Comparison of the interaction of dihydrocholesterol and cholesterol with sphingolipid or phospholipid Langmuir monolayers

Colloids and Surfaces B: Biointerfaces 59 (2007) 81–86

Comparison of the interaction of dihydrocholesterol and cholesterol with sphingolipid or phospholipid Langmuir monolayers Elo¨ıse Lancelot 1 , Christine Grauby-Heywang ∗ Centre de Physique Mol´eculaire Optique et Hertzienne (CPMOH), UMR CNRS 5798, Universit´e Bordeaux 1, 351 cours de la lib´eration, 33405 Talence Cedex, France Received 29 March 2007; received in revised form 20 April 2007; accepted 25 April 2007 Available online 1 May 2007

Abstract We report here a study of the interaction of dihydrocholesterol (DChol) with palmitoyl-oleoyl-phosphatidylcholine (POPC) or sphingomyelin (SM) in Langmuir monolayers. DChol and cholesterol (Chol) have very close chemical structures, and DChol is often used in place of Chol because of its better stability. Surface pressure measurements and experiments of desorption induced by ␤-cyclodextrin show that POPC–DChol monolayers behave similarly to POPC–Chol ones: condensing effects of DChol and Chol on POPC and desorption percentages are in the same range. Moreover Brewster angle microscopy (BAM) experiments performed on these monolayers show that on the whole they are both homogenous. The analysis of mean molecular areas versus DChol percentage shows that this sterol is also able to induce SM condensation at low surface pressure. The condensation of SM molecules is particularly strong at 30 mol% of DChol. At higher surface pressure, the condensation efficiency of DChol decreases and monolayers behave more ideally, even if an inflection point is always observed at 30 mol% of DChol. However, desorption percentages, clearly lower than those obtained with POPC–DChol monolayers, show that DChol is kept at the interface. At last BAM images show also differences in the behaviour of SM–DChol and SM–Chol monolayers. These differences could be due to the different compressibility and conformation of the A/B rings in the two sterols and the rigidity of the sphingosine chain. They suggest that the use of DChol in place of Chol has to be done carefully in the presence of SM. © 2007 Elsevier B.V. All rights reserved. Keywords: Dihydrocholesterol; Cholesterol; Lipid; Monolayer; Brewster angle microscopy

1. Introduction Cholesterol (Chol) is a major lipid of cellular membranes of mammalian cells. Numerous studies show that this sterol plays a crucial role in the lateral organization of lipids in membranes, by inducing phase separation and the formation of a “liquid ordered” phase (lO ) intermediate between gel and fluid phases [1–3]. Chol is also involved in the formation of membrane domains called rafts, by interacting more specifically with sphingolipids [4]. A lot of studies dealing with the behaviour of Chol use simple membrane models, because of the too complex composition of membranes, which includes more than 2000 lipid species [5]. ∗

Corresponding author. Tel.: +33 5 40 00 89 97; fax: +33 5 40 00 69 70. E-mail address: [email protected] (C. Grauby-Heywang). 1 UMR 5248 CBMN, CNRS-Universit´ e Bordeaux 1-ENITAB, IECB, 2 rue Robert Escarpit, 33607 Pessac, France. 0927-7765/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.colsurfb.2007.04.017

In some of these studies, dihydrocholesterol (DChol) is used in place of Chol because of its better resistance to oxidation. Indeed Chol is sensitive to oxidation induced for instance by oxygen in atmosphere, and this can be problematic in particular for experiments performed on long time range. Benvegnu and McConnell followed the effect of Chol oxidation in DMPC monolayers by monitoring the shape transition pressure as a function of time under different atmospheres [6]. They showed that the transition pressure decreases gradually with time under air atmosphere. This decrease is clearly enhanced in the presence of ozone, but partially inhibited if nitrogen is added in the atmosphere above the monolayer. A partial inhibition of oxidation is also observed by adding a free radical scavenger into the subphase. At last, the decrease of pressure is totally inhibited if Chol is replaced with DChol in DMPC monolayers, showing that DChol is insensitive to oxidation contrary to Chol. Under these conditions, some studies have been performed on different DChol–phospholipid systems. For instance Radhakrishnan and McConnell did not observe any difference in phase

82

E. Lancelot, C. Grauby-Heywang / Colloids and Surfaces B: Biointerfaces 59 (2007) 81–86

diagrams and mean molecular areas, when Chol or DChol were embedded in DMPS/DMPC and DPPC/DMPC monolayers [7]. Kodama et al. studied also mean molecular areas, elasticity and phase diagrams of DPPC monolayers containing Chol, DChol or stigmasterol, and showed that these monolayers behave on the whole similarly [8]. 31 P-NMR and DSC experiments performed on phosphatidylethanolamine vesicles showed at last a similar decrease of the transition temperature induced by DChol and Chol [9]. However, only a few experiments concern the interaction of DChol with sphingolipids, such as sphingomyelin [7,10], whereas this kind of lipid is supposed to interact strongly with Chol in cellular membranes. Therefore, in this work, we studied the behaviour of DChol in palmitoyl-oleoyl-phosphatidylcholine (POPC) or sphingomyelin (SM) monolayers, and compared it to the behaviour of Chol, studied under the same conditions in a previous work [11]. POPC and SM were chosen to be representative of lipids present in the fluid phase of membrane or in rafts, respectively. Surface pressure measurements and experiments of DChol desorption were used to characterize these monolayers, whereas Brewster angle microscopy (BAM) enabled us to observe their morphology. The similar results obtained in surface pressure measurements and BAM confirm that DChol can be used in place of Chol in POPC monolayers. However, we found some differences in the behaviour of Chol and DChol in SM monolayers. 2. Materials and methods 2.1. Materials Dihydrocholesterol (3␤-hydroxy-5␣-cholestane, DChol, Fig. 1) was approximately 95% pure, and was purchased from Sigma (France). SM from egg yolk, POPC and ␤-cyclodextrin (␤-CD) were also purchased from Sigma (France), and were approximately 99% pure. SM chains are mainly palmitic. Chloroform and ethanol (both HPLC grade) were purchased from Sigma and Prolabo (France), respectively. All these molecules were used without further purification. Ultrapure Millipore water (pH 5.5, resistivity >18 M cm) was used as subphase. 2.2. Methods 2.2.1. Surface pressure measurements Surface pressure measurements were carried out in air by using a 601M Nima Langmuir through (approximately

Fig. 1. Chemical structures of DChol and Chol.

19 cm × 12 cm × 0.5 cm) equipped with a Wilhelmy balance (Nima). Briefly, lipids were dissolved in chloroform/ethanol 1/1 (v/v) at a concentration around 1 mM. Mixtures of lipids were prepared with these solutions to the required molar ratio. After spreading of the solution and evaporation of solvents (15 min), lipids were compressed continuously at a rate of 5 cm2 min−1 . The temperature of the subphase was kept constant to 20 ± 1 ◦ C. Each experiment was repeated at least three times with monolayers prepared from different solutions. Mixed lipid monolayers were prepared similarly in the case of DChol desorption experiments. Lipid monolayers were spread at the air–water interface, with a same total number of lipid molecules at each experiment, and compressed at a final surface pressure of 30 mN/m, in the range of surface pressure estimated in membranes [12]. ␤-CD, dissolved in water, was injected in the subphase at a final concentration of 0.5 mM in the trough. This injection induced a temporary decrease of the surface pressure, immediately compensated by a decrease of the spreading surface, thanks to a pressure-control system, which maintained the surface pressure at 30 mN/m. A desorption percentage was estimated 90 min after the ␤-CD injection. 2.2.2. BAM experiments A NFT BAM2plus setup (G¨ottingen, Germany) was mounted on a Langmuir trough. The microscope was equipped with a frequency doubled Nd:Yag laser (532 nm, 50 mW), a ×10 objective, polarizer, analyser and a CCD camera. The spatial resolution of the BAM was about 2 ␮m. After spreading of the monolayer and evaporation of solvents, molecules were compressed until the required surface pressure. This value was kept constant thanks to a pressure-control system during the image acquisition. Calibration of intensity versus grey levels of the video camera was performed before each experiment. In BAM images, condensed phase appears brilliant as compared to the liquid expanded phase appearing as a black background. 3. Results 3.1. Surface pressure measurements (␲-A) isotherms of DChol, SM and their mixtures at different molar ratios are reported in Fig. 2A. DChol isotherm is similar to those already reported [8,13]. As in the case of Chol, the sudden increase of surface pressure for mean molecular area higher than ˚ 2 shows that it is in a solid-phase during the compression, in 43 A relation with the planar and rigid structure of the molecule. The orientation and the mean molecular area do not change drastically during the compression. For instance, mean molecular ˚ 2 and 38 A ˚ 2 at 10 mN/m and areas are estimated around 41 A 30 mN/m, respectively. The isotherm of SM is also very close from those already reported [14–16], and shows that this lipid is in a liquid expanded phase at the beginning of the compression. Even if it is not very clear on the isotherm, a transition to a condensed phase occurs at a surface pressure around 10–18 mN/m, as shown by BAM or fluorescence microscopy [17–19].

E. Lancelot, C. Grauby-Heywang / Colloids and Surfaces B: Biointerfaces 59 (2007) 81–86

83

patches or being completely separated, experimental molecular areas will respect the Eq. (1). Any deviation from this Eq. (1) provides evidence of miscibility of the two molecular species (because of attractive or repulsive forces, leading to lower or higher mean molecular areas, respectively) [20]. Fig. 2B shows that the behaviour of mixtures depends on the surface pressure and the DChol percentage. At 10 mN/m and DChol percentage between 10 and 60 mol%, experimental points are under the straight line. This suggests that DChol and SM molecules are miscible under these conditions, and that mixed monolayers are more condensed than ideal ones. The condensation is particularly strong at DChol percentage around 30 mol% (inflection point in Fig. 2B). On the contrary, at high DChol percentage (80 mol%), the mean molecular area of mixture obeys to the Eq. (1), suggesting that molecules are not miscible anymore. In the case of SM monolayer containing 10–60 mol% of DChol, the condensation efficiency of the sterol can be estimated by determining the condensation percentage of SM. This percentage is based on the hypothesis that the decrease of mean molecular areas as compared to ideal ones is only due to changes in mean molecular area of SM, since DChol is rigid. From Eq. (1), one can extract the mean molecular area of SM in each mixed monolayers: ASM =

(Amix − ADChol × XDChol ) . (1 − XDChol )

(2)

It is then possible to compare this area to the SM area at the same surface pressure in a pure monolayer and to estimate the condensation percentage thanks to the equation: %condensation =

Fig. 2. (A) (␲-A) isotherms of SM, DChol and mixed monolayers spread on water (pH 5.5, T = 20 ± 1 ◦ C, compression rate 5 cm2 min−1 ). Percentages of DChol in each monolayer are indicated. (B) Mean molecular areas measured at 10 mN/m (䊉) and 30 mN/m () on SM–DChol isotherms vs. the percentage of DChol in the monolayer. Straight lines correspond to ideal mixtures as defined by the Eq. (1). Other lines are added to guide the eye.

(␲-A) isotherms of DChol–SM mixtures are shifted to lower mean molecular areas as compared to the SM isotherm. Fig. 2B shows mean molecular areas estimated from these isotherms at 10 mN/m and 30 mN/m. These mean molecular areas are compared, for both surface pressures, to the areas of ideal mixtures (Aideal-mix ) estimated thanks to the equation: Aideal-mix = ADChol × XDChol + ASM × (1 − XDChol ),

(1)

where ADChol and ASM are the mean molecular area of DChol and SM at 10 mN/m or 30 mN/m, respectively, and XDChol the molar fraction of DChol in the monolayer. According to Gaines [20], different cases have to be considered. If the two molecular species present in the monolayer are immiscible, forming

(ApureSM − ASM ) × 100 . ApureSM

(3)

In the case of condensed SM–DChol, this percentage is on the whole in 4–20% range, increasing with the amount of DChol until 80 mol%, except at 30 mol% of DChol. At this DChol amount, the condensation percentage increases to 34%. We perfomed previously the same kind of study with SM–Chol monolayers [11]. In this case, the condensation percentages are higher, in the range of 25–40%, increasing also with the amount of Chol in the monolayer. At high surface pressure (30 mN/m, Fig. 2B), the condensing effect of DChol is only observed around 20–30 mol%, with the same inflection point than previously, even if it is less pronounced. At this point, the condensation percentage is 26%. In the other mixtures, experimental points are on the straight line or slightly above. These changes could be related to a reorganization of molecules induced by compression. In particular, at high DChol percentages, mean molecular areas higher than those expected in ideal mixtures suggest that DChol and SM are miscible and repulsing each other. For comparison, in the case of SM–Chol monolayers, a strong condensation effect of Chol on SM molecules is observed at 30 mN/m, with condensation percentages increasing regularly until 17% and saturating at Chol percentages higher than 50 mol% [11,15]. Moreover, even if the effect of Chol seems to be maximal at 30 mol%, there is no inflection point.

84

E. Lancelot, C. Grauby-Heywang / Colloids and Surfaces B: Biointerfaces 59 (2007) 81–86

our previous study around 12% in SM monolayers and 32–43% in POPC monolayers). 3.2. BAM experiments

Fig. 3. Mean molecular areas measured at 30 mN/m on POPC–DChol (䊉) and POPC–Chol () isotherms vs. the percentage of sterol in the monolayer. Straight line corresponds to ideal mixtures as defined by the Eq. (1).

Same experiments were performed with POPC. Fig. 3 shows the mean molecular areas of mixtures estimated from the POPC–DChol (␲-A) isotherms at 30 mN/m (data not shown). Similar diagrams showing the condensation of POPC molecules are obtained at 10 mN/m and 20 mN/m (data not shown). Corresponding condensation percentages estimated, thanks to the Eqs. (2) and (3), increase with increasing amounts of DChol in the monolayer from 5 to 27%. Saturation occurs at DChol percentages higher than 40 mol%. Results obtained with POPC–Chol monolayers are very similar and superimposed in Fig. 3. Corresponding condensation percentages increase regularly until 28% and saturate at Chol percentages higher than 50 mol% [11]. We also performed experiments of DChol desorption, induced by ␤-CD, from SM or POPC monolayers, in order to have some information on the strength of the DChol–lipid interaction. Desorption percentages, measured at 30 mN/m, were estimated as follows: %desorption =

(Si − Sf ) × 100 , Si

(4)

where Si and Sf are the surfaces occupied by lipids between the two compression barriers before injection of ␤-CD and after 90 min, respectively. These percentages are summarized in Table 1. They are on the whole clearly lower in the case of SM–DChol monolayers (percentages of Chol desorption were in Table 1 Percentages of desorption of DChol from SM or POPC monolayers kept at 30 mN/m, induced by ␤-CD (0.5 mM) after 90 min (mean value of three different experiments)

SM monolayers POPC monolayers

10 mol% of DChol

20 mol% of DChol

50 mol% of DChol

22 ± 2 25 ± 2

16 ± 3 37 ± 4

18 ± 4 42 ± 5

Fig. 4 shows BAM images of pure SM, SM–DChol 10 mol% and SM–DChol 30 mol% monolayers performed at different surface pressures. Images of SM monolayers (Fig. 4a) show the presence of domains with typical shapes, appearing brilliant on a black background. These domains correspond to SM molecules in a condensed phase, their shape being likely related to intermolecular hydrogen bonds [17]. They are similar to those that we observed by fluorescence microscopy [19]. At the lateral resolution of the BAM setup, images of SM–DChol 10 mol% monolayers are homogenous for surface pressures in 0–10 mN/m range. When surface pressure increases above this value, small brilliant domains appear on a dark background (Fig. 4b). These domains can be assigned, as previously, to molecules, which are more condensed than in the surrounding phase. In the case of SM–DChol 30 mol% monolayers, images show the presence of some domains at low surface pressure (Fig. 4c), but become homogenous with further compression (Fig. 4d). Same experiments were performed with POPC–DChol 10 mol% monolayers. In this case, monolayers are homogenous, at the lateral resolution of the setup, during all the compression (data not shown). Previous experiments performed by fluorescence microscopy showed that pure POPC monolayers are homogenous at all surface pressures, in agreement with its liquid expanded isotherm [11], whereas POPC–Chol monolayers are homogenous at surface pressure above 10 mN/m [19]. 4. Discussion Surface pressure measurements show that DChol induces the condensation of POPC chains, with a similar efficiency than Chol [11,15]. In both cases, the sterol desorption is also strong and more or less in the same range, even if these values have to be used carefully (a previous study showed a desorption rate of Chol from phospholipids monolayers 10–15% higher than with DChol, assigned to the fact that DChol is a little more hydrophobic than Chol [7]). Moreover, POPC–DChol monolayers studied in this work are homogenous as POPC–Chol monolayers above 10 mN/m [11]. These observations suggest that DChol and Chol behave on the whole similarly in POPC monolayers, and are in agreement with previous studies dealing with DChol–phosphatidylcholine or phosphatidylethanolamine interaction. For instance the compressibility of DChol–DPPC and Chol–DPPC monolayers are almost the same [21], and DChol promotes the formation of DPPC-enriched domains as well as Chol [22]. DSC experiments show also that DChol and Chol decrease in the same extend the transition temperature of POPE [9]. Moreover DChol presents all the characteristics of “a membrane-active sterol”, as described by Barenholz [5]: a fused and flat ring system connected to a small polar head and a “Chol-like” tail, and a small mean molecular area. Under these conditions, the low

E. Lancelot, C. Grauby-Heywang / Colloids and Surfaces B: Biointerfaces 59 (2007) 81–86

85

Fig. 4. BAM images of SM and SM–DChol monolayers: (a) SM, 22 mN/m; (b) SM–DChol 10 mol%, 15 mN/m; (c) SM–DChol 30 mol%, 5 mN/m; (d) SM–DChol 30 mol%, 25 mN/m. Dimensions of all images are 430 ␮m × 535 ␮m.

selectivity of the interaction between hydrophobic sterol rings and phospholipid chains explains the similar behaviours of Chol and DChol. In particular, the absence of the C(5)-C(6) double bond in the ring system (Fig. 1) does not have a major importance in the case of POPC, likely because of the C9-C10 cis double bond on one chain of this lipid. The bend induced by this double bond hinders a close contact between sterol rings and POPC chains and limits interaction. Finally, DChol can favourably replace Chol, since it behaves similarly and is more stable. This is in agreement with the classification of DChol in “promoter sterols”, as described by Beattie et al. [23]. In the case of SM monolayers, DChol induces also the condensation of SM molecules, more particularly at low surface pressure, as shown by results obtained at 10 mN/m (Fig. 2B). Fig. 2B shows also the presence of an inflection point for DChol percentages around 30 mol%. When surface pressure increases, the condensing effect decreases or even disappears, but the inflection point remains at 30 mol% of DChol. This point has been already reported in a previous study, in the case of egg–SM–DChol monolayers at 22 mN/m, and is indicative of a complex formation [7]. However, no data is available to our knowledge for other pressure values. In our study, monolayers

seem to behave more or less ideally, or are slightly expanded at 30 mN/m, except at 20 or 30 mol% of DChol (Fig. 2B). These results are thus different from those reported with SM–Chol monolayers [11,15,17], where Chol induces SM condensation during all the compression and at all percentages, the maximum condensing effect being also obtained around 30 mol% of Chol [15]. Differences between DChol–SM and Chol–SM monolayers are also observed in BAM images. Previous studies reported that SM–Chol monolayers are heterogenous at low surface pressure, with the presence of small domains. These domains fuse together to form a stripe phase with further compression, and finally monolayers become homogenous [11,17]. Such a stripe phase has not been observed in this study, but has been reported in DMPC/DMPS/DChol and DMPC/DChol monolayers [24,25]. In the case of SM–DChol 10 mol% monolayers, BAM images are homogenous at surface pressures lower than 10 mN/m. This is in agreement with the results presented in the Fig. 2B, showing a lower mean molecular area than the area expected in the case of immiscible molecules. Small bright domains appear in BAM images of SM–DChol 10 mol% monolayers with further compression (Fig. 4b). Principle of BAM is to take advantage of the reflectivity properties of the air–water interface. In particular this

86

E. Lancelot, C. Grauby-Heywang / Colloids and Surfaces B: Biointerfaces 59 (2007) 81–86

reflectivity depends on the monolayer thickness and density [26]. Thus, the small bright domains can be assigned to lipids in a more condensed phase than the surrounding dark phase, and could be enriched in SM or DChol. This would explain the ideal behaviour of this mixture at 30 mN/m (Fig. 2B). Finally, ␤-CD experiments show a low desorption of DChol from these monolayers, suggesting that DChol is kept in the monolayer by some interactions. Since monolayers of pure DChol are strongly condensed (Fig. 2A), DChol–DChol interaction can explain the low desorption of this sterol. At last, BAM images of SM–DChol 30 mol% monolayers are homogenous during almost all the compression. This observation is in agreement with the inflection point systematically observed at this DChol percentage (Fig. 2B), which suggests that SM and DChol molecules are miscible. Thus, it seems that the small difference in chemical structures of DChol and Chol has a non-negligible influence on the interaction with SM. A recent Monte Carlo simulation performed on hydrated C18:0 sphingomyelin–Chol bilayers showed that the double bond on the sphingosine chain increases the rigidity of the molecule, when it is roughly perpendicular to the bilayer plane [27]. We can expect that this rigidity also occur during the compression of SM molecules at the air–water interface. Moreover, it is postulated that the double bond in B ring of Chol (Fig. 1) induces a cis form of A/B rings, whereas the single bond in DChol induces rather a trans one [21]. At last, elasticity measurements suggest also that Chol is more compressible than DChol [8]. The difference of conformation and compressibility between the two sterols and the rigidity of SM molecules could limit the “adjustment” of SM molecules with DChol at high surface pressure. 5. Conclusion DChol has a major advantage as compared to Chol in studies dealing with sterol–lipid interaction, since its chemical stability is better. Our results, in agreement with previous studies, suggest that the use of DChol in place of Chol is possible in cases where the sterol–lipid interaction is “non-specific” and only based on van der Waals interaction. However, if the interaction is more specific, this replacement can be problematic, as shown by results obtained with SM monolayers. In such cases,

a preliminary comparative study has to be performed between Chol and its analogue to check if they are really interchangeable. Acknowledgment We are grateful to Dr. Bernard Desbat from the CBMN (University Bordeaux 1) for lending us the BAM experimental setup. References [1] J. Fantini, N. Garmy, R. Mahfoud, N. Yahi, Expert Rev. Mol. Med. (2002) 1. [2] J.R. Silvius, Biochim. Biophys. Acta 1610 (2003) 174. [3] T.P.W. McMullen, R.N.A.H. Lewis, R.N. McElhaney, Curr. Opin. Colloid Interface Sci. 8 (2004) 459. [4] B. Ramstedt, J.P. Slotte, Biophys. J. 76 (1999) 908. [5] Y. Barenholz, Prog. Lipid Res. 41 (2002) 1. [6] D.J. Benvegnu, H.M. McConnell, J. Phys. Chem. 97 (1993) 6686. [7] A. Radhakrishnan, H.M. McConnell, Biochemistry 39 (2000) 8119. [8] M. Kodama, O. Shibata, S. Nakamura, S. Lee, G. Sugihara, Colloids Surf. B 33 (2004) 211. [9] P. Leppim¨aki, J. Mattinen, J.P. Slotte, Eur. J. Biochem. 267 (2000) 6385. [10] A. Radhakrishnan, T.G. Anderson, H.M. McConnell, PNAS 97 (2000) 12422. [11] C. Grauby-Heywang, J.-M. Turlet, Colloids Surf. B. 54 (2007) 211. [12] A. Blume, Biochim. Biophys. Acta 557 (1979) 32. [13] E. Takano, Y. Ishida, M. Iwabashi, Langmuir 13 (1997) 5782. [14] S. Lund-Katz, H.M. Laboda, L.R. MsLean, M.C. Phillips, Biochemistry 27 (1988) 3416. [15] J.M. Smaby, H.L. Brockman, R.E. Brown, Biochemistry 33 (1994) 9135. [16] S.R. Shaikh, A.C. Dumaual, L.J. Jenski, W. Stillwell, Biochim. Biophys. Acta 1512 (2001) 317. [17] S. Lalitha, A. Sampath Kumar, K.J. Stine, D.F. Covey, J. Supramolec. Chem. 1 (2001) 53. [18] M. Kuikka, B. Ramstedt, H. Ohvo-Rekil¨a, J. Tuuf, J.P. Slotte, Biophys. J. 80 (2001) 2327. [19] C. Grauby-Heywang, J.M. Turlet, Chem. Phys. Lipids 139 (2006) 68. [20] G.L. Gaines Jr., in: I. Prigogine (Ed.), From insoluble monolayers at liquid–gas interfaces, Intersciences, New York, 1966, Chapter 6. [21] H. Yamauchi, Y. Takao, M. Abe, K. Ogino, Langmuir 9 (1993) 300. [22] X. Xu, E. London, Biochemistry 39 (2000) 843. [23] M.E. Beattie, S.L. Veatch, B.L. Stottrup, S.L. Keller, Biophys. J. 89 (2005) 1760. [24] A. Radhakrishnan, H.M. McConnell, Biophys. J. 77 (1999) 1507. [25] S.L. Keller, T.G. Anderson, H.M. McConnell, Biophys. J. 79 (2000) 2033. [26] S. H´enon, J. Meunier, J. Chem. Phys. 98 (1993) 9148. [27] G.A. Khelashvili, H.L. Scott, J. Chem. Phys. 120 (2004) 9841.