the inner compartment at 50 "C. This result suggests a stabilizing effect towards temperature-induced membrane permeability exerted by the membrane-spanning lipid in mixtures with natural phospholipids. Polymerization of Asymmetric Diyne Amphiphiles in Monolayer Liposomes The symmetric dipolar diyne amphiphiles reported here could not be polymerized by U.V. irradiation of their aqueous dispensions.Owing to the high fluidity of these vesicle membranes, the topotactical diacetylene polymerization does not take place at low temperatures and in the presence of cholesterol. Difficulties in packing the outer headgroups of membrane-spanning amphiphiles in a space-filling arrangement in small vesicles lead to the suggestion that asymmetric amphiph.iles would orientate preferentially in a such a way that the bulkier headgroup is located in the outer membrane sphere. This was proved by n.m.r. experiments using the line-broadening effect of externally added manganese ions on the aromatic proton absorption. This arrangement of the diyne amphiphiles also allows the correct spacing of these monomers for topotactical diacetylene polymerization. The clear colourless vesicle suspensions from the asym- metric diynes (13)-(15) become deep blue on U.V.irradiation at 254 nm, an indication of the fully conjugated polymer backbone of polydiacetylenes? Outlook Membrane-spanning lipids from natural and synthetic sources are still considered a curiosity among the well known conventional lipids' and their potential cannot yet be evaluated. From our preliminary work with these amphiphiles we have reason to believe that their unusual features will make them useful membrane constituents in monolayers,Plate 1. Liposomes obtained from 11 by the swelling method (light microscope). ( To face p. 336)Plate 2. Freeze-fracture electron micrograph of a 1 : 1 mixture of 11 and DMPC.H. Bader and H. Ringsdorf 337 black lipid membranes and liposomes. If monolayers of dipolar amphiphiles can be transferred to solid supports with the correct orientation of the lipids, integral membrane proteins should also be transferable when embedded in such an arrangement. Thus the construction of biosensors based on ordered receptor protein-lipid assemblies seems to be feasible. In vesicles double-chain membrane-spanning amphiphiles are the best candidates for synthetic channel or pore molecules, when one hydrocarbon chain is substituted by a hydrophilic oligoethylene oxide chain for instance. References 1 Y. Okahata and T. Kunitake, J. Am. Chem. SOC., 1979 101, 5231. 2 T. A. Langworthy, Curr. Top. Membr. Transport, 1982, 17, 45. 3 K. Ring, Biochim. Biophys. Acta, 1984, 778, 74. 4 H. Bader and H. Ringsdorf, J. Polym. Sci., Polym. Chem. Ed., 1982, 20, 1623. 5 E. Baumgartner and J-H. Fuhrhop, Angew. Chem., Znt. Ed. Engl., 1980, 19, 550; J-H. Fuhrhop, H. H. 6 J-H. Fuhrhop and J. Mathieu, Angew. Chem., 1984, 96, 124. 7 B. Neises and W. Steglich, Angew. Chem., 1978, 90, 556. 8 L. D. Bergel’son, t. G. Molotkovskii and M. M. Shemyakin, Zh. Obshch. Khim., 1962, 32, 58. 9 H. Bader, K. Dorn, B. Hupfer and H. Ringsdorf, Ado. Polym. Sci., 1985, 64, 1. David, K. Ellermann and J. Mathieu, Angew. Chem., In?. Ed. Engl., 1982, 21, 440. Received 13th January, 1986

ISSN:0301-7249    

DOI:10.1039/DC9868100329

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

GENERAL DISCUSSION Prof. P. Fromherz (University of ulm, West Germany) said: Stretching and breaking of a macroscopic elastic sheet is described by a force/elongation curve. The integral of this relation is some measure of the energy required to create a fracture. By analogy is it possible to estimate from Prof. Evans’ data the energy required to form the open edge of a lipid bilayer? Prof. E. Evans (University of British Columbia, Vancouver, Canada) replied: At first glance, it does appear that our empirical fracture energy model could be viewed as a crack propagation model for failure of a solid surface which is characterized by an ‘edge’ energy as you have indicated. However, careful examination of the origin of the crack model (see Landau and Lifshitz, Theory of Elasticity) shows that this model is a poor model for bilayer rupture.Specifically, the crack propagation problem is analysed for shear failure in a solid, whereas the lipid bilayer above the acyl chain transition has no shear rigidity (in other words, membrane rupture is more like cavitation than crack propagation). Secondly, the crack failure problem is based on the assumption that there exists a stationary crack distribution (i.e. an empirical length L ) from which the failure emanates. Again, this is a solid-like property not representative of the lipid bilayer in the liquid state. Dr D. S. Dimitrov (Bulgarian Academy of Sciences, Sojia, Bulgaria) said: The very interesting experimental fact for me is that the critical membrane tension of rupture is proportional to the square root of the elastic area compressibility modulus.This is in agreement with the theoretical formula’ for the critical membrane tension To: T:, = 24uGh where u is the membrane surface tension, G = E / 3 for incompressible bodies, E being Young’s modulus. In the case of an incompressible membrane subjected to expansion in one direction it is equivalent to the elastic compressibility modulus K multiplied by the membrane thickness. Therefore, the slope of the dependence Tc,K1I2 is equal to (80)’’’. The experimental value for the slope is 0.4 (dyn cm-’)’’2. This leads to a surface membrane tension equal to 0.03 dyn cm-’, which is a high but reasonable value. I would like to point out, however, that this does not prove that the theoretical formula and especially the underlying mechanism of action of the surface tension is entirely correct.It is interesting to note that the critical surface energy corresponds to the membrane surface tension. 1 D. S. Dimitrov, J. Membr. Biol., 1984, 78, 53; D. S. Dimitrov and R. K. Jain, Biochim. Biophys. Acta, 1984, 779, 437. Prof. E. Evans (University of British Columbia, Vancouver, Canada) said: Your comment brings to light an interesting possibility for the use of your theory of failure for thin liquid films to model bilayer rupture behaviour. However, a key point to keep in mind is that your model is not an appropriate representation of the stable (condensed) bilayer structure, since no comtant tension exists as for a film, only an area dependent tension. What is appealing is that the empirical ‘fracture energy density’ derived from fig.3 could represent a transition from our condensed state to a liquid-like film modelled by your theory, i.e. similar to a transition for an ideal plastic from a solid to a liquid. 339340 General Discussion Prof. J. F. Nagle (Carnegie-Mellon University, Pittsburgh, PA) said: These are very important measurements to which theories sbould be compared. Here let me just make contact with one point. In your paper you mention that the enthalpy of expansion is often ca. 100 erg cm-*, and you then note that this is just twice the surface free energy of hydrocarbon-water interfaces. However, on the next page you note that the enthalpy of expansion is elevated for ten degrees above T,, and from table 1 one would calculate ca.210 erg cm-2. My statistical-mechanical model (specifically case 6 in my table 1) gives an enthalpy of expansion that starts out ca. 500 erg cm-2, just above T,. As temperature is increased by 35 K, the enthalpy of expansion gradually decreases by a factor of two. The fact that the values are too large overall is a reflection of the fact that the values of area thermal expansion in these models are generally too large. This should not be allowed to obscure the important temperature dependence, which is a manifestation of what I call post-critical phenomena (which we have also seen experi- mentally by a decreasing volume expansivity aT in DMPC). The calculation also shows that all the interactions in the system, not just the interfacial term, contribute to the enthalpy of expansion. In the model over 70% of the enthalpy of expansion at constant temperature comes from the van der Waals interaction between chains, ca 20% comes from the trans-gauche energy, and only ca.5% comes directly from the interfacial term, although it might be noted that this latter percentage increases by a factor of about two, from 4 to 7%, as the temperature increases by 35 K from T,. Also, if I use the larger y that you prefer in my interfacial term in my eqn (4), these latter percentages will be roughly tripled. Therefore, while I would not wish to suggest that the model is capable of predicting precise numbers for these quantities, nevertheless it strongly suggests that the interfacial water term is not the exclusive source of the enthalpy of expansion.This agrees with my intuition that a lateral expansion of the membrane at constant temperature is also going to increase the volume, thereby doing work against the van der Waals cohesive energy, and that more gauche conformations will have to be introduced in order for the chains to spread out to fill the larger area, even at constant temperature. Since these interactions dominate the enthalpy of transition, at which area increases by ca. 25%, it is not surprising that they would continue to contribute significantly to the enthalpy of expansion in the single-phase region. Prof. E. Evans (University of British Columbia, Vancouver, Canada) said: It is important to recognize that there are two components of the heat of expansion of a bilayer: the surface pressure plus the heat of expansion of the hydrocarbon-water interface.Thus, if the hydrocarbon-water interaction is constant, changes in the heat of expansion (given by the product of temperature x elastic area compressibility modulus x thermal expansivity) would represent variations in surface pressure which would be expected to decrease with temperature in the La phase (as shown for DMPC in table 2). However, interfacial interactions could be temperature-dependent, which makes the comparison vulnerable. Prof. J. F. Holzwarth (Fritz-Haber Institiit, Berlin, West Germany) said: In your paper you examine the influence of cholesterol and peptides on the compressibility of lipid bilayer systems. We have examined similar preparations by looking into the dynamics of these systems between 1 ps and 100ms. One important result of these experiments was that the relaxation phenomenon around 10 ps is broadened, the relaxation around 1 ms is weakened, the relaxation around 25 ms becomes broader and its amplitude decreases by a factor of two (see figures in our contributed paper).As can be seen from the equilibrium fluorescence anisotropy measurements in fig. 1, we explain this behaviour by the preference of cholesterol for lipids in an intermediate state of order. This intermediate state is clearly preserved even above the phase-transition temperature. When we incorporated a similar peptide (24) mentioned in your paperGeneral Discussion c 341 I I I I I 3 0 35 40 45 50 T/"C Fig. 1,. Temperature dependence of the equilibrium fluorescence anisotropy rss = ( I - I J / ( I + 21,) for unilamellar vesicles of DPPC containing diphenylhexatrian (DPH) as a probe without further additives (-) and with 15% CHOL incorporated (- - -).A,,, = 360 nm, Aobs = 430 nm, cDPPC = 2.7 mmol dm-3. into the bilayer we found a much smaller effect, but this might be due to the low concentration of the peptide which can be dissolved in the bilayer structure without causing precipitation. We could handle preparations of SUVs containing up to 20% CHOL but only 2% peptide could be incorporated without precipitation.' From our results we conclude that the peptide is 10 times more effective in causing cohesion in SUVs than CHOL is. If one incorporates a channel-forming substance like Gramicidin into DPPC vesicles one finds a very dramatic effect on the slowest relaxation process around 20 ms which almost disappears at a ratio of 15/1.We are still puzzled about the dramatic differences between CHOL, peptide 24 and Gramicidin A. Could you give an explanation from your results? 1 A. Gem, J. F. Holzwarth and T. Y. Tsong, Chem. Phys. Lipids, 1986, in press. Prof. E. Evans (University of British Columbia, Vancouver, Canada) said: In our experiments, the addition of cholesterol (0-50 mol%) to PC lipids caused a decrease in bilayer compressibility over a wide temperature range. Also, tensions for vesicle lysis (rupture) increased with cholesterol concentration. Together, these results indicate an increase in bilayer cohesion via the formation of a 1 : 1 lipid cholesterol complex.At all concentrations above 12.5 mol%, the bilayers behaved in a liquid-like manner (no resistance to shear deformation). In this respect, the lipids appear to be in an 'intermedi- ate state of order' as you put it, i.e. low compressibility due to head-group and outer-chain condensation, but still fluid as a result of peripheral chain expansion at the bilayer centre. Our lipid-polypeptide system showed a slightly higher compressi- bility and lower lysis tension than for pure SOPC bilayers. This result indicated a decrease in cohesion, consistent with a lipid wetting model where the shorter peptide causes chain compression in the adjacent lipid. In your experiments, the mismatch between peptide-lipid hydrophobic length is opposite to our case; so, the larger peptide would be expected to decrease compressibility by extension of the shorter acyl chains.Your observation that the peptide is 'ten times more effective' in increasing cohesion than cholesterol may reflect the different stoichiometry and nature of the two interactions.342 General Discussion For example, cholesterol appears to form a 1:l complex with lipid whereas each polypeptide molecule interacts with several nearby lipids (ca. 7-10) in each monolayer. Dr G. Cevc (University of Essen, West Germany) asked Prof. Sackmann three questions. (1) Some of the bilayer surface invaginations that your elegant experiments have demonstrated look very much like closed vesicles. Surely, if they were such their formation would be irreversible; could you give us an estimate of the degree to which the opposing bilayers come towards a complete closure? (2) The vesicle shape-transformations which you report are characterized by a steep temperature dependence, similar to that of lipid chain-melting. Is the relative rate at which these macroscopic transformations occur similar to that of the lipid chain fluidiz- ation; have you considered performing T-jump experiments to investigate this or to obtain further insight into the molecular origin of the vesicle shape transformations? (3) How do the rate of vesicle shape-transformation and the final vesicle form depend on the degree of 'surface polymerization'? Prof.E. Sackmann (University of Munich, West Germany) replied. (1) The neck connecting the two vesicles and the distance of approach of the opposing bilayers are beyond the limit of resolution of the microscope. (2) The rate of the shape changes is an interesting point since it is directly related to the question whether the shape changes are first order transitions. The transitions are indeed instantaneous within the response time of our camera (ca.0.01 s). (3) Since it is very difficult to determine the progress of polymerization of the individual vesicles we cannot answer this question yet. Dr W. S. Bont (Amsterdam, The Netherlands) said: In the introduction it is stated that endocytosis takes place via invagination of the plasma membrane and exocytosis by protrusion. In general this is not correct, as demonstrated by the secretion of hormones. Hormones and many other products are stored in vesicles inside the cells.The 'export' of these substances takes place by fusion of the membrane of the vesicles with the plasma membrane, resulting in an omega-shaped membrane; the material is now outside the cell. This demonstrates that both endocytosis and exocytosis occur via invagination of the plasma membrane. The protrusion shown in fig. l(6) leads to shedding of vesicles, a phenomenon observed with tumour cells. Prof. E. Sackmann (University of Munich, West Germany) replied: I agree with your remark that the initial step of exocytosis is preceded by the fusion process. Thus fig. 1( b) refers to shedding of vesicles or to the process of budding. Prof. B. Zeki (Institute of Biophysics, Ljubljana, Yugoslavia) addressed Prof. Sackmann.(1) You have clearly demonstrated that the variation of temperature can induce shape transformations of vesicles made of charged lipids. As one can see in plate 2 the vesicle attains a two-vesicle shape with large average curvature at increased temperature (43 "C), goes through a discoid shape at intermediate temperature (42 "C) and develops an internal vesicle with small average curvature at lower temperature (41 "C), which agrees with theoretical predictions. I would only like to point out that the shapes with external and internal vesicles, respectively, do not belong to the same class of shapes, and therefore the transition between these two shapes cannot be expected to be con- tinuous. Somewhere between the shapes ( a ) and ( b ) in plate 2 one would expect to observe a discontinuous shape transformation analogous to a first-order phase transition.On the other hand it would be interesting. to know what happens with a qesicle if the temperature is increased above 43 "C or decreased below 41 "C.General Discussion 343 (2) I would expect similar but smaller effects for uncharged lipids. Because of the anisotropic thermal expansion of the two monolayers the relative area difference AA/ AA, should change with temperature and therefore induce shape transformations. (3) Have you observed some changes of shapes with time? A non-spherical shape with given AA has a larger bending energy than the corresponding sphere, and this energy difference represents a driving force for water flow into a cell coupled to lipid flip-flop processes.Such shape transformations are expected to be slow because of large flip-flop relaxation times and because of small membrane permeability for water. Nevertheless, this is in my opinion a process which makes small vesicles spherical and could also be important for large vesicle shape changes which do not require large volume changes. Prof. E. Sackmann ( University ofkfunich, West Germany) replied: ( 1 ) The transitions between shapes d and e or b and c, respectively, occur spontaneously within the response time of our TV camera, and the same holds for the discocyte-to-stomatocyte transition. These are thus of first-order. The transition from shape 6 to the discocyte is difficult to observe since the two states are hard to distinguish in the microscope.(2) I agree with you. However, we did not succeed in observing the transitions for giant uncharged vesicles. It may be that the range of temperature over which the shape changes occur is too large. (3) The giant vesicles of non-spherical shapes transform into spherical shapes after some hours which may be due to a flip-flop process as suggested by you. Prof. T. W. Healy (University of Melbourne, Australia) said to Dr Jones: The conclusion that the authors arrive at, viz. that the phospholipids are adsorbed with their head groups uppermost, is perhaps too simplistic in the light of earlier studies on adsorption of alkylethoxylate non-ionic surfactants by Ottewill's group at Bristol and in my own group. Thus for adsorption of such non-ionics on hydrophobic silica (contact angle 85 ") and hydrophilic silica (contact angle 0 ") the detailed isotherm is a high-affinity type on the hydrophobic surface and an S-shaped (autocatalytic) type on the hydrophilic silica.The plateau adsorption is identical on both substrates at a notional coverage of ca. 'one monolayer'. The adsorption on hydrophobic silica is 90% of the plateau value at C/c.m.c. = 1 and ca. 15% for the hydrophilic surface at C/c.m.c. = 1. At C/c.m.c. > 1 the adsorption process generates the same contact angle of ca. 28 " for both surfaces. Given that the forces controlling micelle (or vesicle) formation and transfer into an adsorbed layer at the solid-water interface are essentially the same, all one can conclude is that the surfactant forms a layer at the solid-liquid interface which best balances the hydrophobic- hydrophilic energies of the surfactant in water.With the hydrophilic surface it appears more difficult to nucleate adsorbed areas or (hemimicelles) into which the hydrophobic chains can transfer. Further analysis of the low-coverage part of the isotherms of Jackson et al. may indeed reveal S-shaped behaviour. I refer to the points above the line at low concentration in fig. 3. In contrast, the adsorption of CTAB on hydrophilic silica, mica etc. is able to produce an ordered monolayer anchored electro- statically followed by bilayer formation with the head groups out. For non-ionic alky1ethoxy:ates and zwitterionic phospholipids adsorbed on hydro- philic silica, the adsorbed layer is perhaps better described as a partly ordered, part hydrophobic, part hydrophilic layer of an area per molecule corresponding to apparent monolayer coverage.Dr M. N. Jones (University ofkfanchester) replied: These are interesting comments and we agree that the apparent monolayer coverage we find for DPPC and the other phosphatidylcholines could have an alternative interpretation as you suggest. We have deposited phospholipid films from monolayers by the Langmuir- Blodgett technique and from vesicle dispersions onto glass slides and measured the advancing contact angle344 General Discussion of water. Both techniques gave very similar contact angles. The contact angle for DPPC deposited on hydrophilic (8 = 0 ”) soda-lime glass is 45 O, suggesting either a monolayer with heterogeneous orientation or the formation of islands or clusters of DPPC inter- spersed by ‘clean’ surface.However, deposition of lipid onto glass from mixed DPPC-PI monolayers results in a decreasing contact angle with increase in PI to DPPC ratio, and for pure PI deposition the contact angle is ca. 9 O. This suggested to us that the orientation of the lipid was changing with PI content in the system and in the extreme case (pure PI) the head groups were largely uppermost. In this case island formation perhaps seems less likely in view of the negative charge on the phosphate group and the hydrophilic nature of the pentahydroxyinositol group. Both these factors would inhibit the nucleation of PI clusters with head groups down on a negatively charged glass surface. It should, however, be noted that the phospholipids probably initially adsorb with their head groups onto the glass and then subsequently change their orientation in aqueous media as was postulated many years ago by Langmuir.’ With regard to the possibility that the adsorption isotherms are S-shaped at low surface coverage, our most recent more detailed analysis of the shape of the isotherms has given no indication that they are S-shaped at low concentrations.1 I. Langmuir, Science, 1935, 87, 493. Prof. J. K. Thomas (University of Notre Dame, I N ) commented: In some recent work we have constructed ‘bilayers’ of various quaternary ammonium detergents on the surface of laponite clays. The detergent should be adsorbed in close proximity on the surface, yet no evidence of ‘true’ bilayers is found’ i.e.no phase transaction in d.s.c., and the pyrene excimer formation is indicative of micelles, and not that explicit for bilayers as shown in our paper. An organic hemimicelle or similar structure is formed. Could similar structures form in your system? Such explicit behaviour in physical studies (pyrene excimer) lends much evidence to our earlier paper. Dr M. N. Jones (University of Manchester) replied: This is a similar point to that raised by Prof. Healy, and we refer to the previous reply. However, with regard to the specific application of pyrene monomer-excimer equilibrium to the characterisation of the structure of adsorbed lipid, to observe a high excimer to monomer ratio requires restriction of the pyrene to localised regions of the surface (e.g. as in hemimicelles) or alternatively to pyrene-rich regions in-a rigid bilayer structure which could arise as a result of lateral phase separation below the phase transition temperature.It is possible that a high excimer to monomer ratio could also occur by lateral phase separation in a tightly packed monolayer where excimer dissociation and pyrene diffusion is restricted. Adsorption at a solid surface has been shown to incease lipid phase transition tem- peratures indicative of restricted mobility [see ref. ( l)]. While hemimicelle formation cannot be excluded as a possible explanation of our data, more detailed information on pyrene in other types of ordered lipid system (monolayers) is perhaps required before the technique can give firm evidence of hemimicelle formation. Dr L.R. Fisher (CSIRO, Sydney, Australia) commented: It seems to me that your finding of monolayer adsorption on barium titanate is perfectly explicable if barium titanate glass has a hydrophobic surface. Have you determined the contact angle of water on this surface to check whether this is the case? Horn’s results were obtained, of course, on mica, which is a hydrophilic surface with 6 == 6 O as measured by our methods.’ 1 L. R. Fisher, J. Colloid Interface Sci., 1981, 72, 200. Dr M. N. Jones (University of Manchester) answered: We have no direct measure- ments of the contact angle of water on the barium titanate beads. However, MinginsGeneral Discussion 345 and Scheludko' have studied the behaviour of these beads by attachment to the surface of a pendant drop.They report that the clean beads will not form a three-phase contact at the drop-air interface in the absence of adsorbed surfactant (in their case C,TAB), which does suggest that the surface of the clean beads is completely wetted and hydrophilic. (The cleaning procedures used by us and by Mingins and Scheludko are essentially the same.) Perhaps what is more significant is that the contact angle for a glass surface with a deposited monolayer of phospholipid varies with the phospholipid composition (DPPC:PI ratio) which implies a change in lipid orientation on the same surface. Thus the nature of the lipid is critical in determining the orientation at the glass-aqueous solution interface, which suggests to us that even with a hydrophilic surface it is possible to orientate some lipids with their headgroups uppermost.1 J. Mingins and A. Scheludko, J. Chem. Soc., Faraday Trans. 1, 1979, 75, 1. Dr L. R. Fisher (CSIRO, Sydney, Australia) continued: The extent of disruption of your vesicles (e.g., fig. 5 ) seems much higher than is needed to provide the material for monolayer coverage of the bead surfaces. Can you suggest any mechanism to account for the additional disruption? Dr M. N. Jones (University of Manchester) replied: This is indeed what we find, and it does appear that the beads catalyse vesicle disruption. The results could be interpreted by assuming that exchange between adsorbed lipid and vesicular lipid continues after surface saturation, resulting in disruption. Alternatively there is no reason to assume that adsorption of all the lipid from a single vesicle occurs, but rather that the build-up of surface coverage arises by transfer of a proportion of lipid from a larger number of vesicles with concomitant disruption.Dr D. S. Dimitrov (Bulgarian Academy of Sciences, Sojia, BuZgaria) commented in introducing his own paper: I would like to point out several additional results we have obtained: We have studied the effects of ax. fields for two cases: first, on ready-formed liposomes and second, on liposome formation. We have observed that liposomes and other structures formed during lipid swelling in water vibrate with the same frequency as the applied a.c. field. On increasing the amplitude of the field the amplitude of the vibrations increases.Unfortunately, we do not have appropriate equipment to record the amplitude as a function of the voltage and the frequency. Instead, we have used the observations that at fixed voltage the amplitude decreases with increasing frequency and that above certain frequency visually one cannot see any change in the position of the liposome, i.e., above this frequency the vibrational amplitude is very small, at least smaller than the optical resolution of the microscope (in the worst case say 1 pm). This characteristic frequency depends on the applied voltage. Increasing voltage leads to an increase of this characteristic frequency. Fig. 2 ( a ) - ( d ) shows the dependence of the characteristic frequency on the applied voltage for different conditions.The main conclusion is that this dependence does not depend on the lipid and the solution osmolarity. Probably this is a pure electrokinetic effect. Fig. 3 shows the patterns of the liquid motion. The most important [plate 1 ( a ) ] observation is that without field there is no liposome formation from egg lecithin. The application of an ax. field (frequency 10 Hz, voltage 2V) leads to formation of giant thin-walled liposomes and many small vesicles [plate l ( b ) and ( c ) ] . The neutral lipid DMPC forms just a few giant thin-walled liposomes [plate 2 ( a ) ] . Application of an a.c. field leads to an increase of the number of these liposomes by at least ten times [plates 2 ( b ) and ( c ) ] .346 General Discussion 34 32 30 28 26 24 22 20 $ 18 '1 16 14 12 10 8 6 4 2 0 1 34 32 30 28 26 24 22 $ 18 16 14 12 10 8 6 4 g 2 0 0 2 4 6 a 10 V l V 0 2 4 6 8 10 V l V Fig.2. ( a ) Egg lecithin-water, T = 30 "C V = 2 V, f = 10 Hz, n = 80, t = 1 h, 2% Ficoll, 0, inner side; + , outer side. (6) Egg lecithin-DMPC-water, T = 30 "C, V = 2 V, f= 10 Hz, n = 80, 24, t = 1 h, 2% Ficoll; 0, egg lecithin; +, DMPC. ( c ) egg lecithin-water, n = 80, T = 30 "C, t = 1 h, 2% Ficoll, 0, V = 0, water; + , f = 10 Hz V = 2 V, water; 0, V = 0 V, 0.5 mol sucrose (d) DMPC-water, n = 24, T = 30 "C, V = 2 V, f = 10 Hz, 2% Ficoll, 0, 60 min swelling; + , 30 min swelling. The negatively charged lipid PS does not form liposomes by swelling either in the presence or in the absence of an electric field. The lipid layer breaks into flakes and detaches from the electrode surface (plate 3).(The bar denotes 50 pm.) It seems that there is a certain similarity between the action of the a.c. field and the effects of sonication when producing small vesicles. In both cases mechanical stresses may play a role in separating and destabilizing the membranes to form liposomes.Plate 1. Egg lecithin swelling: ( a ) without external field; (6) and (c) in an a.c. field off= 10 Hz, v=2v. (To face p. 346)Plate 2. DMPC swelling: (a) without external field; (b) and (c) in an a.c. field o f f = 10 Hz, v=2v. 50p.m Plate 3. PS swelling: no liposome formation either without or in the presence of an external electric field.General Discussion 347 34 32 30 26 24 22 20 18 16 14 1 2 10 8 6 4 2 0 28 0 2 4 6 8 10 VIV 30 28 26 24 22 20 18 2 16 2 14 12 10 18 6 4 2 0 0 2 L 6 8 10 V l V Fig.2. (continued) Prof. P. Fromherz (University of Ulm, West Germany) said: You have mentioned that cholate facilities the formation of vesicles by electrical fields. As cholate is a typical edge-active agent lowering the edge-tension1 I suggest that the electrical field strength (the drop of the applied voltage in the lipid multilayer) is effective through its interaction with the edge.* According to Gibbs’ equation the edge-tension is lowered with an applied electrical field if there exists a change of dipole moment across the membrane with increasing length of edge. Considering the dipolar structure of the headgroup of lipids and a simple model of the edge3 Gibbs equation may be integrated for simultaneous presence of electrical field and edge-actant.2 A mutual reinforcement of the two protagonists is predicted.The description would be consistent with your hypothetical348 General Discussion ( 0 ) ( b ) Fig. 3. ( a ) Vibrations of the lipid structure follow the electric field alternations; ( b ) patterns of the electro-osmotic liquid motion in a.c. fields. sketch of primary formation of open edges with subsequent closure of the fragments to vesicles. 1 P . Fromherz, Faraday Discuss. Chem. SOC., 1986, 81, 39. 2 P. Fromherz, Chem. Phys. Lett., submitted for publication. 3 P . Fromherz, in Reoerse Micelles, ed. P. L. Luisi and B. E. Straub (Plenum Press, New York, 1984), p. 55. Dr D. S. Dimitrov (Bulgarian Academy of Sciences, SoJia, Bulgaria) said: It is a very interesting suggestion which should be checked thoroughly.Presently, we are performing a number of experiments to understand the relationships between cholate and electrical fields. Prof. E. Sackmann (University of Munich, West Germany) said to Dr Dimitrov: (1) I am intrigued by your electroformation experiment. Since you apply a d.c. electric field I wonder whether the formation of vesicles is not caused by local hydrody- namic flows due to electrohydrodynamic effects in the liquid-crystalline layer on the electrode? (2) In your model you assume that the multilayer system is composed of platelets of bilayers which implies a high edge energy. On the other side, in freeze-fracture experiments we always find that multilayers are made up of closed (compressed) vesicles, so that only edge dislocations are formed.Do you have other experimental evidence for your stack model? Dr D. S. Dimitrov (Bulgarian Academy of Sciences, SoJia, Bulgaria) replied: (1) Hydrodynamic flows exist and have been observed when applying d.c. and a.c. fields. For d.c. fields the flow velocity is high in the beginning and sharply decreases after a couple of seconds, probably due to charging effects. For a.c. fields the liquid motion depends on the frequency. The liposome yield is different at different frequencies. Certainly, hydrodynamic effects contribute to liposome formation. It seems, however, that electrostatic and double layer effects layer effects dominate. We are not able to check for local electrohydrodynamic effects in the very liquid-crystalline layer.Such effects could cause or facilitate liposome formation. (2) We do not have any experimental evidence for the stack model. It is just a hypothesis. This model is for the structure of the dried lipid layer and at the very beginning of the hydration.General Discussion 349 Dr A. Nelson (IMER, Plymouth) said to Dr Dimitrov: I was interested to read your paper since it addresses itself to the behaviour of phospholipids in an electric field. We have also studied the effect of electrical potential on phospholipids so our work has some relevance to yours. We perform our experiments on phospholipid monolayers adsorbed on mercury electrodic surfaces and examine the effects of both d.c. and a.c. fields. Two significant transitions are observed in the cathodic potential domain. The first of these is an increase in permeability of the layer to ions and the second at more negative potentials represents a major but reversible disruption of the membrane.' Both transitions are sensitive to the presence of counter-ions. The method of adsorbing phospholipid layers on to mercury electrode surfaces was first developed by Miller.' It seems to be that the electric field induced transitions relate in some way to the formation of liposomes which you observe since they occur in the negative potential domain.First I must ask you what is the electrolyte composition which you use in your experiments? You are not very clear about this in your paper. Indeed, you mention that you use distilled water and that you found that high ionic strength inhibits liposome formation. However, the use of distilled water will create a large resistance between your electrodes and lower the magnitude of electric field across the phospholipid layers.On the other hand, you speak of the redistribution of counter-ions as being a contributory factor. Have you investigated this experimentally? 1 A. Nelson and A. Benton, J. Electroanal. Chem. 1986, 202, 253. 2 I. R. Miller, J. Rishpou and A. Tenenbaum, Bioelectrochem. Bioenerg., 1976, 3, 528. Dr Nelson then added: In this instance one immediately asks why liposomes do not form in the presence of significant ionic strength electrolytes although you suggest that this is due to osmotic forces. I would therefore like to suggest that your liposome formation experiments could be conducted under more precise conditions of electrolyte concentration, pH buffering and potential control. In this way, electrochemical measurements of current and capacit- ance would aid in the elucidating the mechanism of liposome formation which you are still not clear about in your paper.Dr D. S. Dimitrov (Bulgarian Academy of Sciences, Sofia, Bulgaria) replied to Dr Nelson: Your results about effects of electrical fields applied across phospholipid monolayers will undoubtedly be helpful in understanding mechanisms of liposome electroformation. One of the problems is that in our experiments the potential across the lipid layers is practically zero because the lipid is deposited as separate spots on the electrodes and its thickness is very small.We have used distilled water with conductivity 4 ,US cm-' and pH 6.8, as well as PBS (0.3 mol dm-3). As I mentioned above, the magnitude of the electric field across the lipid layers is very small, independent of the resistance of the medium. We have not investigated the redistribution of the counter-ions. We think that the double layers around the electrode and between the membranes are changed due to the external electrical field. At the negative electrode the positive counter-ions are attracted and the double-layer thickness is decreased. This leads to an increase of the intermem- brane repulsive forces and therefore to membrane separation and liposome formation. It is very difficult to develop a theory for electrostatic intermembrane interactions in external electrical field and near to the electrodes.Presently, we are making some estimates. Liposomes do not form in solutions of high ionic strength because of the decrease of the repulsive intermembrane electrostatic forces. We have stated that the increased osmolarity can decrease the rate of swelling because of the decreased osmotic forces.350 General Discussion -25 t Fig.4. The relation between AG" (A), AHtr (W) and TASt' for the transfer of alkylphenols (m-C,) from cyclohexane to liposomes. Dr B. Kronberg (Institute for Surface Chemistry, Stockholm, Sweden) addressed Dr Anderson: When studying the ordering in liquids through transfer functions it is customary to use a non-structured liquid as a reference liquid. In this work the results of transferring probes from water to lipid membrane are discussed in terms of the destruction of order in the lipid membrane.The choice of water as a reference liquid in this work is, however, inappropriate, since water itself is structured in different ways by the presence of different probes. Thus the transfer functions in table 3 also reflect the influence on the structuring of the water by the probes. This problem can, however, be circumvented by using cyclohexane as the reference liquid. Thus the transfer function of interest is - AX&lohexane -., liposome - AXEater -+ liposome - AXEater 4 cyclohexane where X is either G, H or S. These transfer functions are, of course, sensitive to the chemical dissimilarity between the probes, and hence only probes in a homologous series can be compared.From tables 2, 3 and 4 we obtain the relation between AGt', AH" and TASt' for alkylphenol probes as shown in fig. 4. A large TAS", or AH", is associated with the destruction of order in the lipid membrane. The figure reveals the typical enthalpy-entropy compensa- tion, which is inherent in the process of destruction, or build-up, of order. The intercept, at T A S f ' = O , is due to the chemical difference the probe experiences when transferred. Thus it can be attributed to hydrogen bonding of the phenyl group in the lipid bilayer. It should be interesting to investigate larger n-alkyl groups in the p-position on the phenol group, in order to obtain an ordering upon transfer from cyclohexane, i.e. AS" < 0. Such a net ordering should be experienced by the probe in the lipid bilayer, since in the reference liquid the probe is completely disordered.General Discussion 351 Dr N.H. Anderson (Long Ashton Research Station, Bristol) replied: The objective of our work was to compare the partitioning of solutes between water and liposomes with that between water and organic solvents, as indicated in the paper. Thus the thermodynamic analysis was based on the comparison of water-organic phase transfer, but Dr Kronberg is correct in pointing out that the thermodynamic transfer functions for cyclohexane-liposome can be derived from our data and these functions are of interest. In the paper we explain that it is unwise to use plots of enthalpy against entropy to test for enthalpy-entropy compensation, but for the alkylphenols I have confirmed that compensation does exist for the transfer from cyclohexane to liposomes.Because the liposome structure is complex the intercept is likely to represent the sum of several interactions including that due to hydrogen bonding. Prof. J. F. Nagle (Carnegie-Mellon University, Pittsburgh, PA) said: I was pleased to see that your AH of ca. 5.7 kcal mol-' for pure DMPC, obtained with a Perkin Elmer DSC-2 instrument, agrees with the values obtained with the more sensitive, slower scanning rate Privalov and Microcal calorimeters. In contrast, DSC-2 measurements over the years have tended to give 6.6 kcal mol-'. I am curious if you also measured AH for other lipids such as DPPC and if your values agree with the commonly obtained value of ca.8.7 kcal mol-'. Dr N. H. Anderson (Long Ashton Research Station, Bristol) replied as follows: The reported AH of 5.7 kcal mol-' for the DMPC liposomes (prepared by hand-shaking) was obtained using a scanning rate of 5 Kmin-'. However, a previous determination using DMPC liposomes prepared by mechanical agitation gave AH = 7.5 kcal mol-'.' AH values for DPPC and DSPC were 9.51 and 12.31 kcal mol-', respectively,' supporting the view that the Perkin Elmer DSC-2 instrument normally gives a higher AH than slower scanning, more sensitive instruments. Possibly the precise method of preparing the liposomes and hence their size distribution affects the enthalpy of transition. 1 M. Ahmed, J. Hadgraft, J. S. Burton and I. W. Kellaway. Chem. Phys. Lipids, 1980, 27, 251.Dr G. J. T. Tiddy ( Unilever Research, Port Sunlight) commented: A simple calculation based on the Clausius-Clapeyron equation suggests that a solute mole fraction of 0.1 in the lipid should reduce the L,/gel transition temperature by ca. 3 K. In fact the dispersion without solute (two components) is a two-phase sample which must have a sharp transition according to the phase rule. With the solute present there are three components, hence three phases (dilute aqueous phase, L,, gel) can coexist over a range of temperatures. In this case the compositions are expected to vary with tem- perature. I would expect the solute to be preferentially solubilised in L,, hence stabilising this phase as more and more gel phase (with a much lower concentration of solute) is formed.Thus the L, phase would become richer in solute at lower temperatures. It could result in the partition being strongly temperature dependent. The three phase co-existence would cause a reduction of the observed transition heat and increased hysteresis between the heatinglcooling d.s.c. curves. The changes in temperature would be most difficult to detect in heating curves. Can you rule out the presence of L, in your systems? Dr N. H. Anderson (Long Ashton Research Station, Bristol) replied: The Clausius- Clapeyron equation is only applicable to ideal systems and the strong solute-phos- pholipid interactions believed to be present in the liposomes of the system we studied limit its usefulness. Using a mole fraction of 0.1 4-methylphenol in the DMPC liposomes, no change was detected from the normal transition temperature of 23.5 "C, using a differential scanning calorimeter (d.s.c.) of limited sensitivity.352 General Discussion Partitioning experiments involved a maximum mole fraction of 0.02 solute in liposome and d.s.c.showed a reduction in enthalpy of transition from 23.9 to 19.5 kJ mol-' with this mole fraction of 4-methylphenol. The half-height width of the transition peak was increased by 1.56 times under these conditions. Even at higher solute mole fraction than those used by us, a plot of log K against reciprocal temperature showed a clear discontinuity at the normal phase transition temperature.' Taken together, these data suggest that the solute concer trations used did not cause a major perturbation to the phospholipid gel phase or phase transition.However, as stated in the paper, the enthalpies of solute partitioning are exceptionally large and this is interpreted as due to a local perturbation of the phospholipid by the solute; the proportion of phospholipid molecules affected would depend on the nature and concentration of the solute in the liposome. It is not possible to test this model without spectroscopic (e.g. n.m.r.) data, which would show whether the phospholipid adjacent to the solute could be regarded as a localised domain and thus at present we cannot say whether such domains are present or not. 1 Ref. (16) of our paper at this Discussion. Prof. J. K. Thomas ( University of Notre Dame, I N ) said: You indicate that 7'' values were measured in your work, I cannot find them in the paper. Could you give me some values ? Dr N.H. Anderson replied: The transition temperature of the liposomes was measured by d.s.c. and is recorded in the paper as 23.5 "C. Dr A. Nelson (IMER, Plymouth, Devon) said: (1) As I understand it from your paper, you did not measure the rate of establishment of equilibrium partition of the solute between the liposome and aqueous phase. Is this correct? (2) In terms of toxicological studies the rate of permeation of a xenobiotic into the membrane is highly significant. We measure the rate of penetration of hydrocarbons into mercury adsorbed lipid membranes. Preliminary findings show the rate of permeation to correlate with the in vivo toxicity of these compounds. Dr N. H.Anderson (Long Ashton Research Station, Bristol) said: I take Dr Nelson's questions in order. (1) Preliminary experiments (not reported) showed that complete equilibration of solute between the aqueous phase and the liposomes was achieved within the 48 h time allowed. (2) We agree that rates of membrane permeation as well as partitioning equilibria are important in interpreting biological effects caused by xenobiotics. Prof. E. Sackmann (University of Munich, West Gerinany) said: You report an astonishingly high solubility of the substituted phenols in bilayers of DMPC at 22 "C. This is quite surprising since the lipid is in the Ppz phase at this temperature. Moreover, you find only minor changes of the heat of transition. In lateral diffusion measurements we found' that the solubility limit of fatty acids in the Ppg phase is ca.1 mol %, whereas higher concentrations lead to lateral phase separation. Do you know whether the phenol is really incorporated into the bilayer or could it also be intercalated between the stacks of bilayers? Did you do lateral diffusion measurements to see whether the lipid is fluidized by the incorporation of the phenols? 1 H. G. Kapitza, D. A. Riippel, H-J. Golla and E. Sackmann, Biophys. 1, 1984, 45, 577. Dr N. H. Anderson (Long Ashton Research Station, Bristol) said: The fatty acid referred to by Prof. Sackmann is glycophorin, a protein of molecular weight ca. 30 000General Discussion 353 with three parts: a hydrophobic aminoacid core, a hydrophilic carboxylic end and an NH,-terminal segment to which 16 oligosaccharides are linked.We find it difficult to relate the effects of this protein on DMPC bilayers to those observed by us, since the solutes we used were in the 100-150 molecular weight range. We would expect such low-molecular-weight solutes to be considerably more soluble than glycophorin in the DMPC bilayer Ppf phase and the effect of 1-2 mol '/o solute on the enthalpy of transition to be much less than that of 1 mol YO glycophorin. It has been reported that alcohols such as octan-1-01 are relatively insoluble in the phospholipid gel phase and reduce the phase transition temperature. In contrast, we found that 4-methylphenol has little effect on the transition at 1-2 mol O/O , and this may be related to its ability to hydrogen-bond to the phospholipid head groups.Our results are consistent with the view that the solutes we used did partition into the bilayer, rather than intercalating as Prof. Sackmann suggests, but we do not have any direct evidence on this point. We did not make any measurements of lateral diffusion. Dr A. M. Howe (AFRC Institute of Food Research, Norwich) commented: You have compared the partitioning of solute molecules between liposomes and bulk water with that between organic solvents and water. Is not the surface/interfacial activity of the solute molecules likely to be an important factor in determining the partitioning of some molecules between surfactant aggregates, such as liposomes, and bulk water, particularly in the case of strongly surface-active molecules? Prof.E. Sackmann (University of Munich, West Germany) said: Vesicles of normal diacetylene containing phospholipids are quite unstable in the crystalline state and tend to aggregate into multilayers. Do Drs Bader and Ringsdorf have any information concerning the stability of the vesicles of their bipolar diyne lipids in the solid state? Prof. P. Fromherz (University of Ulm, West Germany) said: The stability of vesicles is a wide field. Drs Bader and Ringsdorf have mentioned the stabilization as attained by applying membrane spanning and polymerized lipids with respect to spontaneous leakage as tested by carboxy-fluorescein. What about other mechanisms of destabiliz- ation, such as the effect of detergents, fusogens, electrical fields, enzymes and biological fluids? Dr K.Kurihara (Institute for Surface Chemistry, Stockholm, Sweden) commented: Concerning the stability of polymerized vesicles against their fusion, we have found that photopolymerization of vesicles prepared from the mixture of dipalmitoylphos- phatidylcholine (DPPC) and a styrene-containing surfactant (1) completely obviates the growth of 300 A diameter unilamellar vesicles (fig. 5).' Each vesicle contained ca. 40% of the polymerizable surfactant (l), and such partial polymerization was enough to stabilize the vesicles in this case. 1 K. Kurihara and J. H. Fendler, J. Chem. Soc., Chem. Curnmun., 1983, 1188. Prof. D. A. Haydon (Cambridge University) asked: What proportion of diyne lipids polymerize in the systems described and what is known about the molecular weight of the polymers formed? Prof.J. F. Holnvarth ( Fritz-Haber-Institut, Berlin, West Germany) (communicated) I wish to make some extensive comments on the dynamics of the phase transition in phospholipid vesicles.354 General Discussion 900 Cgl;PO--r--.rO~-~O=)-~=II=EI-II-O- 20 28 days days j o 0 p 0 50 100 150 200 fG--JJ- time/h Fig. 5. Spontaneous growth of DPPC (0, 1.02 mmol dm-3; ., 0.5 mmol dm-3) and non- polymerized DPPC-(1) {A, [DPPC], 0.74 mmol dmP3 + [(l)], 0.44 mmol dm-3; A, [DPPC], 0.47 mmol dmU3 + [ (l)], 0.28 mmol dm-3; V, [ DPPC], 0.34 mmol dm-3 + [ (l)], 0.23 mmol dmU3} vesicles as a function of incubation time. Polymerized DPPC-( 1) vesicles are seen to retain their sizes for extended periods {V, [DPPC], 0.53 mmol dm-3+[(1)], 0.35 mmol dm-3; 0, [DPPC], 0.36 mmol dm-3 +[(1)], 0.25 mmol dm-3; 0, [DPPC], 0.26 mmol dm-3+ [(l)], 0.16 mmol dm-3}.Plotted are the hydrodynamic diameters (D,) of the vesicles, determined by dynamic light scattering, against incubation time at 23 "C. In contrast to thermodynamic studies only a few kinetic investigations of well defined lipid bilayer systems have been carried out. E.s.r.' and n.m.r.2 studies as well as fluorescence polarization lifetime measurements3 have been favourite techniques, but these instruments only cover the nanosecond to picosecond time range. Pressure-jump techniques4 and conventional Joule-heating temperature jump' have been applied but either strong field effects or the release of pressure-limited both methods from 300 ps to 100 ms. Only our iodine-laser temperature-jump ( ILTJ)6 technique covers the whole time range from to 10's without producing unwanted physical or chemical effe~ts.~ Because of the so far unmatched time range we applied our ILTJ to well defined unilamellar vesicles ( UVs) from dipalrnitoylphosphatidylcholine (DPPC) or dirnyri- stoylyphosphatidylcholine (DMPC).The major targets of our investigations were the dynamic processes associated with the main phase transition (PT) which occurs in a temperature range of ca. 10 "C. By covering the dynamic changes from to 10-o s we hoped to be able to compare the kinetic results with thermodynamic data either from turbidity/temperature or microcalorimetric measurements (d.s.c.). UVs containing Cholesterol (CHOL) synthetic peptides (PP) or channel-forming units like Gramicidin (G) or functional proteins like the light-driven proton pump bacteriorhodopsin (bR) were also investigated with respect to marked changes of the lipid dynamics caused by their incorporation into the lipid membrane.To monitor the kinetics of the crystalline-fluid transition after a fast temperature jump in the UVs we used three different techniques: light scattering or turbidity, light absorption from especially tailored lipids and fluorescence anisotropy changes observed through special probe molecules. In this way we could rule out any specific misleading effect from one of the detection methods. The lipids DPPC and DMPC as well as gramicidin A' (GA') were of the purest grade available from Fluka, Switzerland; the fluorescence probe molecules diphenylhexatriene (DPH) and trimethy1an;minodiphenyIhexatriene (TMADPH) were supplied by Molecular Probes, Texas. The specially tailored lipids 1{3[ p-(6-phenyl-1,3,5- hexatrienyl)-phenyl]-propionyl}-2-palmito~l-3-phosphatidylcholine (DPHPC) was synthesized by E.Thomas, University of Salford, and { 2-[ 3,6-bis( dimethylamino)- 10- acridinolethy1)-( 2,3-dipalmitoyl-~,~- 1 -glyceryl)-phosphate ( AOL) was a gift of Prof.General Discussion 355 CH3 ,' I , C ' ' CH3 c ti* c u* I I 0 0- c ~ 00 0 0 0 I I oc co I I R R c holine ,m, acridineorange CH3 H3C-N N-CHJ CI" 2 c y " 2 o= 6 - O Q y 2 n bc,n2 0 0 I 0 0 I I oc co I I A R 1 turbidity absorption fluorescence Fig. 6. Bond structure of (1) lecithin, (2) acridine orange lecithin, (3) diphenylhexatriene pal- mitoylphosphatidylcholine, (4) diphenylhexatriene and (5) trimethylamminodiphenylhexatriene.Zimmermann, University of Freiburg (fig. 6). The synthetic peptide PP24 with the amino-acid sequence 1ys2-g1y-1eu2,-1ys2-a1a-amide was a generous gift from R. S. Hodges, Edmonton and M. Bloom, Vancouver. All other chemicals were purchased from Merck, West Germany, either p.a. or Suprapure if available. Only triply distilled water from quartz instruments was used. The vesicles were prepared by a 'modified injection method'.**'" 20-50 pmol dmP3 of lipid were dissolved in 1 cm3 of ethanol and slowly injected (10-20 min) using a Hamilton syringe into 10cm3 of pure buffer at a temperature 10°C above the phase transition temperature ( T,) while the emulsion was carefully stirred.Afterwards the solution was dialized for 8 h against pure buffer to remove the alcohol." The vesicle preparations were characterized by electron microscopy, laser light scattering as well as their temperature dependence of turbidity, absorption or fluorescence;' differential scanning calorimetry (d.s.c.) was also used. l 1 Fluorescence and absorption labelling was achieved by adding a few mm3 of the probe dissolved in methanol to the UVs, giving the desired probe/lipid ratio. After an incubation time of 1 h at a temperature above T, the methanol was completely removed. DPHPC and AOL were incorpor- ated9-12 into the UVs together with the lipid in the original ethanolic solution. Fig. 6 gives the bond structure of the lipid and probe molecules. Fluorescence anisotropy and d.s.c.measurements are reported in ref. (9), (1 1) and (12). Phase diagrams of pure lipid water systems are shown in ref. (4) for DPPC and ref. (10) for DMPC. Our ILTJ was used either in the mode-locked mode,6 the free-run mode" or the oscillator m0de,6-~.'~ producing pulses at 1.315 pm of halfwidth, either 100 ps to 3 ns or 80ns as well as 2.4ps containing an energy of 1 J to produce a temperature-jump of 1 K in 150 mm3 of s~lution.~ The relaxation processes were monitored spectroscopically. The signals from especially designed photomultiplier were stored in transient recorders from Tektronix or Biomation and processed in Hewlett Packard computers for sampling and relaxation time as well as amplitude calculations. As detection light sources we used 150 W XBOs from Osram or Hannovia.Details of the optics, electronic circuits and the data registration and processing system are given e l ~ e w h e r e . ~ ~ ~ ~ ' ~ ' - ~ ~ In fig. 7 the schematic experimental arrangement is shown. Equilibrium phase-transition-temperature dependences for DPPC obtained from turbidity or fluorescence anisotropy measurements are given in the l i t e r a t ~ r e ~ ' ~ ' ~ ~ ' ',12 and in our comment to the papers of Prof. Fromherz and Evans. In fig. 8 d.s.c. results are summarized for UVs of DPPC with increasing CHOL content and fig. 9 shows the turbidity and absorption changes with temperature of UVs containing AOL as a probe356 General Discussion [ ~ computer I recorder multiplier mirror 2-2OJ laser .'p detect io Fig.7. Schematic arrangement of the iodine-laser temperature-jump (ILTJ) experiment: time resolution 1 0 ~ - 1 0 - ' ~ s. molecule. A size distribution from electron micrographs is included in our question to Dr Cornell's paper and in fig. 10. In fig. 11 the five relaxation processes which could be monitored after a 1 ns temperature jump are shown together with their corresponding amplitudes. We did not monitor changes faster than 4 ns and slower than 100 ms. The five relaxations are clearly separated; by changing the hydrocarbon chains and the head groups it was possible to attribute the fastest process to the formation of simple rotational isomers (kinks) in single molecules because the 4 ns signal called 1 is not sensitive to the length of the hydrocarbon chains or the nature of the head groups.4s1o Relaxation 1 also showed only a very weak maximum of T, and A, near T,, and its amplitude decreased steadily.Signal 2 around 300 ns showed a maximum at the midpoint of the phase transition in its relaxation times as well as the corresponding amplitudes. It also changed with the nature of the head groups.1o We therefore believe that process 2, having only a moderate cooperativity, can be explained by the onset of the free rotation of the head groups, since they are no longer fixed with respect to the hexagonal lattice of the hydrocarbon chains of the all- trans conformation. If we now proceed along the time axis we can see a strongly cooperative relaxation around 10-20 ps; this third process showed strong maxima of the relaxation times r3 and the amplitudes A3 which were due to an increase in complexity of the molecular interactions involved.By applying the absorption probe-lipid AOL we could prove that lateral diffusion has already started in this time range because we could monitor the monomer/aggregate equilibrium of AOL; this can only occur if lateral diffusion has started (fig. 12). An important condition for lateral diffusion in the plane of both monolayers forming the bilayer membrane is that complex rotational isomers such as gauche forms are already formed which shorten the hydrocarbon chains so that they are no longer hindered by the chains of the opposite monolayer in their lateral movement. We could not see much lateral diffusion in the nanosecond time window. At longer times processes 4 and 5 are observed.Their explanation is more difficult than for the three faster dynamic phenomena. From the time scale, the strong cooperativity, and the influence of additives such as CHOL and peptides as well as proteins, we concludeGeneral Discussion 357 t 30 40 (4 t *O ---z - Fig. 8. Differential scanning calorimetric measurements of DPPC vesicles containing different amounts of cholesterol: (a) 0, (b) 7.5 and (c) 16.5%, cDPPC-CHOL = 2.7 x mol dm-3, 0.9 cm3 probe volume in MC1. that process 4 is the formation of lipids in a more liquid state surrounding clusters of more crystalline states of order, with intermediate lipids as buffer. This discontinuity in the state of order of the lipids exists until we reach the 10-30ms time range, where it disappears leaving the bilayer in the fluid state.Fig.13 summarizes all five relaxations in a schematic way. Fig. 14 shows how the relaxation amplitudes are distributed over the whole time range (equilibrium sum) and proves the following: (1) The five relaxation processes represent the whole crystalline-fluid transition dynamics because they repro- duce the equilibrium dependence of fig. 9 perfectly. (2) The major part (>8O%) of the relaxation amplitude is monitored in the ps to ms time range; this is also true for the fluorescence anisotropy time dependence around T,, as fig. 15 shows. Details of these measurements are given in the literature.’”* DMPC preparations of UVs show similar behaviour.’*’’ In fig. 16 we show the kinetic results of DPPC preparations containing 16.5% CHOL.The relaxation amplitudes A5 are decreased by a factor of two, which also holds for process 4. Besides a slightly broader profile above T, nothing dramatic has happened. However, relaxation 3 is strongly affected; the relaxation times 73 have lost their maximum and can still be monitored even 10 K above T,. If we summarize the influence of CHOL on DPPC UVs we find that the phase-transition relaxation is shifted from 74 and 75 towards the 10 ps time range of T~ and the transition is broadened,358 General Discussion 0.55. c a .- Y 2 % 0.5. x x x X X X 7 X X X X x x x x x x x 35 40 45 T I T 30 35 40 45 50 T / "C Fig. 9. Temperature/absorption (484 nm) ( a ) or turbidity (300 nm) (6) measurements representing the main phase transition of UVs from DPPC containing ca.1% AOL: X , T increasing; A, T decreasing. R == 50 nm; clipid = 2.7 x mol dm-3. similar to the d.s.c. measurements in fig. 8. Further details are included in the literature." If we investigate preparations containing the channel forming peptide gramicidin A (GA', the dimer spans the bilayer) we find a similar broadening and increase in the amplitudes for signal 3, as is demonstrated in fig. 17.13 Signal 5 has almost disappeared and signal 4 is reduced by 40%. Reconstituted UVs of DMPC containing ca. 1% bacteriorhodopsin (bR) showed behaviour like (GA') with an even stronger decrease of signal 4 as well as signal 5. Fig. 18 shows the only remaining strong relaxation signal, T ~ , caused by the lipids and a signal TbR which results from the change in absorption around 550 nm belonging to the photocycle of bR.The latter represents the change in the absorption of the chromophore retinal by pumping a proton through the membrane. From the Arrhenius plot in fig. 18 we learn that the activity of bR is strongly influenced by the state of the lipid bilayer. Only when the lipids have switched from the crystalline to the fluid state, bR can effectively pump protons using light as an energy source.General Discussion 3 59 20 16 20 40 60 80 100 120 140 160 180 200 size/nm 20 40 60 80 100 120 140 160 180 200 size/nm Fig. 10. Size distribution of vitrified DPPC UVs, taken from electron micrographs using a fast freezing method ( lo4 K s-') without applying contrast chemicals. ( a ) Injection method: total no.of vesicles = 554, maximum no. = 14.3; size at maximum = 41.7 nm; direct width parameter = 1.5. (6) Injection method plus sonification; total no. of vesicles = 1046; maximum no. = 37; size at maximum = 37 nm; direct width parameter = 1.41. In conclusion, we have measured five well separated relaxation processes representing the whole main crystalline-fluid transition in pure UVs of DPPC and could develop a model on the basis of molecular changes to explain the different steps. New theoretical model calculations using Monte Carlo ~irnulations~~ have reached similar conclusions about the existence of states of different order in a bilayer system if intermediate states are acting as buffer. They could not give a timescale for their transition model so far but this could be achieved using our experimental data.The incorporation of substances like CHOL or Gramicidin shifts the relaxation from longer times into the 10 ps time range and favours lipids in an intermediate state. Measurements with UVs containing reconstituted bacteriorhodopsin show a clear preference for the relaxation in the 10 ps regime and prove that the lipids have to be rather mobile to help the protein pumping protons. Further work will probably show if the relaxation around lops can be360 General Discussion T / "C 8 1 TYPT 35 40 45 50 T / "C v1 f \ h" 400 5001 30011 * O O j l 100 f 1 JMPT t L 35 40 45 50 T/ "C ( a ) Fig. 11. Five relaxation times 7 and their corresponding amplitudes A, representing the whole crystaliine-fluid transition in UVs of DPPC.( a ) cDPPC = 2.7 X mol dmP3; hobs = 360 nm; AT = 0.8 K . ( 6 ) cDPPC = 2.8 x mol dm-3; hobs = 360 nm; AT = 0.9 K. connected with a functionally important movement as proposed by Frauenfelder for haemoglobin. References 1 D. Marsh, in Membrane Spectroscopy, ed. E. Grell (Springer-Verlag, Berlin, 1982), pp. 51-137. 2 R. J. Smith and E. Oldfield, Science, 1984, 225, 280. 3 L. Brand, J. R. Knutson, L. Davenport, J. M. Beechem, R. E. Dale, D. G. Walbridge and A. A. Kowalczyik, in Spectroscopy and the Dynamics of Molecular Biological Systems, ed. P. Bayley and R. E. Dale (Academic Press, New York, 1985), pp. 259-305. 4 J. F. Holzwarth, W. Frisch and B. Gruenewald, in Microemulsions, ed. 1. D. Robb (Plenum, New York, 5 N. I . Kanehisa and T. Y. Tsong, J. Am. Chem. SOC., 1982, 100, 424. 6 J. F. Holzwarth, A. Schmidt, H. Wolff and R. Volk, J. Phys. Chem., 1977, 81, 2300; J. F. Holzwarth, in Techniques and Applicarions of Fast Reactions in Solurion, ed. W. J. Gettins and E. Wyn-Jones (Reidel, Dordrecht, 1979), pp. 47-70. 7 J. F. Holzwarth, V. Eck and A. Genz, in Spectroscopy and the Dynamics of Molecular Biological Systems, ed. P. Bayley and R. E. Dale (Academic Press, London, 1985), pp. 351-377. 8 J. M. Kremer. M. W. Esker. C. Pathmamanocharan and P. H. Wiersema, Biochemistty, 1977, 16, 3932. 9 A. Genz and J. F. Holzwarth, Colloid Polym. Sci., 1985, 263, 484. 10 V. Eck and J. F. Holzwarth, in Surfactants in Solution, ed. K. L. Mittal and B. Lindman (Plenum, New 11 A. Genz, J. F. Holzwarth and T. Y. Tsong, Biophys. J., 1986, 13, 323. 12 A. Genz and J. F. Holzwarth, Eur. Biophys. J., 1986, in press. 13 A. Genz, J. F. Holzwarth and T. Y. Tsong, Chem. Phys. Lipids, 1986, in press. 14 0. G. Mouritsen, Computer Studies of Phase Transitions and Critical Phenomena (Springer-Verlag, 1982), pp. 185-205. York, 1984), vol. 3, pp. 2059-2079. Berlin, 1984).General Discussion A 36 1 1% . , , , , I I n U r r ) N - d . . . . . J r n o L n o I n t 4 w iz In 'rr) rn U362 General Discussion Y e - * . , . . . m u rr) N - rc (c !-0 ( cross - sect ion) crystalline - ( b ) cross -section 0 headgroups temp.>> transition temp. equilibrium second temp. <<transition temp. \ \ - hydrocarbon chains fluid @ -2xlo-cs -a // - top - plan / view cluster lipids D fluid lipids @ intermediate Fig. 13. Schematic molecular model of the main phase transition in UVs between and 10's. ( a ) Initial 0 and final @ states. ( b ) Model of the dynamic processes. w rn w364 General Discussion T / "C L--.- 0- 40 45 c 32 35 T / "C Fig. 14. Normalized amplitude sum of the five relaxation processes around T, of DPPC UVs and the equilibrium turbidity/temperature dependence (solid line). R = 60 nm; pH 7.5. x, C.5 relaxations (10-~-10-l s); 0, C.3 relaxations (10-~-1O-' s). T / "C Fig. 15. Fluorescence anisotropy rss = (11 - IL)/( Ill + 21,) temperature dependence for the probe DPHPC in UVs of DPPC between 1 ps and 100 ms. A,,,=360 nm; A0,,=430nm; cDPPC= 2.7 x lop3 mol dm-3. +, T increasing; 0, T decreasing.General Discussion 365 cc -- I 3 PJ c 4 - I In -U 0 h 1 .o .j c rl cc cu 1:: - I t I; P -5 1 h - -0 U - % In U k qi= m 3366 General Discussion ru 0 ro. 4 4 -0- c(k -o-. 4 Lo1 01 In U chGeneral Discussion 367 Fig. 18. Arrhenius dependence of the lipid relaxation T~ and the protein relaxation TbR in the temperature range of the phase transition in DMPC UVs, containing 1.1 % bacteriorhodopsin. cDMPC = 3 x mol dm-3. 17, lipid signal; 0, protein signal.

ISSN:0301-7249    

DOI:10.1039/DC9868100339

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

LIST OF POSTERS Thermodynamics of Ion Binding to Phospholipid Bilayers as studied by Reaction Calorimetry A. Blume, University of Freiburg, West Germany Statistical Thermodynamics of Association Colloids F. A. M. Leermakers, P. P. A. M. van der Schoot, J. M. H. M. Scheutjens and J. Lyklema, Agricultural University, Wageningen, The Netherlands Structural Polymorphism of Membranes made from Lipopolysacharides and Free Lipid A K. Brandenburg and U. Seydel, Forschungsinstitut Borstel, West Germany Proteins in Water-in-oil Microemulsions D. Chatenay, W. Urbach and A. M. Cazabat, Ecole Normale Superieure, Paris, France and C. Nicot, M. Vacher and M. Waks, CNRS, Paris, France Modelling of Temperature Effects on the Ion Permeability of Bilayer Lipid Membranes L. Cruzeiro, P. L. Christiansen and 0.G. Mouritsen, Technical University of Denmark, Lyngby The Influence of the Membrane on the Electron-transfer Kinetics of Cytochrome b5 bound to Phospholipid Vesicles D. M. Davies and J. M. Lawther, Newcastle upon Tyne Polytechnic Morphological Changes induced by Melittin on Lipid Structures as detected by 2H-N.M.R., Freeze Fracture, Gel Filtration and QELS J. F. Faucon, E. J. Dufourc, J. Dufourcq, G. Fourche and J. L. Dasseux, Domaine University, Talence and M. Le Maire and T. Gulik-Kryzywicki, CNRS Gif-sur- Yvette, France Two-dimensional Monte-Carlo Studies of Lipid Molecules in a Bilayer Membrane D. P. Fraser, R. W. Chantrell and D. Melville, Lancashire Polytechnic and D. J. Tildesley, University of Southampton Kinetics of the Phase Transition in Phospholipid Bilayers A.Genz, R. Groll and J. F. Holzwarth, Fritz-Haber-Institut der Max-Plunck-Gesellschaft, West Germany Quinone- and Tocopherol-mediated Redox Reactions through Lipid Membranes A. Ilani and T. Krakover, Hebrew University, Jerusalem, Israel Multilamellar Haemoglobin Liposomes Hw. Jaroni, V. Pirkl, R. Schubert and K. H. Schmidt, Chirurgische Klinik, Tiibingen, West Germany The Effect of Chemical Modification of Gramicidin on its Lipid Structure-modulating Activity J. A. Killian, C. W. van den Berg, H. Tournois, J. W. Timmermans, S. Keur and B. de Kruijff, State University of Utrecht, The Netherlands E.S.R. Contribution to the Study of Model Membrane Phase Transitions C. Lecompte, P. Bonnet, A. Vachon, V. Roman and F. Berleur, CENSaclay, Gif-sur-Yvette, France Mechanistic Aspects of the Reconstitution of Phosphatidylcholine Vesicles by Dilution of Phos- phatidylcholine-Sodium Cholate Mixed Micelles S.Almog, Tel Aviv University, Israel, T. Kushnir, Chaim Sheba Medical Center, Israel, S . Nir, University of Jerusalem, Israel and D. Lichtenberg, Tel Aviv University, Israel Permeability of Polymerised Vesicles using Small-angle Neutron Scattering A. G. Muddle, P. G. Cummins and E. J. Staples, CIBA-GEIGY Pharmaceuticals, Horsham Electrochemistry of Phospholipid Monolayers at the Mercury-Water Interface A. Nelson, IMER, Plymouth Micelle Vesicle Transitions of the Egg pc-Octylglucoside System. Application to Membrane Reconstitution M. Ollivon, 0. Eidelman, A. Walter and R. Blumenthal, NCI, Bethesda, USA and CNRS Thiais, France 369370 List of Posters Studies of the Interaction of Ubiquinone- 10 with Phospholipid Model Membranes using Lan- thanide Shift Reagents M.Ondarroa and P. J. Quinn, King’s College, London Computer-simulation Studies of Polymerized Lipid Bilayers D. A. Pink and D. J. Laidlaw, St Francis Xavier University, Canada Liposomes in a Net H. Ringsdorf and B. Schlarb, Johannes Gutenberg- Universitat, Mainz, West Germany Polymerizable Lipids with Hydrophilic Spacer Groups and Liposome Formation from Polymer Lipids R. Elbert, A. Laschewsky and H. Ringsdorf, University of Mainz, West Germany Peptide Liposomes from Amphiphilic Amino Acids R. Neumann and H. Ringsdorf, University of Mainz, West Germany Determination of Vesicle Size Distributions by Intensity Fluctuation Spectroscopy IFS.Analysis by a Smoothed Inverse Laplace Transform H. Ruf, E. Grell and Y. Georgalis, Max Planck-Institut, Frankfurt, West Germay Structural Changes in Membranes of Unilamellar Lipid Vesicles by Interaction with Cholate R. Schubert, H. Wolburg and K. Schmidt, Tiibingen University, West Germany and K. Beyer, Munich University, West Germany X-Ray Diffraction Calorimetric Studies of Hydrated Phosphatidylcholine-Fatty-acid Mixtures J. M. Seddon, University of Southampton, G. Cevc, Uniklinikum Essen, West Germany and D. Marsh, Max-Planck-Institut, Gottingen, West Germany Lipid Vesicles in Human Bile. A Quasielastic Light-scattering and Chromatographic Study G. J. Somjen and T. Gilat, Tel Aviv University, Israel The Influence of Extrusion on the Number of Lamellae and Size Distribution of a Dispersion of Multilamellar Vesicles (MLV) H. Talsma and D. J. A. Crommelin, University of Utrecht, The Netherlands, H. Jousma and H. E. Junginger, University of Leiden, The Netherlands and J. G. H. Joosten, University of Utrecht, The Netherlands Bilayer Couple Hypothesis and Phospholipid Vesicle Shapes B. Zeks and S. Svetina, University of Ljubljana, Yugoslavia Molecular Dynamics and Structure of Peptide Hormones at Membrane Interfaces M. Vincent, M. Waks and J. Galley, CNRS, Paris, France The Mechanism of Bilayer Fusion. Liposome Fusion and Solubilization in the Presence of Surfactants or Polyethyleneglycol A. Alonso and F. M. Goiii, University ofthe Basque Country, Spain

ISSN:0301-7249    

DOI:10.1039/DC9868100369

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

INDEX OF NAMES* Afshar-Rad, T., 239 Anderson, N. H., 313, 351, 352 Angelova, M. I., 303 Arrondo, J. L. R., 117 Bader, H., 329 Bailey, A., 239, 262, 263, 264 Berleur, F., 143 Boden, N., 191,217, 218 Bont, W. S., 206, 342 Brand, L., 81 Brandenburg, K., 147 Carvell, M., 223 Cevc, G., 66, 141, 179, 201, 210, 212, 258, 342 Chapman, D., 107, 144, 145, 239 Cornell, B. A., 142, 163, 206, 207, 209, 210 Corns, S. B., 208 Dale, R. E., 76, 137 Davenport, L., 81, 137, 138, 139 Davis, S. S., 313 De Kruijff, B., 70, 148 Dimitrov, D. S., 74, 264, 303, 339, 345, 348, 349 Drummond, C. J., 95 Duwe, H-P., 281 Engelhardt, H., 281 Evans, D. F., 1 Evans, E., 209, 262,267, 339, 340, 341 Fisher, L. R., 72, 146, 249, 263, 266, 344, 345 Fromherz, P., 39,71,72,73,76,139,206,339,347,353 Gershfeld, N.L., 19, 63, 64, 65, 66, 202 Grieser, F., 95 Gruner, S. M., 29 Goiii, F. M., 117, 148 Hall, D. G., 65, 214, 215, 223 Harrop, R., 149 Haydon, D. A., 67,249,353 Hayward, J. A., 107 Healy, T. W., 95, 140, 141, 142, 144, 263, 343 Hill, P. A., 208 Holzwarth, J. F., 74, 77, 145, 209, 340, 353 Howe, A. M., 145, 353 Jackson, S., 291 James, M. J., 313 Jones, M. N., 65, 144, 202, 291, 343, 344, 345 Knutson, J. R., 81 Korstanje, L. J., 49 Kronberg, B., 350 Kurihara, K., 353 Lee, D. C., 107 Leermakers, F. A. M., 66, 139, 140, 260 Levine, Y. K., 49, 76, 79, 137 Lindblom, G., 71, 77 Luckham, P., 142,239, 262, 263, 264 Lyklema, J., 66, 139, 140, 260 Lyle, I. G., 223, 258, 291 MacNaughtan, W., 239 Marsh, D., 63, 68, 179 Middlehurst, J., 163 Murtagh, J., 127 Nagle, J. F., 63, 151, 201,202,203,204, 210,215,340, Needham, D., 267 Nelson, A., 349, 352 Ninham, B. W., 1 Nossal, R. J., 19 Parker, N. S., 249 Parsegian, V. A., 29, 67, 68, 69, 70, 71, 140, 207, 212, 218, 257, 266 Rand, R. P., 29, 257 Reboiras, M. D., 291 Ringsdorf, H., 329 Rocker, C., 39 Ruppel, D., 39 Sackmann, E., 203,212, 281, 342, 348, 352, 353 Scheutjens, J. M. H. M., 66, 139, 140, 260 Schubert, R., 73 Seddon, J. M., 69, 179 Separovic, F., 163 Siegel, D. P., 65, 149, 203 Sixl, F., 191, 218 Stevens Jr, W. B., 19 Thomas, J. K., 72, 127, 138, 149, 344, 352 Tiddy, G. J. T., 70, 145, 146, 149, 223, 258, 262, 351 Van Ginkel, G., 49 van Langen, H., 49 Yarwood, J., 145, 259 Young, D. A., 208 ZekS, B., 342 351 * Page numbers in bold type denote papers submitted for discussion. 371

ISSN:0301-7249    

DOI:10.1039/DC9868100371

出版商:RSC    

年代:1986

数据来源: RSC