INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Cyclodextrins (CD) are water-soluble natural cyclic oligosaccharides composed of 6, 7, or 8 glucose units (, , and  cyclodextrins, respectively). Owing to their torus-shaped structure, they delimit an internal rather hydrophobic cavity although the external part of the molecule is hydrophilic. They are therefore able to include hydrophobic compounds leading to water-soluble inclusion complexes. This property has been used extensively in a wide number of applications in fields such as pharmaceutical, cosmetics, or food industries (Duchêne, 1990). Chemical modifications of the natural oligosaccharidic core allow orientation and modulation of the properties of these cage-molecules toward specific applications. A promising field concerns the addition of one or several highly hydrophobic moieties on the hydrophilic carbohydrate, giving amphiphilic compounds prone to self-organization in aqueous media or to insertion in supramolecular assemblies such as liposomes or micelles. It was observed that the grafting of one cholesteryl unit on natural CD (the coupling can be direct or through a spacer) leads to amphiphilic molecules that can be efficiently inserted as guests in phospholipid liposomes. Using small-angle x-ray scattering, we have shown that the insertion of cholesteryl--cyclodextrin in multibilayers of dimyristoyl phosphatidylcholine (DMPC) results in the formation of two different lamellar phases, one is made of pure DMPC, the other one being strongly loaded by the cholesteryl cyclodextrin guest (Auzély-Velty et al., 1999). This would indicate that the polar CD moieties experience strong lateral interactions, sequestrating lipid molecules in a two-dimensional network. Thus, it appears that besides its biological interest, the cholesteryl cyclodextrin/DMPC system offers also an interesting approach of the molecular-induced lateral segregation of phospholipids within bilayers membranes.

To gain detailed molecular information on the cyclodextrin-induced lateral segregation process, we have carried out a comprehensive deuterium magnetic nuclear resonance (²H NMR) study with DMPC membranes containing derivatives of cholesteryl--cyclodextrin, differentiated by the presence (A) or the absence (B) of a succinyl spacer linking the cholesteryl moiety and the cyclodextrin headgroup. Chain-deuterated DMPC probing the cholesteryl cyclodextrin-induced perturbations, allowed the various phases to be monitored separately on a large range of temperatures (12°C-37°C) and at various cholesteryl cyclodextrin concentrations, providing detailed information on their nature and the order of the lipids.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Synthesis of cyclodextrin derivatives

6^I-(cholest-5-en-3-ylamido)succinyl-amido-6^I-deoxy-cyclomaltoheptaose (CC_A) and 6^I-(cholest-5-en-3-yloxycarbonyl)amino-6^I-deoxy-cyclomaltoheptaose (CC_B) (Scheme 1) were synthesized as described previously (Auzély-Velty et al., 1999). The deuterated analogues of -cyclodextrin and of CC_A, regio-specifically labeled on the C-2 carbon of all glucose units, were obtained according to a published procedure (Djedaïni et al., 1990). All compounds were fully characterized by high field proton NMR and high resolution mass spectrometry.

View larger version (14K): [in this window] [in a new window]   Scheme 1

Sample preparations

DMPC and chain deuterated DMPC-d54 were purchased from Avanti Polar Lipids (Alabaster, AL) and the cholesterol from Sigma (St. Louis, MO). Multilayered liposomes were prepared by mixing chloroform lipid solutions and methanol solutions of the appropriate cyclodextrin derivative. The solvent was then removed by evaporation under N₂. The solid residues were dried under vacuum (10² mm Hg) for 12 h and dispersed by continuous vortexing at 20°C in 100 to 300 µl of deuterium depleted water (Eurisotop, France) equilibrated at pH 8.0 giving ~200 mM lipid dispersions.

²H NMR experiments

²H NMR spectra were recorded at 46 MHz on a Bruker DMX 300 spectrometer equipped with a probe specifically designed for solid state deuterium NMR experiments (Morris Instruments Inc., Gloucester, ON, Canada). Spectra were acquired from 12°C to 37°C with a dwell time of 2 µs, 4-K data points, and a recycling time of 200 ms. A quadrupolar echo pulse sequence (Davis et al., 1976) was used with pulse length of 3 µs and pulse separation, , of 40 µs. The phase was adjusted to obtain no signal in the imaginary channel, which was then discarded before the Fourier transform of the echo. When necessary, the free induction decay was shifted by a fraction of the dwell time using an orthogonal polynomial interpolation routine so that the Fourier transform could start at the top of the echo (Davis, 1983). Oriented ²H NMR spectra (0°) were obtained by the numerical dePake-ing procedure described by Sternin et al. (1983). The method of moments (Davis, 1979; Davis et al., 1979) was also applied to the chain deuteron spectra. The narrow (<3 Hz) dePaked methyl resonances found in the 10 to 10 kHz range were simulated with a Gaussian line shape after baseline correction of the data. Each resonance was fitted with three independent parameters namely the frequency, the line width, and the intensity.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

²H NMR of deuterated phospholipids provides a suitable tool for studying the lipid membrane organization through the direct measurement of the local orientational order parameters of the C---D bonds of the lipid acyl chains and polar headgroups (Davis, 1983). In particular gel-to-liquid crystalline phase transitions of phospholipid membranes can be precisely monitored by following the line shape changes of ²H NMR spectra of the chain-deuterated lipid derivatives (Davis, 1979; 1983). In Fig. 1 A, the line shapes of ²H NMR spectra of saturated DMPC-d54 change dramatically between 20°C and 19°C from a well-resolved, axially symmetric distribution of quadrupolar splittings typical of the fluid L phase above 20°C, to a much broader distribution of lipid molecules in the gel P_' phase below 20°C. At higher temperatures, the quadrupolar splitting distribution is typical of phospholipid bilayers in the fluid state with a large quadrupolar splitting, or plateau, containing the methylene groups near the membrane interface, smaller resolved quadrupolar splittings for the methylene closer to the bilayer center, and a narrow doublet due to the methyl group at the end of the chain. The gel-to-liquid crystalline transition of perdeuterated DMPC occurs at lower temperature (20°C) than the value of ~23°C, measured by various techniques with pure unlabeled DMPC membranes (Marsh, 1990; Koynova and Caffrey, 1998). This effect is due to the chain-chain deuteron interactions that lower the phase transition temperatures of deuterated phospholipids (Peterson et al., 1975).

View larger version (19K): [in this window] [in a new window]   FIGURE 1   Temperature dependence of the 2H NMR spectra of DMPC-d54 membranes. Pure (A) and with 20% (B) CCA in molar %. The spectra were recorded at 37, 25, 20, 19, 10, and 0°C. The thin line in the 40- to 75-kHz region shows the spectra scaled by a factor of 10.

DMPC-d54 bilayers with cholesteryl--cyclodextrin A

The ²H NMR spectra recorded in the presence of 20% molar CC_A (Fig. 1 B) significantly differ from those obtained with pure DMPC. Above 20°C, the lipids are all in the fluid state as probed by the liquid crystalline line shape of the ²H NMR powder patterns, and apparently there is only one component at 37°C. A shoulder is detected at 25°C, on each side of the large quadrupolar splitting of the plateau methylene groups, suggesting that at least two components coexist. This shoulder appears more clearly when the temperature is decreased. At 20°C, the resonances of the methyl terminal deuterons are also split in two signals (±1.5 and ± 2.0 kHz), indicating the coexistence of two distinct types of DMPC molecules in the fluid state, in slow exchange on the ²H NMR timescale. At 19°C, a gel phase component is detected with signal intensity out to ±63 kHz, and also ±6 kHz, where the signal of methyl groups in the gel phase appears on pure DMPC-d54 spectra at the same temperature. Now we can detect three components at 19°C, the DMPC molecules in the gel phase coexisting with at least two species of DMPC in the fluid-like state monitored by the outer and inner splittings, which will be referred as component (I) and (II) in the following, of the methylene deuterons of the plateau region and of the terminal methyl groups. On cooling the sample further, the amount of the gel phase spectrum increases while the maxima of the fluid component (I) detected on both methyl and plateau resonances collapse. In contrast, the intensities of the other fluid component (II) remain constant, although a progressive broadening of the NMR lines occurs below 10°C.

The effect induced by CC_A on the DMPC-d54 acyl chains can be analyzed in more detail after dePake-ing of the ²H NMR data (Fig. 2, bottom spectra). As shown on the spectrum recorded with pure DMPC (Fig. 2, trace a), this procedure allows clear monitoring of the individual quadrupolar splittings of the myristoyl acyl chains, i.e., the terminal methyl groups (1), the methylene groups localized at the end of the acyl chains and those contributing to the plateau region (2). The two components (I) and (II) observed in the presence of CC_A appear clearly on the dePaked spectrum recorded with 20% (Fig. 2 b) or 30% (Fig. 2 c) of the cyclodextrin derivative. An estimation of the relative intensity of the methyl resonances of component (I) and (II) gives ~30% and 50% of (II) at, respectively, 20% or 30% molar CC_A. The ratio of the two methyl components was found to be independent of the pulse separation used in the quadrupolar echo sequence (for values of  between 40 and 200 µs; see Materials and Methods). This indicates that the measurement is not distorted by differences in echo decay times for the methyl groups in the two lipid phases.

View larger version (25K): [in this window] [in a new window]   FIGURE 2   2H NMR dePaked spectra of DMPC-d54 membranes recorded at 20°C with 20%, 30%, and 40% CCA (b, c, d) or CCB (e, f, g) expressed in molar %. The bold letters on the pure DMPC spectrum a point to the quadrupolar splittings attributed to the methyl deuterons (1) and to the methylene deuterons of the plateau region (2). The components (I) and (II) described in the text are indicated on spectrum b. The dePaked spectra were scaled with normalization factors obtained by area-normalization of their related Fourier transform spectra shown in Fig. 1. The intensity of spectrum a obtained without cholesteryl cyclodextrin (0%) is rescaled by a factor of 0.7.

Thus, the signal ratio of the two components is not equal to 1 and changes with the guest/lipid ratio. This indicates that the two lipid forms characterized by the splitting of the methyl resonances do not result from the magnetic inequivalence of the sn-1 and sn-2 chains, as observed in the presence of cholesterol (Sankaram and Thomson, 1990). When 40% of the guest is added to the bilayers (Fig. 2 d), we observe that component (I) has almost disappeared, so that most of the signal of DMPC deuterons are now under component (II). The dePaked deuterium NMR spectra displayed in Fig. 2 show that for all membranes containing CC_A, the quadrupolar splittings of the methyl and plateau methylene groups of component (II) do not depend on the CC_A-to-DMPC molar ratio, i.e., the order parameters of the DMPC acyl chains in the new phase are approximately invariant whatever the guest concentration in the membrane.

DMPC-d54 with cholesteryl--cyclodextrin B

In the B form of cholesteryl--cyclodextrin (CC_B) the cholesteryl moiety is directly linked to the amino group of cyclodextrin without the succinyl spacer found in the A form. As seen in Fig. 2 (e-g), the dePaked ²H NMR spectra of DMPC membranes obtained at 20°C with the two derivatives are quite different. Apparently, there is only one fluid component with CC_B, when the membrane is in the liquid crystalline state. The quadrupolar splittings of the plateau and methyl deuterons have approximately the same values than those measured with pure DMPC-d54, and are not decreased as observed with CC_A.

The dePaked lines are broadened in the presence of CC_B, while they are still narrow and resolved with CC_A (Fig. 2, b-d). This line broadening is quite pronounced when the membrane CC_B concentration reaches 40% (Fig. 2 g).

Temperature dependence of DMPC-d54 bilayers with cholesteryl cyclodextrins

The temperature dependence of the DMPC/cholesteryl cyclodextrin interaction was probed on large range of temperature from 37°C to 12°C. The dePaked spectra obtained with 20% CC_A or CC_B are respectively shown in the stacked plots of Fig. 3, A and B. The line widths and quadrupolar splittings measured for the methyl deuterons from these spectra are plotted in Fig. 4.

View larger version (60K): [in this window] [in a new window]   FIGURE 3   2H NMR dePaked spectra of DMPC-d54 membranes recorded with 20% molar of CCA (A) or CCB (B). The stacked plots show the spectra recorded at 37 and 30°C, and from 25 to 12°C (1°C step, going from the upper traces to the lower traces; selected temperatures are indicated by the bold traces). The dePaked spectra were scaled with normalization factors obtained by area normalization of their related Fourier transform spectra shown in Fig. 1. (A) The letters indicate the quadrupolar splittings of component (I) in the fluid state at 20°C (b), in the gel phase at 18°C (c), in the lamellar gel phase at 9°C (e), and the quadrupolar splittings of component (II) in the fluid LCD phase at 20°C (a) and in a gel-like state at 10°C (d) (see Discussion). The methyl resonances of component II are also indicated by a thin solid line connecting their maxima, showing the transition occurring between 5 and 5°C, from the fluid LCD phase to an ordered LCD phase (see Discussion).

View larger version (26K): [in this window] [in a new window]   FIGURE 4   Quadrupolar splittings (A and C) and line widths (B and D) of the DMPC-d54 methyl resonances, measured from the dePaked spectra shown in Fig. 3, obtained in the presence of CCA (A and B) or CCB (C and D) as follow: (,) 20% CCA; (×) 40% CCA; () 20% CCB; and (+) 40% CCB. In the presence of the inhomogeneous line broadening occurring below 20°C with CCB-containing membranes (C and D), line widths, and quadrupolar splitting were determined from the inner shoulder of the methyl resonances of the dePaked spectra of Fig. 3 B. The measurements below 15°C were found to be unreliable.

Temperature above 0^o C

'

View larger version (27K): [in this window] [in a new window]   FIGURE 5   Moment M1 of the 2H NMR spectra of DMPC d54 membranes pure () and with 5% (), 20% (), 30% (), and 40% (×) (molar %) of CCA or CCB.

Temperatures below 0^oC

'

'

'

'

'

'

DMPC-d54 bilayers with low concentrations of cholesteryl--cyclodextrins

To test further the ability of CC_A to induce the formation of a second lipid environment (II), we have also obtained data at low concentration (5%) of CC_A. It appears that, despite this membrane dilution, a second component is still detected, as shown in Fig. 6. Due to the low amount of CC_A, this component is not resolved on the DMPC-d54 spectra recorded in the fluid phase (data not shown), but it appears clearly on the ²H NMR spectrum recorded at 19°C, just below the fluid-to-gel phase transition (Fig. 6 a). The analysis of the methyl region indicates very clearly the coexistence of two fluid lipid environments with gel phase lipids, as observed on the corresponding NMR spectra recorded with 20% of CC_A. To isolate these fluid components, we have removed the gel signal by subtracting a fraction (~ 62%) of the pure DMPC-d54 gel phase spectrum recorded at the same temperature. We were able to obtain a difference spectrum (Fig. 6 b) with no signal left beyond ±40 kHz, containing only the signals of the fluid lipids. The same procedure was applied to the NMR data recorded at lower temperature, i.e., the same fraction (62%) of the pure lipid gel phase spectrum at the corresponding temperature was subtracted. The difference spectrum derived from the data measured after lowering the temperature of one degree from 19°C to 18°C indicates clearly that the fraction of the pure lipid remaining in the fluid state has now moved into the gel state (Fig. 6 c). After dePake-ing of the obtained difference spectra (Fig. 6 h), it can be shown that the temperature dependencies of the line widths and quadrupolar splittings of the methyl signal follow closely those obtained at 20% and 40% CC_A. Similar difference spectra were obtained at 10% of CC_A (data not shown), and lead to the same conclusions.

View larger version (40K): [in this window] [in a new window]   FIGURE 6   2H NMR difference spectra of DMPC-d54 membranes with 5% of CCA (a, b, c, and g) or CCB (d, e, f, and h). Original (a and d) and difference (b, e, c, and f) spectra at 19°C (b and e) and 18°C (c and f). The expanded methyl region is shown for each spectrum (a-f). The stacked plot (g and h) show the dePaked spectra obtained from the difference spectra at 19, 18, 17, 16, 15, 5, and 0°C.

²H NMR spectra were also obtained with 5% (Fig. 6, d-g) and 10% (data not shown) of the B form of cholesteryl--cyclodextrin. The analysis of the difference spectra obtained with the same procedure, as described above, confirms the results obtained at higher concentrations, showing that there is no additional component in the presence of this cholesteryl--cyclodextrin derivative.

Deuterated CC_A in DMPC bilayers

The interaction of cholesterol cyclodextrin with zwitterionic DMPC bilayers was probed with CC_A-d7 deuterated on the oligosaccharide headgroup, at the level of the C₂ carbon of the seven glucose units. The ²H NMR spectra obtained with this derivative, shown in Fig. 7, contain a very large quadrupolar splitting (~120 kHz) with a broad featureless signal spread over the whole frequency range. The larger splitting has no thermal dependence, retaining the same value at all temperatures. On the other hand, the intensity of the central signals decreases progressively when the temperature is lowered, revealing a well-defined quadrupolar splitting at 0°C. The broad signal distribution observed at high temperatures could either reflect differences in the order parameters of the glucose units or different environments of the CC_A molecules. We have also recorded ²H NMR spectra with native deuterated -cyclodextrin-d7 (without the cholesteryl moiety) and obtained a single narrow isotropic resonance, which indeed confirms that the hydrophilic guest does not interact spontaneously with phosphatidylcholine liposomes in the absence of a hydrophobic anchor (data not shown). There is a 200-Hz difference between the chemical shifts of the isotropic signal of free -cyclodextrin-d7 and the central narrow resonance found on the spectra of membrane-bound CC_A shown in Fig. 7. In the latter case the isotropic signal is due to the deuterons of residual heavy water.

View larger version (19K): [in this window] [in a new window]   FIGURE 7   Temperature dependence of the 2H NMR spectra of 20% molar of CCA-d7, deuterated on the C2 carbon of the seven glucose units, incorporated in DMPC membranes.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Our study shows that the A form of cholesteryl--cyclodextrin, which contains a spacer between the sterol and the cyclodextrin moiety, induces a lateral separation of lipid molecules in two distinct phases containing pure DMPC and DMPC sequestrated between CC_A molecules. We report two components of the DMPC-d54 spectra, with exchange rates that are slow on the ²H NMR timescale (>10⁵ s¹), probing the occurrence of two different lipid environments. Component (I) has the same thermal behavior as pure DMPC membranes, namely 1) a line broadening at the gel-to-fluid transition temperature of pure DMPC-d54 with the appearance of spectral intensities characteristic of lipids in the gel state and 2) the same quadrupolar splitting temperature dependence. Thus, this component is associated with a pure DMPC phase, co-existing with a cyclodextrin-rich phase appearing as component (II) on the NMR traces. These ²H NMR results are in good agreement with our previous data obtained for the DMPC/CC_A system using small angle x-ray scattering and differential scanning calorimetry, which indicate also the coexistence of two lamellar phases at 30°C (Auzély-Velty et al., 1999). In the present study, we have obtained NMR data providing detailed molecular information on the DMPC/cholesteryl--cyclodextrin system on a large range of concentrations and temperatures. A phase separation was found to occur with only 5% of CC_A. The overall perturbation of the bilayer is weak at such concentration (almost probe level) so the membrane order is close to that of pure DMPC, as shown in the corresponding moment M1 curve (Fig. 5). Under these conditions, we can partially dissociate discrete effects from global membrane perturbations induced by larger concentrations of the guest. We have found that there are at least three lipid environments at 19°C just below the fluid-to-gel transition of DMPC-d54. A first one, with the smaller splitting, is associated with lipids interacting with CC_A, in a cholesteryl cyclodextrin-rich phase, which we will refer to as the L_CD phase in the following, and appearing as component (II) on the ²H NMR spectra. A second one is associated to the fluid signal with the larger quadrupolar splitting and should correspond to lipids located at the boundary of the L_CD phase, remaining in the fluid state at 19°C and a third one is that of pure lipids in the gel phase, not interacting with the L_CD phase. These three lipid environments are clearly distinguished on the difference spectrum obtained with 5% CC_A at 19°C (Fig. 6). A quantitative analysis of the dePaked NMR spectra has been attempted by simulating the fluid dePaked methyl resonances of component (I) and (II) with Gaussian line shapes, and the results are shown in Table 1. For the sample containing 5% of CC_A at 19°C, we have estimated that for 10 molecules of CC_A, the amount of DMPC in the L_CD phase, at the boundary of the L_CD phase and in the gel state are (within the experimental error, respectively), 16, 57, and 118. As shown in Table 1, the number of lipids in the gel phase is quickly reduced when the concentration of CC_A is increased, and for 30% CC_A there is no more lipids in the gel phase. Indeed, the total number of fluid lipid per molecule of CC_A is also reduced, but it is mostly at the expense of the fluid boundary lipids, whereas the number of lipid molecules found in the L_CD phase remains approximately constant in the range of 1 to 1.5 DMPC per CC_A molecule. A similar ratio is also found at temperatures above the fluid-to-gel transition at either 20% or 30% of CC_A. At 40% CC_A, which corresponds already to a global ratio of 1.5 molecules of DMPC per cholesteryl--cyclodextrin, almost all the lipids should be in the L_CD phase. The calculated line width values are indeed consistent with the data plotted in Fig. 4, showing that the pure lipid resonances start to broaden at 19°C at the onset of the fluid-to-gel transition, whereas those of the lipids in the L_CD phase are not affected by this transition. Thus, the DMPC-to-CC_A ratio in the L_CD phase appears to be remarkably invariant, i.e., independent on both the temperature variations and the total amount of cholesteryl--cyclodextrin. This strongly suggests that the two phases are fully separated with DMPC either pure or mixed with CC_A. A possible explanation could be that these two phases are in fact macroscopically distinct. In such case, the two ²H NMR components would just reflect the heterogeneity of the sample with membrane aggregates enriched in cholesteryl--cyclodextrin and pure lipid particles. This assumption can be ruled out by the observation that, at all CC_A concentrations, the pure lipid phase does interact with the L_CD phase through the boundary fluid lipids detected at the onset of the fluid-to-gel transition.

View this table: [in this window] [in a new window]   TABLE 1   Lipid distribution at different temperatures and for various molar % of CCA, among the gel phase, the fluid phase, and the LCD phase

It appears that the lateral separation of the L_CD phase, detected at membrane concentrations as low as 5% of CC_A, has to be associated with strong interactions between the cyclodextrin headgroups. The  form of cyclodextrin alone, without cholesterol, is actually well known to form aggregates in water and is 10-fold less soluble than the  and  derivatives. Interestingly, the chemical modifications of the cyclodextrin hydroxyls, for instance by methylation, disrupt these interactions and lead to completely different behaviors (Auzély-Velty et al., 2000). In our case, the CC_A cyclodextrin headgroups incorporated in DMPC bilayers appear to be very rigid, because the deuterons of the cyclodextrin headgroup exhibit a very large quadrupolar splitting, independent of the temperature variations, giving a high order parameter, corresponding to an almost complete absence of motion.

Thus, the emerging picture is that the stability of the L_CD phase should be governed primarily by the hydrophilic interaction between cyclodextrin headgroups. In this case hydrophobic interaction between the sterol and the phospholipid acyl chains could be of secondary importance in maintaining the cohesion of the L_CD phase. The L_CD phase seems to contain a maximum of 1.5 DMPC per CC_A. Considering the ratio of the specific area of the cyclodextrin headgroup and of the cholesteryl hydrophobic anchor, it can be roughly estimated that, in the case of a packed network of cyclodextrin headgroups, the volume left in between two adjacent cholesteryl cyclodextrin molecules should effectively accommodate for no more than two to three phospholipid molecules. The acyl chains of these sequestrated lipids are more disordered than in the pure lipid phase. This is shown by the decrease of the first moment M₁ and the reduction of the quadrupolar splittings of the methyl and plateau deuterons observed in the presence of CC_A. This effect is thus completely opposed to the ordering of the phosphatidylcholine fluid phase, or "condensing" effect, induced by cholesterol alone, which leads in similar experimental conditions, and with the same cholesterol to phospholipid ratio, to an average ~70% increase of the DMPC-d54 quadrupolar splittings (data not shown; Vist and Davis, 1990). The acyl chains quadrupolar splittings are also remarkably insensitive to temperature variations in the L_CD phase, as opposed to what is observed in the pure lipid phase, and remain fluid well below Tc. The above remarks support the model in which the packing order of the L_CD phase is not primarily determined by chain-chain or cholesterol-chain interactions, as it occurs in DMPC bilayers, but is rather dominated by the interactions between the large cyclodextrin moities. However, the chain deuterons become temperature sensitive between 5°C and 5°C, showing that the DMPC molecules do undergo a phase transition from a fluid phase to more a ordered environment, as shown in Fig. 3 A and Fig. 4. Yet, the methyl quadrupolar splittings are still smaller than those measured in DMPC gel phase, indicating that the orientational order of this new L_CD state is smaller than in the P_' gel phase of the pure lipid. This ordered L_CD phase seems to be quite stable and appears to be preserved at low temperatures, even at 9°C when the pure lipids appear to be in the rigid L_' lamellar gel phase. However, it apparently breaks down at 10°C, where the lipids associated to component (II) seem to undergo another transition. The lipids should have evolved into a gel-like state, considering that the quadrupolar splittings of their acyl chain methyl groups are now comparable with those measured for the component (I) of the pure lipid in the gel phase (Fig. 3 A). This transition could be due to the freezing of the bulk water molecules observed simultaneously at 10°C, which probably lead to disorganization or even a disruption of the cholesteryl--cyclodextrin headgroup network and a rearrangement or a suppression of the L_CD phase.

The role of the spacer introduced between the sterol and the cyclodextrin is certainly decisive by increasing the conformational space offered to the CC_A headgroups anchored at the membrane interface. Such a conformational flexibility is probably required to achieve head to head interactions efficient enough to induce the clustering effect leading to the sequestration of the DMPC molecules in the L_CD phase. The results obtained with the B form of cholesteryl -cyclodextrin, lacking the succinyl spacer, support this model. In the presence of this derivative, only one phase is detected with spectrum line shapes indicating that the bilayers are still stable. In the fluid phase, there is a decrease of the average orientational order as probed by the first moment M1 values, which are approximately similar to those obtained with bilayers containing CC_A. In fact, the global disordering effects induced by these two cyclodextrin derivatives on the DMPC acyl chains in the fluid phase are quantitatively similar, although the molecular mechanisms involved in the interaction of CC_A and CC_B with the lipid membrane appear to be qualitatively different. In the absence of succinyl spacer, the B form of cholesteryl--cyclodextrin must be distributed in the whole bilayers, perturbing almost each phospholipid molecule, whereas with the A form, two laterally segregated distinct phases are observed. This appears quite clearly on the ²H NMR dePaked spectra obtained with 20% of the cholesteryl--cyclodextrin derivatives displayed in Fig. 3. With CC_B there is a smoothed gel-to-fluid phase transition involving the ensemble of the lipids, whereas sharp phase transitions concerning only a fraction of lipids separating from the CC_A-induced L_CD phase are observed at 19°C and also at 8°C. At high concentrations of CC_A, there is formation of an almost pure L_CD phase without cooperative fluid-to-gel transition, whereas a sigmoid transition is still detected with CC_B, indicating that the DMPC molecules are still able to undergo a cooperative, although smoothed, transition to the gel state (Fig. 5).

The ensemble of our NMR data provide conclusive evidences that there is globally one lipid phase with CC_B, whereas a pure lipid phase and a laterally segregated cholesteryl cyclodextrin-rich phase are detected with CC_A. Without the succinyl link to the cholesterol moiety, the cyclodextrin headgroup of CC_B must be constrained at the membrane surface with a reduced conformational space, hindering probably the adequate positioning of adjacent headgroups and the formation of the L_CD phase observed with the CC_A. Thus, the cholesteryl cyclodextrin/DMPC system gives a straightforward example on how an amphiphilic molecule can affect lipid membranes with the local formation of a two-dimensional molecular network within the bilayer, through finely-tuned intermolecular headgroup interactions at the membrane interface.

The ability of bilayer membranes to incorporate the cholesteryl--cyclodextrin derivatives should permit the liposome transportation of hydrophobic cyclodextrin-bound drugs, standing out at the bilayer surface toward the external medium, facilitating their interaction with circulating molecules or macromolecular assemblies. Indeed, the primary pharmacological interest of the cholesteryl--cyclodextrin lies in the inclusion property of the cyclodextrin cavity. Further efforts will thus be dedicated to investigate the effect of the inclusion of a guest compound on the cholesteryl--cyclodextrin liposome insertion and on the formation of the L_CD phase. Special attention will be given to pharmacologically active compounds to design new formulations for the administration of drugs combining the size-specificity of the cyclodextrin moiety and the carrier properties of phospholipid liposomes. It will be in particular interesting to evaluate the relative potencies, if any, of the CC_A and CC_B derivatives and determine whether the formation of the laterally segregated L_CD phase is pharmacologically relevant.