6956
Langmuir 2007, 23, 6956-6965
Interactions of Novel, Nonhemolytic Surfactants with Phospholipid Vesicles Per E. G. Thore´n,*,† Olle So¨derman,§ Sven Engstro¨m,† and Christian von Corswant‡ Department of Chemical and Biological Engineering, Chalmers UniVersity of Technology, SE-412 96 Gothenburg, Sweden, AstraZeneca R&D Mo¨lndal, SE 431 83 Mo¨lndal, Sweden, and Department of Physical Chemistry 1, Lund UniVersity, Box 124, SE 221 00 Lund, Sweden ReceiVed December 21, 2006. In Final Form: March 28, 2007 PEG-12-acyloxystearates constitute a novel class of pharmaceutical solubilizers and are synthesized from polyethylene glycol and 12-hydroxystearic acid, which has been esterified with a second acyl chain. The hemolytic activity of these surfactants decreases drastically with increasing pendant acyloxy chain length, and surfactants with an acyloxy chain of 14 carbon atoms or more are essentially nonhemolytic. In this paper, the interactions of PEG-12-acyloxystearates (acyloxy chain lengths ranging from 8 to 16 carbon atoms) with phosphatidylcholine vesicles, used as a model system for erythrocyte membranes, were studied in search of an explanation for the large variations in hemolytic activity. Surfactant-induced alterations of membrane permeability were investigated by studying the leakage of vesicle-entrapped calcein. It was found that all of the surfactants within the series interact with the vesicle membranes and cause slow leakage at elevated surfactant concentrations, but with large variations in leakage kinetics. The initial leakage rate decreases rapidly with increasing pendant acyloxy chain length. After prolonged incubation, on the other hand, the leakage is not a simple function of acyloxy chain length. The effect of the surfactants on membrane integrity was also investigated by turbidity measurements and cryo-transmission electron microscopy. At a surfactant/lipid molar ratio of 0.4, the vesicle membranes are saturated with surfactant. When the surfactant/lipid molar ratio is further increased, the vesicle membranes are progressively solubilized into mixed micelles. The rate of this process decreases strongly with increasing acyloxy chain length. When comparing the results of the different experiments, it can be concluded that there is no membrane permeabilization below saturation of the vesicle membranes. The large variations in the kinetics suggest that several steps are involved in the mechanism of leakage induced by PEG-12-acyloxystearates and that their relative rates vary with acyloxy chain length. The slow kinetics may in part be explained by the low critical micelle concentrations (CMCs) exhibited by the surfactants. The CMCs were found to be in the range of 0.003-0.025 µM.
Introduction Surfactants are widely used in pharmaceutical applications as solubilizers for sparingly soluble active compounds. When the concentration of a surfactant in aqueous solution is above the critical micelle concentration (CMC), micelles, which have the ability to increase the solubility of sparingly soluble active compounds, are formed. The use of micelles as drug delivery vehicles also helps to minimize drug degradation and adverse drug responses.1 Certain types of micelles may have long circulation times in the bloodstream, allowing gradual accumulation of the drug within the required area, and their size permits them to accumulate in areas with affected and leaky vasculature, such as tumors and inflammations.2,3 Micellar systems are most commonly used for parenteral or oral administration, but they can also be useful for ophthalmic, topical, rectal, and nasal delivery.1 For intravenous administration, the generally low viscosity exhibited by solutions containing spherical micelles makes them particularly suitable, since injection of solutions with high viscosity may cause pain.1 Surfactants may, however, cause a variety of adverse side effects when used as pharmaceutical solubilizers.4,5 In particular, surfactants used for parenteral formulation may induce hemolysis, * To whom correspondence should be addressed. E-mail: thoren@ chalmers.se. † Chalmers University of Technology. ‡AstraZeneca. § Lund University. (1) Lawrence, M. J. Chem. Soc. ReV. 1994, 23, 417-424. (2) Torchilin, V. P. J. Controlled Release 2001, 73, 137-172. (3) Lukyanov, A. N.; Torchilin, V. P. AdV. Drug DeliVery ReV. 2004, 56, 1273-1289.
that is, rupture of red blood cells resulting in the release of hemoglobin, as a result of surfactant incorporation into the erythrocyte membrane. The hemolytic activity of a number of surfactants has been investigated.6-10 It is generally accepted that two types of surfactant-induced hemolysis exist.11-15 Osmotic hemolysis occurs at low surfactant concentrations, where membrane absorption of surfactant monomers may lead to increased selective permeability to small solutes, followed by water penetration and cell swelling. At higher surfactant concentrations, the surfactant causes complete or partial solubilization of membrane lipids and proteins by the formation of mixed micelles. In general, nonionic surfactants have exhibited less negative side effects than ionic surfactants, and surfactants based on poly(ethylene glycol) (PEG) are frequently employed. (4) Atwood, D.; Florence, A. T. Surfactant systems. Their chemistry, pharmacy and biology; Chapman and Hall: London, 1983. (5) Sweetana, S.; Akers, M. J. PDA J. Pharm. Sci. Technol. 1996, 50, 330342. (6) Reinhart, T.; Bauer, K. H. Pharmazie 1995, 50, 403-407. (7) Galembeck, E.; Alonso, A.; Meirelles, N. C. Chem.-Biol. Interact. 1998, 113, 91-103. (8) So¨derlind, E.; Wollbratt, M.; von Corswant, C. Int. J. Pharm. 2003, 252, 61-71. (9) So¨derlind, E.; Karlsson, L. Eur. J. Pharm. Biopharm. 2006, 62, 254259. (10) Ross, B. P.; Braddy, A. C.; McGeary, R. P.; Blanchfield, J. T.; Prokai, L.; Toth, I. Mol. Pharmacol. 2004, 1, 233-245. (11) Rideal, E. K.; Taylor, F. H. Proc. R. Soc. London, Ser. B 1957, 146, 225-241. (12) Isomaa, B. Biochem. Pharmacol. 1979, 28, 975-980. (13) Bielawski, J. Biochim. Biophys. Acta 1990, 1035, 214-217. (14) Shalel, S.; Streichman, S.; Marmur, A. Colloids Surf., B 2002, 27, 215222. (15) Shalel, S.; Streichman, S.; Marmur, A. J. Colloid Interface Sci. 2002, 255, 265-269.
10.1021/la063700b CCC: $37.00 © 2007 American Chemical Society Published on Web 05/22/2007
Interactions of NoVel Surfactants with Lipid Bilayers
Langmuir, Vol. 23, No. 13, 2007 6957
croscopy. Large variations in the kinetics of surfactant/lipid interactions are demonstrated, which may help to explain the differences observed in the hemolysis study. Figure 1. Structure of PEG-12-acyloxystearates.
However, most commercially available nonionic surfactants for pharmaceutical use are very complex mixtures and exhibit large batch to batch variations, which complicates chemical analysis, characterization, and product specification.16-18 Recently, a simple and efficient method using enzyme catalysis to produce well-defined fatty acid esters of PEG was presented.17 This method has been used in combination with traditional chemical synthesis to obtain a novel series of surfactants, PEG-12-acyloxystearates (Figure 1 and Table 1).18 The surfactants used in the present study were synthesized using PEG with an average molecular weight of 1500 g/mol, which means that the average number of ethylene oxide units is 34. The size of the hydrophobic part of the surfactants has been modified by esterification of the hydroxyl group of 12-hydroxystearate with a fatty acid of varying length. In contrast to commercially available PEG-based surfactants, PEG-12-acyloxystearates are pure and well-defined. They exhibit excellent solubilization capacity and have high potential as pharmaceutical solubilizers.18 The hemolytic activity of PEG-12-acyloxystearates has been found to vary significantly depending on the number of carbon atoms in the hydrophobic part of the molecules.18 This has been studied by incubating an erythrocyte suspension with a surfactant solution for 40 min at 37 °C and then determining the concentration of released hemoglobin by absorption spectroscopy. For the shorter surfactants (PEG 1500-C18C8, PEG 1500-C18C10, and PEG 1500-C18C12), hemolysis is observed at concentrations in the millimolar range. This is comparable to what has been observed for dodecylmaltoside and sucrose monododecanoate using a similar experimental setup, but is significantly higher than what has been observed for Triton X-100.9 The longer surfactants in the series (PEG 1500-C18C14 and PEG 1500-C18C16), on the other hand, are essentially nonhemolytic. The influence of PEG-12-acyloxystearates on the transepithelial electrical resistance (TEER) measured across a cell layer has also been examined.18 The results are in analogy with the hemolysis study; only the surfactants with short pendant chains exhibit a reduction in TEER. For use as drug delivery vehicles, micelles formed by surfactants with long acyloxy chains thus appear to be the most promising candidates. In search of an explanation for the large differences in hemolytic activity exhibited within the series, the micellar properties and phase behavior of PEG 1500-C18C12, PEG 1500-C18C14, and PEG 1500-C18C16 have been examined.19 The various surfactants all exhibit a transition from the micellar to cubic phase at ∼20 wt % at room temperature and a transition to the hexagonal phase at ∼45-50 wt %. The micelles of the surfactants within the series were found to be spherical with a similar hydrodynamic radius ranging from 65 Å for PEG 1500-C18C12 to 75 Å for PEG 1500-C18C16. Thus, no correlation between micellar properties and hemolytic activity was found. In this paper, the interactions of PEG-12-acyloxystearates with phospholipid vesicles are studied using fluorescence spectroscopy, absorption spectroscopy, and cryo-transmission electron mi(16) Strickley, R. G. Pharm. Res. 2004, 21, 201-230. (17) Viklund, F.; Hult, K. J. Mol. Cat. B: Enzym. 2004, 27, 51-53. (18) Viklund, F. Ph.D. Thesis, Royal Institute of Technology, Stockholm, 2003. (19) McNamee, C. E.; Nilsson, M.; von Corswant, C.; So¨derman, O. Langmuir 2005, 21, 8146-8154.
Materials and Methods Materials. Soybean phosphatidylcholine (Epikuron 200) was obtained from Lucas Meyer Co. Ethylenediaminetetraacetic acid (EDTA) (titriplex III), N-(2-hydroxyethyl)-piperazine-1-ethanesulfonic acid (HEPES), calcein, octaethylene glycol n-hexadecyl monoether (C16E8), and poly(ethylene glycol) 4-tert-octylphenyl ether (Triton X-100) were purchased from Sigma. 1,6-Diphenyl-1,3,5hexatriene (DPH) was obtained from Fluka. Deionized water from a Milli-Q system (Millipore) was used. The PEG-12-acyloxystearates were synthesized as described previously.17 Preparation of Large Unilamellar Vesicles (LUVs). Vesicles were prepared by dispersion of dry lipid in buffer (10 mM HEPES, 100 mM NaCl, 4.8 mM NaOH, pH 7.4) by heavy vortexing. The dispersion was subjected to five freeze-thaw cycles20 before extrusion 21 times through two stacked 100 nm polycarbonate filters on a LiposoFast-Pneumatic extruder (Avestin, Canada) to obtain large unilamellar vesicles (LUVs). A unimodal size distribution (diameter: ∼100 nm) was confirmed by dynamic light scattering analysis. Vesicles with encapsulated calcein were prepared as described above, except that the lipid was dispersed in a solution of 10 mM HEPES, 10 mM NaCl, 70 mM calcein, 1 mM EDTA, and 264 mM NaOH (pH 7.4). Induced Leakage Assay. Vesicles loaded with calcein were separated from nonentrapped dye on a Sephadex G-50 column (Amersham Pharmacia Biotech), using an isoosmolar buffer (10 mM HEPES, 125 mM NaCl, 5 mM NaOH, 1 mM EDTA, pH 7.4). The osmolarity of the solutions was measured on an Advanced Micro osmometer model 330 (Advanced Instruments Inc, MA). The efflux experiments were performed on a Spex Fluorolog τ-2 spectrofluorometer (JY Horiba). The temperature in the sample holder was kept at room temperature (22 °C) using an external water bath. The bandpass of the excitation slit was optimized to obtain an optimal signal-to-noise ratio without photodegradation. Vesicles were diluted to a lipid concentration of 25 µM with isoosmolar buffer in a 1 × 1 cm quartz cell to a final sample volume of 3 mL. The time course of calcein fluorescence intensity was monitored using excitation and emission wavelengths of 490 and 520 nm, respectively, and a time increment of 1 s. First, the initial fluorescence intensity in the absence of surfactant was established. After 5 min, surfactant was added from the appropriate stock solution to yield a final surfactant concentration ranging from 0.5 to 800 µM. The dilution of the sample was typically 1%, and the fluorescence intensity was corrected accordingly. Immediately after surfactant addition, the cuvette contents were mixed using a pipet. At the end of each measurement, the vesicles were completely lysed by addition of 20 µL 10% (w/v) Triton X-100. Turbidity Measurements. Vesicle suspensions for turbidity measurements were incubated with surfactant at room temperature, and samples were collected at various time points. The absorbance at 350 nm was recorded on a GBC UV/vis 920 spectrophotometer (GBC, Australia) at room temperature, using a quartz cell with a path length of 1 cm. For each sample, a given volume of buffer was first added to the cuvette and the background intensity was recorded. The same volume of the sample was then added. The absorbance was measured immediately after mixing, and the background was subtracted. Each sample was prepared in triplicate, and the standard deviation was always less than 5%. Dynamic Light Scattering Measurements. Dynamic light scattering measurements were performed using an ALV-6010/EPP multiple tau digital correlator and an ALV/CGS-8F goniometer at a wavelength of 632.8 nm (HeNe laser). The scattering angle was maintained at 90°, and the temperature was kept at 25 °C. Vesicle samples were mixed with surfactant and incubated at room (20) Mayer, L. D.; Hope, M. J.; Cullis, P. R. Biochim. Biophys. Acta 1986, 858, 161-168.
6958 Langmuir, Vol. 23, No. 13, 2007
Thore´ n et al.
Table 1. PEG-12-acyloxystearates Used in the Present Studya
a
R
name
abbrev
(CH2)6CH3 (CH2)8CH3 (CH2)10CH3 (CH2)12CH3 (CH2)14CH3
PEG 1500 mono-12-capryloyloxy-stearate PEG 1500 mono-12-caproyloxy-stearate PEG 1500 mono-12-lauroyloxy-stearate PEG 1500 mono-12-myristoyloxy-stearate PEG 1500 mono-12-palmitoyloxy-stearate
PEG 1500-C18C8 PEG 1500-C18C10 PEG 1500-C18C12 PEG 1500-C18C14 PEG 1500-C18C16
See Figure 1 for the overall structure.
temperature. Surfactant/lipid molar ratios and incubation times were chosen based on the results of the turbidity measurements. Cryo-Transmission Electron Microscopy (Cryo-TEM). CryoTEM samples were prepared in a controlled environment vitrification system.21 The temperature in the climate chamber was 27-28 °C, and the relative humidity was kept close to saturation to prevent evaporation from the sample during preparation. Lacey carbon filmed copper grids were used, which were made hydrophilic by glow discharge. A 5 µL droplet of the sample was placed on the grid. By blotting the sample with a filter paper, a thin film (50-100 nm) was obtained. The grid was then rapidly plunged into liquid ethane (-180 °C) and transferred to liquid nitrogen. Vitrified samples were stored in liquid nitrogen and transferred to a Philips CM120 BioTWIN cryo-microscope, equipped with an energy filter imaging system (Gatan GIF 100) and digital multiscan CCD cameras (Gatan 791), using an Oxford CT 3500 cryo-holder. The acceleration voltage was 120 kV, and the working temperature was kept below -182 °C. Images were recorded with an underfocus of ∼700-800 nm. Determination of the Critical Micellar Concentration (CMC). The CMCs of the PEG-12-acyloxyatearates were determined using the fluorescent probe 1,6-diphenyl-1,3,5-hexatriene (DPH).22 Surfactant stock solutions in H2O were prepared the day before the measurements. Before the measurements (5-6 h), the surfactant and DPH stock solutions were mixed to yield a final DPH concentration of 5.0 × 10-9 M. The samples were kept in the dark at room temperature. The fluorescence of DPH was measured on a Perkin Elmer LS55 spectrofluorometer using a 1 × 1 cm quartz cell. The temperature in the sample holder was kept at 22 °C ( 0.1 using an external water bath. The excitation wavelength was 360 nm, and the emission wavelength range was 425-435 nm. The emission intensity peak at 430 nm was used to observe the incorporation of DPH in the micelles. The excitation and emission slits were both set at a bandwidth of 5 nm. The CMC was defined as the surfactant concentration at which the fluorescence intensity started to increase rapidly. It was determined as the intersection point between a straight line through the fluorescence at low surfactant concentrations and a straight line through the fluorescence values in the region of rapid intensity increase.
Results Surfactant-Induced Dye Leakage from Vesicles. Vesicles loaded with the fluorescent dye calcein were employed to study the effect of PEG-12-acyloxystearates on membrane permeability. Starting with a self-quenching concentration of calcein (70 mM) inside the vesicles, any leakage of dye into the surrounding solution can be detected as an increase in fluorescence intensity according to
% leakage ) 100
(
)
I(t) - I(0) IT - I(0)
(1)
where I(t) is the fluorescence intensity at time t, I(0) is the fluorescence intensity before surfactant addition, and IT is the fluorescence intensity recorded after completely lysing the vesicles by addition of Triton X-100 to a final concentration of ∼1 mM. (21) Bellare, J. R.; Davis, H. T.; Scriven, L. E.; Talmon, Y. J. Electron Microsc. Tech. 1988, 10, 87-111. (22) Chattopadhyay, A.; London, E. Anal. Biochem. 1984, 139, 408-412.
The above equation is analogous to the one used previously to determine the degree of hemolysis:18
(
% hemolysis ) 100
)
Hb - Hb0 Hbtot
(2)
where Hb is the amount of hemoglobin released after incubation with surfactant, Hb0 is the the amount released as a result of basal hemolysis, and Hbtot is the total amount of hemoglobin in each sample, determined by complete disruption of the erythrocytes by dilution of the erythrocyte suspension 500 times in pure water. In Figure 2, the time course of leakage of calcein upon addition of PEG 1500-C18C12 at various concentrations is shown. The lipid concentration was kept at 25 µm in all experiments. Only minute leakage of calcein is observed at 1 µM surfactant concentration (surfactant to lipid molar ratio of 1:25), even though this is 2 orders of magnitude above the CMC (Vide infra). At higher concentrations, on the other hand, the surfactant causes the release of calcein from the vesicles. Regardless of surfactant concentration, the leakage curves all have a slow initial phase (Figure 2, inset), followed by more rapid leakage, the rate of which is concentration-dependent. Prolonged incubation leads to virtually complete dye release if the surfactant concentration is 200 µM or higher, corresponding to a surfactant to lipid molar ratio of 8:1. The time required for complete leakage varies from 3 h at a surfactant concentration of 800 µM to 9 h at 200 µM. At lower surfactant concentrations, only partial release of calcein is observed. The spontaneous leakage of vesicle-entrapped dye in the absence of surfactant is negligible under the time period studied (24 h) (data not shown). In Figure 3, leakage is compared for all of the surfactants within the series at a surfactant concentration of 800 µM. Note that the release of calcein is obtained in all cases, but the leakage kinetics varies dramatically between the different surfactants. The initial leakage rate (see inset) is strongly dependent on acyloxy chain length and decreases in the following order: PEG 1500C18C8, PEG 1500-C18C10, PEG 1500-C18C12, PEG 1500-C18C14, and PEG 1500-C18C16. For PEG 1500-C18C12, PEG 1500-C18C14, and PEG 1500-C18C16, a lag time is observed before leakage becomes detectable, which in the case of PEG 1500-C18C16 is as long as 2.5 h. The overall leakage is, however, not a simple function of acyloxy chain length. After the initial lag, PEG 1500C18C12-induced leakage is more rapid than that induced by PEG 1500-C18C10, which causes the two curves to intersect. Interestingly, the leakage curve for PEG 1500-C18C8 enters a much slower phase after the initial steep slope and complete release of calcein takes more than 24 h under the present conditions, even longer than the case for PEG 1500-C18C16. For comparison, the nonionic surfactant C16E8 was examined under the same conditions (Figure 4). The kinetics of leakage was, however, found to be completely different. After surfactant addition, a very rapid leakage of calcein takes place. At low concentrations, only a fraction of the entrapped dye is released. At concentrations above 15 µM, complete leakage is obtained
Interactions of NoVel Surfactants with Lipid Bilayers
Figure 2. Leakage of vesicle-entrapped calcein induced by PEG 1500-C18C12 at various concentrations. The lipid concentration in the samples was 25 µM. Surfactant concentrations (from top to bottom): 800, 400, 200, 100, 50, 10, and 1 µM. Inset: The leakage curves during the first 30 min after surfactant addition.
Figure 3. Leakage of vesicle-entrapped calcein induced by PEG12-acyloxystearates at 800 µM concentration plotted on a logarithmic time scale. The lipid concentration in the samples was 25 µM. The curves show leakage for PEG 1500-C18Cn, where n ) 8, 10, 12, 14, or 16, as indicated in the figure. Inset: The leakage curves during the first 30 min after surfactant addition, plotted on a linear time scale.
within seconds. At intermediate concentrations, the rapid initial leakage is followed by a slower, continuous phase. Turbidity and Dynamic Light Scattering Measurements. In connection to studies of membrane permeability in the presence of a surfactant, it is important to consider the impact of the surfactant on membrane integrity. The phenomena associated with the addition of surfactant to a liposome dispersion can generally be described in terms of a three-stage model.23-28 In the first stage, at low surfactant concentrations, surfactant monomers partition between the lipid bilayers and the aqueous solution. As the surfactant concentration is increased, mixed micelles begin to form at a critical surfactant/lipid molar ratio. This is often referred to as saturation of the vesicle bilayers. Further addition of surfactant results in partial transformation of the vesicles into mixed micelles of an almost constant composition (stage two). Finally, at high surfactant/lipid molar ratios, the vesicles are completely solubilized and only mixed micelles remain. The solubilization process can conveniently be followed by monitoring the turbidity of the vesicle dispersion at various (23) Helenius, A.; Simons, K. Biochim. Biophys. Acta 1975, 415, 29-79. (24) Stubbs, G. W.; Litman, B. J. Biochemistry 1978, 17, 215-219. (25) Jackson, M. L.; Schmidt, C. F.; Lichtenberg, D.; Litman, B. J.; Albert, A. D. Biochemistry 1982, 21, 4576-4582. (26) Lichtenberg, D.; Robson, R. J.; Dennis, E. A. Biochim. Biophys. Acta 1983, 737, 285-304. (27) Lichtenberg, D. Biochim. Biophys. Acta 1985, 821, 470-478. (28) Lichtenberg, D.; Opatowski, E.; Kozlov, M. M. Biochim. Biophys. Acta 2000, 1508, 1-19.
Langmuir, Vol. 23, No. 13, 2007 6959
Figure 4. Leakage of vesicle-entrapped calcein induced by C16E8 at various concentrations, as indicated in the figure. The lipid concentration in all samples was 25 µM.
surfactant/lipid molar ratios. Below saturation of the vesicles, an increase in turbidity is often observed due to an increase in the size of the vesicles. For surfactant concentrations above the saturation limit, a decrease in turbidity is observed due to partial solubilization of the bilayers and the formation of mixed micelles, which scatter much less light than the vesicles. At sufficiently high surfactant concentrations, the turbidity is low since only mixed micelles are present in the solution. Figure 5A shows the optical density at 350 nm for vesicles incubated at a lipid concentration of 50 µM in the presence of PEG 1500-C18C12 at varying surfactant concentrations and after various incubation times. At low surfactant concentrations, there is a moderate increase in turbidity after 30 min, which is stable over the investigated time period (1 week). Dynamic light scattering measurements of the corresponding samples show that this is due to an increase in the mean size of the vesicles, from an initial hydrodynamic radius of 45 nm to a radius of ∼50 nm (data not shown). At higher concentrations, the turbidity first increases and then subsequently decreases, on a time scale of hours. Above 200 µM, the turbidity eventually approaches that of the corresponding pure micellar solution of PEG 1500-C18C12 (dotted line). The final hydrodynamic radius of the aggregates was determined by dynamic light scattering to be ∼9 nm (data not shown). The same trends are observed for PEG 1500-C18C16 (Figure 5B), but the turbidity of the suspension changes much more slowly than that for PEG 1500-C18C12. Thirty minutes after surfactant addition, the turbidity is virtually the same as that for pure vesicles (dashed line) for concentrations below 200 µM. Only after 8 h is a significant increase in turbidity observed. As for PEG 1500-C18C12, the turbidity eventually decreases at high surfactant concentrations. The difference in kinetics between PEG 1500-C18C12 and PEG 1500-C18C16 is further illustrated in Figure 5C, where the time course of the turbidity at 200 µM is plotted for both surfactants. Due to the very slow structural rearrangements of the system, it is difficult to deduce at which time point a thermodynamic equilibrium has been reached. This is especially problematic for intermediate surfactant/lipid molar ratios, that is, the region where only partial solubilization of the membranes takes place. From the turbidity measurements, however, it seems reasonable to assume that 72 h of incubation is sufficient to establish a phase boundary. In Figure 5, the curves corresponding to a 72 h incubation of vesicles exhibit a maximum in the optical density at a surfactant concentration of 20 µM. In the three-stage model discussed above, this maximum corresponds to saturation of the bilayers with surfactant (and the onset of solubilization) according to
6960 Langmuir, Vol. 23, No. 13, 2007 sat sat Ssat tot ) Sw + Re L
Thore´ n et al.
(3)
sat where Ssat tot is the total surfactant concentration, Sw is the concentration of surfactant monomers in the aqueous solution, Rsat e is the effective molar ratio of surfactant to lipid in the coexisting mixed bilayers, and L denotes the lipid concentration.27,28 Thus, a plot of Stot versus L at concentrations corresponding to the onset of solubilization is expected to be sat linear with an intercept of Ssat w and a slope of Re . In Figure 6, the PEG 1500-C18C12 concentration corresponding to the maximum in the turbidity curve after 72 h at various lipid concentrations is plotted. For PEG 1500-C18C12, the intercept is found to be equal to zero; the free monomer concentration is thus below the detection limit of the experiment. Rsat e is found to be ) 0.64 reported for Triton 0.4, which is comparable to Rsat e X-100 with egg phosphatidylcholine (EPC) liposomes and 0.62 for C12E8/EPC.29 Interestingly, the same results as for PEG 1500C18C12 are obtained for PEG 1500-C18C8, PEG 1500-C18C10, PEG 1500-C18C14, and PEG 1500-C18C16 (data not shown), demonstrating that, within the accuracy of the assay, the lipid bilayers are able to incorporate equal amounts of all the surfactants within the series before the bilayer structure starts to disintegrate. In a similar fashion, Rsol e , the effective molar ratio required for complete solubilization of the vesicles may be determined from the turbidity data.29 As seen in Figure 5, however, for PEG12-acyloxystearates, the transition from intermediate structures to mixed micelles is extremely slow and no sharp change in the optical density is discernible. It was thus not possible to establish this boundary. Cryo-TEM. A cryo-TEM study was performed to investigate the structures formed when soybean phosphatidylcholine vesicles are incubated with PEG 1500-C18C12 (Figure 7). Compared to the cases of the fluorescence and turbidity measurements, a much higher lipid concentration (1 mM) was required for the cryoTEM samples. Three different surfactant/lipid molar ratios (0.4: 1, 4:1, and 32:1) were selected for investigation based on the results of the turbidity study (cf. Figure 5A). Two sets of samples were collected: one after 15 min of incubation and the second after 24 h. The lowest ratio used, 0.4:1, corresponds to the maximum in the turbidity curve obtained after long incubation times, that is, Rsat e , the onset of solubilization (see above). After 15 min (Figure 7A), the vesicles exhibit no obvious differences from the pure vesicle sample (Figure 7H). Several of the vesicles appear nonspherical or oligolamellar, as reported earlier for samples containing extruded vesicles.30 After 24 h, spherical vesicles with two or more bilayers dominate the sample (Figure 7B). At surfactant/lipid molar ratios just above Rsat e , the initial increase in turbidity is larger, but over the following week, the turbidity decreases to levels well below those measured for the pure vesicle suspension. In the cryo-TEM sample taken after 15 min of incubation at a ratio of 4:1 (Figure 7C), several vesicles appear to be deformed, that is, elongated rather than spherical. Globular micelles are visible as black dots in the images. Furthermore, threadlike micelles are occasionally observed. After 24 h, much fewer vesicles, predominantly unilamellar, are found in the sample (Figure 7D). The time course of the turbidity increase and subsequent decrease is faster at higher surfactant/lipid ratios, and at the highest ratio investigated (32:1), the optical density has reached a
(29) Paternostre, M. T.; Roux, M.; Rigaud, J. L. Biochemistry 1988, 27, 26682677. (30) Walter, A.; Vinson, P. K.; Kaplun, A.; Talmon, Y. Biophys. J. 1991, 60, 1315-1325.
Figure 5. Turbidity of vesicles incubated at a lipid concentration of 50 µM in the presence of (A) PEG 1500-C18C12 and (B) PEG 1500-C18C16 at varying concentrations. The dashed and dotted lines indicate the turbidity of the vesicles (in the absence of surfactant) and the pure micellar solution, respectively. Samples were incubated for (O) 30 min, (b) 4 h, (0) 8 h, (9) 24 h, (4) 72 h, or (2) 1 week. (C) Turbidity as a function of time for (b) PEG 1500-C18C12 and (0) PEG 1500-C18C16 at a surfactant concentration of 200 µM and a lipid concentration of 50 µM.
minimum after 24 h. The cryo-TEM sample taken after 15 min of incubation (Figure 7E) appears to consist of oligolamellar vesicles of varying shape as well as of globular micelles. In some areas of the sample, giant oligolamellar structures, up to 1 µm in diameter, are occasionally observed (Figure 7F). After 24 h, the sample no longer contains any vesicles; only globular micelles are observed (Figure 7G). Determination of the CMC. A critical parameter in the characterization of a surfactant is its CMC. A commonly used method to determine the CMC is based on monitoring the shape of the fluorescence spectrum of pyrene, which changes as probe molecules are transferred from aqueous solution to the hydrophobic environment of the micelles.31 For PEG-12-acyloxystearates, however, the sensitivity of the pyrene assay was found (31) Kalyanasundaram, K.; Thomas, J. K. J. Am. Chem. Soc. 1977, 99, 20392044.
Interactions of NoVel Surfactants with Lipid Bilayers
Langmuir, Vol. 23, No. 13, 2007 6961
Figure 6. Total concentration of PEG 1500-C18C12 corresponding to saturation of the membranes (Ssat tot ) as a function of lipid concentration (L). Ssat tot was derived from the maximum in turbidity observed after 72 h of incubation at varying lipid concentrations and surfactant concentrations (cf. Figure 5A).
to be too low to obtain the CMC (data not shown). Instead, the fluorescent probe 1,6-diphenyl-1,3,5-hexatriene (DPH) at very low concentratiοns was used to measure the CMCs. DPH is virtually nonfluorescent in water, but the fluorescence intensity is substantially enhanced as the molecule is incorporated in a hydrophobic environment. The DPH assay thus relies on an increase in the absolute fluorescence intensity rather than a change in the spectral shape. This investigation will be discussed in more detail in a forthcoming separate publication. From measurements of the DPH fluorescence in the presence of PEG12-acyloxystearates (data not shown), the CMC values were found to be 0.025 µM for PEG 1500-C18C8, 0.015 µM for PEG 1500-C18C10, 0.011 µM for PEG 1500-C18C12, 0.005 µM for PEG 1500-C18C14, and 0.003 µM for PEG 1500-C18C16. Using the same method, the CMC of C16E8 was measured to be 0.25 µM (literature value: 0.5 µM32).
Discussion A number of important observations regarding the novel class of pharmaceutical solubilizers denoted PEG-12-acyloxystearates have been made in the present study. First, the CMCs of these surfactants were found to be extremely low. Second, when compared to other nonionic surfactants such as C16E8, the kinetics of the interactions of PEG-12-acyloxystearates with phospholipid membranes were found to be very slow. The slow kinetics of solubilization of phospholipid membranes is likely to be highly relevant for the use of PEG-12-acyloxystearates in biological applications and may help to explain the generally very low hemolytic activity exhibited for these surfactants.18 Furthermore, when surfactants with varying acyloxy chain lengths were investigated, large differences in the kinetics of surfactant/lipid interactions were observed. These observations are important for our understanding of why the hemolytic activity of PEG12-acyloxystearates is so strongly dependent on the acyloxy chain length,18 as discussed below. Vesicles containing the fluorescent dye calcein were employed as a model system for erythrocytes. An advantage of using this simple system is that the kinetics of leakage can be monitored carefully. A relatively low lipid concentration (25 µM) was employed to allow investigation of the impact of surfactant concentrations in the lower micromolar range while maintaining a relatively high surfactant/lipid molar ratio. As seen in Figure 2, there is no connection between the CMC and the onset of leakage, since only minute leakage is observed at a 1 µM (32) Berthod, A.; Tomer, S.; Dorsey, J. G. Talanta 2001, 55, 69-83.
Figure 7. Cryo-TEM images of soybean phosphatidylcholine vesicles incubated with PEG 1500-C18C12 at varying surfactant/ lipid molar ratios and incubation times. (A and B): Surfactant/lipid molar ratio of 0.4:1 for 15 min and 24 h, respectively. (C and D): Surfactant/lipid molar ratio of 4:1 for 15 min and 24 h, respectively. (E and F): Surfactant/lipid molar ratio of 32:1 for 15 min. (G) Surfactant/lipid molar ratio of 32:1 for 24 h. (H) Soybean phosphatidylcholine vesicles in the absence of surfactant.
concentration of PEG 1500-C18C12, which is 2 orders of magnitude higher than the CMC (0.011 µM). Much higher concentrations are required to induce leakage. The same observation was made for PEG 1500-C18C8, PEG 1500-C18C10, PEG 1500-C18C14, and PEG 1500-C18C16 (data not shown). Even when very high surfactant concentrations are used, the leakage rate is very low, and several hours are required for completion of the leakage process. The leakage curves have very different appearances for the various surfactants within the series (Figure 3), and fitting all of them to one simple biexponential function is not possible. The large variations in the kinetics suggest that several steps are
6962 Langmuir, Vol. 23, No. 13, 2007
involved in the mechanism of leakage induced by PEG-12acyloxystearates and that their relative rates vary with acyloxy chain length. The results stand out against the leakage curves presented in Figure 4, showing the same experiments performed with the nonionic surfactant C16E8. For C16E8, the kinetics of surfactant-induced leakage is completely different, with a virtually instantaneous release of vesicle-entrapped dye already at low concentrations, suggesting a different mechanism of permeabilization of lipid membranes for the two types of surfactants. Further information is obtained by studying surfactant-induced solubilization of vesicles by turbidity measurements and cryoTEM. In the case of PEG-12-acyloxystearates, the solubilization process is found to be very slow (Figure 5). The addition of PEG-12-acyloxystearates initially causes a moderate increase in turbidity. The magnitude of surfactant-induced vesicle growth is generally found to be strongly dependent on the type of vesicles used. Small unilamellar vesicles (SUVs) formed from EPC or soybean phosphatidylcholine, which are frequently employed in studies of surfactant/lipid interactions, have a high curvature and are prone to form large vesicles in the presence of subsolubilizing concentrations of surfactant.28,29 The formation of these larger vesicles has been attributed to the surfactant-induced breakup of the bilayer structure and fusion of the resulting open bilayer fragments.29,33-35 Large unilamellar vesicles (LUVs), on the other hand, have a small curvature and their tendency to form larger structures in the presence of membrane-binding molecules is generally much less pronounced.28,29 When LUVs are employed, as in the present study, the increase in surface area due to surfactant incorporation into the lipid bilayers may account for at least part of the observed size growth.28,36 In the present study, the small increase in turbidity observed at low surfactant/lipid molar ratios and short incubation times is accompanied by a rather small increase in the mean hydrodynamic radius (from 45 to 50 nm) and the size distribution is still unimodal (data not shown). Accordingly, no large differences in vesicular shape could be detected compared to the pure vesicle dispersion (Figure 7A and H). However, in samples of higher turbidity, that is, above Rsat e at short incubation times (Figure 7C, E, and F), the cryo-TEM images of vesicles in the presence of PEG 1500-C18C12 suggest that the number of bilamellar and oligolamellar vesicles as well as elongated vesicular structures in the sample has increased compared to that of the original vesicle dispersion (Figure 7H). These observations suggest that the surfactant induces shape changes and possibly the formation of transient openings in the vesicle membranes, which upon closing may permit the formation of oligolamellar structures.30 Prolonged incubation times lead to a decrease in turbidity above Rsat e due to the solubilization of the lipid bilayers into mixed micelles. For high surfactant/lipid molar ratios (e.g., 32:1), this results in complete solubilization of the lipid bilayers and the formation of mixed micelles, as demonstrated by dynamic light scattering measurements and cryo-TEM (Figure 7G). The close-packed globular mixed micelles were determined from the cryo-TEM images to have a diameter of ∼13 nm. This is comparable to that of the earlier diffusion NMR studies of pure PEG 1500-C18C12 micelles.19 When comparing the leakage experiments with the turbidity measurements and the cryo-TEM study, it is clear that PEG12-acyloxystearates do not permeabilize lipid bilayers to a great extent when the surfactant/lipid molar ratio is below Rsat e . In the (33) Edwards, K.; Almgren, M. Prog. Colloid Polym. Sci. 1990, 82, 190-197. (34) Edwards, K.; Almgren, M. J. Colloid Interface Sci. 1991, 147, 1-21. (35) Johnsson, M.; Edwards, K. Langmuir 2000, 16, 8632-8642. (36) de la Maza, A.; Parra, J. L. Biochem. J. 1994, 303, 907-914.
Thore´ n et al.
leakage experiments, this corresponds to surfactant concentrations below 10 µM (Figure 2). At these low surfactant/lipid molar ratios, the leakage of vesicle-entrapped calcein is very slow and only partial release is obtained within the time period studied (24 h). In contrast, the leakage curves in Figure 4 suggest an entirely different mechanism for C16E8, used as a reference surfactant in the present study. For a number of surfactants,for example, Triton X-100,33,37-40 C12E8,33,34,41,42 octyl glucoside,29,39,43 and sodium cholate,40,44,45 a similar pattern as for C16E8 has been reported: the leakage rate has been found to increase strongly with concentration in this region to become virtually instantaneous above the concentration where vesicle growth is observed, that is, well below Rsat e . A plausible explanation for such observations is that, as surfactant monomers are incorporated within the bilayer structure, structural rearrangements occur which may result in transient openings or pores, through which entrapped solutes can escape.33,40,42,46,47 The observation that only partial release is observed for, for example, C16E8 at low surfactant/lipid molar ratios has been attributed to the resealing of the induced membrane pores due to the equilibration of surfactant monomers across the bilayer.46,48 At elevated surfactant concentrations, surfactant monomers may serve to stabilize membrane pores, rendering the membranes permanently permeabilized. One could speculate that the rate of formation of such pores is very different for PEG12-acyloxystearates on the one hand and, for example, C12E8 and Triton X-100 on the other. There are examples in the literature of slow solubilization of phospholipids by surfactants, for example, sodium dodecyl sulfate (SDS),49-51 dodecylmaltoside,41,49 and the commercially available solubilizers Tween 20 (C12-sorbitan E),41,46 Tween 80 (C18-sorbitan E),41,52 and Lubrol WX (C16-18-E).41,53 As pointed out in ref 49, there could be an interesting relationship between surfactant solubility in organic solvents and the rate of solubilization of lipid vesicles. While the efficient solubilizer C12E8 has a relatively high degree of lipophilicity and has been shown to have a high solubility in a variety of organic solvents,54 SDS and dodecylmaltoside have headgroups with a marked hydrophilic character and can therefore be anticipated to have a relatively low solubility in the hydrophobic interior of the lipid membrane.49 In correlation with these findings, the flip-flop rate of C12E8 across phospholipid bilayers has been (37) Alonso, A.; Villena, A.; Goni, F. M. FEBS Lett. 1981, 123, 200-204. (38) Nilsson, K.; Almgren, M.; Brown, W.; Jansson, M. Mol. Cryst. Liq. Cryst. 1987, 152, 181-203. (39) Ruiz, J.; Goni, F. M.; Alonso, A. Biochim. Biophys. Acta 1988, 937, 127-134. (40) Memoli, A.; Annesini, M. C.; Petralito, S. Int. J. Pharm. 1999, 184, 227-235. (41) de Foresta, B.; Lemaire, M.; Orlowski, S.; Champeil, P.; Lund, S.; Moller, J. V.; Michelangeli, F.; Lee, A. G. Biochemistry 1989, 28, 2558-2567. (42) Edwards, K.; Almgren, M. Langmuir 1992, 8, 824-832. (43) Ueno, M.; Hirota, N.; Kashiwagi, H.; Sagasaki, S. Colloid Polym. Sci. 2003, 282, 69-75. (44) Schubert, R.; Beyer, K.; Wolburg, H.; Schmidt, K. H. Biochemistry 1986, 25, 5263-5269. (45) de la Maza, A.; Parra, J. L. Colloids Surf., A 1997, 127, 125-134. (46) Annesini, M. C.; Memoli, A.; Petralito, S. J. Membr. Sci. 2000, 180, 121-131. (47) Lesieur, S.; Grabielle-Madelmont, C.; Menager, C.; Cabuil, V.; Dadhi, D.; Pierrot, P.; Edwards, K. J. Am. Chem. Soc. 2003, 125, 5266-5267. (48) Gubernator, J.; Stasiuk, M.; Kozubek, A. Biochim. Biophys. Acta 1999, 1418, 253-260. (49) Kragh-Hansen, U.; le Maire, M.; Moller, J. V. Biophys. J. 1998, 75, 2932-2946. (50) Deo, N.; Somasundaran, P. Langmuir 2003, 19, 7271-7275. (51) Yasuhara, K.; Ohta, A.; Asakura, Y.; Kodama, T.; Asakawa, T.; Miyagishi, S. Colloid Polym. Sci. 2005, 283, 987-993. (52) Simoes, S. I.; Tapadas, J. M.; Marques, C. M.; Cruz, M. E. M.; Martins, M. B. F.; Cevc, G. Eur. J. Pharm. Sci. 2005, 26, 307-317. (53) le Maire, M.; Champeil, P.; Moller, J. V. Biochim. Biophys. Acta 2000, 1508, 86-111. (54) le Maire, M.; Moller, J. V.; Champeil, P. Biochemistry 1987, 26, 48034810.
Interactions of NoVel Surfactants with Lipid Bilayers
shown to be very fast (on the order of milliseconds), while SDS and dodecyl maltoside transport across lipid bilayers is extremely slow (hours or more).49,54 These results correlate well with a solubilization mechanism in which the breakup of the bilayer structure and the formation of mixed micelles are dependent on the presence of surfactant on both the outer and the inner leaflet of the membrane. For ethylene oxide-based surfactants, it has been suggested that an inverse relationship between the size of the surfactant headgroup and the flip-flop rate exists.41 For C16E8, which would accordingly be expected to have a relatively high flip-flop rate, a rapid leakage process was observed in the present study (Figure 4). In contrast, it is reasonable to assume that PEG-12-acyloxystearates, which have a large headgroup with an average number of 34 ethylene oxide units, have very low flip-flop rates. Also, membrane pores stabilized by the presence of PEG-12-acyloxystearates would rely on the surfactant PEG chains being confined in a narrow aqueous channel, which would reduce the configurational entropy of the relatively long polymer chain. This may be the reason why the bilayer structure remains intact and relatively nonleaky in the presence of low fractions of PEG-12-acyloxystearates, but not when, for example, C12E8 is incorporated. This applies also to the Tween surfactants and Lubrol WX mentioned earlier, which also are based on relatively long PEG chains. Another type of molecule which structurally resembles PEG-12-acyloxystearates consists of PEG grafted onto phosphatidylethanolamine (PE). Such charged lipids are frequently included in liposome preparations to improve the circulation time of liposomes to be used as drug delivery vehicles, and they have also been suggested as drug carriers.3,55 In particular, lipids formed from PEG 2000 grafted onto 1,2-distearoylphosphatidylethanolamine (PEG 2000-DSPE) have been extensively studied. The aggregate structure of liposomes formed from a mixture of EPC and PEG 2000-DSPE with an increasing fraction of PEG 2000-DSPE has been examined by cryo-TEM, giving further support for the proposed importance of surfactant flipflop.56 It was found that, at PEG 2000-DSPE concentrations above 10 mol %, the vesicle preparations contain not only intact vesicles but also open bilayer discs and wormlike micelles. However, when PEG 2000-DSPE was added externally to preformed EPC vesicles, the vesicles were found to remain intact at PEG 2000-DSPE concentrations up to 23 mol %, corresponding to a PEG 2000-DSPE/EPC molar ratio of 0.30. As for PEG12-acyloxystearates, the ability of PEG 2000-DSPE to break up the lipid membranes thus appears to be very low. There are several other factors that should be taken into account. For grafted PEG chains, it is usually assumed that the conformation is a function of the interaction between neighboring polymer chains.57-59 When the surface density of PEG chains is low, there is little interchain interaction and the polymer is in a “mushroom” conformation. With increasing surface density, the PEG chains begin to overlap and repulsive interactions cause them to extend away from the surface, forming a “brush”. The surface area covered by the polymer is expected to be different for the two conformations. PEG chain interactions are likely to affect the overall kinetics of surfactant-induced membrane permeabilization and solubilization, although it is unclear to what degree. It should be noted that a relatively high fraction of PEG(55) Ashok, B.; Arleth, L.; Hjelm, R. P.; Rubinstein, I.; Onyuksel, H. J. Pharm. Sci. 2004, 93, 2476-2487. (56) Edwards, K.; Johnsson, M.; Karlsson, G.; Silvander, M. Biophys. J. 1997, 73, 258-266. (57) Gennes, P. G. Macromolecules 1980, 13, 1069-1075. (58) Majewski, J.; Kuhl, T. L.; Gerstenberg, M. C.; Israelachvili, J. N.; Smith, G. S. J. Phys. Chem. B 1997, 101, 3122-3129. (59) Marsh, D.; Bartucci, R.; Sportelli, L. Biochim. Biophys. Acta 2003, 1615, 33-59.
Langmuir, Vol. 23, No. 13, 2007 6963
12-acyloxystearates may be incorporated into the vesicle membranes without solubilization (Rsat e ) 0.4, corresponding to a concentration of ∼29 mol %). Another plausible explanation for the generally slow kinetics of PEG-12-acyloxystearates is PEG chain heterogeneity. PEG12-acyloxystearates, as well as Tween surfactants, Lubrol WX, and PEG 2000-DSPE, are heterogeneous in respect to PEG chain length, and multiple equilibria exist in the system. One could speculate that surfactant monomers with long PEG chains are adsorbed first when lipid vesicles are introduced due to their relatively high solubility. With time, a redistribution of surfactant monomers between the membrane and solution should take place, leading to an increase in the fraction of adsorbed monomers with short PEG chains (and thus lower solubility). Thus, it may take a long time before equilibrium is reached. On the other hand, it has previously been shown that the CMC for nonionic surfactants varies little with the number of ethylene oxide units, particularly for long PEG chains.60 It is thus unclear to what extent PEG chain heterogeneity influences the rate of surfactantinduced leakage. However, while slow equilibration of surfactant monomers across the lipid bilayers, conformational changes, and/or slow redistribution of molecules with varying PEG chain length would at least in part account for the generally slow permeabilization and solubilization kinetics of PEG-12-acyloxystearates, one should bear in mind that the PEG chain distribution is identical for the various surfactants within the series. Other factors must thus be considered to explain the observed variations in kinetics between, for example, PEG 1500-C18C8 and PEG 1500-C18C16. Due to the very low CMCs of PEG-12-acyloxystearates, the solution added to the vesicles in the experiments consists mainly of surfactant micelles and very few monomers. The mean aggregation number is in the range of 144-158.19 In the initial stage following surfactant addition, monomers in solution are incorporated into the vesicle membranes, depleting the solution of free surfactant. This may be reflected in the observation that the initial leakage rate for a given acyloxy chain length is the same regardless of surfactant concentration (exemplified by PEG 1500-C18C12 in Figure 2, inset). To provide the necessary amount of monomers to build up a large fraction of surfactant within the lipid bilayer, extensive disintegration of the micelles must occur. It is likely that micellar kinetics play an important role in the kinetics of the interactions of PEG-12-acyloxystearates with phospholipid bilayers. In a micellar solution, the surfactant monomers are constantly being exchanged between the micelles and bulk solution in a fast process (with a relaxation time denoted as τ1). Moreover, the micelles are continuously disintegrating and reforming in a much slower process (with a relaxation time denoted as τ2). The generally accepted model for the kinetic process of micelle formation and disintegration61-63 has been developed for pure surfactant systems near equilibrium and is thus not directly applicable in the present system. Nevertheless, it has been shown in studies of foaming processes and emulsification processes that surfactants that exhibit slow micellar kinetics in pure aqueous systems also exhibit slow kinetics in more complex, dynamic systems.64 In the absence of relaxation time data for PEG-12-acyloxystearates, it is of interest to review the relaxation times for similar surfactants found in the literature. (60) Van Alstine, J. M.; Sharp, K. A.; Brooks, D. E. Collloids Surf. 1986, 17, 115-121. (61) Aniansson, E. A. G.; Wall, S. N. J. Phys. Chem. 1974, 78, 1024-1030. (62) Aniansson, E. A. G.; Wall, S. N. J. Phys. Chem. 1975, 79, 857. (63) Evans, D. F.; Wennerstro¨m, H. The Colloidal Domain, 2nd ed.; Wiley: New York, 1999. (64) Patist, A.; Oh, S. G.; Leung, R.; Shah, D. O. Colloids Surf., A 2001, 176, 3-16.
6964 Langmuir, Vol. 23, No. 13, 2007
It is noteworthy that particularly slow relaxation times are observed for nonionic surfactants due to the absence of ionic repulsion between the head groups.64 In studies of micellar relaxation times by stopped flow dilution, using total surfactant concentrations in the range 0.4-0.8 mM, τ2 was found to be 3.5 s for Triton X-100, while even longer relaxation times were obtained for surfactants with an increasing number of ethylene oxide units, for example, Tween 20 (τ2 ) 6 s), Tween 80 (τ2 ) 8-10 s), and Brij 35 (τ2 ) 80 s).64 For Synperonic A7, an ethoxylated fatty alcohol with C13 and C15 alkyl chains and an average of seven ethylene oxide units, τ2 was found to be as long as 150 s.64 Relaxation times were found to be much smaller for monodisperse nonionic surfactants such as C12E8 (τ2 ) 4 s), showing the importance of the molecular weight distribution. For PEG-12-acyloxystearates, one would expect τ2 to be long due to the low CMCs61,62 and, to a certain degree, due to the heterogeneity of the PEG head group. Within the series, the micellar kinetics are likely to follow the variations in the CMC; the lower the CMC, the slower the micellar kinetics. It has previously been observed for PEG-grafted PE lipids that membrane incorporation is slow and dependent on the length of the hydrocarbon tails.65 These observations may also be related to variations in the CMC. Slow incorporation has also been reported for 1,2-distearoyl-sn-glycero-3-phosphoethanolamineN-[methoxy(polyethylene glycol)-1900] (MPEG 1900-DSPE).66 An interesting question is why the leakage rate decreases after the initial, fast phase for the surfactants with short acyloxy chains, particularly PEG 1500-C18C8. One could speculate that the surfactant transbilayer flip-flop rate is determined not only by the properties of the headgroup but also by the propensity of the surfactant to perturb the bilayer packing. Possibly, the surfactantinduced bilayer distortions become larger with longer acyloxy chains, due to the mismatch of the length of the hydrophobic tail vis-a`-vis the intrinsic lipid bilayer thickness. Such distortions of the bilayer packing may facilitate surfactant flip-flop. In combination with the variations in micellar kinetics, this could lead to the complex behavior observed in Figure 3. One may also consider the possibility that when liposomes are mixed with surfactant at concentrations above the CMC, both surfactant monomers and micelles can interact with the lipid bilayer, as has been suggested by some authors.49,67,68 As a possible explanation for the slow solubilization of vesicles induced by SDS, it was suggested that surfactant micelles may cause permeabilization and solubilization of membranes by the extraction of lipid molecules from the outer leaflet of the lipid bilayer.49 During this process, the outer leaflet would be partially depleted of lipid, resulting in vesicle reorganization and redistribution of lipid from the inner to the outer leaflet. As a consequence of these changes, vesicles slowly open up, leading to fragmentation and finally to solubilization by the surfactant micelles. To our knowledge, no direct evidence has been provided for such a mechanism. However, the interactions of monooleoylphosphatidylcholine (MOPC) with lipid bilayers have been investigated by micropipet manipulation of single giant vesicles, where the uptake of surfactant into the bilayer can be monitored by microscopy as an increase in vesicle surface area.67,68 The initial rate of uptake was found to increase with surfactant concentration, even above the CMC, which is ∼1 µM for MOPC.67,69 This suggests that, above the CMC, the surfactant (65) Li, S. H.; Yamazaki, M. Biophys. Chem. 2004, 109, 149-155. (66) Uster, P. S.; Allen, T. M.; Daniel, B. E.; Mendez, C. J.; Newman, M. S.; Zhu, G. Z. FEBS Lett. 1996, 386, 243-246. (67) Needham, D.; Zhelev, D. V. Ann. Biomed. Eng. 1995, 23, 287-298. (68) Needham, D.; Stoicheva, N.; Zhelev, D. V. Biophys. J. 1997, 73, 26152629.
Thore´ n et al.
may be taken up into the lipid bilayer by two different pathways: by the incorporation of monomers or by the disintegration/fusion of adsorbed micelles at the membrane surface. Interestingly, the uptake of MOPC strongly decreased when vesicles sterically stabilized by the incorporation of lipids grafted with PEG 750 were used.68 When MOPC concentrations above the CMC were used, the amount of MOPC uptake was reduced to levels close to those obtained when surfactant was added at a concentration around the CMC, suggesting that micellar uptake was inhibited by the presence of PEG-grafted lipids. Biological membranes contain, besides phospholipids, a large fraction of proteins that extend beyond the bilayer surface, obstructing the close approach of surfactant micelles. Surfactant-induced hemolysis based on such a mechanism would therefore be expected to be slow and inefficient, as in the case of PEG-12-acyloxystearates. One could speculate that both PEG-12-acyloxystearate monomers and micelles interact with lipid bilayers, leading to permeabilization and solubilization. As the acyloxy chain length is increased, the monomeric contribution to the uptake of surfactant would be smaller due to a lower CMC. In erythrocyte membranes, the direct contribution from micelles would be strongly impeded, thus rendering the longer surfactants essentially nonhemolytic. In the lipid model system used in the present study, there are, however, no proteins present. Since the micelles are similar in size and the PEG chains are the same, regardless of acyloxy chain length, one would not expect large differences in the rate of leakage if micellar adsorption on the vesicle surface were the predominant mechanism. Since large variations in the leakage rate were observed, a significant contribution from micellar adsorption is unlikely. It is more probable that the overall behavior of the PEG-12-acyloxystearates is governed by the micellar kinetics discussed above.
Conclusion In this paper, the interactions of PEG-12-acyloxystearates with phospholipid membranes were examined in search of clues as to why the hemolytic activity of these surfactants is so strongly dependent on acyloxy chain length. While hemolysis has been observed at relatively high surfactant concentrations for the shorter surfactants (PEG 1500-C18C8, PEG 1500-C18C10, and PEG 1500C18C12), the longer surfactants in the series (PEG 1500-C18C14 and PEG 1500-C18C16) are essentially nonhemolytic. Since surfactant-induced hemolysis is believed to result from surfactant incorporation into the cell membrane, soybean phosphatidylcholine vesicles were used as a simple model system, allowing careful examination using spectroscopic techniques. The leakage of a vesicle-entrapped fluorescent probe was used to model the release of hemoglobin from erythrocytes, and solubilization of the vesicles as a result of surfactant incorporation was studied using turbidity measurements and cryo-TEM. The most striking result of the present study is that the major differences between the surfactants lie in the kinetics of their interactions with phospholipid vesicles. However, the general series of events that follow upon mixing of the surfactant with vesicles is the same regardless of acyloxy chain length. When mixed with vesicles at low surfactant/lipid molar ratios, surfactant monomers partition into the lipid bilayers. This process is, however, slow and not associated with extensive permeabilization of the vesicle membranes, as shown by the lack of vesicle-entrapped dye leakage. This suggests that the incorporation of sublytic molar fractions of surfactant does not have a large destabilizing effect on the lipid packing. Only at higher surfactant concentrations do (69) Marsh, D. In Handbook of Lipid Bilayers; CMC Press: Boca Raton, FL, 1990; p 387.
Interactions of NoVel Surfactants with Lipid Bilayers
the vesicles slowly begin to leak. Examination of the vesicle integrity showed that the leakage process is associated with the disintegration of the vesicles, that is, gradual solubilization into mixed micelles. The leakage characteristics are different from those of many other nonionic surfactants that have been examined in similar lipid systems, where leakage has been shown to be rapid and to take place at surfactant concentrations below saturation of the vesicle membranes. The large variations in the kinetics of leakage and solubilization found within the series of PEG-12-acyloxystearates indicate that the process is not governed by a single rate-limiting step in the mechanism. It is, however, reasonable to assume that the surfactant CMC is a critical parameter. The CMCs were found to be in the range of 0.003 µM for PEG 1500-C18C16 to 0.025 µM for PEG 1500-C18C8. When the surfactant and vesicles are mixed, the solution thus contains very few surfactant monomers. Not surprisingly, the initial leakage rate appears to be proportional to the variations in the CMC. To build up a significant fraction of surfactant in the membrane, there must be a net transport of surfactant monomers from the micelles to the vesicles via the aqueous solution. This process is governed by micellar kinetics, which depends on the CMC. However, the picture becomes more complicated when one considers the slower leakage phase that follows. This is particularly obvious for PEG 1500-C18C8 and PEG 1500-C18C10 and causes the curves to intersect. A plausible explanation is that, for leakage to occur, the surfactant must flip-flop across the bilayer. This process could certainly be expected to be slow for surfactants with a large hydrophilic headgroup such as PEG 1500, but it could also be affected by the size of the acyloxy tail group. In summary, the kinetic patterns observed in the present study point toward several mechanistic steps, and the relative contribu-
Langmuir, Vol. 23, No. 13, 2007 6965
tions of these steps to the kinetic patterns cannot be distinguished at this point. From these experiments alone, it is not possible to fully explain the observed lack of hemolysis for PEG-12acyloxystearates with long acyloxy chains. The very slow kinetics of surfactant/lipid interactions that has been observed here is, however, likely to be highly relevant for their use as pharmaceutical solubilizers. Ideally, surfactants intended for use in micellar drug delivery systems should fulfill certain criteria to be considered suitable.1,2 First, they should exert minimal toxicological effects upon administration. Furthermore, they should preferably form spherical micelles in the presence of active compounds, given the low viscosity of solutions containing spherical micelles and the possibility for such small aggregates to penetrate various tissues. The CMC should be low to avoid precipitation of the drug due to dilution below the CMC upon introduction into the body. A low CMC is also desirable to maximize the amount of drug that can be solubilized for a given quantity of surfactant. After administration, the micelles should be stable and, to a sufficient degree, be able to coexist with cell membranes without disrupting them. The present study shows that PEG-12-acyloxystearates have the desired properties, particularly when a long acyloxy chain is used. Acknowledgment. This work was financially supported by the Centre for Surfactants Based on Natural Products (SNAP). The authors thank Professor Karl Hult and Seongsoon Pak for synthesis of the surfactants, Gunnel Karlsson (cryo-TEM), and Camilla O ¨ hgren (CMC measurements). Professor Staffan Wall is acknowledged for fruitful discussions about micellar kinetics. LA063700B
