J. CHEM. SOC. FARADAY TRANS., 1994, 90(2), 305-309 Effect of Glycerol on the Translational and Rotational Motions in Lipid Bilayers studied by Pulsed-field Gradient 'H NMR, EPR and Time-resolved Fluorescence Spectroscopy Greger Oradd, Goran Wikander, Goran Lindblom and Lennart B-A. Johansson" Department of Physical Chemistry, University of Umed, S-90187 Umed, Sweden Glycerol can replace water in both lipid vesicles and lyotropic liquid-crystalline phases. 1,2-Dioleoylsn- glycero(3)phosphocholine (DOPC) forms a lamellar (La)liquid-crystalline phase in arbitrary mixtures of glycerol and water (Biochim. Biophys. Acta, 1993,1149,285.). Monoolein (MO) forms La and also cubic liquidcrystalline phases in glycerol-water mixtures. The present study is focussed on characterizing the influence of glycerol on the molecular dynamics in the lipid bilayer.By EPR and time-resolved fluorescence spectroscopy we measure the rotat i onaI mo bi Ii ty of spin-l abe1led fatty acids [243carboxypropyl)-4,4-d i methy 1-24ridec y Ioxazol id i n-3-y lox y l (5-DS) and 2-(14carboxytetradecyl)-2-ethyl-4,4-dimethyloxazolidin-3-yloxyl (16-DS)] and a hydrophobic fluoro- phore, 2,5,8,1l-tetra-tert-butylperylene(TBPe), respectively. The translational diffusion of MO in the cubic phase is obtained by pulsed-field gradient 'H FT NMR experiments. The rotational rate of 16-DS and TBPe decreases continuously with increasing glycerol concentration, being a factor of 2-3 lower at 100% glycerol. A continuous decrease in the lipid translational diffusion coefficient, D, is also found with increasing glycerol content, so that D = 12.6 x 10-l2 m2 s-' at 0% and D = 1.9 x 10-l2 m2 s-' at 100% glycerol. The effects of glycerol on both the translational diffusion of the lipid in the bilayers and the rotational dynamics of the probe molecules residing in the interior of the hydrophobic regions are ascribed to changes of the viscosity in the interbilayer regions.It has been shown that lipid bilayers form not only with water, but also in other polar solvents such as glycerol, for- mamide and ethylene glycol. 1-5 In biological systems, gly- cerol is often used as a cryoprotectant and it substitutes very well for water.6-' Recently, we have found that 1,2-dioleoyl- sn-glycero(3)phosphocholine(DOPC) forms a lamellar liquid- crystalline phase (La) in arbitrary mixtures of glycerol and water.5 The phase was characterized by means of X-ray dif- fraction, 31P NMR spectroscopy and differential scanning calorimetry (DSC).Moreover, it was found that unilamellar vesicles could be prepared from a diluted suspension of the lamellar phase, as was shown by means of 31PNMR, EPR and fluorescence spectro~copy.~The last two methods revealed a strong dependence of the rotational relaxation rates of fluorescent and spin probes with glycerol concentra- tion, although these probes resided in the interior of the lipid bilayer. In the present paper we have investigated the molecular dynamics occurring in the La phases of DOPC and mono- olein (MO) and in the cubic liquid-crystalline phase of MO, which is built up of bilayer aggregate units.6 Cubic phases are convenient systems for studies of the motion of lipids and probes, since such phases are optically isotropic, which may simplify the interpretations of fluorescence depolarization and EPR experimental data.Furthermore, the conventional 'H NMR diffusion technique can be applied straightfor- wardly on cubic phases.' The rotational relaxation and molecular ordering of spin labels and fluorophores localized in the lipid matrix have been determined, as well as the lipid translational diffusion coefficients. Experimental DOPC was purchased from Avanti Polar Lipids (USA). The purity of the lipid was better than 99%, as checked by thin- layer chromatography at our laboratory.MO was purchased from Sigma (Missouri, USA) and was used without further purification, and glycerol was purchased from BDH (Canada). 5-DS[S-doxylstearic acid, 2-(3-~arboxypropy1)-4,4-dimethyl-2-tridecyloxazolidin-3-yloxyl]and 16-DS [16-doxyl-stearic acid, 2-(14-carboxytetradecyl)-2-ethyl-4,4-dimethyl-oxazolidin-3-yloxyl] (see Fig. 1) were purchased from Molec- ular Probes Inc. (USA) and were used directly without any further purification. 2,5,8,1l-Tetra-tert-butylperylene(TBPe, see Fig. 1) was synthesized by Friedel-Crafts alkylation of perylene.' Vesicles for EPR experiments were prepared by sonication according to the following procedure. Appropriate amounts of the dry lipid powder were dissolved in a mixture of chloroform-methanol (2 :1 v/v).A suitable amount of the spin label, dissolved in a chloroform-methanol mixture (2 :1 v/v), was added to the former solution. The label :lipid molar ratio for the vesicle samples was 1: 100 in order to prevent spin-spin interactions. The solvent was evaporated and dried at 320 K and 0.1 Torr for at least 2 h. Afterwards, 3 ml of a glycerol-H,O mixture were added and the suspension was sonicated eight times for 5 min each time. During the sonica- tion the sample was cooled to ca. 283 K. The sonicator was a Soniprep 150 (MSE Scientific Instruments, England) supple- mented with an exponential microprobe. The level of the amplitude used was 10-14 pm. Between the different sonica- tion steps the lipid-glycerol-water mixture was frozen with liquid nitrogen and thawed several times in order to obtain as large aggregates as possible.Glass capillary tubes were so-s1 5-DS,n= 3,m=12 16-DS,n=14,m= 1 Fig. 1 Structures of 5-DS, 16-DS and TBPe. The electronic dipoles of the So++ S, transitions are polarized in the plane of the molecule, as indicated by the arrows. filled with the optically clear solution and thereafter flame- sealed. EPR spectra of the vesicle solutions were recorded at 274 or 298 K. In the preparation of cubic phases of MO for EPR experi- ments, a suitable amount of the spin label dissolved in a chloroform-methanol mixture was transferred to glass vials. The solvent was pumped off under an N, atmosphere.Appropriate amounts of the amphiphile and a glycerol-water mixture were added to the thoroughly dried label. The ratio between MO and glycerol-H,O was always 70: 30 (wt.%). Thereafter, the glass vials were sealed. The samples were thoroughly mixed by repeated centrifuging at 313 K for several days. They were stored in darkness for equilibration for some days. Small amounts of the samples were sucked into glass capillary tubes, sealed and thereafter measured on the EPR spectrometer. The label :amphiphile ratio was always kept at 1 :600 on a molar basis for the MO samples. The samples were run in the temperature range 298-328 K with a temperature increment of 5 K between each point. Temperature regulation (within k0.5 K) was achieved by means of a Bruker ER 41 11 VT variable-temperature regula- tor.Each sample was checked between crossed polarizers before and after every temperature variation experiment. The samples never displayed any birefringence. This indicated that an optically isotropic phase remained during the variable-temperature experiment. The EPR spectra were recorded with a Bruker model ESP 300E X-band spec- trometer (9 GHz). The modulation frequency was 100 kHz and the modulation amplitude was always less than 0.5 times the linewidth of the central peak in each spectrum. For the NMR experiments the samples were made by weighing appropriate amounts of MO, glycerol and D20 into 7 mm glass tubes which were then flame-sealed. The samples were mixed by centrifugation and left for some days to equilibrate. The glass tubes were then opened and the samples transferred to 5 mm NMR tubes.The self-diffusion coefficient was measured with the Fourier-transform pulsed magnetic field gradient spin-echo technique.' The pulse sequence (RWn/2-z-n-z), was utilized in the diffusion experiment, where RD is the waiting time between repetitive scans, and N is the number of scans collected in each experi- ment. The magnetic field gradient pulses of width 6 and strength g are separated by a time interval A and placed at each side of the n pulse. In order to obtain a steady-state condition of residual gradient tails, three dummy gradient pulses were separated by A, and applied before each pulse sequence.l4 Two dummy scans were rejected at the beginning of each experiment.The attenuation of the signal measured through the peak height is described by eqn. (1) and the fitting of the experimental data to this equation was made on a personal computer by using the program Sigma Plot for curve fitting. A(t) = A(O)exp[ -y2g2a2A(A -6/3)]exp[ -2z/T] (1) The diffusion experiments were performed on a Bruker ACP-250 spectrometer equipped with an HR-50 high-resolution VT diffusion probe for 5 mm samples (Cryomagnet Systems Inc., Indianapolis). The magnetic field gradient pulses were generated with a home built gradient unit driven by a Kenwood PD35-20D power supply. The temperature was kept constant to within 1 K by a heated air stream around the sample and was measured by means of a thermocouple placed close to the sample.The lipid diffusion was monitored through the peak height of the chain terminal methyl groups. Several experiments were made with different (constant) values of RD, z, A and g while 6 was varied. Typical settings on these parameters for the investigation were, RD = 2 s, T = 200-400 ms, A = 200-400 ms, g = 0.12-J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 0.5 T m-I. The observed diffusion coefficients did not depend on a variation of these parameters. The steady-state fluorescence spectra and anisotropies were obtained using a Spex Fluorolog 112 instrument (Spex Ind., New Jersey), equipped with Glan-Thompson polarizers. The spectral bandwidths were 5.6 and 2.7 nm for the excitation and emission monochromators, respectively.The fluorescence spectra were corrected. The fluorimeter was calibrated by using a standard lamp from the Swedish National Testing and Research Institute, Borgs, Sweden. A PRA 3000 system (Photophysical Research Ass. Inc., Canada) was used for single-photon-counting measurements of the fluorescence decay. The excitation source is a thyratron-gated flash lamp (Model 510 C, PRA) filled with deuterium gas and operated at ca. 30 kHz. The excitation wavelengths were selected by interference filters (Omega/ Saven AB, Sweden) centred at 409.4 nm [half band width (HBW) = 13.0 nm]. The fluorescence emission was observed above 470 nm through a long-pass filter (Schott KV 470, Schott). The maximum absorbance of all samples was kept below 0.08 which corresponds to a total concentration of less than mol dm-3.The time-resolved polarized fluorescence decay curves were measured by repeated collection of photons during 200 s, for each setting of the polarizers. The emission polarizer was fixed and the excitation polarizer rotated periodically. In each experiment the decay curves Fll(t)and F,(t) were col- lected. The subscripts 11 and Irefer to an orientation of the emission polarizer parallel and perpendicular with respect to the excitation polarizer. From these a sum curve S(t) = FIl(t) + 2GF,(t) (2) and a difference curve 4)= FllW -GF,(t) (3) were calculated. The correction factor, G, was obtained by normalizing the total number of counts, FII and F, ,collected in Fl,(t)and F,(t), respectively, to the steady-state anisotropy, Is, as G = (1 -rJ(1 + 2rs)-1F,l(F,)-' (4) The data were analysed with a MINC-11/03 computer using the deconvolution software (DECAY V3.0 a, ATROPY, V1.0) developed by PRA.Linear dichroism (LD) spectra were recorded on a Jasco 5-720 supplemented with an Oxley device and the absorption spectra on a Cary 119 spectrophotometer. Details of studying macroscopically aligned lamellar liquid crystals and the inter- pretation of data are given elsewhere." Results DOPC vesicles EPR spectra of 5-DS or 16-DS solubilized in DOPC vesicles have been recorded at different ratios of glycerol and water. The lineshapes of 5-and 16-DS in DOPC vesicles at various glycerol-water mixtures, are typical for reorientational rates in the time domain of intermediate and slow motions.Under these conditions it is not possible to calculate unambiguously the rotational correlation time from the EPR spectrum. Instead, computer simulations must be performed. Such an analysis gives values of the order parameter, S, and the rota- tional correlation time, z,. The N-0 label is assumed to undergo Brownian reorientations with the rotational diffu- sion constants of RII and R, parallel and perpendicular to the N-0 bond, respectively. For this case the correlation time, z, is defined by z, = 6-' (RIIR,)-'j2. Details concerning these J. CHEM. SOC. FARADAY TRANS., 1994, VOL.90 Fig. 2 Experimental (-) and simulated (. .) EPR spectra of 5-DS solubilized in vesicles of DOPC at different mixtures of glycerol and water. Composition of the solvent mixture (percentage glycerol) : (a) 0, (b)50, (c) 70, (d) 90. 298 K, total scan range, 8 mT. theoretical models are given elsewhere.16 Fig. 2 shows the simulated (dotted lines) and experimental (solid lines) spectra of 5-DS in DOPC vesicles. Table 1 displays the best fit values of S and z,. Vesicles containing the hydrophobic fluorophore TBPe were studied by using steady-state and time-resolved fluores- cence spectroscopy. The rotational motions of TBPe in DOPC vesicles formed in water and in a glycerol-water mixture (91 : 9, wt.%) were compared. The values of the steady-state fluorescence anisotropies are 0.21 and 0.32 in the water and the solvent mixture, respectively.Since the fluores- cence lifetimes of TBPe in these systems are very similar (about 4.8 ns), the different values of the fluorescence aniso- tropy suggest that the rotational rates are considerably slower in the DOPC vesicles formed in the glycerol-water mixture. In these studies (as well as in the EPR experiments) note that the influence of rotational motion of the vesicles can safely be neglected. A direct proof for slower rotational motions, in the presence of glycerol, is obtained from the time-resolved fluorescence anisotropy. For both kinds of system the decay of the fluorescence anisotropy is biexponen- tial with the rotational correlation times about four times 307 smaller in the vesicles suspended in water, than in those in the glycerol-water mixture.On a nanosecond timescale, the actual anisotropy decays at 274 K are given by r(t) = 0.11 exp(-t/3.5) + 0.21 exp( -t/25) and r(t) = 0.10 exp(-t/ 10.8) + 0.27 exp( -t/124). Cubic Phases EPR experiments reveal that the rotational motions of the spin label (16-DS) occur in the intermediate time domain. Table 1 displays z, obtained from lineshape simulations for cubic phases of monoolein at different glycerol : water ratios. The correlation time can be interpreted as the reciprocal rate of rotational relaxation of the label, having a temperature dependence given by z, = A' exp(-EJRT) In eqn. (5), E, is the activation energy for the reorientation process, A' is a pre-exponential factor (of dimension frequency-'), R is the universal gas constant and T is the temperature.From eqn. (5) an estimate of the activation energy of reorientation of the spin label is obtained. The acti- vation energy is about 35 kJ mol-and changes only slightly with glycerol content (Table 1). The hydrophobic fluorophore, TBPe, was solubilized in the cubic phase at the same ratios of glycerol and water as were used in the EPR experiments. The fluorescence relaxation of TBPe is monoexponential (with the lifetime zfgiven in Table 2) as is expected, if all TBPe molecules are homogeneously distributed in the lipid bilayer. We find a small, but signifi- cant decrease of zf with increasing glycerol content, which means that the physico-chemical properties change in the vicinity of the TBPe molecules.The zfvalues observed in the absence of glycerol are very similar to those found in other liquid-crystalline phases composed of non-ionic or ionic detergents in water.'* From the time-resolved and steady- state fluorescence anisotropies we obtain the (second-rank) rotational correlation function of TBPe. This function is fitted to exponential functions, as was done for the vesicle systems. We find that the rotational correlation function does not fit to a single-exponential function, but does fit very well to a biexponential function. The two correlation times ob- tained are summarized in the Table 3. One of the correlation Table 1 crystals MO (b)in different mixtures of glycerol and water' (a) Vesicles of DOPC in mixtures of glycerol and water Results from lineshape analysis of EPR spectra obtained for 5-DS and 16-DS solubilized in vesicles of DOPC (a) and cubic liquid glycerol (%) glycerol (Oh) glycerol (%) content ?,Insb content t,/nsb content S r,/ns 16-DS at 298 K 16-DS at 274 K 5-DS at 298 K 90 1.32 90 2.56 90 0.44 4.18 70 1.11 70 2.08 70 0.42 2.41 50 1.04 50 1.85 50 0.36 1.60 30 0.98 30 1.75 0 0.34 1.39 10 0.93 10 1.67 0 0.83 0 1.59 ~ (b)Cubic phases of MO (70%)in glycerol and water (1 6-DS at 298 K) glycerol : water temperature ratio rr/nsb EJcJ mol-' range/K 1 :o 1.15 34.3 298-328 2:l 1.04 34.1 298-328 1 :2 0.88 37.8 298-328 0:l 0.56 35.5 298-328 The parameters S, 7, and E, denote the order parameter, rotational correlation time and the activation energy of rotation, respectively.The order parameter, S = (P,(cos B)), describes the average orientation (B) of the long axis of the probe molecule with respect to the local director of the bilayer. The errors in S and 7,are estimated to be within 5%. S = 0. Table 2 Fluorescence lifetime (zc)of TBPe in cubic phases of MO at various mixtures (given as molar ratios) of glycerol with water glycerol : water ratio %Ins xz 0: 1 4.7 1.15 2:1 4.7 1.05 1 :2 4.8 0.97 l:o 5.0 1.17 The MO content is kept constant at 70 wt.%. The parameter xz is a statistical test of the curve fit to experimental data.The accuracy of q is within kO.05 ns. times (41)is very long compared with tf. Consequently, cannot be determined with any accuracy and is therefore con- sidered to be equal to infinity on the timescale of fluores- cence. The second rotational correlation time, 42, increases by a factor of ca. 3, upon increasing the glycerol content from 0 to 100 mol% in the cubic phases. The second-rank order parameter (S) of TBPe was obtained from linear dichroism experiments, performed on a macroscopically aligned lamellar phase of MO and water. These lamellar phases contain glycerol and water of the same molar ratios, as were used for the cubic phases. The order parameters, shown in Table 3, are very small, which strongly indicates a nearly isotropic orientation of the TBPe mol- ecules with respect to the director of the lipid bilayers.More- over, these values are very similar to those found in lamellar phases composed of DOPC and other detergent^.'^.' Taken together, these data are compatible with a localization of TBPe in the hydrophobic interior of the monoolein aggre- gates, where their molecular tumbling is considerably ham- pered by increasing the glycerol content outside the lipid bilayer. The S values change sign, implying that the orienta- tional distribution of TBPe changes with the glycerol concen- tration. Hence, the lipid orientation and/or lipid packing are influenced by the glycerol content. The self-diffusion coefficient, D,,, of MO was determined from pulsed field gradient ‘H NMR spin-echo measurements. The values of D,, obtained at the different glycerol concen- trations and temperatures are summarized in Fig.3. The lipid translational diffusion increases by a factor of six as the gly- cerol content is decreased from 100 to 0% in the cubic phase built up of MO and a mixture of glycerol and water. From the temperature dependence of D,, the activation energy of the translational motion can be estimated. The activation J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 -23.5 -24.0 -24.5 -25.0 a C--25.5 -26.0 -26.5 -27.0 0.0030 0.0031 0.0032 0.0033 0.0034 K/T Fig. 3 Natural logarithms of the translational diffusion coefficient (DmJ of MO as a function of inverse temperature in cubic phases with different molar ratios of glycerol and water.The activation ener- gies (E&J mol-’) of translational motion are: (V) 2: 1 Gly :D,O, 38; (0) 1:2 Gly :D,O, 30.1:0 Gly :D,O, 41 ;(A)0 :1 Gly :D,O, 30; (0) Discussion In the cubic phase of MO, the dynamics of reporter mol- ecules (16-DS and TBPe) residing in the hydrophobic part of the lipid region is significantly reduced upon replacing water with glycerol. Similarly, a decrease in the translational diffu- sion of the lipid itself is observed with increasing glycerol content. The relative decreases in the rotational rates of 16-DS and the fluorescent label TBPe, at 0 and 100% glyc- erol in the cubic phase, are similar.The rotational correlation times of TBPe are slower than those for 16-NS, which is to be expected since the molecular volume of TBPe (ca. 500 A3) is much greater. energy decreases from about 40 kJ mol-’ to about 30 kJ In dispersions of unilamellar vesicles of DOPC, the rota- mol-upon decreasing the glycerol concentration from 100 tional motions of the probes are hampered with increasing to 0%. glycerol content. However, the relative decrease is slightly Table 3 Steady-state (T,) and time-resolved [r(t)] fluorescence anisotropy data of TBPe in the cubic phase of MO and various mixtures of glycerol with water ~~~ glycerol : water ratio rs a1 41 a2 42 xz S 0:l 0.256 0.08 cc 0.26 6.6 1.16 -0.02 2:l 0.214 0.06 00 0.28 6.4 1.51 -0.02 1.2 0.184 0.05 cc 0.27 4.8 1.45 0.02 l:o 0.126 0.08 co 0.14 2.3 1.20 0.04 The MO content is kept constant at 70 wt.%.The time-resolved fluorescence anisotropy was fitted to experimental data according to r(t) = a, exp(-t/4 + a, exp( -t/&) where #1+ tf,i.e. a, exp(-t/$l) is essentially constant on the timescale of the experiment. The parameter X’ is a statistical test of the curve fit to experimental data. Data were collected at 298 K. The order parameter (S) was calculated from linear dichroism and absorption spectra of TBPe in the L, phases of MO formed with different mixtures of glycerol and water. The MO content was kept constant at 90 wt.%. The L, phase was macroscopically aligned between quartz plates.The order parameter, S = (P,(cos /3)), describes the average orientation (/3) of the transition dipole moment of the probe molecule, with respect to the director of the bilayer. The errors in 4, and S are within 10 and 5%, respectively. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 larger for 5-DS than for 16-DS. The values of the order parameters of 5-DS (S = 0.4) and 16-DS (S = 0), estimated from the lineshape analysis, are compatible with 16-DS being located deeper in the interior of the lipid bilayer. Hence, the influence of glycerol on the dynamics decreases towards the bilayer interior. A comparison of the relative decrease in the dynamic parameters between 0 and 100% of glycerol in the cubic phase, shows that the lipid translational diffusion is much more reduced than the rotational motions of 16-DS and TBPe.Previously,” it has been concluded that inter- actions in the polar head groups of the lipids in the aggregate play a dominant role for lipid lateral diffusion. It seems rea- sonable, therefore, that the relative effect of glycerol on the dynamics should be larger on the lipid translation diffusion than on the rotational motions of probe molecules solubilized within the interior of the lipid bilayer. The most simple mechanistic explanation for the present observations is that the increase of the macroscopic viscosity, exerted by the addition of glycerol to the polar region of the liquid-crystalline phase, affects the dynamics of all molecules in the phase structure.The viscosity of pure glycerol is ca. 1500 times greater than that of water at 293 K. Thereby, the motional friction in the interface of the lipid aggregates increases and hampers both the acyl chain motions and the lipid translational diffusional motion. Note also that the vis- cosity of glycerol is more temperature dependent than that of water, which should influence the activation energy of the lipid translational diffusion. The present experimental results clearly demonstrate that glycerol influences the dynamics in the bilayer and, most likely, also the packing of the lipid molecules in the aggre- gates. In a previous st~dy,~ X-ray diffraction studies of DOPC show that the bilayer thickness decreases, while the polar head group area increases with increasing glycerol content in the L, phase.The order parameter of 5-DS in DOPC vesicles changes significantly with glycerol content, which also suggests an influence on lipid packing. Further- more, the order parameter of TBPe, solubilized in lamellar phases of MO, changes sign with increasing glycerol concen- tration. Taken together, these observations are all compatible with a change in the lipid packing caused by the replacement of water with glycerol. Thus, both the dynamics and the lipid packing are significantly influenced upon replacing water with glycerol in vesicles, and lamellar and cubic liquid crys- tals. We are grateful to Mrs. Eva Vikstrom for skilful technical assistance and to Mr. Stein-Tore Bogen for performing the light spectroscopic experiments.This work was supported by the Swedish Natural Research Council. References I R. V. McDaniel, T. J. McIntosh and S. A. Simon, Biochim. Biophys. Acta, 1983,731,97. 2 J. L. Green and C. A. Angell, J. Phys. Chem., 1988,93,2880. 3 M. A. El-Nokaly, L. D. Ford and S. E. Friberg, J. Colloid Inter- face Sci., 1981,84, 228. 4 B. A. Bergenstihl and P. Stenius, J. Phys. Chem., 1987,91,5944. 5 L. B-A. Johansson, B. Kalman, G. Wikander, A. Fransson, K. Fontell, B. Bergenstihl and G. Lindblom, Biochim. Biophys. Acta, 1993, 1149, 285. 6 G. Lindblom, K. Larsson, L. B-A. Johansson, K. Fontell and S. Forsen, J. Am. Chem. SOC., 1979, 101, 2204; G. Lindblom and L. Rilfors, Biochim. Biophys. Acta, 1989,998, 221.7 G. Lindblom and H. Wennerstrom, Biophys. Chem., 1977,6, 167. 8 R. J. Taylor, G. D. J. Adams, C. F. B. Boardman and R. G. Wallis, Cryobiology, 1974, 11,430. 9 P. Mazur, R. H. Miller and S. P. Leibo, J. Membrane Biof., 1974, 15, 137. 10 A. A. Newman, 1968, in Gfycerol,ed. A. A. Newman, CSC Press, Cleveland, p. 91. 11 B. Schobert, 2. Naturforsch., Teil C, 1979,34,699. 12 L. B-A. Johansson, J. G. Molotkovsky and L. D. Bergelson, J. Am. Chem. SOC., 1987,109,7374. 13 E. 0.Stejskal and J. E. Tanner, J. Chem Phys., 1965,42, 288. 14 S. J. Gibbs and C. S. Johnson Jr., J. Magn. Reson., 1991,93, 395. 15 B. Norden, I. Jonas and G. Lindblom, J. Phys. Chem., 1977,81, 2084; L. B-A. Johansson, A. Davidsson, G. Lindblom and B. Norden, J. Phys. Chem., 1978, 82, 2604; L. B-A. Johansson and A.Davidsson, J. Chem. SOC.,Faraday Trans. I, 1985,81, 1373. 16 D. J. Schneider and J. H. Freed, in Biological Magnetic Reson- ance: Spin Labelling Theory and Applications, 1989, ed. L. J. Ber-liner and J. Reuben Plenum Press, New York, vol. 8, ch. 1; G. Wikander P-0. Eriksson, E. E. Burnell and G. Lindblom, J. Phys. Chem., 1990, 94, 5964; Z. Liang, P-0.Westlund and G. Wikander J. Chem Phys., 1993,99, 7098. 17 B. Kalman, L. B-A. Johansson, M. Lindberg and S. Engstrom J. Phys. Chem., 1989, 93, 8371; B. Kalman, L. B-A. Johansson, J. Phys. Chem., 1992,%, 185. Paper 3/03708H; Received 29th June, 1993