Differential Scanning Calorimetry to Study Lipids and Lipid Membranes - CHEMICAL BIOLOGY

CHEMICAL BIOLOGY

Differential Scanning Calorimetry to Study Lipids and Lipid Membranes

Ruthven N.A.H. Lewis, David A. Mannock, and Ronald N. McElhaney, University of Alberta, Edmonton, Alberta, Canada

doi: 10.1002/9780470048672.wecb049

Differential scanning calorimetry (DSC) is a relatively rapid direct and nonperturbing thermodynamic technique for studying the thermotropic phase behavior of hydrated lipid dispersions and of reconstituted lipid model and biological membranes. DSC can accurately and reliably determine the temperature, enthalpy, entropy and cooperativity of a wide variety of lipid phase transitions and how these parameters are influenced by variations in hydration, and in the pH and the ionic strength and composition of the aqueous phase. Also, the effects of the presence of membrane-associated sterols, peptides and proteins, as well as toxins, drugs, and other agents, on the thermotropic phase behavior of lipid membranes can be determined. Under appropriate conditions, DSC can also characterize the kinetics of some lipid phase transitions. The thermodynamic data provided by DSC, therefore, can provide valuable information about the phase state and organization of lipid assemblies and about how the structure and physical properties of lipid model and biological membranes are modulated by other membrane constituents and by the environment. However, because DSC is a thermodynamic and not a structural technique, it is most valuable when applied in conjunction with a direct structural technique, such as X-ray diffraction, and with nonperturbing spectroscopic methods, such as nuclear magnetic resonance and Fourier transform infrared spectroscopy.

Biological Background

The central structural feature of almost all biological membranes is a continuous and fluid lipid bilayer that serves as the major permeability barrier of the cell or intracellular compartment (1) and as a scaffold for the attachment and organization of other membrane constituents (2, 3). In particular, peripheral membrane proteins are bound to the surface of lipid bilayers primarily by electrostatic and hydrogen-bonding interactions, whereas integral membrane proteins penetrate into, and usually span, the lipid bilayer, and are stabilized by hydrophobic and van der Waal’s interactions with the lipid hydrocarbon chains in the interior of the lipid bilayer as well as by polar interactions with the glycerol backbone and polar headgroup regions of the host lipid bilayer. In addition to playing a structural role in determining the topology and stabilizing the active conformation of peripheral and integral membrane proteins, the physical properties of the lipid also markedly influence the activity and thus presumably the conformation and dynamics of many membrane proteins (4-8, but see 9). Specifically, the physical state (lamellar gel versus liquid-crystalline), fluidity, hydrophobic thickness, lipid lamellar/nonlamellar phase propensity (lipid shape), surface charge and surface-charge density, as well as various mechanical properties of the lipid bilayer (10), all modulate the thermal stability and activity of many membrane-associated enzymes, transporters and receptors. Therefore, understanding the thermotropic phase behavior and organization and thus the specific functions of the large number of lipid classes and molecular species that comprise biological membranes, remains a major challenge in membrane biology generally. In this brief review, we consider the applications of DSC to lipid model and biological membranes to address in particular the role of lipid fluidity and phase state and to some degree the role of lipid lamellar/nonlamellar phase propensity in membrane structure and function.

Lipid Mesomorphic Phase Behavior

Membrane lipids are invariably polymorphic; that is, they can exist in a variety of kinds of organized structures, especially when hydrated. The particular polymorphic form that predominates depends not only on the structure of the lipid molecule itself and on its degree of hydration, but also on such variables as temperature, pressure, ionic strength and pH (see References 11 and 12 and article Lipids, Phase Transitions of). However, under physiologically relevant conditions, most (but not all) membrane lipids exist in the lamellar or bilayer phase, usually in the lamellar liquid-crystalline phase but sometimes in the lamellar gel phase. It is not surprising, therefore, that the lamellar gel-to-liquid-crystalline or chain-melting phase transition has been the most intensively studied lipid phase transition and is also the most biologically relevant. This cooperative phase transition involves the conversion of a relatively ordered gel-state bilayer, in which the hydrocarbon chains exist predominantly in their rigid, extended, all-trans conformation, to a relatively disordered liquid-crystalline bilayer, in which the hydrocarbon chains contain several gauche conformers and exhibit greatly increased rates of intra- and intermolecular motions. The gel-to-liquid-crystalline phase transition is accompanied by a pronounced lateral expansion and a concomitant decrease in the thickness of the bilayer, as well as by a small increase in the total volume occupied by the lipid molecules. Evidence also shows that the number of water molecules bound to the surfaces of the lipid bilayer increases during hydrocarbon chain melting. Thermodynamically, the gel-to-liquid-crystalline phase transition occurs when the entropic reduction in free energy that results from hydrocarbon chain isomerism counterbalances the decrease in bilayer cohesive energy that results from the lateral expansion and from the energy cost of creating gauche rotational conformers in the hydrocarbon chains.

Gel-to-liquid-crystalline phase transitions can be induced by changes in temperature and hydration, as well as by changes in pressure and in the ionic strength or pH of the aqueous phase. In this article, we will concentrate on thermally induced phase transitions because these have been studied most extensively and are of direct biological relevance, particularly for organisms that cannot regulate their own temperature. However, hydration-induced (lyotropic) and pressure-induced (barotropic) phase transitions also occur, and these may also be biologically relevant under special environmental circumstances. Finally, phase transitions induced by alterations in pH and in the nature and quantity of ions in the aqueous phase that surrounds the bilayer are also possible, and these transitions may also be of importance in living cells. However, a detailed discussion of these types of lipid phase transitions is beyond the limited scope of the current article, and interested readers should consult appropriate reviews for detailed information on this topic (11, 12).

Pure synthetic lipids often exhibit gel-state polymorphism, and phase transitions between various forms of the gel-state bilayer can occur. Although we will illustrate this behavior for a common phospholipid, dipalmitoylphosphatidylcholine (DPPC), gel-state transitions will not be emphasized here because with only one known exception (13), they do not seem to occur in the heterogeneous collection of lipid molecular species found in biological membranes. Moreover, certain synthetic or naturally occurring lipid species can exist in liquid-crystalline nonlamellar phases, especially three-dimensional reversed cubic and hexagonal phases. Although the actual existence of nonbilayer lipid phases in biological membranes has never been demonstrated under physiological conditions, the propensity to form such phases likely plays major roles in membrane fusion and other processes (see Reference 14). Moreover, evidence suggests that the relative proportion of bilayer-preferring and nonbilayer-preferring lipids may be biosynthetically regulated in response to variations in temperature and membrane lipid fatty acid composition and cholesterol content in some organisms. Thus, lipid species that in isolation may form nonlamellar phases may have important roles to play in the liquid-crystalline bilayers found in essentially all biological membranes. The transitions between lamellar and nonlamellar lipid phases have been reviewed in detail by us and others elsewhere (see References 14 and 15).

Differential Scanning Calorimetry

As mentioned earlier, the technique of DSC has been of primary importance in studies of lipid phase transitions in model and biological membranes (see References 16-18). The principle of DSC is comparatively simple. A sample and an inert reference (i.e., a material of comparable thermal mass that does not undergo a phase transition within the temperature range of interest) are simultaneously heated or cooled at a predetermined constant rate (dT/dt) in an instrument configured to measure the differential rate of heat flow (dE/dt) into the sample relative to that of the inert reference. The temperatures of the sample and reference may either be actively varied by independently controlled units (power compensation calorimetry) or be passively changed through contact with a common heat sink that has a thermal mass that greatly exceeds the combined thermal masses of the sample and reference (heat conduction calorimetry). For our purposes, the sample would normally be a suspension of lipid or membrane in water or an aqueous buffer, and the reference cell would contain the corresponding solvent alone. At temperatures distant from any thermotropic events, the temperatures of the sample and reference cells change linearly with time, and the temperature difference between them remains zero. The instrument thus records a constant difference between the rates of heat flow into the sample and reference cells, which, ideally, is reflected by a straight, horizontal baseline. When the sample undergoes a thermotropic phase transition, a temperature differential between the sample and reference occurs, and the instrument either actively changes the power input to the sample cell to negate the temperature differential (power compensation calorimetry) or passively records the resulting changes in the rate of heat flow into the sample cell until the temperature differential eventually dissipates (heat conduction calorimetry). In both instances, a change develops in the differential rates of heat flow into the sample and reference cells, and either an exothermic or endothermic deviation from the baseline condition occurs. On completion of the thermal event, the instrument either re-establishes its original baseline condition or establishes a new one if a change in the specific heat of the sample has occurred. The output of the instrument is thus a plot of differential heat flow (dE/dt) as a function of temperature in which the intensity of the signal is directly proportional to the scanning rate (dT/dt).

The variation of excess specific heat (dE/dt) with temperature for a simple two-state, first-order endothermic process, such as the gel-to-liquid-crystalline phase transition of a single, highly pure phosphatidylcholine (PC), is illustrated schematically in Fig. 1. From such a DSC trace, several important parameters can be determined directly. The phase transition temperature, usually denoted Tm, is that temperature at which the excess specific heat reaches a maximum. For a symmetrical curve, Tm represents the temperature at which the transition from the gel-to-liquid-crystalline state is one-half complete. However, for asymmetric traces, which are characteristic of certain pure phospholipids and many biological membranes, the Tm does not represent the midpoint of the phase transition, and a T1/2 value may be reported instead. Once normalized with respect to the scan rate, the peak area under the DSC trace is a direct measurement of the calorimetrically determined enthalpy of the transition, ∆Hcal, usually expressed in kcal/mol. The area of the peak can be determined by planimetry or by the cutting and weighing technique; alternatively, the calorimeter output can be digitized, and the Tm and ∆Hcal can be calculated by a computer. Because at the phase transition midpoint temperature the change in free energy (∆G) of the system is zero, the entropy change associated with the transition can be calculated directly from the equation:

where ∆S is normally expressed in cal/K-1mol-1.

Figure 1. The variation of excess specific heat with temperature during a two-state, endothermic lipid phase transition. The symbols are explained in the text.

The sharpness or cooperativity of the gel-to-liquid-crystalline phase transition can also be evaluated from the DSC trace. The sharpness of the phase transition is often expressed as the temperature width at half-height, ∆T1/2, or as the temperature difference between the onset or lower boundary of the phase transition, Ts, and the completion or upper boundary, Tl, or ∆T = Tl — Ts. The ∆T1/2values may range from <0.1°C for very pure synthetic phospholipids to as much as 10-15°C for biological membranes. From the Tm and ∆T1/2 values determined for a particular phase transition, the van’t Hoff enthalpy, ∆HvH, can be approximately determined from the relationship:

From the ratio ∆HvH/∆Hcal, the cooperative unit size (CUS) (in molecules) can be determined. The CUS is a measure of the degree of intermolecular cooperation between phospholipid molecules in a bilayer; for a completely cooperative, first-order phase transition of an absolutely pure substance, this ratio should approach infinity, whereas for a completely noncooperative process, this ratio should approach unity. Although the absolute CUS values determined should be regarded as tentative, because this parameter is markedly sensitive to the presence of impurities and may be limited by instrumental parameters, carefully determined CUS values can be useful in assessing the purity of synthetic phospholipids and in quantitating the degree of cooperativity of lipid phase transitions.

It must be stressed that the thermodynamic parameters derived from DSC measurements will be valid only if measurements are performed under conditions where the instrument response is true to the properties of the sample (so-called high-fidelity DSC) and if the thermotropic process being studied is at equilibrium throughout the measurement. In practice, this statement means that the measurement must be made with a high-sensitivity instrument operating with a modest thermal load at scan rates that are slow relative to the thermal time constant of the instrument and to both the width and half-life of the thermotropic process under investigation. For processes such as the gel-to-liquid-crystalline phase transition of certain single, pure synthetic phosphatidylcholines, this condition is rarely a problem because such processes are usually rapid enough to be effectively free of kinetic limitations even at moderate scan rates. However, processes such as the pretransition and the subtransition of synthetic disaturated phosphatidyl-cholines are known to be kinetically limited at all temperatures and scan rates at which calorimetric observation is feasible (16-18). For such processes, their thermodynamic parameters cannot be reliably measured by DSC.

Another aspect of the thermodynamic equilibrium problem that should be considered is the question of whether the system is at equilibrium before the calorimetric scan is initiated. In many DSC studies of model and biological membranes, the sample is placed in the calorimeter and cooled fairly rapidly to a low temperature, and a calorimetric heating scan then is begun relatively quickly. Because, as was mentioned above, the kinetics of lipid phase transitions in complex systems are not well studied and because the rates of reversible lipid phase transitions are generally considerably slower when proceeding from a higher-temperature to a lower-temperature state than the reverse, the possibility exists that the system under study may not be at thermodynamic equilibrium when the calorimetric run is begun. This possibility can be the case even if no exothermic events are observed during heating. For this reason, it is always advisable to cool the sample slowly and to investigate the effect of variations in the “annealing” time at low temperatures on the DSC results obtained.

We stress here that although DSC is in principle a relatively straightforward physical technique, its theoretical thermodynamical and kinetic basis is not trivial but should be well understood as it applies to equilibrium and nonequlibrium thermotropic lipid phase transitions of various types and to either heat conduction or power compensation instruments. Moreover, some care must be taken in sample preparation, selection of sample size, and sample equilibration before data acquisition; in the choice of suitable scan rates, starting temperatures, and ending temperatures during data acquisition; and in the analysis and interpretation of the DSC thermograms obtained. An adequate treatment of these issues is not possible in this brief article, and for a full treatment of these and other issues, the reader is referred to a recent review (18).

Mention should be made here of the recently developed technique of pressure perturbation calorimetry (PPC), which measures the temperature-dependent volume change of a solute or colloidal particle in aqueous solution. PPC can also be used to detect thermotropic phase transitions in lipid model membranes and to characterize the accompanying volume changes and the kinetics of the phase transition. PPC essentially measures the heat change that results from small pressure changes at a constant temperature in a high-sensitivity isothermal calorimeter. For an excellent recent review on PPC as applied to lipid systems, the reader is referred to Heerklotz (19).

DSC Studies of Lipid Model Membranes

Dipalmitoylphosphatidylcholine

A DSC heating scan of a fully hydrated aqueous dispersion of dipalmitoylphosphatidylcholine (DPPC), which has been annealed at 0°C for 3.5 days, is displayed in Fig. 2. The sample exhibits three endothermic transitions, termed (in order of increasing temperature) the subtransition, pretransition, and main phase transition. The thermodynamic parameters associated with each of these lipid phase transitions are presented in Table 1. The presence of three discrete thermotropic phase transitions indicates that four different phases can exist in annealed, fully hydrated bilayers of this phospholipid, depending on temperature and thermal history. All of these phases are lamellar or bilayer phases differing only in their degree of organization.

Figure 2. A typical high-sensitivity DSC heating thermogram of a multilamellar, aqueous suspension of DPPC which has been annealed at 0-4°C dor 3-5 days prior to commencement of heating. The substransition, pretransition and main phase transition temperatures are denoted by Ts, Tp and Tm respectively.

The low-temperature gel phase corresponds closely to that of the crystalline dihydrate and is thus denoted the Lc’ phase. The detailed structure of this phase and the other phases discussed below is treated in detail elsewhere (see article Lipids, Phase Transitions of). The Lc’ phase is characterized by extended hydrocarbon chains that are tilted slightly with respect to the bilayer normal. These chains are packed very tightly, and rotation about their long axes is very severely restricted. The polar headgroup contains only a few bound water molecules, and its motion is also severely restricted.

With increasing temperature, the steric and van der Waal’s interchain interactions that favor a crystalline-like packing of the DPPC molecules are progressively overcome by thermally induced rotational excitations of the hydrocarbon chains. Thus, at the subtransition temperature, the Lc’ phase converts to the lamellar gel or Lβ’ phase. In this phase, the extended hydrocarbon chains are tilted more strongly from the bilayer normal, are packed in a distorted orthorhombic lattice, and undergo relatively slow, restricted rotational motion about their long axes. The polar headgroup now contains about 15-18 waters of hydration and exhibits slow, hindered rotation on the NMR timescale. The subtransition results in a small increase in the lipid hydrocarbon chain cross-section area and a larger increase in the interfacial area. Thus, the Lβ’ phase is less tightly packed and much more strongly hydrated than is the Lc’ phase. The Lc’ of DPPC forms very slowly when cooling to low temperatures and requires about 3.5 days at 0°C to fully form; if a heating run on a DPPC sample not annealed at low temperature is performed, no subtransition will be detected, which indicates the importance of lipid samples being at thermal equilibrium before analysis by DSC. In fact, the DPPC subtransition was discovered by Sturtevant and coworkers (20) when a DPPC sample was inadvertently cooled and left over the weekend before a DSC heating scan was initiated.

Additional increases in temperature result in a marked increase in the long-axis rotational rates of the hydrocarbon chains, and at the pretransition temperature, the Lβ’ phase converts to the so-called ripple or Pβ’ phase. In the Pβ’ phase, the extended hydrocarbon chains seem to remain tilted with respect to the normal to the local bilayer plane but behave as if they are rotationally symmetric, packing into a hexagonal lattice. The cross-section areas of the hydrocarbon chain thus show a small increase at the pretransition temperature. The interfacial area increases much more substantially, however, because of the displacement of each lipid molecule along its long axis with respect to its neighbor. The increased area occupied by the polar headgroups allows them to rotate almost freely, although the degree of hydration does not seem to change. In contrast to the Lc’ and Lβ’ phases, the bilayer is no longer planar but exists as a series of periodic, quasi-lamellar segments.

Table 1. Thermodynamic characteristics of the three phase transitions exhibited by multilamellar aqueous suspensions of dipalmitoylphosphatidylcholine after annealing (~3-5 days) at 0-4°C before heating

Transition type

T(°C)

∆T1/2 (°C)

∆Hcal(k cal/mol)

∆S(cal/kmol)

Subtransition

16.5*

3.0*

3.5

11.7

Pretransition

34.2

1.8*

1.1

3.6

Main Transition

41.4

0.1

7.8

24.8

*The phase transition temperature of the subtransition and the ∆T1/2 values of the subtransition and pretransition are overestimated because of kinetic limitations, even at the lowest heating scan rates feasible. The data listed were obtained from Reference 47.

With increasing temperature, the formation of gauche rotational conformers in the hydrocarbon chain becomes increasingly favorable until at the gel-to-liquid-crystalline phase transition temperature, chain melting occurs. Spectroscopic and thermodynamic studies have shown that the hydrocarbon chains of DPPC in the melted or La phase contain about four gauche bonds per chain, mostly, but not entirely, in the form of kink (gauche+-trans-gauche-) sequences. As the melting of the hydrocarbon chains produces a marked increase in cross-section area and effectively shortens the length of the chains, the bilayer expands laterally and thins at the main phase transition. Although the hydrocarbon chains exhibit rapid flexing and rotation in the Lα phase, they are on average oriented normally to the bilayer plane and pack in a loose hexagonal lattice. This increase in the cross-section area per molecule results in an increase in the area available to the polar headgroup, with the result that rotational motion becomes fast on the NMR timescale, and the hydration at the bilayer interface increases, in part because of the partial exposure of more deeply located polar residues, such as the carbonyl oxygens of the fatty acyl chains, to the aqueous phase.

The pattern of thermotropic phase behavior exhibited by an aqueous dispersion of any lipid molecular species will vary considerably, depending on the length and structure of the hydrocarbon chains, the structure and charge of the polar headgroup, the nature of the link (ester or ether) of the hydrocarbon chains to the glycerol or sphingosine backbone, and other chemical features of the lipid under study. Also, the degree of hydration and the pH and ionic composition of the aqueous phase can affect lipid thermotropic phase behavior profoundly. However, even a cursory discussion of this topic is beyond the scope of this article, and the reader is referred to recent reviews for more detailed information (16-18).

Phospholipid mixtures

Although studies of the thermotropic phase behavior of singlecomponent multilamellar phospholipid vesicles are necessary and valuable, these systems are not realistic models for biological membranes that normally contain at least several different types of phospholipids and a variety of fatty acyl chains. As a first step toward understanding the interactions of both the polar and apolar portions of different lipids in mixtures, DSC studies of various binary and ternary phospholipid systems have been carried out. Phase diagrams can be constructed by specifying the onset and completion temperatures for the phase transition of a series of mixtures and by an inspection of the shapes of the calorimetric traces. A comparison of the observed transition curves with the theoretical curves supports a literal interpretation of the phase diagrams obtained by DSC. For a summary of the first high-sensitivity DSC studies of binary phospholipid-phospholipid and phospholipid-cholesterol mixtures and a description of how phase diagrams can be constructed from DSC data, the reader is referred to an early review by Mabrey and Sturtevant (21); for a compilation of the results of later DSC and other studies on other phospholipid mixtures, the reader is referred to Marsh (22).

The effect of cholesterol

The occurrence of cholesterol and related sterols in the membranes of eukaryotic cells has prompted many investigations of the effect of cholesterol on the thermotropic phase behavior of phospholipids (see References 23-25). Studies using calorimetric and other physical techniques have established that cholesterol can have profound effects on the physical properties of phospholipid bilayers and plays an important role in controlling the fluidity of biological membranes. Cholesterol induces an “intermediate state” in phospholipid molecules with which it interacts and, thus, increases the fluidity of the hydrocarbon chains below and decreases the fluidity above the gel-to-liquid-crystalline phase transition temperature. The reader should consult some recent reviews for a more detailed treatment of cholesterol incorporation on the structure and organization of lipid bilayers (23-25).

Recent high-sensitivity DSC studies of cholesterol/PC interactions have revealed a complex picture of cholesterol/DPPC interactions (26). At cholesterol concentrations from 0 to 20-25 mol %, the DSC endotherm consists of two components (see Fig. 3). The sharp component exhibits a phase transition temperature and cooperativity only slightly reduced from those of the pure phospholipid, and the enthalpy of this component decreases linearly with increasing cholesterol content, becoming zero at 20-25 mol %. In contrast, the broad component exhibits a progressively increasing phase transition temperature and enthalpy with a progressively decreasing cooperativity over this same range of cholesterol content. Above cholesterol levels of 20-25 mol %, the broad component becomes progressively less cooperative, the phase transition midpoint temperature continues to increase, and the transition enthalpy continues to decrease, eventually approaching zero only at cholesterol concentrations near 50 mol %. These results suggest that at low cholesterol concentrations, cholesterol-poor and cholesterol-rich domains coexist, with the former decreasing in proportion to the latter as cholesterol concentrations increase. In fact, a cardinal point in the cholesterol/DPPC phase diagram at about 22 mol % had been predicted from the earlier model-building studies, which calculated that the cholesterol molecule could interact with a maximum of 7 adjacent phospholipid hydrocarbon chains (or 3.5 phospholipid molecules) and thus that free phospholipid would exist only at cholesterol concentrations below this value. This model also explains the decreasing enthalpy of the broad component observed above 22 mol % cholesterol because an increasing proportion of phospholipid molecules would interact with more than one cholesterol molecule rather than with the more flexible hydrocarbon chains of adjacent phospholipids and, thus, progressively decrease and eventually abolish the cooperative chain-melting phase transition.

Figure 3. Typical high-sensitivity DSC heating thermograms of multilamellar, aqueous suspensions of DPPC containing various amounts of incorporated cholesterol. The amount of cholesterol present (in mole %) is indicated near each thermogram.

McMullen and coworkers (26) have studied the effects of cholesterol on the thermotropic phase behavior of aqueous dispersions of a homologous series of linear saturated PCs, using high-sensitivity DSC and an experimental protocol that ensures that the broad, low-enthalpy phase transitions at high cholesterol concentrations are accurately monitored. They found that the incorporation of small amounts of cholesterol progressively decreases the temperature and the enthalpy, but not the cooperativity, of the pretransition of all PCs exhibiting such a pretransition and that the pretransition is completely abolished at cholesterol concentrations above 5 mol % in all cases. Moreover, the incorporation of increasing quantities of cholesterol was found to alter the main or chain-melting phase transition of these phospholipid bilayers in both hydrocarbon chain length-dependent and hydrocarbon chain length-independent ways. The temperature and cooperativity of the sharp component are reduced only slightly and in a chain length-independent manner with increasing cholesterol concentration, an observation ascribed to the colligative effect of the presence of small quantities of cholesterol at the domain boundaries. Moreover, the enthalpy of the sharp component decreases and becomes zero at 20-25 mol % cholesterol for all PCs examined. In contrast, the broad component exhibits a chain length-dependent shift in temperature and a chain length-dependent decrease in cooperativity but a chain length-independent relative increase in enthalpy over the same range of cholesterol concentrations. Specifically, cholesterol incorporation progressively increases the phase transition temperature of the broad component in PCs that have hydrocarbon chains of 16 or fewer carbon atoms and decreases the broad-component phase transition temperature in PCs that have hydrocarbon chains of 18 or more carbon atoms, an effect attributed to hydrophobic mismatch between the cholesterol molecule and its host PC bilayer. The best match between the effective length of the cholesterol molecule and the mean hydrophobic thickness of the PC bilayers is obtained with the diheptadecanoyl PC molecule. Moreover, cholesterol decreases the cooperativity of the broad component more rapidly and to a greater extent in the shorter-chain as compared with the longer chain PCs. At cholesterol concentrations above 20-25 mol %, the sharp component is abolished, and the broad component continues to manifest the chain length-dependent effects on the temperature and cooperativity described above. However, the enthalpy of the broad component decreases linearly and reaches zero at about 50 mol % cholesterol, regardless of the chain length of the phosphatidylcholine.

The effect of cholesterol on the thermotropic phase behavior of PC bilayer also varies significantly with the structure, particularly the degree of unsaturation, of the hydrocarbon chains, with more highly unsaturated PCs exhibiting a reduced miscibility with cholesterol and other sterols. Moreover, the structure of the lipid polar headgroup is also important in determining the effect of cholesterol on the host lipid, as is the structure of the sterol molecule itself. For more information on the application of DSC to the biologically important area of lipid-sterol interactions, the reader is referred to recent reviews (23-25).

The effect of small molecules

Several lipid-soluble small molecules, including drugs like tranquilizers, antidepressants, narcotics, and anaesthetics, produce biological effects in living cells. Although some of these compounds are known to produce their characteristic effects by interacting with specific membrane proteins, others seem to interact rather nonspecifically with the lipid bilayer of many biological membranes. The effect on the gel-to-liquid-crystalline phase transition profile of synthetic PCs of over 100 hydrophobic small molecules that produce biological effects have now been studied by DSC (27). At least four different types of modified transition profiles can be distinguished: In so-called type C profiles, the addition of the additive shifts Tm usually (but not always) to a lower temperature while having little or no effect on the cooperativity (∆T1/2) or ∆Hcal of the transition; other physical evidence suggests that additives that produce this behavior are usually localized in the central region of the bilayer, which interacts primarily with the C9-C16 methylene region of the phospholipid hydrocarbon chains. Type A profiles are characterized by a shift in Tm usually to a lower temperature, an increase in ∆T1/2, and a relatively unaffected ∆Hcal during the addition of the appropriate small molecules; these additives seem to be partially buried in the hydrocarbon core of the bilayer, which interacts primarily with the C2-C8 methylene region of the hydrocarbon chains. In type B profiles, a shoulder emerges on the main transition, the area of which increases in conjunction with a corresponding decrease in the area of the original peak as the concentration of additive increases. The total area of both peaks is relatively unchanged, at least at low additive concentration. Additives that produce type B profiles generally reside at the hydrophobic-hydrophilic interface of the bilayer and interact primarily with the glycerol backbone of the phospholipid molecules. Finally, type D profiles exhibit a discrete a new peak that grows in area at the expense of the parent peak as the additive concentration increases; normally, however, the final ∆Hcal and ∆T1/2 values of the new and original peaks are not greatly different. Type D additives usually seem to be located at the bilayer surface and interact with the phosphorylcholine headgroup.

Although this classification is useful, not all small molecules produce one of these four types of DSC profiles. Whether a consistent relationship exists between the type of transition profile produced by a small molecule and its physiological effects remains to be determined.

The effect of transmembrane peptides

DSC has been used to great effect to study the effect of the incorporation of α-helical transmembrane peptides on the thermotropic phase behavior of various phospholipid bilayers. Because most integral membrane proteins contain one or more α-helical transmembrane segments, such studies are relevant to the mechanisms by which the physical properties of the membrane lipid bilayer modulate the structure and activity of such proteins. In this regard, several investigations have been carried out using DSC and many other physical techniques to understand how the presence of such transmembrane peptides effect the organization and dynamics of the host lipid bilayer and vice versa. Such studies have examined the effects of systematic variations in the length and structure of model α-helical transmembrane peptides on lipid bilayer organization and dynamics, and how the effects of such peptides are themselves affected by the hydrophobic thickness and chemical composition of the host phospholipid bilayer. These important studies are ongoing, and the reader should consult recent reviews for more information (28, 29).

The effect of membrane antimicrobial peptides

DSC has also been used to study the effects of a wide variety of antimicrobial peptides on the thermotropic phase behavior of different lipid bilayers. These studies again are highly biologically relevant because the primary mode of action of most antimicrobial peptides is the perturbation and permeabilization of the lipid bilayers of the target membrane, and these agents have considerable promise as antibiotics, especially to treat multiple drug-resistant pathogenic bacteria. Again, the reader should consult recent reviews for more information on this topic (30, 31).

The effect of membrane proteins

Because of their obvious relevance to biological membranes, the effect of several peptides and proteins on the thermotropic phase behavior of single synthetic phospholipids or phospholipid mixtures has been studied by many groups (see 16, 17). It was originally proposed by Papahadjopoulos et al. (32) that polypeptides and proteins could be considered as belonging to one of three types according to their characteristic effects on phospholipid gel-to-liquid-crystalline phase transitions. Type 1 proteins typically produce no change or a modest increase in Tm, a slight increase or no change in ∆T1/2and an appreciable and progressive increase in ∆Hcal as the amount of protein added is increased. These proteins normally do not expand phospholipid monolayers nor alter the permeability of phospholipid vesicles into which they are incorporated. Type 1 proteins are “hydrophilic” proteins that are thought to interact with the phospholipid bilayer exclusively by electrostatic forces and, as such, normally show stronger effects on the phase transitions of charged rather than zwitterionic phospholipids. Type 2 proteins produce a decrease in Tm, an increase in ∆T1/2 and a considerable and progressive decrease in ∆Hcal; phospholipid monolayers are typically expanded by such proteins, and these proteins normally increase the permeability of phospholipid vesicles. These proteins, which are also hydrophilic, are believed to interact with phospholipid bilayers by a combination of electrostatic and hydrophobic forces, initially adsorbing to the charged polar headgroups of the phospholipids and subsequently partially penetrating the hydrophilic-hydrophobic interface of the bilayer to interact with a portion of the hydrocarbon chains. Finally, type 3 proteins usually have little effect on the Tm or ∆T1/2 of the phospholipid phase transition, but ∆Hcal decreases linearly with protein concentration. Type 3 proteins are “hydrophobic” proteins that markedly expand phospholipid monolayers and increase the permeability of phospholipid vesicles. These proteins are thought to penetrate deeply into or to span the hydrophobic core of anionic or zwitterionic lipid bilayers and, thus, to interact strongly with the phospholipid fatty acyl chains and essentially to remove them from participation in the cooperative chain-melting transition. It should be noted, however, that some type 3 proteins may also interact electrostatically with phospholipid polar headgroups, particularly with those bearing a net negative charge.

The results of more recent DSC and other studies of lipid-protein model membranes clearly indicate that the classification scheme originally proposed is not completely appropriate for naturally occurring membrane proteins (see Reference 17). Thus, none of the water-soluble, peripheral membrane-associated proteins studied thus far exhibit classical type 1 behavior (no change or a modest increase in Tm, a slight increase in ∆T1/2 and an increase in the AH of the phospholipid phase transition). Therefore, it seems doubtful whether natural membrane proteins ever interact with phospholipid bilayers exclusively by electrostatic interactions. However, a few examples of membrane proteins do exhibit more-or-less-classical type 2 behavior. These examples include the myelin basic protein and cytochrome c, all of which usually reduce the Tm, increase the ∆T1/2 and substantially reduce the AH of the chain-melting transition of anionic phospholipids. Strictly speaking, few if any membrane proteins actually exhibit classical type behavior as originally defined (no change in the Tm or ∆T1/2 and a progressive linear reduction in the ∆H of both neutral and anionic phospholipid phase transitions with increasing protein concentration). This is because, with the advent of high-sensitivity calorimeters and the availability of pure phospholipids, it has become clear that all integral membrane proteins reduce the cooperativity of gel-to-liquid-crystalline phase transitions, as indeed would be expected from basic thermodynamic principles. Moreover, some type 3 proteins exhibit a nonlinear decrease in ∆H with changes in protein levels, whereas others can produce at least moderate shifts in the Tm of phospholipid phase transitions. However, if we relax the original type 3 criteria somewhat, then several integral, transmembrane proteins can be said to exhibit “modified” type 3 behavior.

The classification scheme of Papahadjopoulos et al. (32), appropriately modified for type 3 proteins, is still of some use in studies of lipid-protein interactions, although some proteins, at least under certain conditions, do not fall neatly into any of these three categories. It seems that all naturally occurring membrane proteins studied to date interact with lipid bilayers by both hydrophobic and electrostatic interactions and that different membrane proteins differ only in the specific types and relative magnitudes of these two general classes of interactions. It is also clear that the behavior exhibited by any particular membrane protein can depend on its conformation, method of reconstitution, and relative concentration, as well as on the polar headgroup and fatty acid composition of the lipid bilayer with which it is interacting (see Reference 17).

Although DSC and other physical techniques have made considerable contributions to the elucidation of the nature of lipid-protein interactions, several outstanding questions remain. For example, it remains to be definitively determined whether some integral, transmembrane proteins completely abolish the cooperative gel-to-liquid-crystalline phase transition of lipids with which they are in direct contact or whether only a partial abolition of this transition occurs, as is suggested by the studies of the interactions of the model transmembrane peptides with phospholipids bilayers (see above). The mechanism by which some integral, transmembrane proteins perturb the phase behavior of very large numbers of phospholipids also remains to be determined. Finally, the molecular basis of the complex and unusual behavior of proteins such as the concanavalin A receptor and the Acholeplasma laidlawii B ATPase is still obscure (see Reference 17).

Lipid lamellar/nonlamellar phase transitions

The mixture of lipids present in all biological membranes studied to date seems to exist exclusively in the liquid-crystalline lamellar phase under physiologically relevant conditions of temperature and hydration. However, individual membrane lipids potentially can form a variety of liquid-crystalline normal, lamellar, or reversed phases when dispersed in water, depending primarily on their effective molecular shapes. For these rod-like amphiphilic lipid molecules, the relative effective sizes of their polar and nonpolar regions are important elements in determining their molecular shapes, in particular, the relative cross-section areas occupied by their polar headgroups and nonpolar hydrocarbon chains. The effective cross-section area of a lipid polar headgroup seems to depend primarily on headgroup volume, whereas the effective cross-section area of the hydrocarbon chains depends primarily on the length and degree of unsaturation of the chains. If the effective cross-section area of the polar headgroup exceeds that of the nonpolar region, then the lipid molecule will have a conical shape and will tend to aggregate in water to form normal micelles or related structures. Conversely, if the relative cross-section of the polar headgroup is less than that of the hydrocarbon chains, then the lipid will have an “inverted” conical shape and will tend to aggregate in water to form either a reversed cubic or a reversed hexagonal phase. If, however, the relative areas occupied by the polar headgroup and the hydrocarbon chains are roughly equal, then the molecules will be cylindrical in shape and will tend to form a lamellar or bilayer phase. Because the effective area of the hydrocarbon chains in the liquid-crystalline state increases to a much greater extent with temperature than does that of the polar headgroup, increases in temperature favor the formation of lamellar over normal and reversed over lamellar phases (see Reference 14).

It is relatively straightforward to determine the types of phases formed by aqueous dispersions of individual membrane lipids over a range of temperature and thus to infer something about the lipids’ overall effective shape. In fact, the structures of the various phases formed by the individual lipids of the membrane of A. laidlawii have been studied extensively (see Reference 8). It is difficult, however, to quantitate the relative strengths of the phase preferences of a series of different lipids because the effective shapes of the lipid molecules cannot be directly determined in their various liquid-crystalline phases. However, Epand (33) has shown that small amounts of lipids with small (large) polar headgroups decrease (increase) the liquid-crystalline, lamellar-reversed, hexagonal phase transition temperature (Th) of the host dielaidoylphosphatidylethanolamine bilayer (DEPE), and Janes and coworkers (34) have recently shown that the intrinsic headgroup volumes of seven synthetic dioleoyl glycerolipids correlate well with the ability of these lipids to alter the Th of a 1-palmitoyl-2-oleoyl phosphatidylethanolamine matrix. In fact, both groups have presented evidence that Th probably varies linearly with the effective size of the lipid polar headgroup at the lipid/water interface. This approach thus seems suitable for quantitating the relative phase preferences of any series of lipids based on differences in their effective shapes, which will be determined largely by effective head- group size when the structures of their fatty acyl chains are identical. However, this method would generally not be applicable to the lipids of most biological membranes because the fatty acid compositions of the individual lipids are usually quite different. However, the ability to manipulate the fatty acid composition of the membrane lipids of the simple, cell wall-less prokaryote A. laidlawii B has permitted us to determine the relative effective headgroup sizes and, thus, the relative strength of the phase preferences of all quantitatively significant membrane lipids of this organism by determining the effect of the incorporation of small amounts of these lipids on the Th of a phos- phatidylethanolamine matrix of identical fatty acid composition (35). We found that the incorporation of small amounts of these lipids produced effects ranging from a moderate depression to a marked elevation of the Th of the phosphatidylethanolamine. Thus, although the total membrane lipids from this organism form only lamellar phases under physiological conditions, the individual membrane lipids seem to exhibit a wide range of phase preferences. Phosphatidylglycerol and diglucosyldiacyl-glycerol seem to have relatively strong and weak preferences for the lamellar liquid-crystalline phase, respectively, whereas monoglucosyldiacylglycerol and, especially, acyl polyprenyl glucoside strongly prefer the reversed hexagonal phase. Most notable in this regard is the phase preference of glycerylphos-phoryldiglucosyldiacylglycerol, which strongly destabilizes the reversed hexagonal phase and which actually prefers the normal micellar phase in isolation (36) (see Fig. 4). The presence of normal, lamellar, and reversed phase-preferring lipids in a single membrane has important implications for understanding the physical basis of lipid organization and biosynthetic regulation in this organism and possibly in other organisms. We also showed that the characteristic effect of the individual A. laidlawii membrane lipids on the lamellar/reversed hexagonal phase transition temperature of the phosphatidylethanolamine matrix is not well correlated with their polar head- group intrinsic volumes. This result indicates that the effective cross-section area of the polar headgroups of these lipid species must be strongly influenced by factors such as charge, hydration, orientation, and motional freedom as well as by intrinsic headgroup volume.

The approach discussed above to determine quantitatively the effect of various membrane phospho- and glycolipids on the lamellar/nonlamellar phase behavior of a host phosphatidylethanolamine or similar matrix has been applied to determine the relative shape, and thus the effect on the monolayer curvature of the host bilayer, of several agents, including sterols, peptides, detergents, and drugs (see Reference 14). Such studies can be very useful in providing insight into the function and mechanism of action of these agents on biological membranes.

Figure 4. High-sensitivity DSC heating thermograms of aqueous multilamellar dispersions of DEPE containing (A) 0 mol %, (B) 2.5 mol % > (C) 5.0 mol and (D) 10.0 mol % of the glycerylphosphoryldiglucosyl diacylglycerol from elaidic-acid homogeneous Acholeplasma laidlawii B membranes. Only the lamellar liquid-crystalline (Lα) to liquidcrystalline reversed hexagonal (H||) phase transition is illustrated. Note that the strong upward shift in the Lα/H|| phase transition temperature indicates that this membrane lipid strongly stabilizes the La phase and destabilizes the H|| phase of DEPE, indicating that it has a conical shape with a large polar headgroup volume relative to the volume of the hydrocarbon chains. In fact, this lipid forms a normal micellar phase in water in isolation from the other membrane lipids (31).

DSC Studies of Biological Membranes

The A. laidlawii membrane was used by Steim and colleagues (37) to show for the first time that biological membranes can undergo a gel-to-liquid-crystalline lipid phase transition similar to that previously reported for lamellar phospholipid-water systems. These workers demonstrated that when whole cells or isolated membranes are analyzed by DSC, two relatively broad endothermic transitions are observed on the initial heating scan. The lower-temperature transition is fully reversible, varies markedly in position with changes in the length and degree of unsaturation of the membrane lipid fatty acyl chains, is broadened and eventually abolished by cholesterol incorporation, and exhibits a transition enthalpy characteristic of the mixed-acid synthetic phospholipids. Moreover, an endothermic transition that has essentially identical properties is observed for the protein-free total membrane lipid extract dispersed in excess water or aqueous buffer, which indicates that the presence of membrane proteins has little effect on the thermotropic phase behavior of most membrane lipids. In constrast, the higher-temperature transition is irreversible, is independent on membrane lipid fatty acid composition or cholesterol content, and is absent in total membrane lipid extracts, which indicates that the higher temperature transition results from an irreversible thermal denaturation of the membrane proteins. A comparison of the enthalpies of transition of the lipids in the membrane and in water dispersions indicates that at least 75% of the total membrane lipids participate in this transition. Evidence was also presented that the lipids must be predominantly in the fluid state to support normal growth. These results were later confirmed and extended by Reinert and Steim (38) and by Melchior et al. (39), who showed that the gel-to-liquid-crystalline lipid phase transition is a property of living cells and that about 85-90% of the lipid participates in the gel-to-liquid-crystalline phase transition. These studies provided strong, direct experimental evidence for the hypothesis that lipids are organized as a liquid-crystalline bilayer in biological membranes, a basic feature of the currently well-accepted fluid-mosaic model of membrane structure.

Representative high-sensitivity DSC initial heating scans of viable cells, isolated membranes, and total membrane lipid dispersions are shown in Fig. 5; in this instance, cells, membranes, and lipids were made nearly homogeneous in elaidic acid (13). The fully reversible gel-to-liquid-crystalline lipid phase transitions observed in cells and membranes essentially have identical phase transition temperatures, enthalpies, and degrees of cooperativity, which suggests that membrane lipid organization in these two samples is very similar or identical. In contrast, the midpoint of the chain-melting transition of the membrane lipid dispersion is shifted to a higher temperature, exhibits a greater enthalpy, and is considerably less cooperative than in cells or membranes, which suggests that native membrane lipid organization has been perturbed during extraction and resuspension of the membrane lipids in water. The thermal denaturation of the proteins in the cells and membranes has absolutely no effect on the peak temperature or cooperativity of the lipid phase transition. However, about 15% of the lipids do not participate in the cooperative gel-to-liquid-crystalline phase transition in both the native and heat-denatured membranes, presumably because their cooperative phase behavior is abolished by interaction with the transmembrane regions of integral membrane proteins. Alternatively, a larger proportion of the membrane lipids may interact with the membrane proteins but have their cooperative melting behavior only partially perturbed, which thereby leads to the 15% reduction in the transition enthalpy observed. The fact that the gel-to-liquid-crystalline lipid phase transition in cells and membranes exhibits a similar temperature maximum and a higher cooperativity than does the membrane lipid dispersion favors the former interpretation.

Figure 5. High-sensitivity DSC heating scans of Acholeplasma laidlawii B elaidic acid-homogeneous intact cells, isolated membranes and extracted total membrane lipids dispersed as multilamellar vesicles in water.

The presence of high levels of cholesterol in many eukaryotic membranes, particularly plasma membranes, abolishes a discrete cooperative gel-to-liquid-crystalline membrane lipid phase transition in these systems. Thus, no lipid phase transitions could be detected by DSC in the cholesterol-rich erythrocyte (40) or myelin (41) membranes. The thermotropic behavior of rat liver microsomal membranes, which contain moderate levels of cholesterol, has been studied by DSC. An early study using conventional DSC revealed a single reversible, broad phase transition occurring between — 15°C and +5°C in both intact membranes and isolated lipids (42). A more recent high-sensitivity DSC study confirmed the absence of a reversible phase transition above 0°C (43). Rat liver mitochondrial membranes, which are low in cholesterol, have been studied by several groups using DSC and other techniques. The earliest work with whole mitochondrial revealed a reversible broad gel-to-liquid-crystalline phase transition centered at 0° C in mitochondrial membranes and in extracted lipids (42). A later study of both intact mitochondria and of isolated inner and outer membranes confirmed these results, except that the outer membrane transition seemed to occur at a slightly lower temperature than did the inner membrane transition (44). However, a more recent study of the rat liver inner mitochondrial membrane reported a narrower membrane lipid transition centered near +10°C; by artificially increasing cholesterol content some 10-fold to about 30 mol %, the inner membrane gel-to-liquid-crystalline phase transition could be lowered and broadened, and its ∆Hcal reduced to less than one tenth that of the native membrane (45). It has also been reported that in beef heart mitochondrial inner membranes, a broad reversible endothermic phase transition centered at —10° C occurs.

DSC has been used to study the individual protein components of biological membranes of relatively simply protein composition and the interaction of several of these components with lipids and with other proteins. The red blood cell membrane, which has been most intensively studied, exhibits five discrete protein transitions, each of which has been assigned to a specific membrane protein. The response of each of these thermal transitions to variations in temperature and pH as well as to treatment with proteases, phospholipases, specific labelling reagents, and modifiers and inhibitors of selected membrane activities, has provided much useful information on the interactions and functions of these components in the intact erythrocyte membrane (46-49). Similar approaches have been applied to the bovine rod outer segment membrane (50) and to the spinach chloroplast thylakoid membrane (51).

Acknowledgments

Research performed in the authors’ laboratory has been supported by operating and major equipment grants from the Canadian Institutes of Health Research, operating grants from the Natural Sciences and Engineering Research Council of Canada, and by major equipment and personnel support grants from the Alberta Heritage Foundation for Medical Research.

References

1. McElhaney RN. The effect of membrane lipids on permeability and transport in prokaryotes. In: Structure and Properties of Cell Membranes, Volume 2. Benga G, ed. 1985. CRC Press, Boca Raton, FL. pp. 19-52.

2. Gennis RB. Biomembranes. Molecular Structure and Function. 1989. Springer-Verlag, New York.

3. Yeagle PL. The Structure and Function of Biological Membranes, 2nd edition. 2004. CRC Press, Boca Raton, FL.

4. Sandermann H. Regulation of membrane enzymes by lipids. Biochim. Biophys. Acta 1978; 515:209-237.

5. McElhaney RN. Effects of membrane lipids on transport and enzymatic activities. In: Current Topics in Membranes and Transport, Volume 17. Razin S, Rottem S, eds. 1982. Academic Press, New York. pp. 317-380.

6. McElhaney RN. Membrane lipid fluidity, phase state and membrane function in prokaryotic microorganisms. In: Membrane Fluidity in Biology, Volume 4. Aloia RA, Boggs JM, eds. 1985. Academic Press, New York. pp. 147-208.

7. McElhaney RN. The influence of membrane lipid composition and physical properties on membrane structure and function in Acholeplasma laidlawii. CRC Crit. Rev. Microbiol. 1989; 17:1-32.

8. McElhaney RN. Membrane function. In: Mycoplasmas; Molecular Biology and Pathogenesis. 1992. American Society for Microbiology, Washington, DC. pp. 259-287.

9. Lee AG. Lipids and their effects on membrane proteins: evidence against a role for fluidity. Prog. Lipid Res. 1991; 30:323-348.

10. Andersen OS, Koeppe RE. Bilayer thickness and membrane protein function: an energetic perspective. Annu. Rev. Biophys. Biomol. Struct. 2007; 36:107-130.

11. Cevc G, Marsh H. Phospholipid Bilayers: Physical Principles and Models. 182. John Wiley and Sons, New York.

12. Lewis RNAH, McElhaney RN. The Mesomorphic Phase Behavior of Lipid Bilayers. In: The Structure of Biological Membranes, 2nd edition. Yeagle, Philip, L, ed. 2004. CRC Press, Boca Raton, FL. pp. 53-120.

13. Seguin C, Lewis RNAH, Mantsch HH, McElhaney RN. Calorimetric studies of the thermotropic phase behavior of cells, membranes and lipids from fatty acid-homogeneous Acholeplasma laidlawii B. Israel. J. Med. Sci. 1987; 23:403-407.

14. Epand RM. Lipid Polymorphism and Membrane Properties. 1997. Academic Press, San Diego, CA.

15. Lewis RNAH, Mannock DA, McElhaney RN. Membrane Lipid Molecular Structure and Polymorphism. In: Lipid Polymorphism and Membrane Properties. Epand RM, ed. 1997. Academic Press, San Diego, CA. 25-102.

16. McElhaney RN. The use of differential scanning calorimetry and differential thermal analysis in studies of model and biological membranes. Chem. Phys. Lipids 1982; 30:229-259.

17. McElhaney RN. Differential scanning calorimetric studies of lipid-protein interactions in model membrane systems. Biochim. Biophys. Acta 1986; 864:361-421.

18. Lewis RNAH, Mannock DA, McElhaney RN. Differential scanning calorimetry in the study of lipid phase transitions in model and biological membranes: practical considerations. In: Methods in Membrane Lipids. Dopico AM, ed. 2007. Humana Press, To- towa, NJ. pp. 171-195.

19. Heerklotz PDH. Pressure perturbation calorimetry. In: Methods in Membrane Lipids. Dopico AM, ed. 2007. Humana Press, Totowa, NJ. pp. 197-206.

20. Chen SC, Sturtevant JM, Gaffney BJ. Scanning calorimetric evidence of a third phase transition in phosphatidylcholine bilayers. Proc. Natl. Acad. Sci. U.S.A. 1980; 77:5060-5064.

21. Mabrey S, Sturtevant JM. High-sensitivity differential scanning calorimetry in the study of biomembranes and related model systems. In: Methods in Membrane Biology, Volume 9. Korn ED, ed. 1978. Plenum Press, New York. pp. 237-274.

22. Marsh D. CRC Handbook of Lipid Bilayers. 1990. CRC Press, Boca Raton, FL.

23. Yeagle PL, ed. The Biology of Cholesterol. 1988. CRC Press, Boca Raton, FL.

24. Finegold L, ed. Cholesterol in Model Membranes. 1993. CRC Press, Boca Raton, FL.

25. McMullen TPW, McElhaney RN. Physical studies of cholesterol/phospholipid interactions. Curr. Opin. Colloid Interface Sci. 1996; 1:83-90.

26. McMullen TPW, Lewis RNAH, McElhaney RN. Differential scanning calorimetric study of the effect of cholesterol on the thermotropic phase behavior of a homologous series of linear saturated phosphatidylcholines. Biochemistry 1993; 32:516-522.

27. Jain MK, Wu NW. Effect of small molecules on the dipalmitoyl lecithin liposomal bilayer. III. Phase transitions in lipid bilayers. J. Membr. Biol. 1977; 34:157-241.

28. Lewis RNAH, Zhang Y-P, Liu F, McElhaney RN. Mechanisms of the interaction of a-helical transmembrane peptides with phospholipid bilayers. Bioelectrochemistry 2002; 56:135-140.

29. de Planque MRR, Killian JA. Protein-lipid interactions studied with designed transmembrane peptides: role of hydrophobic matching and interfacial anchoring (Review). Mol. Membrane Biol. 2003; 20:271-284.

30. Lohner K, Prenner EJ. Differential scanning calorimetry and X-ray diffraction studies of the specificity of the interaction of antimicrobial peptides with membrane-mimetic systems. Biochim. Biophys. Acta 1999; 1462:141-156.

31. Prenner EJ, Lewis RNAH, McElhaney RN. The interaction of the antimicrobial peptide gramicidin S with lipid bilayer model and biological membranes. Biochim. Biophys. Acta 1999; 1462:201-221.

32. Papahadjoupoulos D, Moscarello M, Eylar EH, Isaac T. Effects of proteins on thermotropic phase transitions of phospholipid membranes. Biochim. Biophys. Acta 1975; 401:317-335.

33. Epand RM. Diacylglycerols, lysolecithin, or hydrocarbons markedly alter the bilayer to hexagonal phase transition temperature of phosphatidylethanolamines. Biochemistry 1985; 24:7092-7095.

34. Lee Y-C, Taraschi TF, Janes N. Support for the shape concept of lipid structure based on a headgroup volume approach. Biophys. J. 1993; 65:1429-1432.

35. Foht PJ, Quynh MT, Lewis RNAH, McElhaney RN. Quantitation of the phase preferences of the major lipids of the Acholeplasma laidlawii B membrane. Biochemistry 1995; 34:13811-13817.

36. Lewis RNAH, McElhaney RN. Acholeplasma laidlawii B membranes contain a lipid (glycxerylphosphoryldiglucosyl diacylgly- cerol) which forms micellar rather than lamellar or reversed phases when dispersed in water. Biochemistry 1995; 34:13818-13824.

37. Steim JM, Tourtellotte ME, Reinert JC, McElhaney RN, Rader RL. Calorimetric evidence for the liquid-crystalline state of lipids in a biomembrane. Proc. Natl. Acad. Sci. U.S.A. 1969;63:104-109.

38. Reinert JC, Steim JM. Calorimetric detection of a membrane lipid phase transition in living cells. Science 1970; 168:1580-1582.

39. Melchior DL, Morowitz HJ, Sturtevant JM, Tsong TY. Characterization of the plasma membrane of Mycolplasma laidlawii. VIII. Phase transitions of membrane lipids. Biochim. Biophys. Acta 1970; 219:114-122.

40. Labrooke BD, Williams RM, Chapman D. Studies on lecithin- cholesterol-water interactions by differential scanning calorimetry and x-ray diffraction. Biochim. Biophys. Acta 1968; 150:333-340.

41. Ladbrook BD, Jenkinson TJ, Kamat VB, Chapman D. Physical studies of myelin. I. Thermal analysis. Biochim. Biophys. Acta 1968; 164:101-109.

42. Blazyk JF, Steim JM. Phase transitions in mammalian membranes. Biochim. Biophys. Acta 1972; 266:737-741.

43. Mabrey S, Powis G, Schenkman JB, Tritton TR. Calorimetric study of microsomal membranes. J. Biol. Chem. 1977; 252:2929-2933.

44. Hackenbrock CR, Hochli M, Chau RM. Calorimetric and freeze fracture analysis of lipid phase transitions and lateral translational motion of intramembrane particles in mitochondrial membranes. Biochim. Biophys. Acta 1976; 455:466-484.

45. Madden TD, Vigo C, Bruckdorfer KR, Chapman D. The incorporation of cholesterol into mitochondrial membranes and its effect on lipid phase transition. Biochim. Biophys. Acta 1980; 599:528-537.

46. Brandts JF, Erickson L, Lysko K, Schwartz AT, Taverna RD. Calorimetric studies of the structural transitions of the human erythrocyte membrane. The involvement of spectrin in the A transition. Biochemistry 1977; 16:3450-3454.

47. Brandts JF, Taverna RD, Sadasivan E, Lysko KA. Calorimetric studies of the structural transitions of the human erythrocyte membrane. Studies of the B and C transitions. Biochim. Biophys. Acta 1978; 512:566-578.

48. Snow JW, Brandts JF, Low PS. The effects of anion transport inhibitors on structural transitions in erythrocyte membranes. Biochim. Biophys. Acta 1978; 512:579-591.

49. Snow JW, Vincentelli J, Brandts JF. A relationship between anion transport and a structural transition of the human erythrocyte membrane. Biochim. Biophys. Acta 1981; 642:418-428.

50. Miljanich G, Brown MF, Dratz EA, Mabrey SV, Sturtevant JM. Calorimetric studies of the retinal rod outer segment membrane. Biophys. J. 1976; 16:37a.

51. Cramer WA, Whitmarsh J, Low PS. Differential scanning calorimetry of chloroplast membranes; identification of an endothermic transition association with the water-splitting complex of photosystem II. Biochemistry 1981; 20:157-162.

52. Lewis RNAH, Mak N, McElhaney RN. A differential scanning calorimetric study of the thermotropic phase behavior of model membranes composed of phosphatidylcholines containing linear saturated fatty acyl chains. Biochemistry 1987; 26:6118-6126.