INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Cholesterol is a major constituent of plasma membranes of mammalian cells, contributing to ~25 mol % of the lipid fraction in the human erythrocyte membrane, for example. It is now well established that cholesterol causes profound changes in the physical properties of membranes. Investigations on dipalmitoylphosphatidylcholine (DPPC)/cholesterol (chol) mixtures have shown the formation of a new phase at cholesterol concentrations above ~22 mol %, designated as a liquid-ordered phase (Ipsen et al., 1987; Vist and Davis, 1990; McMullen and McElhaney, 1995). This phase shares characteristics with liquid-crystalline and gel phases. It is characterized by axially symmetric motion and lateral diffusion, similar to those of the liquid-crystalline phase, but high orientational order of the acyl chains, similar to the gel phase. Such a liquid-ordered phase appears to exist over a wide temperature range in DPPC containing more than 22 mol % cholesterol (Ipsen et al., 1987; Vist and Davis, 1990) although some structural changes may occur upon heating (Reinl et al., 1992; McMullen and McElhaney, 1995). The formation of a similar liquid-ordered phase has been proposed for other phosphatidylcholines/cholesterol systems (Thewalt and Bloom, 1992; Linseisen et al., 1993). This phase generates a great interest because it is proposed to be more representative of plasma membrane organization.

Therefore, it is important to examine whether the formation of a liquid-ordered phase is a general trend of a phospholipid/cholesterol system or whether it is a singular behavior with phosphatidylcholines (PCs). As phosphatidylethanolamines (PEs) are the second most abundant phospholipid in animal cell membranes, we have investigated the influence of cholesterol on its polymorphism. Previous studies have examined the behavior of PE/chol mixtures by differential scanning calorimetry (DSC) (Blume, 1980; Ghosh and Seelig, 1982; Epand and Bottega, 1987; Cheetham et al., 1989; Takahashi et al., 1996; McMullen and McElhaney, 1997), solid-state nuclear magnetic resonance (NMR) spectroscopy (Brown and Seelig, 1978; Blume and Griffin, 1982; Ghosh and Seelig, 1982; Marinov and Dufourc, 1995; Marinov and Dufourc, 1996), and x-ray diffraction (Cheetham et al., 1989; Takahashi et al., 1996). Because of its molecular properties, unsaturated PE can generally form gel lamellar (L), liquid-crystalline lamellar (L), and inverted hexagonal (H_II) phases. In the lamellar phase, cholesterol is reported to disorder the L phase whereas it orders the phospholipids in the L phase (Brown and Seelig, 1978; Ghosh and Seelig, 1982; Blume and Griffin, 1982). The L-to-L phase transition is broadened and shifted toward low temperatures by cholesterol and is abolished for high cholesterol content for most PEs (between 30 and 50 mol %, depending on the phospholipid) (Brown and Seelig, 1978; Ghosh and Seelig, 1982; Blume and Griffin, 1982; Marinov and Dufourc, 1996; Takahashi et al., 1996; McMullen and McElhaney, 1997). For intermediate cholesterol proportions, the coexistence of a gel and a fluid phase has been reported below the L-to-L phase transition temperature (Tm) of the pure PE (Blume and Griffin, 1982; Takahashi et al., 1996). Cholesterol has also been shown to influence the formation of nonlamellar phases. Previous studies have reported that a limited amount of cholesterol leads to a decrease of the lamellar-to-hexagonal phase transition temperature (Th), but proportions of cholesterol higher than 30 mol % in PE reverse this effect and Th increases (Epand and Bottega, 1987; Cheetham et al., 1989; Takahashi et al., 1996). Cholesterol appears to reduce the intercylinder spacing in the H_II phase (Cheetham et al., 1989; Takahashi et al., 1996). Up to now, very little has been known about the molecular details of the changes induced by cholesterol in PE matrices.

In this study, we use Fourier transform infrared (FTIR) and ²H-NMR spectroscopy to examine 1-palmitoyl-2-oleoyl-sn-phosphatidylethanolamine (POPE)/chol polymorphism. This phospholipid is particularly interesting as both Tm and Th are fairly accessible (Tm = 25°C and Th = 71°C; Epand and Bottega, 1987). In addition, a perdeuterated saturated acyl chain at the sn-1 position of the phospholipid constitutes a nice probe of the orientational order along the lipid chains. We have examined the effect of cholesterol on POPE acyl chain order and carbonyl hydration in the three phases and attempt to relate these molecular changes to the polymorphism. In addition, we have also addressed the question of whether cholesterol induces the formation of a liquid-ordered phase with POPE. These results lead us to propose a new temperature-composition diagram for the POPE/cholesterol system.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

POPE and 1-perdeuteriopalmitoyl-2-oleoyl-sn-phosphatidylethanolamine (POPE-d₃₁) were purchased from Avanti Polar Lipids (Birmingham, AL) and used without further purification. Cholesterol was obtained from Sigma Chemical Co. (St. Louis, MO). Deuterium-depleted water was purchased from Aldrich (Milwaukee, WI).

Samples for FTIR were prepared as follows: 5 mg of POPE dissolved in a benzene/methanol (96:04) mixture was mixed with an appropriate volume of cholesterol in solution in the same solvent to yield the desired molar ratio. The solvent was removed by lyophilization. The powder was dispersed in 20 µl of 20 mM Hepes buffer containing 2 mM EDTA and 100 mM NaCl, pH 7.4. The sample was then shaken on a vortex mixer, heated to 40°C, and cooled down to liquid nitrogen temperature. This cycle was repeated at least three times to ensure sample homogeneity. The sample was put between two CaF₂ windows spaced out by a 5-µm Teflon spacer. The cell was mounted in a brass sample holder the temperature of which was computer controlled using thermopumps (Pézolet et al., 1983). The spectra were recorded on a Bio-Rad FTS-25 FTIR spectrometer equipped with a mercury-cadmium-telluride detector. For each spectrum, 100 scans were collected with 2-cm¹ nominal resolution.

To eliminate water contribution (_O-H  3400 cm¹) in the methylene stretching region, a polynomial was fitted to simulate the edge of the water band and then subtracted in the _C---H region. The cholesterol contribution in this region was also eliminated (Kodati and Lafleur, 1993). In the carbonyl region, correction for the contribution of the water deformation band was made by subtracting the buffer spectrum recorded at the same temperature. The Fourier self-deconvolution algorithm in the GRAMS software (Galactic Industries Corp., Salem, NH) was used to highlight the components of the carbonyl band.

Samples for ²H-NMR spectroscopy were prepared the same way as for FTIR. However, the amount of lipid used was 30 mg and the lipid was perdeuterated on the palmitoyl chain. The buffer was prepared with deuterium-depleted water. ²H-NMR spectra were acquired on a Bruker DSX-300 spectrometer with a probe equipped with a 5-mm solenoid coil. The quadrupolar echo sequence was used with a 90° pulse between 2.7 and 3.0 µs, an interpulse delay of 35 µs, and a recycling time of 0.5 s. After the second pulse, 8192 points were acquired in quadrature with a dwell time of 0.5 µs. The number of scans was at least 20,000. The sample temperature was regulated using a Bruker variable temperature controller. Spectral de-Pakeing was performed as described by Sternin et al. (1983). The quadrupolar splittings obtained from the ²H-NMR spectra provide a measurement of the anisotropy of C---D bond motion. The order parameter of the C---D bond, S_C---D, is expressed by where A_q is the static quadrupolar splitting constant (167 kHz for a C---D bond), _q is the quadrupolar splitting measured for a lipid oriented at 90° relative to the magnetic field, and  is the angle between the C---D bond and the bilayer normal. The brackets represent the averaging over the NMR time scale. The de-Paked spectra were used to determine the smoothed orientational order profile, according to a procedure described elsewhere (Lafleur et al., 1989). This approach assumes a monotonic decrease of order along the chain and therefore reproduces only the smoothed features of the order variation (Lafleur et al., 1989).

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Infrared spectroscopy

Thermotropism as probed by _C---H

1

1

1

View larger version (19K): [in this window] [in a new window]   FIGURE 1   Variations of the symmetric C---H band position as a function of temperature for pure POPE (), POPE/20 mol % chol (), and POPE/45 mol % chol ().

View larger version (17K): [in this window] [in a new window]   FIGURE 2   Influence of cholesterol on the lamellar-to-hexagonal phase transition temperature (±1°C; ) and on the amplitude of the symmetric C---H frequency shift during this transition (±0.03 cm1; ).

Carbonyl bands

CO

1

1

CO

View larger version (14K): [in this window] [in a new window]   FIGURE 3   Carbonyl stretching regions for pure POPE (), POPE/20 mol % chol (- - -), and POPE/45 mol % chol (· · ·) in the gel (5°C), the fluid lamellar (40°C), and the HII (78°C) phase. The spectra have been corrected for water contribution and Fourier self-deconvoluted as described in Materials and Methods.

NMR spectroscopy

²H-NMR spectroscopy has been used to identify the different phases formed by the lipid systems. It is a powerful technique because the lamellar gel, the lamellar fluid, and the H_II phases provide distinct ²H-NMR signatures. Fig. 4 displays typical spectra obtained from our samples. For pure POPE-d₃₁ (bottom line), three different profiles are observed. At 5 and 15°C, broad and almost featureless spectra are obtained; these are typical of the gel phase where the chains are almost all-trans and the rotational diffusion is slow (Davis, 1983). At 32°C, the spectrum is typical of the fluid lamellar phase, where the fast rotation of the lipid along the bilayer normal (fast on the NMR time scale) gives rise to a set of powder patterns typical of axially symmetric systems, representative of the orientational order profile along the lipid acyl chains (Davis, 1983; Lafleur et al., 1989). At 70°C, the spectrum still indicates the axial symmetry of the lipid motion, but the width of the spectrum is reduced by a factor larger than two essentially due to the lipid diffusion around the H_II cylinder and the increased motional freedom (Lafleur et al., 1990b). The resolution of the powder patterns associated with the different CD₂ groups along the perdeuterated chain is reduced due to the more linear decrease of the order along the acyl chain (Lafleur et al., 1990b). The presence of cholesterol induces several changes in the spectra. At 5 and 15°C, cholesterol leads to the transformation of the broad gel-phase spectrum to a broad spectrum characteristic of lipids experiencing axially symmetric motion. Actually, at 45 mol % cholesterol, the spectra recorded at these temperatures are exclusively composed of a fluid component. At 5°C, the spectrum shows a _q of 50 kHz for the outermost doublet. A similar shape change of the NMR signal has also been observed for specifically labeled 2[4,4-²H]dipalmitoylphosphatidylethanolamine (2[4, 4-²H]DPPE) in the presence of cholesterol (Blume and Griffin, 1982). We observe that the transformation toward a spectrum typical of a fluid phase species appears to be completed at lower temperatures for higher cholesterol content (e.g., comparing 10 and 25 mol % at 15°C). At intermediate temperature and cholesterol content, the spectra show the coexistence of a gel-like and a fluid-like component (POPE/25 mol % chol mixture at 5°C and POPE/10 mol % chol mixture at 15°C, for example). The two components of such spectra can be isolated by spectral differences as previously shown for the DPPC/chol system (Vist and Davis, 1990). Fig. 5 illustrates an example of the process. The spectra of POPE-d₃₁ containing 10 and 15 mol % cholesterol, recorded at 15°C, are composed of a gel and a fluid component; this is especially clear by looking at the central signal associated with the terminal methyls. The components obtained by spectral difference are displayed on the right side. From the subtractions, it is possible to estimate the cholesterol content corresponding to each component (Vist and Davis, 1990). At 15°C, our results indicate a cholesterol content of 5 and 20 mol % for the gel and the fluid component, respectively. These values should correspond to the borders of the coexistence region. This spectral difference approach was done at 5, 10, and 15°C.

View larger version (16K): [in this window] [in a new window]   FIGURE 4   2H-NMR spectra of POPE-d31 containing various proportions of cholesterol (as indicated for each row), recorded at different temperatures (as indicated at the top of each column). Note that the frequency scale is different for the right column.

View larger version (14K): [in this window] [in a new window]   FIGURE 5   2H-NMR spectra of POPE-d31/10 mol % chol and POPE-d31/15 mol % chol at 15°C (left) and difference spectra (right). The gel component is obtained from the following subtraction: spectrum of POPE-d31/10 mol % chol  0.53 spectrum of POPE-d31/15 mol % chol. The fluid component is obtained from the following subtraction: spectrum of POPE-d31/15 mol % chol  0.48 spectrum of POPE-d31/10 mol % chol.

The third column of Fig. 4 shows the spectra of the various POPE-d₃₁/chol mixtures recorded at 32°C. All of these are typical of lamellar-phase lipids experiencing axially symmetric motion. Cholesterol leads to an increase of the quadrupolar splittings. For example, the quadrupolar splitting of the outermost doublet increases from 35 to 49 kHz when cholesterol content increases from 0 to 45 mol %. Finally, at 70°C, all of the spectra show a shape typical of the H_II phase. Similar to the behavior in the fluid lamellar phase, the presence of cholesterol leads to an increase of quadrupolar splittings as illustrated by the widening of the signal.

From the spectra in the fluid lamellar and the H_II phases, it is possible to extract smoothed order profiles that describe the variation of orientation order along the lipid acyl chains (Lafleur et al., 1989). The order profiles are shown for the fluid lamellar phase (32°C; Fig. 6 A) and the H_II phase (70°C; Fig. 6 B). The order profiles obtained at 32°C are typical of the lamellar phase, showing a plateau region where the order parameter does not vary considerably followed by a rapid decrease of the order toward the middle of the bilayer. The order profile of POPE-d₃₁ is similar to that previously published (Lafleur et al., 1990a). As it has been observed for PCs (Vist and Davis, 1990; Lafleur et al., 1990a), cholesterol leads to an increase of S_C---D(n) all along the chain of POPE-d₃₁. This result is in agreement with a previous study on 2[4,4-²H]DPPE showing an increase of _q in the presence of cholesterol for a temperature greater than Tm of the phospholipid (Blume and Griffin, 1982). The change of overall order parameter of POPE-d₃₁, S_C---D, as a function of cholesterol content is shown in Fig. 7 A. S_C---D represents the arithmetic average of S_C---D(n) for 2  n  16, S_C---D(16) being linearly extrapolated from S_C---D(14) and S_C---D(15) to avoid the contribution of the additional motion of the terminal CD₃ methyl groups. Results obtained for POPC-d₃₁ in the presence of cholesterol (Lafleur et al., 1990a) are reproduced for comparison. The order of pure POPE-d₃₁ is greater than that observed for POPC-d₃₁ due to the fact that the smaller effective size of the PE polar headgroup leads to a tight lipid packing and, as a consequence, restricted chain motions (Perly et al., 1985; Lafleur et al., 1990a). Cholesterol induces an ordering of the acyl chains for both phospholipids, but the magnitude of the increase of S_C---D is more pronounced for POPC-d₃₁ than for POPE-d₃₁. From 0 to 45 mol % cholesterol, S_C---D increases by ~37% for POPE-d₃₁ and by ~85% for POPC-d₃₁. Similarly, POPE-d₃₁ and POPC-d₃₁/20 mol % chol have very similar orientational chain order at ~32°C, a value of S_C---D of ~0.215. The addition of an extra 10 mol % cholesterol to these two bilayers leads to a value of 0.232 and 0.250 for the POPE-containing and the POPC-containing bilayers, respectively. The chain order becomes very similar for both lipids when the cholesterol content is at ~45 mol %. In these conditions, S_C---D is ~0.29 and the order parameter at the plateau region is ~0.4, indicating very ordered lipid chains.

View larger version (24K): [in this window] [in a new window]   FIGURE 6   Smoothed orientational order profile obtained at 32°C (A) and 70°C (B) for POPE-d31 (), POPE-d31/10 mol % chol (), POPE-d31/25 mol % chol (), and POPE-d31/45 mol % chol ().

View larger version (19K): [in this window] [in a new window]   FIGURE 7   Variation of SC---D as a function of cholesterol content in the lamellar phase (A) for POPE-d31 (32°C; ) and POPC-d31 () (from Lafleur et al., 1990a, at 30°C) and in the HII phase (B) for POPE-d31 (70°C; ).

As can be seen in Fig. 6 B, the variation of order along the phospholipid acyl chain is more linear in the H_II phase, in agreement with previous observations (Lafleur et al., 1990b; Thurmond et al., 1990). Our results show that cholesterol maintains its ordering effect on the lipid acyl chains in the H_II phase. The increase of S_C---D(n) is observed all along the chain. The addition of 45 mol % chol to POPE-d₃₁ at 70°C, i.e., in the H_II phase, leads to an increase of S_C---D by 47%, from 0.056 to 0.081. The increase of S_C---D as a function of cholesterol content in the H_II phase is summarized in Fig. 7 B.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Lamellar phases

At temperatures below Tm of pure POPE, both IR and NMR results indicate that cholesterol perturbs the gel phase. By IR spectroscopy, the increase of _C---H frequency suggests an increase of conformational disorder whereas the _CO region indicates an increased hydration at the carbonyl level. The ²H-NMR results indicate that cholesterol increases the motion rate of the gel-phase lipids, giving rise to spectra associated with axially symmetric motions, similarly to previous observations on PC/chol systems (Vist and Davis, 1990; Thewalt and Bloom, 1992). At temperatures higher than Tm of pure POPE, our results indicate that cholesterol increases the order of the fluid-phase lipids. This is shown by the decrease of the _C---H frequency in the presence of cholesterol and the increase of orientational lipid chain order. This chain ordering effect causes also the reduction of the hydration of the carbonyl groups, in agreement with a previous study on low-hydration POPE samples (Marinov and Dufourc, 1996). Our results on POPE/chol system lead to the same conclusion as that proposed in the pioneer studies on the effect of cholesterol on PC bilayers (Umemura et al., 1980; Cortijq and Chapman, 1981): the sterol disorders the gel phase and orders the fluid phase.

These opposite effects lead to the abolition of the gel-to-liquid crystalline phase transition. In the IR data, the sharp shift of _C---H is no longer observed for POPE/45 mol % chol mixture, and in the NMR data, the spectra of this mixture show axial symmetry over a wide temperature range. The decrease of the transition amplitude and its eventual disappearance in the presence of cholesterol have been previously reported for various unsaturated PEs, using DSC (Epand and Bottega, 1987; Cheetham et al., 1989; McMullen and McElhaney, 1997), x-ray diffraction (Takahashi et al., 1996), and NMR (Ghosh and Seelig, 1982; Blume and Griffin, 1982). At high cholesterol content, the ²H-NMR spectra indicate that the PE molecules in the mixture have axially symmetric motion relative to the normal of the bilayer, like in a fluid phase, but the values of the order parameters are high, indicating ordered chains. These characteristics are strikingly similar to those of the proposed liquid-ordered phase previously described for PCs (Vist and Davis, 1990; Thewalt and Bloom, 1992; Linseisen et al., 1993).

These findings are combined in the temperature-composition diagram shown in Fig. 8. Below Tm, three regions are defined: a gel phase, a fluid-ordered phase, and a coexistence region. The diagram differs from the one derived from DSC data (Epand and Bottega, 1987). This is somehow expected as, at high cholesterol concentration, the transition is abolished and, as a consequence, the calorimetry data provide little information. The lower part of the proposed temperature-composition diagram shows similar trends to the one proposed for the 1,2-dipalmitoyl-sn-phosphatidylethanolamine (DPPE)/chol system (Blume and Griffin, 1982). We have included in the temperature-composition diagram the end points obtained from spectral subtraction (as exemplified in Fig. 5) as well as symbols reporting the phase(s) present in the sample, as determined by visual inspection of the ²H-NMR spectra. In general, both approaches are consistent. This good agreement is shown by the fact that the difference spectrum representative of one phase and the experimental spectrum of a sample with the corresponding amount of cholesterol were always very similar. For the liquidus line, there is a good agreement between the limits obtained from the spectral difference and the limits that can be inferred from the visual inspection of the spectra. For the solidus line, the subtraction of the spectrum of the gel component from an experimental spectrum from a sample with the corresponding cholesterol content led to very little residual signal. However, a close inspection of the experimental spectra with cholesterol content defined by spectral subtraction indicates a slightly different position for the border between the L and the coexistence regions. The case for which the discrepancy was the most pronounced was for the gel end point at 15°C; the cholesterol content for the end point was estimated to be 5 mol % by spectral subtraction, but the experimental spectrum of the sample containing 5 mol % chol displayed a contribution of fluid lipids evaluated to be ~9% based on the relative area. This could be due to the fact that, at this temperature, the sample containing 15 mol % chol is close to the liquidus line; in such a situation, the fast lipid exchange between the phases may affect the signal shape, and the spectra close to the boundaries should be considered with caution (Blume and Griffin, 1982). This effect may lead to some deviations as discussed previously (Morrow et al., 1991; Linseisen et al., 1993).

View larger version (25K): [in this window] [in a new window]   FIGURE 8   Proposed temperature-composition diagram for POPE/chol system. , limit of a transition as detected by IR spectroscopy; , obtained by spectral subtraction as illustrated in Fig. 5; , conditions for which a spectrum typical of a gel phase is recorded; , conditions for which a spectrum typical of a fluid phase is recorded; , conditions for which a spectrum typical of the coexistence of a gel and a fluid phase is recorded, as determined by visual inspection. The temperatures associated to the NMR data were systematically increased by 3.5°C, the difference between the Tm of POPE and POPE-d31, to include on the same figure the NMR and IR data. The error bars represent the composition uncertainty related to the determination of the subtraction factor. The lines on the diagram define regions over which the same phase(s) is(are) identified. These phase boundaries should be regarded with caution considering the limitations of the techniques used, as discussed in the paper; these include the impossibility of identifying the coexistence of fluid lamellar phases and the fact that the indirect information relative to the lamellar-to-HII phase transition obtained from IR spectroscopy leads to a coexistence region that does not respect the phase rules, suggesting a more complex behavior.

Our results in the lamellar phase of the POPE/chol system reveal several features. First, at cholesterol content greater than ~35 mol %, POPE molecules form a fluid phase where the acyl chains are very ordered. From both the NMR and IR spectroscopy point of view, the organization of POPE bilayers containing 35 mol % chol is very similar to that observed with PCs. Inspired by recent papers on PC/chol systems (Vist and Davis, 1990; Thewalt and Bloom, 1992; Linseisen et al., 1993), this region of the temperature-composition diagram is referred to as a liquid-ordered phase. Therefore, the formation of such a phase does not seem to be exclusive to PC species, and this work extends this behavior to POPE. Second, our results suggest that the coexistence region of the gel and the liquid-ordered phases is different compared with that observed with PCs. The data do not suggest the presence of an eutectic, as proposed by Vist and Davis (1990) for the DPPC/chol system. The transition from a gel to a fluid phase of POPE-d₃₁, as probed by the shape of the ²H-NMR spectra, is broadened progressively upon the addition of cholesterol. The transition occurs over 6°C when the sample contains as little as 3 mol % chol and over more than 10°C when the cholesterol proportion is 5 mol %. The proposed phase diagrams for several PC/chol systems (Ipsen et al., 1987; Vist and Davis, 1990; Thewalt and Bloom, 1992; Linseisen et al., 1993) also include a three-phase horizontal line representing the coexistence of the L, L, and liquid-ordered phases. This line stretches roughly from 5 to 20% chol. A similar line is proposed in the temperature-composition diagram of the DPPC/chol system proposed by McMullen and McElhaney (1995) except it begins at 0 mol % chol. Upon heating a sample, crossing this line is associated with the sharp peak observed by DSC. Our results do not suggest the presence of a similar horizontal line for the POPE/chol system. The completion of the gel-to-fluid phase transition is observed at lower temperatures when POPE contains more than 10 mol % chol. Therefore, a three-phase line is much shorter or absent in the POPE/chol temperature-composition diagram. This conclusion is supported by the DSC results obtained with unsaturated PE/chol systems (Epand and Bottega, 1987; Takahashi et al., 1996; McMullen and McElhaney, 1997). The addition of cholesterol to PE shifts the endotherm associated with the gel-to-fluid phase transition toward lower temperatures, and no sharp component is observed at the Tm of the pure phospholipid when cholesterol content is higher than 10 mol %. The liquidus boundary shows also a different shape for cholesterol proportion higher than 10 mol % compared with those in the proposed phase diagrams of PC/chol systems (Vist and Davis, 1990; Thewalt and Bloom, 1992; Linseisen et al., 1993). For the POPE/chol system, the proportion required to eliminate the gel-phase component from the spectrum is dependent on temperature; below Tm of the pure lipid, smaller proportions of cholesterol are needed to obtain exclusively a spectrum typical of a fluid phase at higher temperatures, as previously observed for the DPPE/chol system (Blume and Griffin, 1982). It is somehow different from the behavior reported for cholesterol-containing PC (Vist and Davis, 1990; Linseisen et al., 1993) for which a relatively constant cholesterol content is necessary to lead to a completely fluid (liquid-ordered) sample. This is clearly observed on the 1-stearoyl-2-elaidoyl-sn-glycero-3-phosphocholine/chol partial phase diagram (Linseisen et al., 1993) for which data were recorded 15°C below Tm.

Two comments should be added. First, the part of the proposed temperature-composition diagram at low cholesterol and around Tm of the pure phospholipid is simpler than those for PCs because POPE does not form a P (ripple) phase. Second, we have not included in the temperature-composition diagram regions where fluid phases coexist even though such regions must be present according to the phase rules. Such regions have been proposed for PC/cholesterol systems (Ipsen et al., 1987; Vist and Davis, 1990; McMullen and McElhaney, 1995). Because it is difficult experimentally to infer the existence of such regions, we make no attempt to define such regions in the POPE/chol temperature-composition diagram. Along the same line, coexisting liquid-ordered phases between which fast lipid exchange would occur cannot be detected by our approach.

From a molecular point of view, our results indicate that the ordering effect of cholesterol is less pronounced on POPE than on POPC. The weaker effect of cholesterol on the POPE matrix relative to the POPC one may be at the origin of the differences observed in the temperature-composition diagrams. It is interesting to note that both phospholipids converge toward similar orientational order at high cholesterol content; at 45 mol %, S_C---D is 0.29 and the orientational order in the plateau region is ~0.4, indicating a very ordered lipid. A 1-palmitoyl-2-oleoyl-phospholipid in a fluid bilayer should have, at a given temperature, an order parameter limit corresponding to a very tight chain packing. The S_C---D values calculated for POPE-d₃₁ and POPC-d₃₁ both containing 45 mol % cholesterol may approach this situation. Because of the relationship between S_C---D and bilayer thickness (Ipsen et al., 1990; Thurmond et al., 1991; Douliez et al., 1995), this implies also approaching a limit thickness for a fluid bilayer. Using the approach proposed by Douliez et al. (1995), we have estimated the effect of cholesterol on chain length for POPE and POPC, at 30°C. We have assumed that the molecular order parameter S_mol equals 1. The chain length of POPE-d₃₁ is 14.4 and 16.1 Å for the pure lipid and the lipid containing 45 mol % chol, respectively; this corresponds to a 12% increase. In the case of POPC-d₃₁, the calculated chain length is 13.3 and 15.9 Å for the pure lipid and the lipid containing 45 mol % chol, respectively, an increase of 20%. These results indicate the more pronounced effect of cholesterol on POPC than on POPE, and the convergence toward a thick bilayer.

Lamellar-to-hexagonal phase transition

Cholesterol also has an influence on the lipid propensities to form nonlamellar phases. Between 0 and 30 mol % cholesterol, the Th of POPE, as detected by IR spectroscopy, is shifted toward lower temperatures whereas a larger amount of cholesterol leads to an increase of Th. This behavior has been already observed for various PEs by DSC (Epand and Bottega, 1987; Cheetham et al., 1989) and x-ray diffraction (Takahashi et al., 1996). These results are included in the temperature-composition diagram (Fig. 8). The symbols represent the beginning and the end of the more abrupt variation of _C---H as a function of temperature, this shift being associated with the lamellar-to-H_II phase transition. The NMR data confirm that, at 70°C, H_II phases are formed by POPE-d₃₁ for cholesterol content varying from 0 to 45 mol %. The lines on the diagram should be considered as guidelines linking the points and do not necessarily correspond to the phase boundaries. The situation is likely more complex in this region of the diagram as the coexistence zone does not respect the phase rules. The present results reveal the influence of cholesterol on the lamellar-to-H_II phase transition of POPE and provide molecular details about this transition. First, the results reveal that cholesterol maintains its ordering effect of the phospholipid acyl chains; this is inferred from the increase of quadrupolar splittings observed by ²H-NMR and from the shift of _C---H toward lower frequencies in the IR spectra. The variation in lipid chain length induced by the presence of cholesterol can be estimated from the order parameters, using a similar approach to that proposed in the L phase (Douliez et al., 1995), as the motions leading to the quadrupolar interaction averaging are similar in both phases. However, the order parameters measured in the H_II phase have to be multiplied by a factor of 2 to take into account the additional averaging introduced by the fast diffusion of the lipid molecules around the H_II cylinders (Thurmond et al., 1990; Lafleur et al., 1996). At 70°C, the chain length is estimated to be 12.5 and 13.4 Å for pure POPE-d₃₁ and POPE-d₃₁/45 mol % chol mixture, respectively. The straightening effect of cholesterol is maintained in the H_II phase, but it is not as pronounced as in the L phase. The tighter lipid packing around the H_II cylinder is accompanied by a limited dehydration of the carbonyl region as observed by IR spectroscopy. A headgroup dehydration of POPE in the presence of cholesterol in the H_II phase has been also suggested by the ²H-NMR of D₂O on low-hydration samples (Marinov and Dufourc, 1996). This region of the IR spectrum shows that, for pure POPE, there are fewer carbonyl groups participating in hydrogen bonds in the H_II phase than in the L as indicated by the weak relative intensity of the low-frequency component. This change during the lamellar-to-hexagonal phase transition is concomitant with a shift of the PO₂ antisymmetric stretching band going from ~1221 cm¹ in the lamellar phase to ~1225 cm¹ in the H_II phase (data not shown). Actually, an abrupt increase of the frequency of this vibrational mode is observed during the lamellar-to-H_II phase transition, in agreement with a previous report on egg PE (Castresana et al., 1992). A shift toward high frequencies has been associated with weaker hydrogen bonding of the phosphate (Sen et al., 1988; Wong and Mantsch, 1988), which is consistent with the dehydration inferred from the carbonyl stretching region. It is therefore expected that cholesterol does not lead to a pronounced dehydration of this phase, which has already limited hydration.

The variation of Th as a function of cholesterol content is not monotonous. Two parameters will affect this transition. First, the presence of cholesterol shifts the amphiphilic balance as this molecule has a small headgroup and a considerable apolar core. In terms of the molecular shape model, cholesterol has a high packing parameter (Israelachvili et al., 1976). In terms of the curvature model, cholesterol should reduce the spontaneous radius of curvature of the lipid layer, R₀. This deduction has been verified with DEPE/chol systems, the x-ray diffraction data showing a decrease of the H_II-intercylinder spacing as a function of cholesterol content (Takahashi et al., 1996). This decrease is fairly linear over the cholesterol content varying from 0 to 40 mol %. This contribution should favor the formation of the H_II phase and seems to be the most significant to explain the decrease in Th observed in the presence of cholesterol. Second, we show here that cholesterol has an ordering effect of the phospholipid acyl chains in the H_II phase, despite the different geometry. It is generally accepted that the disordering of the acyl chains (with increasing temperature, introduced by the presence of unsaturations) favors the formation of the H_II phase by reducing R₀ (Rilfors et al., 1984; Lafleur et al., 1996). If an analogous interpretation is made, the ordering effect of cholesterol in the H_II phase should be unfavorable to its formation. This factor does not seem to contribute significantly for cholesterol content lower than 30 mol % as a decrease in Th is observed. It may, however, be involved in the increase of Th observed for a higher proportion of cholesterol. It is interesting to note that the amplitude of the chain order observed during the lamellar/H_II phase transition, as probed with the symmetric _C---H, varies in a concerted manner with Th (Fig. 2). For low cholesterol content, the amplitude increases; this can be rationalized on the basis that the lamellar phase becomes more ordered, and the structural change toward a phase for which the geometry allows considerable space for chain motions leads to the introduction of increasing chain disordering. For cholesterol content between 30 and 45 mol %, the amplitude of the _C---H shift during the lamellar/H_II phase transition decreases. The NMR spectra show that, in these conditions, the relative ordering of the phospholipid acyl chains in the H_II phase becomes more important than in the lamellar phase. This may be related to reaching a limit of chain packing in the lamellar phase whereas the cylindrical geometry of the H_II phase leaves some space for additional ordering. The variation of the enthalpy associated with the lamellar-to-H_II phase transition of POPE in the presence of cholesterol does not vary linearly between 0 and 45 mol % chol (Epand and Bottega, 1987), and it may be influenced by the extent of chain disordering during the transition. Finally, other factors, such as the influence of cholesterol on the curvature modulus Kc and cholesterol phase separation, should be examined. Additional characterization is being currently done.

The present work reports the characterization of the effect of cholesterol on PE acyl chains in the gel, the liquid-crystalline, and the H_II phases. The combined investigation of the phase transitions by IR spectroscopy and the phase characterization by ²H-NMR leads to the conclusion that POPE forms a liquid-ordered phase in the presence of high cholesterol content. The formation of a lipid matrix with such properties is not related to a peculiar behavior of the PC species, and this conclusion reinforces the significance of this phase in biological membranes. It is significant to extrapolate the formation of liquid-ordered phase in the presence of cholesterol to PE matrices as PE is almost as abundant as PC species in certain membranes and it even constitutes the main phospholipid of the internal leaflet of the human erythrocyte plasma membrane. It should be noted, however, that this behavior is linked to the solubilization of a high concentration of cholesterol in fluid phospholipid membranes, and it is not clear that systems with limited cholesterol solubility (McMullen and McElhaney, 1997) would display the same trends. We also report that cholesterol maintains its acyl chain ordering effect in the H_II phase, and this phenomenon must be included in a detailed molecular description of the H_II cylinders formed by PE/cholesterol mixtures.