Self-Organization of Synthetic Cholesteryl Oligoethyleneglycol

pubs.acs.org/Langmuir © 2009 American Chemical Society

Self-Organization of Synthetic Cholesteryl Oligoethyleneglycol Glycosides in Water )

Vincent Faivre,*,† Pierre-Louis Bardonnet,‡,§ Paul Boullanger, ,^,#,3,þ,z Heinz Amenitsch,O Michel Ollivon,† and Franc-oise Falson‡ Laboratoire de Physico-Chimie des Syst
emes Polyphas es, UMR CNRS 8612 - IFR 141, Universit e Paris-Sud, 5 rue J.B. Cl ement, 92296 Ch^ atenay-Malabry, France, ‡Laboratoire de Pharmacie Gal enique Industrielle, ISPB - Universit e Lyon I, 8 avenue Rockefeller, 69373 Lyon cedex 08, France, §Pharmapeptides, Campus Universitaire - Parc d’Affaires International, 74166 Archamps, France, ICBMS, Institut de Chimie et Biochimie Mol eculaires et Supramol eculaires, Chimie Organique 2 - Glycochimie, 43 boulevard du 11 novembre 1918, Villeurbanne, F-69622, France, ^CNRS, UMR 5246, Villeurbanne, F-69622, France, #Universit e de Lyon, Lyon, F-69622, France , 3Universit e Lyon 1, Lyon, F-69622, France , þINSA-Lyon, Villeurbanne, F-69622, France, z CPE Lyon, Villeurbanne, F-69616, France, and OInstitute of Biophysics and Nanosystems Research, Austrian Academy of Sciences, Schmiedlstrasse 6, A-8042 Graz, Austria )

†

Received February 9, 2009. Revised Manuscript Received June 29, 2009 Lectin-sugar recognition systems are of interest in the pharmaceutical field, especially for the development of drug carriers, tailored for selective delivery. This paper deals with the anhydrous and aqueous self-organization properties of a synthetic cholesteryl oligoethyleneglycol glycoside with the aim of their incorporation in liposomes. Successive phases (lamellar, R3m, Im3m, micelles) have been described depending on water content and temperature. As a result of the presence of sugar residues and their hydration ability, this glycolipid shows a large range of packing parameter with increasing water content. However, because of oligoethyleneglycol spacer, a slight dehydration has been observed with increasing temperature from 20 to 60 C.

1. Introduction Drug targeting of therapeutic vector is an interesting approach to increase the pharmacological effect of drugs and to reduce potential side-effects. Among receptor-ligand couples investigated,1 the lectin-sugar recognition systems are of interest in the pharmaceutical field, especially for the development of drug carriers, tailored for selective delivery.1-3 In this context, we focused our work on glycosides of cholesteryl oligoethyleneglycols. A carbohydrate was used as the ligand for molecular recognition; a tetraethyleneglycol moiety was used as a spacer between the carbohydrate and the lipid fractions to favor recognition of the headgroup; cholesterol was selected as the hydrophobic anchor. Such structure has been used to target with success Helicobacter pylori by the mean of glycosyled liposomes.4 The aim of the present work is to establish the phase diagram of this new surface active agent. In presence of water, the behavior of some cholesterol or sterol derivatives has been described in the literature. Among them can be noted short-chain polyoxyethylene cholesterol ether containing 3, 10, or 15 ethylene oxide (EO) units5,6 or polyoxyethylene *Corresponding author. Tel.: þ 33 1 46 83 54 65. Fax.: þ 33 1 46 83 53 12. E-mail: [email protected]. (1) Vyas, S. P.; Singh, A.; Sihorkar, V. Crit. Rev. Ther. Drug Carrier Syst. 2001, 18(1), 1–76. (2) Park, T. G.; Jeong, J. H.; Kim, S. W. Adv. Drug Delivery Rev. 2006, 58(4), 467–486. (3) Lehr, C.-M.; Gabor, F. Adv. Drug Delivery Rev. 2004, 56(4), 419–420. (4) Bardonnet, P.-L.; Faivre, V.; Boullanger, P.; Piffaretti, J.-C.; Falson, F. Eur. J. Pharm. Biopharm. 2008, 69(3), 908–922. (5) Rodriguez, C.; Naito, N.; Kunieda, H. Colloids Surf., A: Physicochem. Eng. Aspects 2001, 181(1-3), 237–246. (6) Sato, T.; Hossain, M. K.; Acharya, D. P.; Glatter, O.; Chiba, A.; Kunieda, H. J. Phys. Chem. B 2004, 108(34), 12927–12939. (7) Folmer, B. M.; Svensson, M.; Holmberg, K.; Brown, W. J. Colloid Interface Sci. 1999, 213(1), 112–120. (8) Folmer, B. M. Adv. Colloid Interface Sci. 2003, 103(2), 99–119.

9424 DOI: 10.1021/la900492j

phytosterol containing 5, 10, or 20 EO units.7,8 In the case of trioxyethylene cholesterol ether, a lamellar phase could be observed for surfactant concentration higher than 75%; at lower concentrations, a phase separation with an excess of aqueous phase occurs.5 When the number of EO units increased to 10 and 15,6 the phase diagrams of the binary water/CholEOn systems displayed successive lamellar, ribbon, hexagonal and cubic phases. Concerning phytosterol surfactants, it was found that the pentaoxyethylene derivative formed a lamellar phase for important surfactant concentration in water;7,8 when the concentration decreased to less than 40%, the lamellar phase separated. For 10-13 EO units, hexagonal phases are formed, followed by large lamellar and reverse micellar regions at increasing surfactant concentration. For longer hydrophilic chains, micellar, discontinuous cubic, and hexagonal phases were found. Other sterol surfactants, generally named “Chol-PEG”, have been extensively reported in the literature, however they have polymeric poly(ethylene oxide) (PEO) chains acting as steric stabilizer.9 As mentioned before, the extremity of the cholesterol derivative develop by us is a carbohydrate. Because of the large variety of glycolipid structures (in terms of carbohydrate head groups and lipophilic parts),10-12 it is not possible to give a general overview of their phase diagrams as a function of concentration or temperature. To our knowledge, cholesteryl glucosides described in the literature8,13-15 have only been studied for their biological (9) Beugin-Deroo, S.; Ollivon, M.; Lesieur, S. J. Colloid Interface Sci. 1998, 202 (2), 324–333. (10) Vill, V.; Hashim, R. Curr. Opin. Colloid Interface Sci. 2002, 7(5-6), 395– 409. (11) Collective, Carbohydr. Res. 1998, 312, 167-175. (12) Chester, A. Eur. J. Biochem. 1998, 257, 293–298. (13) Tannaes, T.; Grav, H. J.; Bukholm, G. Apmis 2000, 108(5), 349–356. (14) Smith, P. F. J. Bacteriol. 1971, 108(3), 986–991. (15) Schulz, J. D.; Hawkes, E. L.; Shaw, C. A. Med. Hypotheses 2006, 66(6), 1222–1226.

Published on Web 07/24/2009

Langmuir 2009, 25(16), 9424–9431

Faivre et al.

Article

Figure 1. Reaction scheme.

aspects and not from a physicochemical point of view. Despite a lack of studies on the cholesteryl glucosides, a strong difference between polyoxyethylene and the sugar-based surfactants should be already noted. The former is very sensitive to the temperature and they become less water-soluble at higher temperatures, whereas the latter exhibit a classical temperature dependence, i.e., their solubility in water increases with temperature. This paper deals with the anhydrous and aqueous self-organization properties of a cholesteryl oligoethyleneglycol glycoside. The temperatures of interest in the present work are between 20 and 60 C. The two temperatures have been chosen because they are the key temperatures of the liposome preparation process we used in previous studies.4 The behavior of these cholesteryl oligoethyleneglycol glycosides with water would impact on some parameters such as vesicle stability and fluidity or glycosylation rate.

2. Materials and Methods 2.1. General Synthetic Methods. The glycolipid 1 (Figure 1)

was prepared by glycosylation of acceptor 316 with donors 2.17 The synthesis has already been reported in the literature.16 2.2. Sample Preparation. Trace of solvent in the glycolipid preparation was removed by drying under vacuum for 12 h. Then the mixtures were prepared by weighing the required amounts of glycolipid and water into glass tubes and mixing with a vortex. Once the samples were homogeneous (visual control), they were left to equilibrate for at least 2 days. 2.3. X-ray Diffraction (XRD). Equilibrium XRD measurements at 20 and 60 C were performed using the following device: XRD patterns were recorded in transmission mode using quartz capillaries (1.5 mm diameter, GLASS W. M€ uller, Berlin, Germany); the X-ray generator was a long line-focus sealed tube (ENRAF NONIUS; Cu anode operating at 40 kV and 20 mA); two gas-filled linear detectors (1024 channels each, filled with argon-ethane mixture) were used to collect the data. With the settings used, scattering vectors q ranging from 0.04 to 0.37 A˚-1 and from 1.24 to 1.85 A˚-1 were accessible. The scattering vector is defined as q = [4π sin(θ)]/λ where 2θ is the scattering angle. From this scattering vector, it is possible to calculate the distances by the use of the following equation q = 2π/d. The calibration of the detectors was carried out on the peaks of the 2Lβ form of pure tristearin (4.61, 3.84, and 3.70 ( 0.01 A˚ and 44.97 ( 0.05 A˚) and that of the silver behenate (58.38 ( 0.01 A˚). In order to determine the peak positions, diffractograms were fitted with the Gaussian model by the use of IGOR pro software (WaveMetrics, Inc.). The full width of the Gaussian curve at half-maximum was calculated with Kaleidagraph 3.6 by using the following equation: width = 2σ(2 ln 2)1/2 in which σ is the standard deviation. For some mixtures, high-resolution data were taken at the Austrian small-angle X-ray scattering (SAXS) beamline of ELETTRA (Trieste, Italy). The calibration of the detectors and peak position determinations were carried out as described for the lab experiments. (16) Lafont, D.; Boullanger, P.; Chierici, S.; Gelhausen, M.; Roux, B. New J. Chem. 1996, 20, 1093–1101. (17) Boullanger, P.; Banoub, J.; Descotes, G. Can. J. Chem. 1987, 65, 1343–1348.

Langmuir 2009, 25(16), 9424–9431

Dynamic X-ray scattering experiments were also realized by modifying the temperature (1 C/min) of the sample during the measurements.18 2.4. Polarizing Optical Microscopy. Sample preparations were examined with a l/4 retarded in white light optical using a Nikon E600 Eclipse microscope (Champigny/Marne, France), equipped with a phase-contrast setup and a long focus objective (LWD 20  0.55; 0-2 mm). The images were recorded with a color Nikon Coolpix 950 camera at a resolution of 1600  1200 pixels. The samples were placed between two thin glass slides (26 mm thick) and observed (20) at room temperature. 2.5. Critical Micellar Concentration (cmc). Experiments were conducted in ultrapure water, obtained by osmosis from a MilliRO6 Plus Millipore apparatus (pH 5.5, surface tension >72 mN/m at 20 C). All glassware were cleaned with Texapon detergent and then abundantly rinsed with distilled water. The cmc of the glycolipid was determined by surface tension measurements performed by the Wilhelmy plate method using a Kr€ uss K10ST tensiometer (Germany). The surface tensions of the glycolipid solutions were measured after 16 h. In the absence of glycolipid, the stability of the system was checked by measuring the surface tension of water for 24 h. 2.6. Surfactant Adsorption Area. The mean molecular area of the adsorbed glycolipid at the air-water interface was calculated using the Gibbs adsorption equation:
 0:4343 dγ Γ ¼ RT d log C where R is the gas constant, T is the temperature, γ is the surface tension, and C is the concentration of glycolipid in the bulk phase. Since the surface tension of the glycolipid solutions was measured after 16 h, all presented values are equilibrium surface tensions, allowing calculation of the mean area per molecule. 2.7. Molecular Graphism. In order to approximate the dimensions of the glycolipids, the DS ViewerPro suite (Accelrys, Inc.) was used. The Dreiding force field was used to minimize the energy. General force constants and geometry parameters for the Dreiding force field are based on simple hybridization rules rather than specific combinations of atoms. The following three-step methodology was used: (i) building the molecular structure from the chemical bonding diagram, (ii) searching for plausible arrangements of the molecule, and (iii) optimizing the generated structure by energy minimization. The zigzag and the helical models have been taken into account as plausible arrangements of the polyethyleneglycol spacer fragment with regard to the literature.19,20

3. Results and Discussion 3.1. Anhydrous Glycolipid. The SAXS spectra of the anhydrous glycolipid at 20 and 60 C are displayed on Figure 2a; no WAXS diffraction peak could be detected (data not shown). Independent of the temperature, three diffraction peaks were (18) Ollivon, M.; Keller, G.; Bourgaux, C.; Kalnin, D.; Villeneuve, P.; Lesieur, P. J. Therm. Anal. Calorim. 2006, 85(1), 219–224. (19) Takahashi, Y.; Isao, S.; Tadokoro, H. J. Polym. Sci.: Polym. Phys. Ed. 1973, 11(11), 2113–2122. (20) Takahashi, Y.; Tadokoro, H. Macromolecules 1973, 6(5), 672–675.

DOI: 10.1021/la900492j

9425

Article

Faivre et al.

Figure 2. Diffractograms of (a) anhydrous glycolipids GlcNAc-E4-cholesterol at 20 C (full line) and 60 C (dotted line) and of (b) 20%

water-containing mixture at 20 C (full line) and 60 C (dotted line). Inset: Focus on the second and third order of the anhydrous glycolipid structure at 20 C. Table 1. Analysis of the Anhydrous Glycolipid SAXS Patternsa Experimental Values at 20 C

repetitive distance (A˚) width (A˚ -1)

60.3 ( 0.6 0.028

at 60 C repetitive distance (A˚) width (A˚ -1)

61.1 ( 0.4 0.014

Molecular Graphism Tool Calculations PEO helical model (A˚) 29.4 PEO zigzag model (A˚) 37.2 a “Width” refers to the full width at half maximum of the Gaussian curve, calculated for the peak corresponding to the first order.

observed and indexed as the three first orders of a lamellar phase. The periods of the organization are reported in Table 1. A similar lamellar arrangement has been described previously for anhydrous polyoxyethylene cholesteryl ethers containing 10 or 15 EO units6 or polyoxyethylene phytosterol containing 5 or 10 EO units.7 No significant increase in size could be noted with the temperature between 20 and 60 C. The most important results obtained by minimization of glycolipid conformations are summarized in Table 1. Assuming that the dimensions of the glycolipid strongly depend on the conformation of the PEO chain, which stands between zigzag19,21 and distorded helical conformations,20,22,23 it is possible to calculate lengths of the whole glycolipid molecule. The calculated (21) Evans, C. C.; Bates, F. S.; Ward, M. D. Chem. Mater. 2000, 12(1), 236–249. (22) Neyertz, S.; Brown, D.; Thomas, J. O. J. Chem. Phys. 1994, 101(11), 10064– 10073. (23) Krishnan, M.; Balasubramanian, S. Chem. Phys. Lett. 2004, 385(5-6), 351– 356.

9426 DOI: 10.1021/la900492j

lengths were 29.4 A˚ in the helical model, and 37.2 A˚ in the zigzag model. The validity of the models has been verified after Drieding minimization with regard to the length of the PEO chain. Indeed, it has been observed in crystals that the length per EO monomer was 2.8 A˚ in the helical arrangement and 3.6 A˚ in the zigzag arrangement. The distances per EO monomer measured after our calculations are also 2.8 A˚ and 3.6 A˚, respectively, suggesting that the energy minimization of the whole molecule did not affect the PEO chain conformation. This correlated well with the lamellar repetitive distances, extracted from the diffraction results. Also, the arrangement of the glycolipid within the lamella could be discussed. Two distinct organizations could be taken into account: (1) interdigitated structures, in which cholesterol anchors or hydrophilic chains melt, (2) a classical bilayer with two cholesterol layers in a sandwich of ethylene glycol plus sugar layers. The “interdigitated” models do not fit with the d-spacing measured (Figure 2) since such layers should have thickness around 45 A˚, which is lower than the 58-60 A˚ experimentally measured. The classical bilayer model is more appropriate and the glycolipid length in the “PEO helical arrangement” fitted very well the experimental values. Such a “PEO zig-zag arrangement” could accommodate a tilt angle to the normal of the bilayer around 30. Infrared spectroscopy data24 on anhydrous GlcNAcE4-cholesterol allow us to partially discriminate between the previous possibilities. The “zig-zag” model would lead to a specific band at 1500 cm-1 (J.B. Brubach, internal report). The lack of such absorption on the previously published data argues against the “zig-zag” model. This result could be confirmed by other infrared measurements, since PEO helices display specific absorptions in the range of 800-1400 cm-1. 3.2. Hydrated Glycolipid. 3.2.1. 20% (w/w) Water Content. For a glycolipid/water 80:20 (w/w) mixture, the lamellar (24) Kemoun, R.; Gelhausen, M.; Besson, F.; Lafont, D.; Buchet, R.; Boullanger, P.; Roux, B. J. Mol. Struct. 1999, 478(1-3), 295–302.

Langmuir 2009, 25(16), 9424–9431

Faivre et al.

Article

Figure 3. Diffractograms obtained at 20 C with a GlcNAc-E4cholesterol/water 75:25 mixture by the use of the Elettra Synchrotron beam. Numbers 1-9 correspond to the peak indexation used in Table 2.

arrangement remains visible (Figure 2). However, two significant changes could be observed: (1) an increase of the d-spacing (from 60.3 ( 0.6 A˚ to 63.7 ( 0.5 A˚ at 20 C and from 61.1 ( 0.4 A˚ to 63.3 ( 0.5 A˚ at 60 C) that could be attributed to the hydration of the sugar headgroup and (2) a decrease of the full width of the Gaussian curve at half-maximum, that could be due to an increase of the PEO chains degree of freedom, allowing a well-ordered arrangement of the glycolipid. 3.2.2. Water Content from 25 to 40% (w/w). From diffractograms described in Figures 3-5, it is apparent that at least two peaks appeared around 0.065 A˚-1 and 0.086 A˚-1 q-values. Because, it was difficult to gain further insight into the organization of the mixture, we studied the glycolipid/water 75:25 (w/w) mixture at 20 C with synchrotron radiation (Figure 3). Nine reflections could be distinguished on this diffractogram. Various intermediate phases, such as rectangular ribbons or bicontinuous cubic phase, have been tested to index the peaks observed on the diffractogram; however, high-order reflection cannot be fitted with these structures.25 Only two lattices, the rhombohedral mesh and the tetragonal mesh, can give a convincing explanation of the SAXS pattern. The results of the indexation are reported in Table 2. The rhombohedral R3m lattice could be described in two ways. In terms of rhombohedral axes, a, b and c are the shortest noncoplanar lattice vectors symmetrically equivalent with respect to the 3-fold axis. In that case, the Bragg peak positions were obtained according to that general equation:

dhkl

Figure 4. Diffractograms between 20 and 60 C obtained with the GlcNAc-E4-cholesterol/water 70:30 mixture. Temperature rate: 1 C/min.

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1 þ 2 cos3 R - 3 cos2 R ¼l 2 2 2 ðh þ k þ l Þ sin2 R þ 2ðhk þ hl þ klÞðcos2 R -cos RÞ

(25) Hyde, S. Identification of lyotropic liquid crystalline mesophases. In Handbook of Applied Surface and Colloid Chemistry; Holmberg, K., Ed.; John Wiley and Sons, Ltd: New York, 2001.

Langmuir 2009, 25(16), 9424–9431

Figure 5. Diffractograms between 20 and 60 C obtained with the GlcNAc-E4-cholesterol/water 60:40 mixture. Temperature rate: 1 C/min. Inset: d101 and d003 intensities vs temperature. DOI: 10.1021/la900492j

9427

Article

Faivre et al.

Table 2. Peak Positions, Tetragonal and Rhombohedral Indexations for Glycolipid 1/Water (75:25) at 20Ca tetragonal position 1 2 3 4 5 6 7 8 9

rhombohedral

Table 3. a and c Parameters of the Rhombohedral Mesh Phase at 20 C for Mixtures Containing 70% and 60% of GlcNAc-E4Cholesterol 1

dobs (A˚) plane dcal (A˚) % error plane dcal (A˚) % error 99.9 71.8 62.5 47.0 41.8 34.1 31.6 29.9 20.7

110 200 101 211 301 330 202 411 303

101.5 71.8 62.5 47.1 39.4 33.9 31.3 31.1 20.8

1.6 0 0 0.2 6.1 0.6 0.9 3.9 0.5

101 110 003 113 030 131 006 042 060

103.7 71.8 62.5 47.2 41.5 33.9 31.3 29.6 20.7

3.7 0 0 0.4 0.7 0.6 0.9 1.0 0

mean error (1.5% (0.8% a (A˚) 143.6 143.6 c (A˚) 69.4 187.6 a The diffractogram was obtained by the use of the Austrian SAXS beamline facilities in the ELETTRA Synchrotron (Trieste, Italy).

In the present study, the rhombohedral R3m phase has been described in terms of a hexagonal lattice in which c lies along the 3-fold symmetry axis, with a and c chosen as for the hexagonal system.26 The calculated Bragg peak positions were then obtained more easily according to the equation for the hexagonal symmetry:27 a dhkl ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 4 2 2 2 a2 3 ðh þ k þ hkÞ þ l c2 in which h, k, and l are the Miller indexes, and a (= b) ¼ 6 c are the unit cell parameters. The condition limiting possible reflections is -h þ k þ l = 3n, with n being an integer. Concerning the tetragonal mesh phase, the following equation has been used:27 a dhkl ¼ rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi  ffi h2 þ k2 þ l 2

a c

2

The condition limiting possible reflections is h þ k þ l = pair number. Both these organizations correspond to stack of “punctured” bilayers, together with an ordered arrangement of punctures within each bilayers.25 The rhombohedral mesophase (R3m space group) contains a hexagonally close-packed array of punctures and the tetragonal mesophase (space group I422) contains a square array of punctures. By using the two indexing schemes, the lattice parameters are a = 14.36 nm and c = 18.76 nm in the R3m space group and a = 14.36 nm and c = 6.94 nm in the I422 space group. The position mean errors are 0.8% and 1.5% in rhombohedral and tetragonal models, respectively. Such error values are very close to that reported in the literature, confirming they are realistic models.28-31 However, on the basis of these position mean errors, the rhombohedral structure seems better (26) Tilley, R. Crystals and Crystal Structures. 1 ed.; John Wiley and Sons: Chichester, U.K., 2006; p 270. (27) Guinier, A. Theorie et Technique de la Radiocristallographie; Dunod: Paris, 1964; p 740. (28) Funari, S. S.; Holmes, M. C.; Tiddy, G. J. T. J. Phys. Chem. 1994, 98(11), 3015–3023. (29) Fairhurst, C. E.; Holmes, M. C.; Leaver, M. S. Langmuir 1996, 12(26), 6336–6340. (30) Fairhurst, C. E.; Holmes, M. C.; Leaver, M. S. Langmuir 1997, 13(19), 4964–4975. (31) Burgoyne, J.; Holmes, M. C.; Tiddy, G. J. T. J. Phys. Chem. 1995, 99(16), 6054–6063.

9428 DOI: 10.1021/la900492j

a (A˚) c (A˚)

70%

60%

141.4 187.5

146.6 188.4

than the other one. Furthermore, an epitaxial relation between the d003 reflection of the rhombohedral structure and the periodicity in the lamellar arrangement has been reported in the literature.32,33 Interestingly, the value of the unit cell parameter c is very close to 3d, where d is the lamellar periodicity of the neighboring phase. Such a correlation, observed for the glycolipid/water 75:25 (w/w) mixture at 20 C, strongly suggests the probable R3m arrangement, which corresponds to a three-layer stacking of punctured bilayers. It is important to note here that most of the studies on mesh phases have been carried out on poly(oxyethylene glycol) containing nonionic surfactants, where they occur over a rather wide range of composition. Furthermore, all the ordered mesh phases observed with these compounds have a rhombohedral structure belonging to the space group R3m.34 Because the three first reflections (d101, d110, and d003) are very similar at 70% and 60% (Figures 4 and 5), we think that the organization in the mixture did not change significantly for these concentrations at 20 C. With this assumption, it is possible to calculate the unit cell parameters at these two concentrations (Table 3). Indeed, d110 and d003 are directly correlated to parameters a and c, respectively. With the 70% glycolipid-containing mixture, a transition to a lamellar phase could be detected around 52 C; the lamellar period is 63.9 ( 0.5 A˚ at 60 C. This R3m to lamellar phase transition is consistent with other observations of a decrease in hydration number, or a decrease in water concentration near the headgroup, at higher temperatures reported for EO chains with spectroscopy and water self-diffusion measurements35 or micellar growth experiments.36 If there is only “turbostratic” interaction between the adjacent layers, Hyde underlined the smectic ordering between them leading to “lamellar” small-angle scattering pattern. Because strong lamellar reflections dominate in the XRD profiles, it is important to exclude the possibility that a lamellar phase would coexist with other phases in the phase diagram. Figure 6 summarizes polarized-light microscopy observations made with the anhydrous and 30% water-containing mixtures. Samples were macroscopically homogeneous, and their polarizing optical microscopy textures were completely different. The two images shown in Figure 6 were observed in all sample domains checked, confirming the homogeneity and the absence of phase coexistence in the 70% glycolipid-containing mixture. With 60% of glycolipid, the intensity of the d101 peak decreases with increasing temperature, while the d003 reflection intensity slightly increases. This confirms the possible dehydration of the EO chains with increasing temperature. 3.2.3. Water Content from 40 to 50% (w/w). For the mixture containing 50% of water, at 20 C, the SAXS pattern significantly changed compared with that recorded at 60% and 70% of water (Figure 7). The peak corresponding to the d110 (32) Leaver, M.; Fogden, A.; Holmes, M. Langmuir 2001, 17(1), 35–46. (33) Ghosh, S. K.; Ganapathy, R.; Krishnaswamy, R.; Bellare, J.; Raghunathan, V. A.; Sood, A. K. Langmuir 2007, 23(7), 3606–3614. (34) Raghunathan, V. A. J. Indian Inst. Sci. 2008, 88(2), 197–210. (35) Jonsson, B.; Lindman, B.; Holmberg, K.; Kronberg, B. Surfactants and Polymers in Aqueous Solution; John Wiley and Sons, Ltd: Chichester, U.K., 1998; p 438. (36) Briganti, G.; Bonincontro, A. J. Non-Cryst. Solids 1998, 235-237, 704–708.

Langmuir 2009, 25(16), 9424–9431

Faivre et al.

Article

Figure 6. Typical polarizing optical microscopy textures of the anhydrous glycolipid (left) and the 30% water-containing mixture (right).

by the following equation: shkl ¼

Figure 7. Diffractograms between 20 and 60 C obtained with the GlcNAc-E4-cholesterol/water 50:50 mixture. Temperature rate: 1 C/min.

reflection of the R3m phase disappeared, those corresponding to the d101 and d003 reflections were displaced toward smaller q values (0.0566 A˚-1 and 0.0937 A˚-1, respectively); at the same time, new reflections appeared at 0.1180 A˚-1 and 0.1762 A˚-1. However, to our knowledge, it is not possible to affect unambiguously the whole pattern to a known structure. Between 20 and 60 C, an organization change could be observed around 40 C (Figure 7). The diffraction peaks at 0.0915, 0.1291, 0.1572, and 0.18202 A˚-1 fit very well with the Im3m space group in which the four first√reflections to the lattice parameter (a) by √ √ are correlated √ factors 2, 4, 6, and 8. The Im3m (or Q229) space group, belongs to the cubic phases, characterized by diffraction patterns in which the reciprocal spacings (shkl) of the Bragg peaks are given Langmuir 2009, 25(16), 9424–9431

qhkl ðh2 þ k2 þ l 2 Þ1=2 ¼ 2π a

where h, k, and l are the Miller indices, and a is the lattice parameter. The indexation (hkl) of the first allowed Bragg reflections in the Im3m space group are 110, 200, 211, 220, 310, 222, 321, 400, etc. In the investigated organization, the lattice parameter (a) could be calculated at 97.5 ( 0.4 A˚. The Im3m cubic phase could be either a bicontinuous or a micellar structure; its topology could be either normal (type I, “oil-in-water”) or inverted (type II, “waterin-oil”). The bicontinuous structure is characterized by orthogonal networks of water channels, connected six-by-six, and separated by a lipid bilayer folded as an infinite periodic minimal surface (P-minimal surface).37,38 The micellar structure is constituted by quasi-spherical micelles closely packed in the bodycentered model.39 As a result of the presence of PEO chains in the glycolipid and the way by which the Im3m cubic phase is obtained (addition of water to a lamellar structure), it could be assumed that the topology of our system is normal (type I). It is also probably possible to discriminate between the bicontinuous and the micellar arrangement. Indeed, the Im3m bicontinuous phase is usually inverted, although one example of type I has been reported.37 Furthermore, the lattice parameter (a = 97.5 A˚) is not compatible with that of a bicontinuous structure in which this parameter should overlap at least the thickness of two bilayers. Then it can be concluded that, at 60 C, for a 50/50 glycolipid/ water mixture, glycolipid self-organizes as micelles packed in the body-centered mode. 3.2.4. Water Content Higher than 60% (w/w). By increasing the amount of water (from 50% to 60%), it is possible to lose the cubic organization of the micelles. These “less organized micelles” are stable with the temperature (data not shown) and seem sensitive to the glycolipid concentration in the range (2040%), as it can be seen on the corresponding SAXS patterns in the inset of the Figure 8. The more concentrated one (60% water) shows two important shoulders around 0.10 and 0.16 A˚-1, respectively. The first shoulder is slightly present in the more dilute sample, while the second one is totally absent. This strongly

(37) Seddon, J. M.; Templer, R. H. Polymorphism of lipid-water systems. In Handbook of Biological Physics; Lipowsky, R., Sackmann, E., Eds.; Elsevier Science B.V.: Amsterdam, 1995; Vol. 1. (38) Delacroix J. Microsc. 1998, 192(3), 280–292. (39) Luzzati, V.; Delacroix, H.; Gulik, A. J. Phys. II France 1996, 6, 405–418.

DOI: 10.1021/la900492j

9429

Article

Figure 8. Surface tension versus concentration of GlcNAc-E4cholesterol. Inset: diffractograms of 60% (upper) and 80% (lower) water-containing mixture at 20 C.

suggests that the glycolipid concentration affects the degree of order in this phase diagram region. Surface tensions versus GlcNAc-E4-cholesterol concentrations in water are reported on Figure 8. The time required to reach surface tension equilibrium was around 6-8 h (data not shown). The value of the cmc could be estimated around 0.091 mM, and the calculated Gibbs area is 55.8 A˚2/molecule. Compared with polyoxyethylene derivatives of fatty alcohol, the sterol-based surfactants have a hydrophobic tail whose characteristics are (i) a higher number of carbon atoms (28 vs 10-14 for the aliphatic fatty alcohol), (ii) a higher steric hindrance, and (iii) a higher rigidity (the four condensed rings constitute a fairly stiff skeleton). The long time required to reach equilibrium surface tension is most likely due to the characteristics of the hydrophobic tail in sterol surfactants in which the alignment at the interface is hindered by the rigid skeleton. The low value of surface tension at the cmc is close to that reported previously for polyoxyethylene phytosterol derivatives containing 10 oxyethylene units.7 However, it should be noted that the cmc value is much higher in the present study (91 μM versus 10 μM). This difference could be explained by the nature of the sterol anchors (29 carbons in phytosterol and 27 in cholesterol), by the length of the polyoxyethylene moieties and by the presence of a hydrophilic sugar residue at the end of the latter in the glycolipid. For polyoxyethylene phytosterol, the shape parameter (s) was reported to be close to 1. This parameter (s) is defined as v0/(a 3 l0), where v0 is the volume of the hydrocarbon tail, a is the headgroup area, and l0 is the extended length of the tail. A shape parameter close to 1 is indicative of a rod-like shape of the molecule unfavorable to micellization for geometrical reasons. Therefore, it was shown that polyoxyethylene phytosterol containing five oxyethylene units was insoluble in water and does not form a single phase until a concentration of approximately 30 wt %, where a lamellar phase formed.7 The hydrophilic headgroup is too small for this surfactant to pack into discrete aggregates in water; hence there is a phase separation into a water 9430 DOI: 10.1021/la900492j

Faivre et al.

Figure 9. Summary of the different phases observed during the hydration of GlcNAc-E4-cholesterol. Lines are guides for the eyes, but the error of their positions in terms of glycolipid percentage is (5%.

phase and a lamellar liquid crystalline region. In the case of our glycolipid, micellization is possible for glycolipid concentrations lower than 50% (see X-ray diffractograms in the inset of Figure 8), thus suggesting that the sugar moiety was able to induce the formation of discrete aggregates. Furthermore, the Im3m organization previously depicted requires quasi-spherical micelles. In that case, by considering a micelle with a core radius r, made up of n molecules, simple geometrical considerations allow one to calculate the volume of the core, v = nv0 = 4πr3/3, the surface area of the core, A = na = 4πr2, and hence r = 3v0/a. In tightly packed micelles, the radius r cannot exceed the length l0 of the tail. Introducing this limit in the expression of r, a range of 0 e s e 1/3 was found for spherical micelles.40 It is generally acknowledged that the surface area occupied by cholesterol at the airwater interface is around 39 A˚2.41 Furthermore, cholesterol adopts only a close-packed arrangement at the interface, as attested by the low-extent of the surface pressure-area isotherm and the high collapse pressure observed in Langmuir films. By using the Gibbs equation, the calculated area per molecule of glycolipid at the air-water interface is 56 A˚2. Molecular graphism tools allow one to calculate the shape parameter for GlcNAc-E4cholesterol in hydrated conditions: the volume v0 (363.7 A˚3), and the length l0 (17.9 A˚) of the hydrophobic tail are close to the data reported in the literature.42 By using these values, a shape parameter of 0.36 could be calculated, thus confirming the quasi-spherical shape of the micelles. Contrariwise to nonglycosylated cholesterol derivatives, the cone shape of the glycolipid is favorable to the formation of discrete structure in solution. For this reason, no phase separation between a water phase and a lamellar liquid crystalline region was observed. Successive transitions resulted in changes in curvature, induced by hydration of glycolipid headgroup. (40) Nagarajan, R. Langmuir 2002, 18(1), 31–38. (41) Mingotaud, A.-F.; Mingotaud, C.; Patterson, L. K. Handbook of Monolayers; Academic Press, Inc.: San Diego, CA, 1993; Vol. 2. (42) Kumar, V. V. Proc. Natl. Acad. Sci. U.S.A. 1991, 88, 444–448.

Langmuir 2009, 25(16), 9424–9431

Faivre et al.

Article

4. Conclusion In this work, the self-organization of a cholesteryl tetraethyleneglycol containing glycolipids in the dry state and in the presence of increasing amounts of water was studied. As summarized in Figure 9, successive structures were observed, from lamellar phase to micelles. In the presence of water, the observed phases were different from those described for cholesteryl oligoethyleneglycols. Such differences are probably due to the presence of the sugar moiety, which increases hydration ability of the hydrophilic headgroup, inducing shape parameters in the range 1 to 0.36. This range of shape parameter is responsible of curvature changes and phase transitions. However, as it is expressed by the R3m to the

Langmuir 2009, 25(16), 9424–9431

LR temperature-dependent transtition, the glycolipid becomes less water-soluble at higher temperatures like cholesteryl oligoethyleneglycol derivatives. This tendency to micellization should be taken into account for the preparation of glycosyled liposomes, and the incorporation of an important amount of glycolipid should be managed carefully to avoid liposome destabilization. Acknowledgment. The authors gratefully acknowledge C. Bourgaux and G. Le Bas for fruitful discussions on XRD and molecular graphism, respectively, and V. Rosilio for the access to the tensiometer. This paper is dedicated to Michel Ollivon.

DOI: 10.1021/la900492j

9431