Calorimetric properties of mixtures of ganglioside GM1 and dipalmitoylphosphatidylcholine.

High sensitivity differential scanning calorimetry has been employed to study the thermotropic behavior, over the range from 12-83”C, of dispersed mixtures of 1,2-dipalmitoyl-sn-glycero-3-phosphocholine (DPPC) with bovine brain ganglioside GM1 of high purity. No thermal phase transitions of pure GM1 alone in excess water could be detected, even at concentrations of 4 mg/ml? over the temperature range of these studies. The excess heat capacity curve for multilamellar dis- persions of pure DPPC was as previously reported, with a pretransition at 32.7”C and a main transition at 41.4”C. As GM1 was added in increasing amounts to DPPC, the curve remained qualitatively similar up to a GM1 mole fraction (xc) of 0.24, but the temperature of the pretransition gradually increased, from 32.7 to about 37”C, and that of the main transition gradually increased, from 41.4-44.3%. When xo was increased above 0.24, several changes were observed: 1) the tur- bidity of the suspensions abruptly decreased, becoming negligible at xo r, 0.40; 2) the pretransition disap- peared; 3) the amplitude of the main transition began to decrease; and 4) a new transition appeared at a higher temperature, 48°C. The calculated enthalpy of the main transition was constant, within the error of the measurement, up to xo = 0.24, and then decreased linearly to zero over the range of xo between 0.24 and 0.63; over the same range the enthalpy of the transition at 48°C increased linearly from zero, so that the sum of the two enthalpies remained constant. All the observed thermotropic transitions were experimentally reversi- ble over a period of at least 2 weeks. These results are interpreted in terms of a model, consistent with pre- vious observations, in which GM1 at concentrations up to xG = 0.24 is incorporated into the multilamellar DPPC structure, but at higher concentrations induces the progressive formation of mixed micelles because of the repulsive interaction of its bulky and negatively charged head groups. In both lamellar and micellar structures, GM1 appears to stabilize the phospholipid lattice, causing the transition temperature to be higher than for pure DPPC. The thermotropic behavior of biomembranes and their lipid components has come increasingly under investigation (1, 2). The calorimetric properties of the numerically predominant

High sensitivity differential scanning calorimetry has been employed to study the thermotropic behavior, over the range from 12-83"C, of dispersed mixtures of 1,2-dipalmitoyl-sn-glycero-3-phosphocholine (DPPC) with bovine brain ganglioside GM1 of high purity. No thermal phase transitions of pure GM1 alone in excess water could be detected, even at concentrations of 4 mg/ml? over the temperature range of these studies. The excess heat capacity curve for multilamellar dispersions of pure DPPC was as previously reported, with a pretransition at 32.7"C and a main transition at 41.4"C. As GM1 was added in increasing amounts to DPPC, the curve remained qualitatively similar up to a GM1 mole fraction (xc) of 0.24, but the temperature of the pretransition gradually increased, from 32.7 to about 37"C, and that of the main transition gradually increased, from 41.4-44.3%. When xo was increased above 0.24, several changes were observed: 1) the turbidity of the suspensions abruptly decreased, becoming negligible at xo r, 0.40; 2) the pretransition disappeared; 3) the amplitude of the main transition began to decrease; and 4) a new transition appeared at a higher temperature, 48°C. The calculated enthalpy of the main transition was constant, within the error of the measurement, up to xo = 0.24, and then decreased linearly to zero over the range of xo between 0.24 and 0.63; over the same range the enthalpy of the transition at 48°C increased linearly from zero, so that the sum of the two enthalpies remained constant. All the observed thermotropic transitions were experimentally reversible over a period of at least 2 weeks. These results are interpreted in terms of a model, consistent with previous observations, in which GM1 at concentrations up to xG = 0.24 is incorporated into the multilamellar DPPC structure, but at higher concentrations induces the progressive formation of mixed micelles because of the repulsive interaction of its bulky and negatively charged head groups. In both lamellar and micellar structures, GM1 appears to stabilize the phospholipid lattice, causing the transition temperature to be higher than for pure DPPC.  (4), are now reasonably well understood. Other lipid components, such as glycolipids, have so far received less attention.
The rich variety of carbohydrate arrangements among the glycolipids equips them well for possible roles in the recognition of molecules by cell membranes. The sialic acid-containing gangliosides (5, 6) are of particular interest, as putative receptors for bacterial toxins (7-lo), peptide hormones (ll-15), viruses (16-18), and neurotransmitters (19). Variations in the oligosaccharide portions of these glycosphingolipids, coupled with variations in their pattern of occurrence in various tissues, may confer upon those tissues specificity to differing classes of biological stimuli.
At present, we know little about how gangliosides or other receptors may communicate stimulus effects on the cell. In particular, we do not know very much about the effects of the gangliosides themselves on local membrane structure. It is of interest,, then, to examine the influence of gangliosides on the physical properties of model membranes. Study of the properties of ganglioside/phospholipid mixtures may also enhance our understanding of the effects of ganglioside accumulation in several gangliosidoses, such as Tay-Sachs disease and GM,' gangliosidosis, in which the lack of specific catabolic enzymes causes gangliosides to accumulate in neural tissues (20, 21). In the present investigation, therefore, we have examined the effects of ganglioside GMI, the cholera toxin receptor, isolated in high purity from bovine brain, on the thermotropic behavior of a well studied synthetic phospholipid system, multilamellar dispersions of dipalmitoyllecithin (I,2-dipalmitoyl-sn-glycero-3-phosphocholine, DPPC) in aqueous buffer. volumes of stock solutions corresponding to a fixed amount, 1 mg, of DPPC and the desired mole fraction of GMI were added to 5 ml pear shaped flasks, the solvents were removed with a stream of dry nitrogen, and the samples were stored in U~CUO overnight at 4O'C. The mixed lipids were resuspended by adding 1.5 ml of phosphate buffer (NaCl, 137 mM; KCl, 2.7 mM; Na2HP04, 6.9 mM; KH2POI, 1.5 mM; pH 7.34) prewarming to above 5O"C, and blending on a Vortex mixer at the same temperature for 1 min. The turbidity of all samples was judged visually.

MATERIALS
The samples were loaded into one cell of a Privalov differential scanning calorimeter, and buffer was loaded into the other cell. Runs were initiated at nominal heating rates of 0.5'C min-' on samples degassed in UQCUO. Scans were made over the temperature range from 12-83°C.
The scan rate, power calibration, and heater voltages of the calorimeter were monitored for each run. For each transition the temperature of the transition maximum, T,, and the width of the transition at half-maximum, P, were measured. The enthalpy AH for each transition was also determined by integrating the scans with a planimeter; for this purpose the base-line was estimated as indicated in Fig. 1. The apparent number of molecules in the cooperative unit, n, was calculated as 29 Z',"/(F AH).  (3), with a pretransition at 32.7"C and a main transition (Fig. 2) at 41.4"C. The enthalpy of the main transition ( Fig. 3)  = 0.30, the temperature of the original main transition remained constant or perhaps decreased slightly (Fig. 2).

Turbidity
Up to xo = 0.24, the calculated enthalpy of the pretransition remained constant, within the error of the measurement, at 3.01 f 0.38 kJ mol-'. The enthalpy of the main transition also remained constant over this range of XC, values (Fig. 3) scans were repeated at a slower heating rate, 0.25"C min-', identical results were obtained. As x. was raised from 0.00 to 0.24, the apparent size of the cooperative unit for the pretransition fell from about 190 molecules to about 135 molecules as the width of the pretransition increased from 3.3-66°C. For the main transition, the calculated cooperative unit size fell rapidly from 102 + 9 molecules to 39 f 3 molecules at xo = 0.07, and remained at this level (Fig. 4). The corresponding value for the higher temperature transition was rather similar to this (45 molecules) at xo = 0.30 to 0.50, rising to 100 or more at higher values of xo.

Absence of Transitions
for Pure GM1-We have been unable to detect any transitions in calorimetric scans of pure Ghll over the temperature range from 12-83°C at concentrations as high as 4 mg/ml. In an earlier calorimetric study of commercial ganglioside mixtures, broad thermotropic transitions at 15 and 39°C were observed (23). If transitions such as these had occurred in our system, they would have been readily detectable. We have not attempted to establish the cause of this apparent discrepancy, but the effects seen in the earlier study might have been due to 1) low water content (less than 85%); 2) possible effects of ganglioside heterogeneity; and, most likely, 3) contamination with other lipid molecules (24 interesting. Hill and Lester (27) prepared phospholipid/ganglioside mixtures by two different methods and compared the resulting structures by centrifugation and electron microscopy. Using the method of preparation employed in our study, they found that only multilamellar structures were formed at low ganglioside concentrations (up to about 30 mole %); between 65 and 80 mole % only cylindrical micelles were found; and above 80 mole % the micelles became spheroidal. Subsequent investigations have supported the view that at low concentrations, gangliosides (29) and other glycolipids (25,30) are incorporated into the phospholipid structure. Ganglioside incorporation occurs up to a ganglioside mole fraction of about 0.25 to 0.30, at which point a marked physical change occurs (27,29).
The structure of phospholipid/glycolipid mixtures is determined by the balance between the interaction of head groups (normally repulsive) and the hydrophobic interaction of tail groups (31,32), which in turn depends upon the mean surface/ volume ratio of the molecules. The available structures include bilayers extended in two dimensions as lamellae (L), in one dimension as cylindrical micelles (C), or in zero dimensions as spherical micelles (S). For a hydrocarbon region of constant thickness, a convenient first order approximation, the surface/volume ratios of L, C, and S structures are in the proportions L:C:S = 1:2:3. If the mean head group area at the site of maximum repulsive interaction is low, an L structure is formed; for increasing head-group area, C and then S structures are required. Mixtures of pure phospholipids and pure glycolipids may form one, two, or all three structures, depending on their composition and the geometry of the individual molecular species.
Our turbidity observations and calorimetric data indicate that mixtures of pure DPPC and pure GM, behave qualitatively like the lecithin/ganglioside mixtures studied by Hill and Lester (27), and as expected from the preceding considerations. At low GM1 mole fractions (up to 0.24), Ghll continuously perturbs the DPPC transitions characteristic of the lamellar structure, but does not introduce any new transitions. The abrupt drop in turbidity at xc between 0.30 and 0.40 reveals a sharp decrease in particle size over this range. The appearance of a higher temperature transition at xo = 0.30, and its growth with increasing GM~ mole fraction, at the expense of the original transition, points to the progressive replacement of the lame&r phase by a different structure, presumably the cylindrical mice&r phase. At xo = 0.63 and 0.68, the persistence of the higher temperature transition in the absence of the one at lower temperature suggests that only micelles are present.
The addition of Ghll to DPPC increases T, for the main transition up to xo = 0.24, and T, for the mice&r transition, where xo is presumed to be around 0.63, is even higher, implying that GM1 stabilizes the DPPC structure. Stabilization of phospholipids by myristic acid (33) and sphingomyelin (34), and by gangliosides (29) and other glycolipids (25,30,35,36) has been observed by both calorimetric (25,30,33,34) and spin-labeling (29,(34)(35)(36) techniques. Such stabilization appears to be due partly to shielding of phospholipid head groups from each other, reducing their repulsive interaction (37). Stabilizing interactions are also thought to take place directly between the ganglioside head groups (29). For mixtures of egg lecithin and sphingomyelin (34) and of DPPC and monoglucocerebroside (25), formation of a 2:l phospholipid. sphingolipid complex at low sphingolipid concentrations has been postulated.
The cooperativity of the main DPPC transition, as reflected in the apparent size of the cooperative unit, n, decreases sharply with the addition of small amounts of GMl, then plateaus (Fig. 4). This behavior closely parallels the effect of glucocerebroside on the cooperativity of DPPC transitions (25). On the other hand, addition of Gwl to the mice&r phase at x. values above 0.54 causes an abrupt and unexplained increase in the cooperativity of the high temperature transition.
We have not explored the region of GM1 mole fractions above 0.68; to do this would require using either very large amounts of GM1 or amounts of DPPC so small as to yield unacceptable signal/noise ratios. Thus our data do not bear directly on the existence of two micellar structures.
Interpretation in Terms of Heterogeneous Equilibrium-The approach of DPPC/G&water systems to thermodynamic equilibrium is dependent on the method of mixing. The penetration of gangliosides into multilamellar structures (27) and the dissociation of ganglioside micelles into monomers (38) are extremely slow processes. Equilibrium DPPC/GM~ mixtures do not readily result from mixing aqueous suspensions of the pure components; but resuspending dried mixtures, as we have done, results in rapid homogeneous dispersion of the ganglioside in the bilayers (27) and produces uniform structures whose physical properties are stable over long periods of time. Our calorimetric scans were completely reproducible after an interval of 2 weeks. Similar stability was noted in a recent calorimetric study of monoglucocerebroside/ DPPC mixtures (25). Also, the limiting concentrations of the GMl-rich and DPPC-rich phases should not be strongly temperature-dependent if they are determined mainly by molecular structure. Consequently, the compositions of the two phases should not vary greatly during a calorimetric run at a given mole fraction of ganglioside. The fact that the curves obtained were independent of scan rate confiis this view.
We conclude that the calorimetric behavior of our system closely approximates equilibrium behavior. A binary system at equilibrium at constant pressure is invariant with respect to the temperature and mole fractions when three phases co-exist (39). Addition of either component to such a system alters the relative amounts of the three phases, but does not affect either their compositions or the equilibrium temperature.
In our experiments, the DPPC/GM~ system behaved as one would expect for an equilibrium system of two partially miscible components whose conjugate solid phases (DPPC-rich and GMl-rich, respectively) melt at slightly different temperatures.
Specifically, the curves in Fig. 2 between xo values of 0.30 and 0.54 may readily be interpreted in terms of two invariant systems of three phases: 1) one at a temperature of 44"C, in which the three phases at equilibrium are ordered bilayers, disordered bilayers, and ordered micelles; and 2) one at a temperature of 48"C, in which the three phases are disordered bilayers, ordered micelles, and disordered micelles.
In summary, we have studied the thermotropic behavior, over the temperature range from 12-83°C of pure ganglioside GM1, pure synthetic dipalmitoylphosphatidylcholine (DPPC), and 10 mixtures of GM~ and DPPC, at G~,J~ mole fractions from 0.04 to 0.68. Our principal observations are that 1) pure GM1 shows no thermal transitions; 2) at low mole fractions (up to 0.24), GM, quantitatively perturbs the excess heat capacity curve for DPPC multilamellar bilayers, which shows a major transition in the temperature range 41-44°C; 3) at moderate mole fractions of Ghll (between 0.30 and 0.54), the transition at 41-44°C gradually diminishes in height and area as the proportion of GM1 increases, being replaced by a new transition at 48"C, whose height and area increase correspondingly; and 4) at still higher GM1 mole fractions (0.63 and 0.68), only the transition at 48°C is observed.
The present data support the interpretation that ganglioside GM1 can be incorporated into bilayers of DPPC only up to mole fractions near 0.25. It appears that at low concentrations GM, can substitute for phospholipid molecules in the bilayer lattice. At xc values above 0.25 Ghll disrupts the bilayer, with the formation of mixed micelles of phospholipid and ganglioside. Phase transitions are detected over the whole range of ganglioside mole fractions reported here (0 5 XG 5 0.68), indicating that GM~ can pack well enough into DPPC structures to exhibit cooperative melting. It further appears that GM~ stabilizes the DPPC lattice, possibly by the formation of a specific complex.