Calorimetric studies of dilute aqueous suspensions of bilayers formed from synthetic L- -lecithins.

Abstract The gel-liquid crystal transitions in lipid bilayers formed from synthetic dimyristoyl, dipalmitoyl and distearoyl l-α-lecithins have been studied by high sensitivity differential scanning calorimetry in dilute aqueous suspensions ranging in concentration from 0.4 to 6.6 mg ml-1. Each lipid shows two endothermic transitions, an extremely sharp main transition and a broader transition accompanied by a smaller heat absorption at a temperature 5–10° below the main transition. The enthalpy increases in the main transition are respectively 6.3, 9.7, and 10.8 kcal per mole of monomeric lipid. The main transition widths, which are not very reproducible, reach a minimum in the case of dimyristoyl lecithin of as little as 0.2° for 10% to 90% conversion, suggesting that these transitions would be truly isothermal with completely pure lipids. The apparent heat capacities of the lipids in the liquid crystal state are, with the possible exception of dipalmitoyl lecithin, no more than 5 cal deg-1 (mole of lipid)-1 larger than in the gel state, indicating that the hydrocarbon chains have much less mobility in the liquid crystalline state than in the corresponding liquid normal paraffins. This conclusion was also reached by Phillips et al. (Phillips, M. C., Williams, R. M., and Chapman, D. (1969) Chem Phys. Lipids 3, 234) on the basis of entropy comparisons.

FUM the Departments of Chemistry and Molecular Biophysics and Biochemistry, Yale CTrLivcrsity, lj-etc Hovel?, Cowleclicut 06520 SUMMARY The gel-liquid crystal transitions in lipid bilayers formed from synthetic dimyristoyl, dipalmitoyl and distearoyl L-W lecithins have been studied by high sensitivity differential scanning calorimetry in dilute aqueous suspensions ranging in concentration from 0.4 to 6.6 mg ml-r. Each lipid shows two endothermic transitions, an extremely sharp main transition and a broader transition accompanied by a smaller heat absorption at a temperature 5-10" below the main transition. The enthalpy increases in the main transition are respectively 6.3, 9.7, and 10.8 kcal per mole of monomeric lipid. The main transition widths, which are not very reproducible, reach a minimum in the case of dimyristoyl lecithin of as little as 0.2' for 10% to 90% conversion, suggesting that these transitions would be truly isothermal with completely pure lipids.
The apparent heat capacities of the lipids in the liquid crystal state are, with the possible exception of dipalmitoyl lecithin, no more than 5 cal degi (mole of lipid)-l larger than in the gel state, indicating that the hydrocarbon chains have much less mobility in the liquid crystalline state than in the corresponding liquid normal paraffins. Current interest in phospholipids stems largely from their widespread occurrence in biological membranes.
The view that lipid bilayers may be an especially important feature of membrane structure dates from 1935 and is still widely held, although much current evidence shows that the unit membrane model involving lipid bilayers originally proposed by Danielli Many reports have been published of studies of phospholipid bilayers in the presence and absence of water, using a wide variety of techniques.
Most of these techniques, when applied to lipid-water mixtures, require high concentrations of lipids.
The technique which at present appears best adapted to working with low concentrations (below 1 yc by weight) is differential scanning calorimetry.
We have applied this method to dilute aqueous suspensions of three L-a-lecithins, dimyristoyl, dipalmitoyl, and distearoyl.
In a previous communication (5) we have given the results of a calorimetric study of aqueous suspensions of mixtures of cholesterol with dimyristoyl lecithin and dipalmitoyl lecithin.

MATERIALS AND METHODS
Lipids were purchased from Analabs, North Haven, Connecticut, Serdary Laboratories, London, Ontario, and Calbiochem, San Diego, California.
Considerable differences, particularly with respect to the sharpness of transitions, were observed with different preparations, to the extent of having no observable transition in one or two cases. All the results reported here were obtained with dimyristoyl lecithin, Lot 100822, dipalmitoyl lecithin, Lot 100653, and distearoyl lecithin, Lot 100296, all obtained from Calbiochem.
These preparations appeared homogeneous in silica gel thin layer chromatography (using chloroform-methanol-water in volume ratios 60 : 30 : 5). Suspensions were in most instances prepared by heating the solid lipid in water to a temperature well above its transition temperature and then shaking the mixture vigorously for 10 to I5 min in a Super-mixer (Lab-Line Instruments, Melrose Park, Illinois).
The suspension was then reheated above its transition temperature and cooled rapidly in ice water. The pH of the suspensions was always within the range 6.5 to 7.2, and underwent no significant change as a result of two or more successive heatings and toolings in the ca1orimeter.r In some experiments lipid concentrations were estimated by phosphate determinations (6) on the suspensions. An equally valid measure of concentration appeared to be afforded by the weight makeup of the suspension, and this more convenient measure was used in many of the experiments.
The lipid preparations were shown to be anhydrous by the fact that 1 Note Added in Proof-It has been found that prolonged sonication of the lipid suspensions leads to disappearance of the transitions discussed in this paper. This matter will be more fully considered in a future publication. The dashed curve is drawn to give a symmetrical specific heat curve. they suffered no significant loss in weight after several hours of heating at 80" in a vacuum (10 to 15 mm Hg).
The differential scanning calorimeter used in these esperiments has been previously described (7,8). Recent modifications of the equipment include provision for digital recording of all pertinent data on punched paper tape for convenient computer work-up.
The calorimeter holds 1.94 ml of suspension, and can detect heat absorption, in excess of that required for the usual heating rate of 18" per hour, amounting to as little as 2 x low5 cal per deg of temperature rise. In a few esperiments a heating rate of 4.5" per hour was employed.

RESULTS
The results of a typical experiment with dipalmitoyl lecithin are illustrated in Fig. 1. Curve A is a plot of the excess specific heat of the lipid suspension as a function of temperature and Curve B is the corresponding integral curve. Both curves are plotted from data recorded digitally at 1-min intervals, the experimental points being shown only for Curve B. A small heat absorption centered at 33.5" precedes the main transition at 41.5". The existence of these two transitions complicates the evaluation of the corresponding enthalpy changes since the excess specific heat does not go to zero between them. In low sensitivity scanning calorimetry, where the base line tends in any case to be rather irregular, the usual procedure adopted is to draw a line such as dashed line b in Fig. 1, and to evaluate the integral above this line. In the present case this gives AH2 = 8.56 kcal mole-1 for the upper transition, in good agreement with the value 8.66 reported by Phillips et al. (9).
An alternative interpretation is suggested by the fact that the excess specific heat is very nearly zero below the first transition and above the second transition.
If the excess specific heat is plotted on a larger scale as in Fig. 2  G Numbers in parentheses refer to Reference 9.
baseline problem is less prominent than in that of dipalmit,oyl lecithin, in the former case because the lower aud upper transitions are more widely separated in temperature and there is a close approach to zero excess specific heat between them, and in the latter case because the transitions are close together so that little enthalpy is absorbed in the region between them. This accounts for the fact that our mean values of AI-T, for dimyristoyl lecithin and distearoyl lecithin agree well with those of Phillips et al. (9) while our value for dipalmitoyl lecithin does not. Although in our earlier report (5) on the effect of added cholesterol on lipid transitions we used the first method considered above for evaluating enthalpies, and thus arrived at a considerably smaller value for the transition enthalpy of dipalmitoyl lecithin than given in the present paper, the relative effects produced by cholesterol are not significantly in error.
Some minor anomalies were observed in some of the experiments. For example it was found that a suspension of distearoyl lecithin freshly prepared and rapidly cooled as described above showed a small, sharp heat absorption between the first and second transitions, as shown in Fig. 3. This phenomenon was absent if the material was heated after having been slowly cooled from above its higher transition temperature.
The samples in the calorimeter were in nearly all cases cooled in situ during a period of 1 to 8 hours and reheated.
The behavior on reheating was always similar to that observed in the first heating except that the transitions were usually somewhat broadened, though without any significant change in total heat absorption. Table I summarizes the results obtained for dilute aqueous suspensions of dimyristoyl lecithin, dipalmitoyl lecithin and distearoyl lechithin.
Data are given for both the lower and upper transitions.
The error estimates shown in the table are standard errors of the mean, indicating primarily the calorimetric reproducibility, and are thus minimum estimates of the uncertainties in the data, particularly those pertaining to the lower transitions.
!I',,,1 and T,z are the temperatures at which the enthalpy changes are half completed.
The evaluation of AH1 and AHz from the experimental data was outlined above; they are expressed in kcal per mole of monomeric lipid.
The sizes for the apparent cooperative units2 given in columns 6 and 9 are simply the ratios of the enthalpies derived from the van't Hoff equation to the corresponding calorimetric enthalpies. If heat absorption is assumed to be a linear measure of the extent of the * The significance of the term cooperative unit can be expressed as follows. The temperature course of the transition is approximately that to be expected for an assemblage of independent units of the indicated size each one of which shows a strictly two state. or all-or-none, transition. The small peak at 52.8" occurred only with suspensions which had been rapidly cooled from above the main transition temperature.
at zero specific heat. The dashed line in Fig. 2 is drawn to give a symmetrical curve for the lower transition, which turns out to have approximately the shape expected for a simple two state transition.
In this interpretation it is appropriate to assign to the lower transition an enthalpy equal to twice that absorbed up to the midpoint of this transition, and the remaining enthalpy to the upper transition. This is the procedure employed to obtain Aq, and Aq, as indicated in Fig. 1, and gives AHz = 9.47 kcal mole+.
The data reported here for all three lipids were obtained by this method of analyzing the excess enthalpy curves. In the case of dimyristoyl lecithin and distearoyl lecithin the transition under observation, the van't Hoff expression for a two state process in which no dissociation takes place can be put in the form dol (6) AIJ,H =-dT T-T", 4 RT,2 (1) where o( is the fractional completion of the enthalpy absorption. In point of fact some of the transition curves deviated significantly from the symmetrical form expected for a two state transition with no change in heat capacity, so that application of Equation 1 is only approximately valid.

DISCUSSION
As mentioned above, the values for TInz and AH2 given in Table  I agree moderately well with those reported by Phillips et al. (9). The largest discrepancy, in the value of AH2 for dipalmitoyl lecithin, presumably results from the base line problem frequently encountered in scanning calorimetry.
The agreement between these independent sets of data is particularly interesting in that the suspensions used in the work quoted by Phillips et al. (9) were as much as four orders of magnitude more concentrated than our most dilute suspensions, and were thermally scanned at rates an order of magnitude or more higher than we employed.
AII outstanding feature of these transitions is their sharpness. The transition width was to some extent dependent on the previous history of a particular sample, and varied considerably with different preparations of the same lipid, SO that the transition widths indicated by the cooperative unit* sizes given in Table  I cannot be considered as quantitatively determined characteristic properties of the lipids.
In some experiments dimyristoyl lecithin was observed to melt (10% to 907,) within about 0.2", although the usual transition width was larger than this. Our observed main transition widths are much smaller than those previously reported on the basis of scanning calorimetry (10) and of observations of changes in the fluorescence of a probe incorporated in the bilayer (11). Even the lower transitions, which are broader than the main transitions, turn out to be characterized by the same high degree of cooperativity as the main transitions, if it is assumed that the lower transition also involves all of the molecules present.
It was shown by experiments at one-fourth the usual scanning rate that the transition of dimyristoyl lecithin is not significantly broadened by calorimetric lags. However, since any impurities would broaden the transitions, it seems likely that the gel to liquid crystal transition in pure lipid bilayers is truly isothermal.
The lower transition, or pretransition peak, was observed by Chapmann et al. (10) and was suggested by Ladbrooke and Chapman (12) to be due to a rotation of the polar head portion of lipid molecules.
As noted by Ladbrooke and Chapman, the temperature separation between the pretransition and the main transition decreases with increasing chain length, extrapolating to zero at dibehenoyl L-a-lecithin with chains 22 carbons long.
Precise heat capacity data (13) for the normal paraffins up to Cl8 show that those having an odd number of carbon atoms greater than seven undergo a phase transition between orthorhombic and hexagonal (14) crystal forms a few degrees below the melting point.
The heat capacities of these compounds in the hexagonal form are higher than in either the or-thorhombic or liquid states, and increase rapidly with temperature. Infrared data (15) and other evidence have led to the view (14) that the paraffin chains, in fully extended form, undergo cooperative rotation about their long axes in the hexagonal crystal. In view of the fact that the enthalpies of these transitions are similar in magnitude to those of the pretransitions observed with phospholipids, the question arises as to whether the pretransitions may not also lead to cooperative rotation of the hydrocarbon chains. If this is indeed the case, and if the intermediate state has a relatively high heat capacity as in the case of the normal paraffins, then the procedure we have employed in evaluating AHi and AH* is incorrect, and the values of AH2 given in Table I are too high by approximately 2, 7, and 4% for dimyristoyl lecithin, dipalmitoyl lecithin and distearoyl lecithin, respectively. Phillips et al. (9) found a linear relation between the length of the hydrocarbon chains in the lecithins dimyristoyl lecithin through dibehenoyl lecithin and the enthalpy of the main transition, whereas our data suggest a nonlinear relation. Consideration of the base-line problem, and of the fact that the transition enthalpy for dibehenoyl lecithin presumably includes an unknown contribution from the pretransition, may alter somewhat the enthalpy-chain-length relation given by Phillips et al. (9).
In nearly all our experiments the excess specific heat returned very closely to zero above the main transition.
In a few experiments with dipalmitoyl lecithin there was a slight continuing apparent absorption of heat above the transition amounting to 10 to 20 cal deg-i (mole of lipid)-', or 5 to 10 cal deg-* (mole of hydrocarbon chain)+. This is illustrated in Fig. 1. In other experiments the excess heat capacity above the transition amounted to no more than a fifth of this amount.
The increase in heat capacity on melting a normal alkane, on the other hand, is of the order of 17 cal deggi mole-1 (13). It is thus evident that many fewer degrees of freedom become excitable as a result of the lipid transition than in the melting of an alkane, and that the hydrocarbon chains in the lipid liquid crystal have considerably less mobility than in a liquid alkane. This conclusion is supported by the corresponding entropy data, as shown by Phillips et al. (9).
Transfer of lipid molecules from the bilayer to the aqueous phase would be expected to be accompanied by a large increase in apparent specific heat because of the structuring of water around the released hydrocarbon chains. An estimate of this effect is difficult to make because of lack of data for appropriate model systems, but it seems safe to conclude that the small values for AC, actually observed in the transitions mean that no more than 1 '%, and probably considerably less than this, of the hydrocarbon chains can leave the bilayer during the transition. 1.