Faraday Discuss. Chem. SOC., 1986, 81, 49-61 The Correlation between Molecular Orientational Order and Reorientational Dynamics of Probe Molecules in Lipid Multibilayers Gijs van Ginkel, Leo J. Korstanje, Herman van Langen and Yehudi K. Levine* Department of Molecular Biophysics, Physics Laboratory, University of Utrecht, Princetonplein 5, 3584 CC Utrecht, The Netherlands The behaviour of probe molecules in lipid systems can be characterized in terms of molecular order parameters and rates of reorientational motion. The correlation between these parameters has been investigated in oriented lipid multibilayer systems on changing the degree of unsaturation of the chains and the cholesterol content. Angle-resolved fluorescence depolarization experiments were carried out as a function of two angles: (1) the angle of incidence of the excitation beam and (2) the angle between the directions of incidence and observation.The e.s.r. spectra from nitroxide spin labels were simulated numerically with a general stochastic Liouville equation formalism. The probe molecules were assumed to undergo stochastic small-step reorientational diffusion subject to an orienting potential. Both techniques indicate that the introduction of unsaturation into the lipid chains lowers both the order parameters and the diffusion coefficients of the probe molecules. The incorporation of cholesterol reverses this trend. These findings are at odds with current ideas about the fluidity of membrane systems. Much of our current understanding of the dynamic structure of biological membranes has been derived from studies of model systems, particularly lipid bilayer systems.It is now generally accepted that the bilayer can be considered as an orientationally anisotropic fluid and that the degree of anisotropy varies across its thickness. The dynamic structure of the system is characterized in terms of order parameters and correlation times (or diffusion coefficients) for the various motions. However, the details of these properties revealed by 2H-n.m.r. seem to differ significantly from those obtained from intrusive probe techniques such as e.s.r. and fluorescence depolarization. As a result, the fidelity of the reporting of the probe molecules has been questioned in the last decade on the grounds that the molecules perturb the structure of their surroundings to such an extent that they do not monitor the intrinsic properties of the bilayer.Nevertheless, these probe techniques are particularly useful tools in membrane studies not only because of their sensitivity, but also because the specific labelling of complex molecules with 2H atoms is not a trivial task. In recent years there have been important new developments in the interpretation of e.s.r. and fluorescence depolarization experiments on liquid-crystalline materials. These treatments not only take explicitly into account the orientational and motional anisotropy, but are also valid in situations where the correlation times of the molecular motions are similar to the intrinsic timescale of the experiment, e.g. the fluorescence lifetime.One important conclusion is that the interpretation must be based on a model for reorientational motion and that the results obtained may only be valid within the context of that model. It is thus imperative to test the validity of the description used in the analysis. To this end one requires the use of macroscopically oriented membrane 4950 Order and Dynamics in Lipid Multibilayers systems which afford the determination of the anisotropic response of the reporter molecules. We shall present here a study of a number of oriented multibilayer systems differing in the degree of unsaturation of the hydrocarbon chains and cholesterol content using fluorescence depolarization and e.s.r. techniques. The results are analysed using the rotational diffusion model in which the probe molecules are assumed to undergo stochastic reorientational motions subject to an anisotropic orienting potential.This model is found to provide a satisfactory description of the behaviour of the probes in both the techniques used. The introduction of unsaturation into the lipid chains lowers both the order parameters and the rates of reorientational motion of the probe molecules. On the other hand, the incorporation of cholesterol into the bilayers has the opposite effect, enhancing the orientational order and the rates of motion. These findings are at odds with current ideas about membrane fluidity. Both techniques yield strikingly similar values for the order parameters and motional rates for probes anchored in the headgroup region of the bilayer.This indicates that the probes perturb the bilayer structure to a lesser extent than has been presumed up to now. Theory Orientation of Molecules in Membrane Systems We shall consider here the orientation behaviour of molecules relative to the bilayer normal, the local director, within the framework of the molecular field approximation. We shall further restrict the discussion to the case of a uniaxial bilayer system containing molecules with an effective cylindrically symmetric form. Thus the orientation of the molecules in the system is specified by the angle p between the molecular symmetry axes and the local director.' The orientational distribution of an average molecule in the system is then characterized by a probability distribution f( p).This distribution can be expressed2 as a series expansion of Legendre polynomials P,(cos p), each of which is weighted by an order parameter (PL) which is the ensemble average of the corresponding term This expansion defines the order parameters (P2) and (P4) as ( P 2 ) = i ( 3 cos2 p - 1) ( p4) = $(35 C O S ~ p - 30 C O S ~ p + 3). ( 2 4 ( 2 6 ) The orientational distribution function f ( P ) is fully characterized if all the order parameters (PL) are known. In practice, however, only (P2) and (P4) are accessible experimentally. For example 2H-n.m.r. and linear dichroism experiments yield ( P2) only, while e.s.r. and two-photon experiments such as Raman scattering and fluorescence depolarization yield both (P2) and (P4). The main difficulty now is obtaining an objective and realistic estimate of the form of f ( P ) from limited knowledge of its moments.This may be accomplished by an information-theoretic a p p r ~ a c h . ~ The essential point is that, given the order parameters ( P2) and ( P4), the most probable values of ( PL), L 3 6, are calculated under the assumption that the informational entropy off@) is a maximum. This corresponds to the construc- tion of the broadest possible distribution function consistent with the known values of (P2) and (P4). The resulting distribution function has the form f ( P ) = A exp [A2P*(cos P ) +A4P4(cos P)I (3)G. van Ginkel, L. J. Korstanje, El. van Langen and Y. K . Levine 51 where A is a normalization constant and h2 and h4 are determined from the known values of (P2) and (P4).If only the order parameter (P2) is known, the distribution function takes the form of eqn ( 3 ) but with h4 = 0. We note here that eqn (3) has the form of a Boltzmann distribution with an angle-dependent orienting potential. It can be seen from eqn (3) that if only ( P2) is known, the reconstructed distribution function either has a maximum at p = 0 and decreases monotonically to a minimum at /3 = 7r/2 or vice versa. Knowledge of (P4) is required for establishing the existence of a collective molecular tilt which is manifested by a maximum of the distribution function at an angle intermediate between 0 and ~ / 2 . On the other hand the observation of a minimum at such an angle may be taken as an indication for a superposition of two or more independent populations of molecules.The available information, however, is too limited to allow their resolution. Model for Reorientational Motion We shall assume that the reorientational motion of the molecule is a stochastic Markov process and neglect any inertial effect^.^ Consequently the Markov process need be described only in terms of the random angle p ( t ) , The master equation for the conditional probability P(Po/Pt) that the molecule has an orientation p relative to the director at time t, given that it had an orientation Po at t = O is (4) a - P(Po/Pt) = r,P(Po/Pt) at where rp is the stochastic operator describing the orientational motion. We shall here only consider the rotational diffusion model in which the average molecule is assumed to undergo small-step stochastic diffusion subject to an orienting potential U ( p ) . The Markov operator rP is now given by4 where D is the rotational diffusion tensor, diag (D,, D,, Dll), and M is the angular momentum ~ p e r a t o r , ~ both defined in the molecular frame.Eqn (4) is solved numeri- cally4 subject to the initial boundary condition P(pO/pO) = S(po-p). In view of the discussion above on the reconstruction of the equilibrium orientation distribution function, we shall choose U ( p ) to take the form U ( p ) = -~T[A,P~(COS p ) + A~P~(cos p ) ] . ( 6 ) We note that our choice of the orienting potential spans all the physically permissible pairs of (( P2), ( P4)) values. Angle-resolved Fluorescence Depolarization (AFD) Experiments The theory of the experiment has been discussed in detail e l ~ e w h e r e ~ - ~ and will only be summarized briefly. A macroscopically aligned bilayer system is subjected to con- tinuous illumination with light of defined wavelength and polarization direction.The geometrical arrangement is shown in fig. 1. The excitation light is incident at an angle 8 relative to the macroscopic director and is polarized in the zy plane. The fluorescence emission is observed at an angle 4 with its polarization either parallel (Ill) or perpen- dicular ( I , ) to the zy plane. In general the polarizations of the exciting light and the fluorescence emission will not be parallel in the zy plane. The polarization ratio Re = I,/ Ill is given by R , = [ I - R , + ( R ~ + R ~ ) sin2 ~ ] [ ~ + R , + ( R ~ - R ~ ) sin2 & + ( R ~ - R ~ ) sin2 e +R, sin2 8 sin2 & + R4 sin 28 sin 241-l.(7)52 Order and Dynamics in Lipid Multibilayers I X I I I I I I Fig. 1. Experimental AFD geometry for a multibilayer sample S lying in the xy plane. 6 and 4 are the angles in air between the macroscopic director z and, respectively, the direction of excitation and observation of the fluorescence. The states of polarization, either parallel or perpendicular to the zy plane, are determined by the polarizer P and the analyser A. The angles 8 and and the intensities Il and Ill in eqn (7) refer to quantities within the sample and not to those measured in the laboratory system. The angles in the sample and the laboratory can be related simply if the refractive index of the sample is known. However, the effects of the sample-air interface on the transmission of light intensity are more complex and depend furthermore on the direction of polarization.The various artefacts have been discussed in detail by Lax and Nelson.' It can be easily shown" that the effects of multiple reflections of the exciting and emitted light within the sample compensate the transmission losses of the fluorescent light at the interface. Furthermore, consideration of the +dependence of the intensities Il and 11, obtained with a normal incidence of the exciting light ( 8 = O O ) , indicate that only solid-angle expansion' distorts the experimental results. This effect cancels out on calculating the polarization ratio. Consequently no correction need be applied to the values of Re determined experimentally. The polarization ratio Re is measured for various combinations of 8 and 4 and affords the determination of five independent quantities Sp, S,, go, g , and g2 from steady-state experiments.Here Sp and S , are, respectively, the second-rank order parameters ( = ( P2)) for the absorption and emission transition moments. The quantities gk, k = 0, 1, 2 are defined by where DiI are Wigner rotation matrix elements" and .Iz, and fi, denote, respectively, the orientation of the absorption moment at time t = 0 and that of the emission moment at time t, relative to the director frame, fig. 1. F ( t ) denotes the normalized intrinsic fluorescence decay function. The time behaviour of the correlation function Gk(t) is obtained numerically from the solution of eqn (4)4,'2 and for many practical situations is found to be a mono- exponential decay.G. van Ginkel, L.J. Korstanje, H. van Langen and Y K. Levine 53 It is important to realize that the values of Gk(t) for t = 0 and t + 00 are model independent as a consequence of the assumption that the motion can be described as a stochastic process. These limits can be expressed solely in terms of the order parameters (P2) and (P4).l3,I4 The correlation functions G,( t ) and G2( t ) decay to zero at long time, whereas the function Go( t ) decays to a constant value given by ( P2)2. This behaviour is in marked contrast to that observed in isotropic systems. F ( t ) can be determined experimentally on observing the fluorescence emission in a direction normal to the sample surface ( 4 = 0') and under an angle of incidence in the sample 6 = sin-' ( l / f i ) .If the polarizer on the emission side is set with its optical axis at 45" to the vertical, then the combined signal Il + Ill is proportional to F ( t ) . Simulations of E.S.R. Spectra The analysis of e.s.r. spectra of nitroxide spin labels embedded in lipid bilayers in terms of the orientational distribution and the reorientational motions of the molecules has been reviewed in detail e1~ewhere.l~ Suffice it to say that the spin Hamiltonian describing the spectral line shapes contains contributions from the anisotropic Zeeman and hyperfine interactions and is given by where g and A are the Zeeman and hyperfine tensors respectively, S the electron spin and I the nitrogen spin operators.The interpretation of the e.s.r. spectra is not straightforward, but determined by the ratio of the time scale of the reorientational motions, expressed as a correlation time T, to the anisotropy of the hyperfine tensor A expressed in frequency units. Three regimes can be distinguished. (i) Fast-motion Regime ( lo-' < r / s < 2 x 1 0-9) In this regime the e.s.r. spectrum consists of three lines whose positions are determined by the time-averaged spin Hamilt~nian'~ and yield the order parameter ( P2). The spectral lineshapes containing the information about ( P4) and the reorientational motion can be calculated using Redfield's (ii) Slow-motion Regime (2 x In this regime the spin Hamiltonian must be considered as an explicit function of the orientation of the spin label with respect to the external static magnetic field B0.16717 The time evolution of the classical stochastic orientation is now coupled with the spin variables and the e.s.r.spectra can be simulated only on solving the stochastic Liouville equation (SLE) for the density matrix operator p ( a , t ) : l 6 > l 7 < r / s < where r p is defined by eqn (4), &(a, t ) is the Hamiltonian describing the interaction between the spins and the applied microwave field and p 0 ( a ) is the equilibrium density matrix operator. The SLE can only be solved numerically, and fast and reliable algorithms have been developed in recent years.17-19 The spectral lineshapes in this regime may become quite complex and often lose their triplet character.20y21 Spectra consisting of a superposition of five lines are fairly common.54 Order and Dynamics in Lipid Multibilayers Fig.2. Simulations of the spectrum of a CSL molecule embedded in a lipid bilayer with the applied magnetic field Bo perpendicular to the bilayer plane. The order parameter ( P2) = 0.45, while (P4) decreases from top to bottom ( a ) 0.20, (6) 0.13, ( c ) 0.10 and ( d ) 0.05; D, = 1.0 x lo7 s-' and Dll = 5D,. (iii) Powder [ Rigid-lattice Limit ( T / S > lop7 )] The reorientational motions are now too slow to affect the form of the e.s.r. spectra, which can thus be described as a superposition of lines arising from a static distribution of spin label molecules. It has not been widely appreciated that all three motional regimes can be treated by the SLE formalism.The drawbacks of this approach are its apparent mathematical complexity and the use of involved numerical algorithms for spectral simulations. Consequently it has been neglected in the past. The vast majority of people preferred the mathematically tractable Redfield theory for the analysis of e.s.r. experiments. This has often led to reports of motional rates quite clearly too slow to fall within the fast motion approximation. The question then arises as to whether the conclusions drawn in those studies are tenable. The adherence to the fast motional limit with its three line spectra, has led many workers to believe that any other spectral lineshape must be due to a superposition of signals from two or more populations of labels differing in their dynamic behaviour.However, as illustrated in fig. 2, a complex lineshape can also be obtained from a dynamically homogeneous population in the slow motion regime.20721G. van Ginkel, L. J. Korstanje, H. van Langen and Y K. Levine 55 Small changes in the order parameters (P4), keeping all other parameters constant, can be seen to induce large changes in the e.s.r. spectra. One is then forced to conclude that the interpretation of experimental e.s.r. spectra exhibiting five or more lines is ambiguous in the absence of other information about the system. The simulations of the experimental e.s.r. spectra presented below were carried out using the SLE formalism and the Lanczos a l g ~ r i t h m ’ ~ ” ~ for matrix diagonalization. In order to restrict the number of parameters, we have used the potential (6) with h4 = 0.The simulated spectra were convoluted with a Gaussian line in order to account for the broadening caused by unresolved electron-proton hyperfine interactions. Experimental Apparatus Angle-resolved fluorescence depolarization experiments were carried out with a home- built fluorimeter described Fluorescence decay measurements were carried out with a similar home-built set up, using an Edinburgh Instruments 399 flash-lamp operating with N2 gas at atmospheric pressure as a light source, and equipped with single-photon-counting detection, The fluorescence decay law as extracted from the measurements using POPOP as a reference compound following Zucker et a1.22 The intensity of the exciting light was kept as low as possible to avoid bleaching of the probe molecules and the intensity of fluorescence, measured at the same scattering geometry, remained constant during the experiments.Control experiments showed that the intrinsic fluorescence signals from the lipids and the coverslips amounted to <1% of the probe fluorescence intensity. E.s.r. measurements were carried out on a Varian E-9 X-band spectrometer equipped with a TM 110 cavity and a Varian V4540 variable-temperature accessory. Most of the spectra reported below were recorded with the applied static magnetic field normal to the plane of the sample. Materials Dimyristoylphosphatidylcholine (DMPC), palmitoyloleoylphosphatidylcholine (POPC), dioleoylphosphatidylcholine (DOPC), egg phosphatidylcholine (EPC) and cholesterol were obtained from Sigma and digalactosyldiglyceride (DGDG) was from Lipid Products.The lipids were used without further purification. The fluorescent probe 1,6-diphenyl- 1,3,5-hexatriene (DPH) was obtained from Fluka and its polar analogue trimethylamino-DPH (TMA-DPH) was purchased from Molecular Probes Inc. The probes were dissolved in absolute ethanol (AR, Baker) and stored in the dark at 4°C. The cholestane nitroxide spin label (CSL) was purchased from Syva. Doubly distilled water was used throughout. Sample Preparation Macroscopically aligned bilayer systems were prepared as described and the alignment monitored under a polarizing microscope equipped with a first-order red-plate. The sample contained ca. 25 ‘/o water unless otherwise stated. The molecular probe to lipid ratio was 1 : 250 for AFD experiments and 1 : 125 for e.s.r.experiments. Oxidation of the Lipids As the unsaturated DOPC and DGDG molecules are highly susceptible to oxidation, the sample preparation was carried out strictly under a nitrogen atmosphere. The lipids came into contact with air for a short time during the alignment procedure under the56 Order and Dynamics in Lipid Multibilayers polarizing microscope. The oxidation of the lipids was monitored after every preparative step by recording their absorption spectrum in the 200-300 nm region.z3 Samples showing traces of peroxide formation were discarded. Results and Discussion AFD Experiments The polarization ratios Re were measured for 56 combinations of 8 and $ for each sample and the parameters R,- R5 were determined from the angle-dependence following ref.(7) and (8). The quantities Sp, S,, go, g , and g , were obtained on taking appropriate linear combinations of R1-R5.6 The same values of Re within the experimental error of 2% were obtained on rotating the samples by an arbitrary angle about the normal to their surface. This indicates that the macroscopic distribution of probe molecules is uniaxially symmetric. The analysis shows that in all the systems studied S >S, in accordance with our previous and the recent finding of JohanssonZ57n other oriented lipid systems. Interestingly, we have found that the values of S, in multibilayers of egg PC increase on reducing the hydration of the bilayers. These observations indicate either that the absorption and emission moments of the DPH and TMA-DPH molecules are not mutually paralle17,R,24 or that the orientational distribution function of the molecules in the excited state differs significantly from that in the ground If the latter explanation is correct, then the analysis of the experiments becomes difficult.However, we have previously argued7 on the basis of the known photophysical properties of DPH 12,26727 that vibronic mixing of the close-lying ' B : and 'A: excited states leads to polarization borrowing effectsz8 and that consequently the two transition moments are not mutually parallel. We have therefore analysed the experimental data on assuming that the absorption transition moment, p, lies parallel to the molecular symmetry axis and that the emission moment, u, is tilted by an angle pV with respect to that axis.Furthermore we assume that all the molecules can be assigned to the same dynamic behaviour. With these assumptions we have It is important to realize that the ensemble averages implied in eqn ( 1 1 ) and (8) entail an average over the multibilayer stack and the derived order parameters and diffusion coefficients are not necessarily properties of a single bilayer. The experimental results were analysed' using the measured fluorescence decay behaviour given in table 1, with D,, h2, h4 and P2(c0s p,) as free model parameters. The derived values of the parameters for the different systems are summarized in table 1. In this connection it is important to note that the values shown change by <5% if the fluorescence decay function was taken to be monoexponential with a lifetime equal to (T): ( T ) = loa tF( t) dt.It is thus possible to analyse the experimental data without prior knowledge of the fluorescence decay and on taking F ( t ) to be a monoexponential function. In this case the model parameter D, is simply replaced by the product D,(T). It can be seen from table 1 that the angle P, for DPH varies with the composition of the system and can reach large values in unsaturated lipid bilayers. In marked contrast p, = 15 *I3" for TMA-DPH molecules embedded in all the lipid systems studied. A further marked difference between the two probes is that the orientation distribution for DPH is characterized by h2 == h4, while for TMA-DPH h2 > h4 in all the systemsC Q 3 Table 1.Model parameters and fluorescence decay obtained from AFD experiments Q 5. lipid + probe a1/a2 Tl/ns 7,/ns PZ A 2 h4 D,/ns-' f ( g / 2 > / f ( o ) T/"C "g DMPC + DPH - - 8.21 24 2.41 0.69 0.045 - 2 35 POPC + DPH 0.05 1.62 7.60 30 1.75 1.06 0.034 - 4 21 fr DOPC + DPH - - - 28 0.95 0.96 - 0.22 13 21 5 DGDG+ DPH - - - 34 1.55 0.75 - 0.04 6 21 2: EPc(24'/0 H20) + DPH 0.59 5.53 8.53 22 1.77 0.53 0.044 - 6 21 s EPC( 10% H20) + DPH - - 7.39 13 2.28 0.6 1 0.037 - 3 21 5. EPC(8.3Yo H20) + DPH 0.27 3.76 8.16 10 2.50 0.60 0.027 - 2 21 POPC/CHOL(20%) + DPH I - 8.0 14 2.93 1.22 0.042 - 0.5 21 3 DOPC/CHOL(20%) + DPH - - - 21 1.83 0.96 - 0.24 4 21 g DGDG/CHOL(20%) + DPH - - - 24 1.89 0.65 - 0.17 4 21 DMPC + TMA-DPH 0.07 0.54 5.8 12 3.24 0.06 0.035 - 0.5 35 s POPC + TMA-DPH 0.24 1 .oo 4.8 18 2.96 0.53 0.014 - 1 21 DOPC + TMA-DPH 1.33 2.45 5.42 19 2.41 0.18 0.022 - 2 21 3 DGDG+TMA-DPH 0.35 1.19 3.55 17 2.27 0.12 0.014 - POPC/CHOL(20%) +TMA-DPH - - 5.5 13 3.58 0.89 0.023 - 0.5 21 & 1 21 3 DPOC/CHOL(20%) +TMA-DPH - - 4.2 15 2.82 0.64 0.024 - 21 k 3 21 A DGDG/CHOL(20°/o + TMA-DPH - - 3.5 17 2.69 0.40 0.017 - 1.558 Order and Dynamics in L -80-60-40-20 0 20 40 '60 00 ipid Mult ibila y ers -80-60-40-20 0 20 40 60 80 P / " P / " 0.11 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 ) 0.0. 0. 0. 0 . -80-60-40-20 0 20 40 60 80 -80-60-40-20 0 20 40 60 80 P / " P l " Fig. 3. The orientational distribution functions of DPH ( a ) and (c) and TMA-DPH ( b ) and ( d ) molecules in multibilayers of POPC.(c) and ( d ) show a blown-up plot to emphasize the form of the distribution for angles near to * ~ / 2 . h @l IL v 4.0 ,&I 3.21 0.5 0 - 4 0 . 1 4.01 1 1 I I I 1 I I 1 I 1 1 I I I I 1 1 3.2 2.4 ;;;' 1.6 0.0 0 -80-6e40-20 0 20 40 60 80 Pl" P l " -80-60-40-20 0 20 40 60 80 P l " -80-60-40-20-0 20 40 60 80 P l " Fig. 4. The orientational distribution functions of DPH ( a ) and (c) and TMA-DPH ( b ) and ( d ) molecules in multibilayers of DOPC. (c) and ( d ) show a blown-up plot to emphasize the form of the distribution for angles near to * t / 2 .G. van Ginkel, L. J. Korstanje, H. van Langen and Y. K. Levine 59 studied. Fig. 3 and 4 show the distribution functionf(P) for the two probes in POPC and DOPC bilayers, respectively. The main difference can be seen to be the appearance of a distinct minimum at p - 60" for DPH molecules and their more pronounced tendency to lie with their axes parallel to the bilayer surface.This is also the case for the other lipid systems. The addition of cholesterol, however, suppresses this effect. Note (table 1) that in EPC bilayers, the population of DPH molecules lying parallel to the bilayer surface increases with increasing hydration. These results strongly suggest a heterogeneous distribution of DPH molecules in lipid bilayers, consistent with their lipophilic nature. On the other hand, TMA-DPH molecules, which are expected to be anchored at the headgroup region of the bilayer appear to form a homogeneous population in all the systems studied. We consider it therefore inappropriate to use DPH molecules as probes of membrane dynamic structure especially in unsaturated systems.The results for TMA-DPH molecules exhibit the known trends of lower order parameters in unsaturated systems than in saturated systems above their phase transition and the increase in molecular orientational order induced by cholester01.~~ However, the results show unambiguously that the diffusion coefficients D, are lower for the unsaturated systems at 21 "C, well above their phase transition, than for DMPC at 35 "C. Furthermore D, increases in the presence of cholesterol. This latter behaviour has been observed earlier using 2H-n.m.r. techniques." These findings run counter to established ideas about membrane fluidity29 and may simply indicate that the only role of unsatur- ation in lipid bilayer systems is the lowering of the phase transition temperature of the chains, E.S.R.Experiments E.s.r. experiments complement AFD measurements on DPH and TMA-DPH in that the spectra of CSL molecules embedded in lipid bilayers reflect not only the orientational ordering and D,, but also Dll characterizing the reorientational motion of the molecules about their long axes.16-19 As CSL molecules are known to be anchored in the head-group region of the bilayers, their behaviour should be analogous to that of TMA-DPH molecules. The three-line e.s.r. spectra of CSL in bilayers of DMPC and POPC above their phase transition were simulated numerically with the general SLE formalism employing the simple potential U ( P ) = -~TA~P*(cos p). This approximation provides an adequate description of the AFD results with TMA-DPH as probe.The temperature dependences of the order parameter (P2) and the diffusion coefficients D, and Dll are shown in fig. 5 and 6, respectively. The values of the diffusion coefficients found fall quite clearly within the slow-motion regime for both systems. This conclusion is supported by the observation of a significant asymmetry in the positions of the low- and high-field lines relative to the centre of the spectrum. We have previously shown that still lower values of the diffusion coefficients are derived from simulations of the spectra of CSL in DGDG bilayer systems.2' The results show that the ordering of CSL molecules in DMPC bilayers is higher than in POPC bilayers, yet in the 30-50 "C temperature range similar values for D, are found.Nevertheless, the CSL molecules appear to rotate faster about their long axes in the DMPC bilayer over the same temperature range. Table 2 shows that the results obtained for the same lipid system with CSL and TMA-DPH probe molecules are strikingly similar. This agreement supports our approach to the interpretation of the experimental results and the conclusions reached.60 1.0 0.9 0 . 8 0 . 7 0 . 6 0 . 5 0.4; Order and Dynamics in Lipid Multibilayers I I 1 I I I - - - 8 - 0 D - 0 0 8 - ,,no c1 - 8 . - 8 " - - I I I I I I Fig. 5. The temperature dependences of (P2) obtained from numerical simulations of the spectra from CSL molecules embedded in multibilayers of DMPC (0) and POPC (W) above their respective phase transitions.1 o9 1 o8 1 o7 1 o6 1 o5 3.0 3.1 3.2 3 . 3 3 . 4 3 . 5 3.6 3.7 lo3 KIT Fig. 6. The temperature dependences of the diffusion coefficients obtained from numerical simula- tions of the spectra from CSL molecules embedded in multibilayers of DMPC (0, Dil; 0, Dl) and POPC (0, Dll; ., Dl) above their respective phase transitions.G. van Ginkel, L. J, Korstanje, H. van Langen and Y. K . Levine Table 2. A comparison of the parameters obtained with TMA-DPH and CSL molecules embedded in different multibilayer systems 61 label POPC, 22 "C CSL 0.60 0.24 1.7 x 10' TMA 0.66 0.35 1.4 x lo7 CSL 0.68 0.26 3.5 x lo7 TMA 0.64 0.29 3.5 x 10' TMA 0.50 0.18 1 . 4 ~ lo7 DMPC, 35 "C DGDG, 22°C CSL" 0.40 0.10 8 x lo6 " From ref. (21).We gratefully acknowledge the contributions made by G. Deinum, D. Engelen and F. Mulders to the experimental work and thank Dr A. J. Dammers for developing the e.s.r. simulation programs. References 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 I C. Zannoni, in The Molecular Physics of Liquid Crystals ed. G. R. Luckhurst and G. W. Gray (Academic Press, London, 1979), chap. 3, pp. 51-83. L. B-A. Johansson and G. Lindblom, Q. Rev. Biophys., 1980, 13, 63. The Maximum Entropy Formalism, ed. R. D. Levine and N. Tribus (M.I.T. Press, Boston, 1979). P. L. Nordio and U. Segre, in The Molecular Physics of Liquid Crystals, ed. G. R. Luckhurst and G. W. Gray (Academic Press, London, 1979), chap. 18, pp. 411-426. E. Merzbacher, Quantum Mechanics (Wiley, New York, 2nd edn, 1970), chap. 9, pp.172-190. B. W. van der Meer, R. P. H. Kooyman and Y. K. Levine, Chem. Phys., 1982, 66, 39. R. P. H. Kooyman, M. Vos and Y. K. Levine, Chem. Phys., 1983,81, 461. M. J. M. van de Ven and Y. K. Levine, Biochim. Biophys. Acta, 1984,777, 283. M. Lax and D. F. Nelson, in Coherence and Quantum Optics, ed. L. Mandel and E. Wolf (Plenum, New York, 1973), pp. 415-445. J. NauB, AUPO, Fac. r. Nar. vol. 76, Physica XXII, 1983, pp. 9-18. M. E. Rose, Elementary Theory of Angular Momentum (Wiley, New York, 1957). C. Zannoni, A. Arcioni and P. Cavatorta, Chem. Phys. Lipids, 1983, 32, 179. P. L. Nordio, in Spin Labelling, Theory and Applications, ed. L. J. Berliner (Academic Press, New York, 1976), chap. 2, pp. 5-52. J. Seelig, in Spin Labelling, Theory and Applications, ed. L. J. Berliner (Academic Press, New York, 1976), chap. 10, pp. 373-410. Spin Labelling, Theory and Applications, ed. L. J. Berliner (Academic Press, New York, 1976). J. H. Freed, in Spin Labelling, Theory and Applications, ed. L. J. Berliner (Academic Press, New York, 1976), chap. 3, pp. 53-132. A. J. Dammers, Ph.D. Thesis (University of Utrecht, 1985). G. Moro and J. H. Freed, J. Chem. Phys., 1981, 74, 3757. A. J. Dammers, Y. K. Levine and J. A. Tjon, Chem. Phys. Lett., 1982, 88, 198. E. Meirovitch, A. Nayeem and J. H. Freed, J. Phys. Chem., 1984, 88, 3454. P. Koole, A. J. Dammers, G. van Ginkel and Y. K. Levine, Biochim. Biophys. Acta, 1984,777, 297. M. Zuker, A. G. Szabo, L. Bramall, D. T. Krajcarski and B. Selinger, Rev. Sci. Instrum., 1985, 56, 14. R. A. Klein, Biochim. Biophys. Acta, 1970, 210, 486. M. Vos, R. P. H. Kooyman and Y. K. Levine, Biochem. Biophys. Res. Commun., 1983, 116, 462. L. B-A. Johansson, Chem. Phys. Lett., 1985, 118, 516. P. C. Alford and T. F. Palmer, Chem. Phys. Lett., 1982, 86, 248. B. E. Kohler and T. A. Spiglamin, J. Chem. Phys., 1984, 80, 5465. A. C. Albrecht, J. Chem. Phys., 1960, 33, 156. 29. D. Chapman and G. Benga, Biological Membranes 5 (Academic Press, London, 1984), chap. 1, pp. 1-56. 30 R. Ghosh and J. Seelig, Biochim. Biophys. Acta, 1982, 691, 151. Received 5 th December, 1985