6328
Langmuir 2006, 22, 6328-6334
“Sponge” Nanoparticle Dispersions in Aqueous Mixtures of Diglycerol Monooleate, Glycerol Dioleate, and Polysorbate 80 Justas Barauskas,*,†,‡ Audrius Misiunas,§ Torsten Gunnarsson,‡ Fredrik Tiberg,†,‡ and Markus Johnsson*,†,‡ Physical Chemistry 1, Center for Chemistry and Chemical Engineering, Lund UniVersity, P.O. Box 124, SE-221 00 Lund, Sweden, Department of Bioelectrochemistry and Biospectroscopy, Institute of Biochemistry, Mokslininkø 12, LT-086 62 Vilnius, Lithuania and Camurus AB, So¨lVegatan 41, Ideon Science Park Gamma 1, SE-223 70 Lund, Sweden ReceiVed January 31, 2006. In Final Form: April 28, 2006 Lipid nanoparticles of nonlamellar lyotropic phases have a wide solubilizing and encapsulating spectrum for a range of substances thanks to their nanostructured interior featuring both lipophilic and hydrophilic domains. As a consequence, these systems have emerged as promising drug delivery systems in various pharmaceutical and diagnostic applications. Here we present the phase behavior and dispersion properties of a novel three-component lipid system composed of diglycerol monooleate (DGMO), glycerol dioleate (GDO), and polysorbate 80 (P80) which shows several advantageous features relating to drug delivery applications including: spontaneous dispersion formation with a narrow size distribution and tunable particle phase-structure. The obtained phase diagram shows the presence of lamellar (LR), hexagonal (H2), and reverse bicontinuous cubic (V2) liquid crystalline phases and an inverse micellar (L2) solution. A particularly interesting observation is the presence of a phase region where two liquid phases coexist, most likely the L2 and L3 (“sponge phase”). These two phase structures appear also to coexist in the submicron particles formed in the dilute water region, where the L3 element appears to stabilize nanoparticles with inner L2 structure. Increasing the fraction of the dispersing P80 component results in the growth of the more water rich L3 “surface phase” at the expense of the size of the inner L2 core.
Introduction Lipids and surfactants in aqueous solutions self-assemble into various nonlamellar liquid crystalline phases of different morphologies.1,2 With decreasing spontaneous curvature, the most common reversed mesophases are L3 (sponge), reversed bicontinuous cubic (V2), reversed hexagonal (H2), and reversed micellar cubic (I2) phases.3 In addition, lipids can also form intermediate self-assemblies with other nonlamellar geometries and topologies, including mesh phases and branched and punctured bilayer morphologies.4 The continuing importance of studies of lipid self-assembly is demonstrated by the formation of nonlamellar and intermediate geometries in membrane-forming lipid systems.5-7 In analogy with vesicles and liposomes, reversed mesophases can be dispersed into biomimicking particle structures in systems comprising a nonlamellar phase in equilibrium with excess water. Such dispersions of reversed mesophases have many potential uses in science and technology.8-11 Suitable applications of these * To whom correspondence should be addressed. (J.B.) Tel.: +46 46 222 8175. Fax: +46 46 222 4413. E-mail:
[email protected]. (M.J.) Tel.: +46 46 286 5745. Fax: +46 46 286 5739. E-mail: Markus.Johnsson@ camurus.com. † Lund University. ‡ Camurus AB. § Institute of Biochemistry. (1) Luzzati, V. In Biological Membranes, Physical Facts and Function; Chapman, D., Ed.; Academic Press: New York, 1968; p 71. (2) Larsson, K. J. Phys. Chem. 1989, 93, 7304-7314. (3) Hyde, S. T. In Handbook of Applied Surface and Colloid Chemistry; Holmberg, K., Ed.; John Willey & Sons: New York, 2001; p 299. (4) Hyde, S. T.; Schro¨der, G. E. Curr. Opin. Colloid Interface Sci. 2003, 8, 5-14. (5) Landh, T. FEBS Lett. 1995, 369, 13-17. (6) Almgren, M. Biochim. Biophys. Acta 2000, 1508, 146-163. (7) Yang, L.; Huang, H. W. Science 2002, 297, 1877-1879. (8) Larsson, K. Curr. Opin. Colloid Interface Sci. 2000, 5, 64-69.
colloidal systems include mimetics of biomembranes,12,13 matrixes for protein crystallization,14,15 applications in material sciences,16 and delivery vehicles of therapeutic agents.17,18 Particles with a reversed bicontinuous structure were first observed in a study of fat digestion.19 It was later found that that these structures could be prepared by dispersing glycerol monooleate (GMO) in water in the presence of bile salt20 or polymeric stabilizers.21-23 After these pioneering reports, a number of studies have been focused on the GMO-based cubic phase colloidal dispersions.24-29 It has also been shown to be (9) Yang, D.; Armitage, B.; Marder, S. R. Angew. Chem., Int. Ed. 2004, 43, 4402-4409. (10) Barauskas, J.; Johnsson, M.; Tiberg, F. Nano Lett. 2005, 5, 1615-1619.. (11) Spicer, P. T. Curr. Opin. Colloid Interface Sci. 2005, 10, 274-279. (12) Lindblom, G.; Rilfors, L. Biochim. Biophys. Acta 1989, 988, 221-256. (13) Luzzati, V. Curr. Opin. Struct. Biol. 1997, 7, 661-668. (14) Landau, E. M.; Rosenbusch, J. P. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 14532-14535. (15) Caffrey, M. Curr. Opin. Struct. Biol. 2000, 10, 486-497. (16) O′Brien, D. F.; Armitage, B.; Benedicto, A.; Bennet, D. E.; Lamparski, H. G.; Lee, Y.-K.; Srisiri, W.; Sisson, T. M. Acc. Chem. Res. 1998, 31, 861-868. (17) Shah, J.; Sadhale, Y.; Chilikuri, D. M. AdV. Drug DeliVery ReV. 2001, 47, 229-250. (18) Drummond, C. J.; Fong, C. Curr. Opin. Colloid Interface Sci. 2000, 4, 449-456. (19) Patton, J. S.; Carrey, M. C. Science 1979, 204, 145-148. (20) Larsson, K. J. Phys Chem. 1989, 93, 7304-7314. (21) Landh, T. J. Phys. Chem. 1994, 98, 8453-8467. (22) Gustafsson, J.; Ljusberg-Wahren, H.; Almgren, M.; Larsson, K. Langmuir 1996, 12, 4611-4613. (23) Gustafsson, J.; Ljusberg-Wahren, H.; Almgren, M.; Larsson, K. Langmuir 1997, 13, 6964-6971. (24) Neto, C.; Aloisi, G.; Baglioni, P.; Larsson, K. J. Phys. Chem. 1999, 103, 3896-3899. (25) Monduzzi, M.; Ljusberg-Wahren, H.; Larsson, K. Langmuir 2000, 16, 7355-7358. (26) Nakano, M.; Sugita, A.; Matsuoka, H.; Handa, T. Langmuir 2001, 17, 3917-3922. (27) Spicer, P. T.; Hayden, K. L.; Lynch, M. L.; Ofori-Boateng, A.; Burns, J. L. Langmuir 2001, 17, 5748-5756. (28) Siekman, B.; Bunjes, H.; Koch, M. H. J.; Westesen, K. Int. J. Pharm. 2002, 244, 33-43.
10.1021/la060295f CCC: $33.50 © 2006 American Chemical Society Published on Web 06/06/2006
“Sponge” Nanoparticle Dispersions
Langmuir, Vol. 22, No. 14, 2006 6329
Figure 1. Molecular structures of diglycerol monooleate (DGMO) (a), glycerol dioleate (GDO) (b), and Polysorbate 80 (P80) (c). Note that all three lipids/amphiphiles are mixtures of different isomers and that the molecular structures represent one of the dominating species in the respective samples.
possible to swell and fragmentize H223 and L3 phases (disordered sponge)30 by admixing small amounts of triglyceride and sodium cholate into the GMO matrix, respectively. However, until now, only a very small number of lipid constituents have been available for making such particle structures. With a few but increasing in number recent exceptions, including glycolipid,31 phytantriol,10 phospholipid,32 and glycerate analogue-based33 systems, only unsaturated monoglycerides in the presence of nonionic block copolymers such as Pluronic F127 have been shown to possess appropriate and straightforward phase behavior and dispersion properties. Alternatively, the latest findings demonstrate the possibilities to produce different nonlamellar particle morphologies with a high degree of internal order and uniformity by choosing suitable lipid combinations.10,34,35 All key particle properties, size, shape, and internal mesophase structure, can be tuned by using two structure forming lipids and an appropriate processing scheme. Herein we present yet a new system for making disordered nonlamellar spongelike lipid nanoparticles with uniform size distribution and internal structure. In a previous study, we showed that aqueous mixtures of diglycerol monooleate (DGMO, Figure 1a) and glycerol dioleate (GDO, Figure 1b) display a multitude of mesophases including lamellar, two V2, H2, and L2 phases.35 Furthermore, monocrystalline H2 phase nanoparticles are easily formed in excess water in the presence of small amounts of the polymeric dispersion stabilizer Pluronic F127.10,35 Here we investigate the aqueous phase behavior of DGMO/GDO mixtures in the presence of another nonionic dispersion stabilizer (or fragmentation agent), Polysorbate 80 (P80, Figure 1c). In contrast to relatively long Pluronic polymers, P80 has a significantly lower molecular weight, shorter hydrophilic poly(ethylene glycol) (29) Esposito, E.; Eblovi, N.; Rasi, S.; Drechsler, M.; Di Gregorio, G. M.; Menegatti, E.; Cortesi, R. AAPS Pharm. Sci. 2003, 5, A30. (30) Gustafsson, J.; Nylander, T.; Almgren, M.; Ljusberg-Wahren, H. J. Colloid Interface Sci. 1999, 211, 326-335. (31) Abraham, T.; Hato, M.; Hirai, M. Biotechnol. Prog. 2005, 21, 255-262. (32) Johnsson, M.; Barauskas, J.; Tiberg, F. J. Am. Chem. Soc. 2005, 127, 1076-1077. (33) Fong, C.; Krodkiewska, I.; Wells, D.; Boyd, B. J.; Booth, J.; Bhargava, S.; McDowall, A.; Hartley, P. G. Aust. J. Chem. 2005, 58, 683-687. (34) Barauskas, J.; Johnsson, M.; Joabsson, F.; Tiberg, F. Langmuir 2005, 21, 2569-2577. (35) Johnsson, M.; Lam, Y.; Barauskas, J.; Tiberg, F. Langmuir 2005, 21, 5159-5165.
(PEG) chains, and a hydrophobic anchor consisting of an oleic acid residue. This makes P80 more miscible with respect to the structure forming lipids and more able to penetrate and swell liquid crystalline phases as well as promote dispersion formation. Indeed, dispersions are easily obtained with a minimum of energy input in the presence of P80. This is an important finding since it alleviates the need for high-energy treatment in preparation of well-defined dispersions and also facilitates the use of preconcentrates (e.g., liquids in capsules) and nanoparticle formation in situ. The particle morphology is investigated by cryo-TEM as a function of P80 content. Well-defined nanoparticles stabilized by a L3 “surface phase” are visualized and the characteristics of the dilute dispersions are discussed and related to the bulk phase behavior of the concentrated lipid/water system. Experimental Section Materials. Glyceroldioleate (GDO) and diglycerolmonooleate (DGMO) were obtained from Danisco (Brabrand, Denmark). The composition of the GDO sample as given by the manufacturer was the following: diglycerides 87.5%, monoglycerides 11%, and triglycerides 1.5%. The main fatty acid component was oleoyl (C18: 1) constituting 91% of the acyl chains. The composition of the DGMO sample was as follows: diglycerol monoester 89% and free glycerol and polyglycerol 5.1%. The balance comprises diglycerol diester and small amounts of mono- and diglycerides. The fatty acid distribution was oleic (C18:1) 90%, saturated 4.7%, and linoleic and linolenic 5.3%. Polyoxyethylene(20) sorbitan monooleate (Polysorbate 80, P80) was purchased from Apoteket (Sweden). Milli-Q purified water was used for all experiments. All other solvents and reagents were of analytical grade and were used as received. Sample Preparation. Bulk samples were prepared by co-melting (50 °C) appropriate amounts of DGMO, GDO, and P80 into 14 mm (i.d.) glass ampules (total lipid amount of ca. 0.5-1 g) and then adding water. The vials were immediately sealed and vortexed for several seconds to distribute water inside the samples. The samples were allowed to equilibrate at 25 °C for at least 3 weeks, with intermittent centrifugation at 1500g for 10 min. To obtain phase homogeneity, samples were centrifuged up and down, before measurements. If needed, the centrifugation was repeated several times. The samples were investigated between crossed polarizers to check sample homogeneity and the presence of birefringent phases. Dispersion Preparation. Spontaneous dispersions were prepared by adding appropriate amounts of the melted lipid mixture (DGMO/
6330 Langmuir, Vol. 22, No. 14, 2006 GDO/P80 with DGMO/GDO ) 50/50 wt/wt) into an aqueous solution. The lipid (DGMO + GDO)/P80 ratio was varied between 9/1 and 6/4 (wt/wt) and the water concentration was generally 95 wt % unless otherwise stated. The total sample volume was usually 50-100 mL. The samples were immediately sealed, hand-shaken, and mixed for 12 h on a mechanical mixing table at 350 rpm and room temperature. In all cases, this procedure resulted in homogeneous dispersions with narrow and monomodal size distributions. Heat treatment procedure was performed in order to improve dispersion properties in terms of reducing the amount of metastable vesicular aggregates. Heat treatment of the spontaneous (mechanically shaken) dispersions was performed using a bench-type autoclave (CertoClav CV-EL, Certoclav Sterilizer GmbH, Traun, Austria) operated at 125 °C and 1.4 bar. The dispersions were filled into Pyrex glass bottles and put into the autoclave. A period of about 12 min was required to vent the entrapped air and to heat up the autoclave. The samples were then subjected to heat treatment for 20 min at 125 °C. After the heat treatment, the samples were allowed to cool to room temperature before analysis. X-ray Diffraction (XRD). XRD measurements were performed on a Kratky compact small-angle system equipped with a positionsensitive wire detector (OED 50M from MBraun, Graz, Austria) containing 1024 channels of width 53.6 µm. Cu KR radiation of wavelength 1.542 Å was provided by a Seifert ID 3000 X-ray generator operated at 50 kV and 40 mA. Bulk LC samples were mounted between mica sheets in a steel body, whereas dispersed samples were filled into a 1 mm (i.d.) quartz capillary in a steel sample holder. To minimize scattering from air, the camera volume was kept under vacuum during the measurements. Temperature control within 0.1 °C was achieved using a Peltier element. The recorded slit-smeared diffraction patterns were desmeared and evaluated using 3D-View software (MBraun, Graz, Austria). Particle Size Measurements. Particle size distributions were measured using a Coulter LS230 laser diffraction particle size analyzer (Beckman-Coulter, Inc., Miami, U.S.A.) which operates on the principles of Fraunhofer diffraction for large particles (0.4-2000 µm) and uses the polarization intensity differential scattering (PIDS) method for small particles (0.04-0.5 µm). The instrument was fitted with a 125 mL volume module. Data were collected during 90 s. A standard model based on homogeneous oil spheres with a refractive index (RI) of 1.46 was used for the particle size calculations. The change of RI to either side only shifts the obtained particle size distributions within a few percents. Note that the model is based on spherical particles and the measured mean particle size is calculated based on this assumption. Cryogenic Transmission Electron Microscopy (cryo-TEM). The samples were prepared in a controlled environment vitrification system. The climate chamber temperature was 25-28 °C and the relative humidity was kept close to saturation to prevent evaporation from the sample during preparation. A 5 µL sample drop was placed on a carbon-coated holey film supported by a copper grid and gently blotted with filter paper to obtain a thin liquid film (20-400 nm) on the grid. The grid was then rapidly plunged into liquid ethane at -180 °C and transferred into liquid nitrogen (-196 °C). The vitrified specimens were stored in liquid nitrogen and transferred into a Philips CM120 BioTWIN microscope equipped with a postcolumn energy filter (Gatan GIF 100) using an Oxford CT 3500 cryo-holder and its workstation. The acceleration voltage was 120 kV, and the working temperature was kept below -182 °C. The images were recorded digitally with a CCD camera (Gatan MSC 791) under low-dose conditions with an underfocus of approximately 1 µm. High-Pressure Liquid Chromatography (HPLC). HPLC analysis was performed on a system consisting of Gilson 321 pump, a Gilson 831 Temperature Regulator, a Dionex degasys DG-1210 degasser, a Gilson 402 Syringe Pump, a Gilson 233XL Fraction Collector, and a Sedex 75 evaporative light scattering detector (ELSD) operating at 30 °C. The column used was a 250 mm stainless steel column (3 mm i.d.) in combination with a 20 µL loop. The column was packed with LiChrospher 100 Diol, 5 µm at 950 bar. The flow rate was 0.5 mL/min. A binary linear gradient was run according
Barauskas et al.
Figure 2. Size distribution of samples with DGMO/GDO ) 50/50 wt/wt and lipid/P80 ) 85/15 wt/wt containing 95 wt % water, after mechanical shaking and after the heat treatment procedure (see text). Note that the distributions nearly coincide. The mean particle size was 143 and 140 nm for mechanically shaken and heat treated dispersions, respectively. to the following scheme: Time 0 min (t0) - 0%B, t3 - 0%B, t34 - 80%B, t35 - 80%B, t40 - 0%B, t60 - 0%B. Solvent mixture A consisted of hexane/1-propanol/acetic acid (93/5/0.8 v/v) and solvent mixture B of hexane/1-propanol/water/acetic acid (16/77/5/0.8 v/v). All standards and samples were dissolved in solvent mixture A without acetic acid.
Results and Discussion 1. Dispersions. The equilibrium phase diagram of the ternary DGMO/GDO/water system has been recently determined.35 With increasing GDO content, the LRf V2f H2 f L2 phase sequence was observed, reflecting the negative spontaneous curvature of the GDO-aqueous interface. It was also found that kinetically stable and structurally well-defined colloidal dispersions of the H2 phase nanoparticles of DGMO/GDO mixtures could be produced by use of a long-chained polymeric stabilizer, Pluronic F127, which due to its molecular characteristics mainly partitions at particle surfaces. In the present study, we investigate the effect of another stabilizer, P80 (Figure 1), on the phase behavior and nanoparticle structure of the aqueous DGMO/GDO mixtures. To this end, the DGMO/GDO ratio was fixed to 1/1 by weight (this mixture forms a H2 phase that swells to about 25-28 wt % water without P80) and the concentration of P80 was varied to obtain different lipid/P80 ratios. The dispersions were prepared by simple mechanical shaking and optionally including a heat treatment procedure to decrease the number of vesicular contaminants.34 Figure 2 shows indeed that dispersions of the three-component system prepared by mechanical agitation only are characterized by a monomodal and narrow size distribution and a mean particle size in the range of 100-150 nm; depending weakly on P80 content and total amphiphile (DGMO+GDO+P80) concentration. The fact that the samples are easily prepared to give such well-defined and distinct particle dispersions is noteworthy since it is usually difficult to disperse reversed liquid crystalline phases unless high energy/shear methods, such as microfluidization or ultrasonication, are applied. Note that the heat treatment procedure has a relatively limited effect on the size distribution and mean particle size of the dispersion. Overall the results are in good agreement with those reported for nanoparticles based on recently
“Sponge” Nanoparticle Dispersions
Langmuir, Vol. 22, No. 14, 2006 6331
Figure 3. Cryo-TEM images of samples with lipid/P80 ) 90/10 (a-c) and 85/15 wt/wt (d-f). All samples contain 95 wt % water. Dispersions were prepared by mixing components for 12 h on a mechanical mixing table followed by heat treatment (125 °C, 20 min).
discovered H2 phase dispersions of DGMO/GDO/F127 which also were prepared by mechanical shaking but required a heat treatment step in order to facilitate creation of monocrystalline particles with narrow size distribution.35 In that case, the effective dispersion formation was explained by the segregation of the Pluronic F127 polymeric stabilizer and the lipid components resulting in an accumulation of the F127 at the particle surface. In the present situation, the smaller amphiphilic P80 molecule would partition more readily between the particle surface and the inner structure and thus the particle stabilization and morphology should be fundamentally different when compared to the same DGMO/GDO system stabilized with F127 block copolymer. Importantly, the colloidal (storage) stability of the DGMO/ GDO/P80 nanoparticle dispersions is very good, and no changes of the mean particle size and/or size distribution could be detected over several months at room temperature (results not shown). To gain further insight about the dispersion formation and particle stabilization mechanism as well as the morphological properties of the nanoparticles, cryo-TEM experiments were performed on a number of samples with different lipid (DGMO+GDO)/P80 ratios (Figures 3 and 4). Figure 3 shows well-defined dispersions comprising particles with intricate, disordered, and rather dense inner cores enclosed by a spongelike surface region composed of intersecting lamellas. Judging from images of hundreds of particles, the inner dense structure region is persistently larger for the dispersion prepared at the lower lipid/P80 ratio of 90/10 (wt/wt) compared to the 85/15 sample. Correspondingly, the surface spongelike region clearly increases with increasing P80 fraction. The outer disordered sponge (L3)like structure apparently segregates from the core and appears to stabilize the reversed phase particle dispersion. It is also obvious that the particle inner structure is not hexagonal as is the case for corresponding dispersions stabilized by surface partitioning of F127 instead of P80.35 The effect of adding P80 is further illustrated by the micrographs displayed in Figure 4 where the P80 content was
increased to 20 (Figure 4a-c) and 30 wt % (Figure 4d-f) with respect to the total lipid content. The size of the outer spongelike region of interconnected lamellas has swollen substantially at the expense of the denser innercore region (compare Figures 3 and 4). It is also clearly observed by comparing Figure 4a-c with Figure 4d-f that increasing the P80 content from lipid/P80 ) 80/20 to 70/30 (wt/wt) results in a decreased volume of the dense innercore and a more swollen surface region. Distances between bilayers also become larger and the spongelike structure contains fewer interlamellar attachments (pores). As previously stated, the samples displayed in Figures 3 and 4 were prepared by mechanical agitation followed by the heat treatment procedure. To clarify the consequences of this procedure on the particle morphology and nanostructure, samples with lipid/ P80 ratios of 85/15, 80/20, and 70/30 (wt/wt) were prepared by mechanical agitation only and examined by cryo-TEM. As seen from the results presented in Figure 5, the influence of the heat treatment is rather limited in the present system, and mechanically vortexed samples display essentially the same particle morphology as the heat treated ones shown in Figures 3 and 4. This result is also in agreement with measured particle size distributions (Figure 2). The only significant difference that can be noted in the cryo-TEM micrographs in Figure 5 is a somewhat larger abundance of vesicular aggregates. We have recently shown that the heat treatment procedure results in the conversion of metastable vesicle structures to nanoparticles with equilibrium cubic phase or hexagonal phase structure and is required for producing well-defined particle dispersions absent of metastable particle structures.10,34 In these cases, the heat treatment effect in terms of structural relaxation was much larger, partly due to the high energy input technique (i.e., microfluidization) used during dispersion formation and the associated large fractions of nonequilibrium vesicular “contaminants” and possibly also due to increased energy barrier for relaxation in the case of stabilized liquid crystalline phases. 2. Bulk Phase Behavior. To better understand the nature of the nanoparticles, the isothermal (room temperature) phase
6332 Langmuir, Vol. 22, No. 14, 2006
Barauskas et al.
Figure 4. Cryo-TEM images of samples with lipid/P80 ) 80/20 (a-c) and 70/30 wt/wt (d-f). All samples contain 95 wt % water. Dispersions were prepared by mixing components for 12 h on a mechanical mixing table followed by heat treatment (125 °C, 20 min).
Figure 5. Cryo-TEM images of samples with lipid/P80 ) 85/15 (a-b), 80/20 (c-d), and 70/30 wt/wt (e-f). All samples contain 95 wt % water. Dispersions were prepared by only mixing components for 12 h on a mechanical mixing table.
behavior of the pseudoternary DGMO/GDO/P80/water system was investigated at a fixed DGMO/GDO weight ratio of 1/1 and varying contents of P80 and water. The determined phase diagram, based on visual observations supplemented by X-ray diffraction (XRD) data of a total of 140 individual samples, is rich, featuring a range of liquid and liquid crystalline phases. At low water content, a reversed micellar phase (L2) exists appearing as an isotropic fluid. At intermediate P80 levels and higher water content, this swells into an isotropic cubic liquid crystalline phase as evidenced by the appearance of two strong Bragg peaks with
relative positions in a ratio of x6:x8. This phase is most likely corresponding to the reversed bicontinuous structure (Q230) based on the G minimal surface given its relative location between the L2/H2 and LR phases. The V2 phase is also found in the ternary DGMO/GDO/water system.35 An H2 phase originating from the ternary DGMO/GDO/water mixtures is shown to hold a maximum of about 10% of P80 and is at higher contents transformed into other phases. This is clearly observed by the disappearance of the three strong characteristic Bragg reflections which for the H2 phase are spaced in the ratio of 1:x3:x4. At even higher P80
“Sponge” Nanoparticle Dispersions
Langmuir, Vol. 22, No. 14, 2006 6333
Figure 7. Photographs obtained under polarized light of samples with lipid/P80 ratio of 80/20 (wt/wt) containing 45 (a), 50 (b), and 80 wt % of water (c). An interface between two isotropic liquids is labeled with arrows.
Figure 6. Pseudoternary DGMO/GDO/P80/water phase diagram. (a) Prepared and investigated sample compositions. Symbols denote L2 (filled reversed triangles), HII (filled hexagons), Q230 (filled squares), LR (filled triangles), HI (open hexagons), L1 (open reversed triangles), region of two coexisting isotropic liquid phases (open circles), and other multiple phase mixtures (crosses). (b) The resulting schematic phase diagram showing one phase and the two coexisting liquid phase (2 liq) regions. The phase boundaries drawn in solid lines are determined within an accuracy of (1-2 wt % whereas dashed lines trace the most uncertain regions. Data points 1-3 are discussed in the text.
concentration, a wide LR phase region is observed that swells extensively, up to about 70 wt % of water. In the binary P80/ water system, another hexagonal liquid crystalline phase is found between 30 and 50 wt % of water surrounded by reversed (L2) and normal micellar phases (L1) at lower and higher hydrations, respectively. The fact that this hexagonal phase is transformed into the L1 phase with increasing water concentration shows that it is a normal, or oil-in-water, hexagonal phase (H1). The most important feature of the determined phase diagram is that at about 50 wt % of water a clearly distinguishable mixture of two coexisting isotropic solutions (denoted as 2liq in Figure 6) between the H2 and LR phases is observed. This mixture of coexisting solutions is particularly interesting for understanding the observed structural properties of the nanoparticles in excess water; since the compositions of the investigated dispersions (Figures 2-5) essentially all fall on dilution trajectories within or very close to this region. We concentrate on samples falling on the dilution line with lipid/P80 ratio of 80/20 (wt/wt). Figure 7 shows the visual
Figure 8. Powder XRD patterns of top (1) and bottom (2) layers of two coexisting isotropic liquids at the (global) sample weight ratio DGMO/GDO/P80/water ) 24/24/12/40.
appearance of these samples containing two isotropic liquids as a function of water concentration (points 1-3 in Figure 6b indicate the exact positions of the samples in the phase diagram). As clearly seen from Figure 7a, sample 1 consists of two layers of isotropic liquids both of which are characterized by relatively low viscosity. On the other hand, XRD measurements of these liquids indicate their disordered structure since both layers are characterized by only one diffuse diffraction peak (Figure 8). Importantly, the characteristic distance for the top layer is almost two times smaller than that for the bottom layer indicating a lower degree of swelling. Further insights into the nature of coexisting solutions are obtained by a semiquantitative determination of the respective composition. To determine the relative water content, the coexisting layers of sample 1 were separated and subjected to freeze-drying. Simple comparison of the sample weight before and after freeze-drying allows for an estimation of the water concentration in the top and bottom layer to be about 14 and 51 wt %, respectively. Furthermore, using HPLC, the relative DGMO/GDO composition in the two layers of sample 1 is estimated to be 36/64 (wt/wt) and 62/38 (wt/wt) in the top and bottom layer, respectively. Although these analyses are semi-
6334 Langmuir, Vol. 22, No. 14, 2006
Barauskas et al.
quantitative, they clearly show that the top layer of the two coexisting liquid phases is rich in GDO but poor in water. Presumably it indicates that this phase is a reversed micellar solution (L2) originating from the water poor side of the phase diagram. Contrary to this, the bottom phase is rich in water and contains a relatively large amount of DGMO. Together with the diffuse XRD pattern and the isotropic and fluid nature, this phase most likely corresponds to a disordered sponge or L3 phase.36 Altogether, it may be concluded that two coexisting isotropic liquids denoted as 2liq in Figure 6 and found between the H2 and LR phases at relatively high hydration appear to correspond to coexisting L2 and L3 (spongelike) phases. It is very important to note that the L2/L3 phase mixture also coexists in excess water (Figure 7b). It can be clearly seen that dilution of sample 1 leads to separation of the excess water phase from the L2/L3 phase mixture (excess water at the bottom of the vial, some birefringence arises from the slight mixing of the L3 phase with water). Further dilution results in a stable milky dispersion (Figure 7c). Another observation from the bulk phase behavior is that with increasing P80 content the relative volume of the L3 phase increases at the expense of the coexisting L2 phase. 3. Relation between Bulk Phase Behavior and Particle Morphology. From the results presented above, we conclude that there is a clear relation between the bulk phase behavior and the spongelike nanoparticles observed in mixtures with P80 content g20 wt % with respect to lipid (Figure 4). The coexisting liquid isotropic phases observed in the region denoted 2liq (Figure 6) are most likely the liquid L2 and L3 phases. According to the particle morphologies elucidated in Figure 4, the dense homogeneous particle core is thus most likely originating from the L2 phase and the outer surface spongelike region is originating from the L3 phase. Importantly, within a single nanoparticle, a kind of phase separation occurs on the submicron scale. Rather than forming distinct particle populations originating from the L2 and L3 phases, respectively, one single particle dispersion population is formed. The outer sponge phase layer encapsulates the L2 droplets and facilitates thereby dispersion formation and longterm colloidal stability. Less obvious but notable is the finding that the morphology of the particle core changes when decreasing the P80 content below 20 wt % (Figure 3). The particle compositions presented in Figure 3 fall onto dilution lines just below the 2liq zone shown in the bulk phase diagram. The particle core of the particles in Figure 3 is still disordered, however, with a clearly observable nanostructure which is absent at higher P80 concentrations (Figure 4). It is likely that at lower P80 concentrations the particle core is more swollen thus making the L2 phase nanostructure visible using cryo-TEM. However, we have not yet been able to elucidate the nature of the gradual structure transition reflected in the images in Figures 3 and 4 but simply note that the small scale structure of the core phase gradually disappears with increasing P80 content.
4. Implications for the Use of the Nanoparticle System in Drug Delivery. Similar to so-called self-emulsifying drug delivery systems (SEDDS) comprised of microemulsions with high levels of polar cosolvents (e.g., ethanol or poly(ethylene glycol) 400),37 the present system yields nanoparticle dispersions when contacted with excess aqueous fluids and with a minimum of energy input. In fact, the nanoparticles are formed easily from the liquid lipid mixture without the need of a polar cosolvent. The self-dispersing property is very useful for constructing waterfree nanoparticle precursor formulations of various therapeutic substances that, for example, can be filled into capsules for oral pharmaceuticals. When the capsules encounter gastro-intestinal fluids and disintegrate, the formulation will spontaneously form nanoparticles containing the formulated active. The nanoparticles thus formed facilitate the transport and subsequent oral absorption of the drug in question. Furthermore, drug species that are sensitive to high shear forces (such as proteins, peptides, and other biomolecules) may be conveniently formulated with the present system. Dispersions of the “sponge” nanoparticles containing such active agents may be prepared with a minimum of shear force with good encapsulation efficiency. Such dispersions can be further manipulated by the addition of cryo-protectants followed by freeze- or spray-drying to form powder precursors of the “sponge” nanoparticles.
(36) Anderson, D.; Wennerstro¨m, H.; Olsson, U. J. Phys. Chem. 1989, 93, 4243-4253.
(37) Gao, P.; Rush, B. D.; Pfund, W. P.; Huang, T.; Bauer, J. M.; Morozowich, W.; Kuo, M. S.; Hageman, M. J. J. Pharm. Sci. 2003, 92, 2386-2398.
Conclusions Well-defined and kinetically stable sponge phase nanoparticles can be easily prepared by self-dispersion (low energy mechanical agitation) of aqueous DGMO/GDO mixtures fortified with small amounts of P80. The particles have a size of about 100-150 nm with intricate, disordered, and rather dense inner cores enclosed by a surface spongelike region composed of intersecting lamellas. The results from the bulk phase behavior have shown that the particle core and outer surface region are most likely originating from the L2 and L3 phase, respectively. The surface coverage of the sponge phase layer on L2 droplets provides unique particle stabilization facilitating dispersion formation and long-term colloidal stability. Such self-dispersing nanoparticle compositions provide a very useful tool for constructing water-free nanoparticle precursor formulations of various active substances for oral pharmaceuticals. Acknowledgment. We are grateful to Gunnel Karlsson for help with Cryo-TEM instrumentation. This work is performed in the “New Principles for Oral Delivery of Peptides and Peptidomimetics” program sponsored by the Swedish Foundation for Strategic Research, Vinnova, and Camurus Lipid Research Foundation. LA060295F
