Diffusion, Molecular Separation, and Drug Delivery from Lipid

Article pubs.acs.org/Langmuir

Diffusion, Molecular Separation, and Drug Delivery from Lipid Mesophases with Tunable Water Channels Renata Negrini and Raffaele Mezzenga* Food and Soft Materials Science, Institute of Food, Nutrition & Health, ETH Zurich, Schmelzbergstrasse 9, CH-8092 Zürich, Switzerland ABSTRACT: Lyotropic liquid crystals characterized by a bicontinuous cubic phase (BCP) have a structure characterized by interpenetrated water channels following triply periodic minimal surfaces, which can be stable in excess water conditions and thus suitable in a multitude of applications. The control of the water channels size in these systems has a direct impact on their use for drug delivery, crystallization, and membrane separation processes. In this work we carry out systematic diffusion studies to show how the control on the water channel dimensions directly correlates with the release and separation performance of bicontinuous cubic phases. Specifically, we tune the water channels diameter of the monolinolein/water system by adding different amounts of sucrose stearate, which, having hydration-enhancing properties, can shift the boundaries of the phase diagram. We then design a model bicontinuous cubic phase lipidic membrane of the Im3m space group, having a sugar ester to monolinolein ratio of 20%, and we follow the diffusion within its water channels, by using molecules that differ systematically in size and molecular conformation, and we demonstrate, for each class of molecules, a diffusionenhanced process upon increase of the water channel diameter. Finally, we also show the ability of the bicontinuous cubic phase to efficiently and selectively separate nanoparticles of a target size, by choosing an amount of sucrose stearate for which the water channel diameter and the nanoparticle dimensions match, demonstrating the possible use of these systems as filtering membranes of tunable molecular cutoff.

1. INTRODUCTION Lyotropic liquid crystals (LLC), also known as mesophases, are gaining interest in different fields of fundamental and applied sciences, such as biomedical, pharmaceutical, and materials science, due to their potential in a multitude of applications ranging from active ingredients encapsulation to stimuli responsive sustained release, biomimetic membranes, membrane technology, protein crystallization, or material templating.1−11 For example, their amphiphilic nature allows them to entrap hydrophilic, hydrophobic, and amphiphilic molecules, which makes them particularly suitable for drug delivery applications, in view of the large spectrum of hydrophilicity/ hydrophobicity of therapeutic drugs. Moreover, their intrinsic highly ordered internal structure, compared to polymers and liposomes, confers them unmatched design opportunities in templating and materials design.12 Different types of mesophases based on self-assembly of lipids in the presence of water exist.13 Among them, much attention has been focusing on the inverse bicontinuous cubic phases (BCP), which consist of two sets of noncommunicating but interpenetrating three-dimensional periodic water channel networks separated by a lipid bilayers. The water channels are organized according to well-defined symmetries that have high specific surfactant/water interfacial area and constant mean curvature and whose most common space groups are of the Ia3d, Pn3m, and Im3m types. Some of these bicontinuous cubic phases (Pn3m and Im3m) exist at thermodynamic equilibrium with excess water, which opens for a vast range of possible applications. The tuning of their internal structure, and features © 2012 American Chemical Society

associated with, is a topic of past and current investigations.12,14−17 Because the control on the water channels diameter is of primary importance in the encapsulation and release of active ingredients18−20 and protein crystallization,21 there is a clear need to understand better the role played by size selectivity and confinement in these applications or, as in biomedical devices, for the selective separation of biomolecules of varying dimensions.22 Angelov et al.18 have showed that it is possible to enlarge the mesophase water channels by the inclusion of the hydrationmodulating agent octyl glucoside (OG) in the monoolein/ water system under excess water conditions. Yaghmur et al.23 have investigated the effect of an amphiphilic molecule, diglycerol monooleate (DGMO), on the internal structure of the monolinolein-based systems, and a significant change in the mean lattice parameter was found, indicating that the presence of DGMO as cosurfactant also increases the hydration of the mesophase. A recent contribution from our group21 has demonstrated the possibility of crystallizing in-meso globular proteins, which exceed in size the water channels of monolinolein/water bicontinuous cubic phases, when the mesophase is swollen by the presence of a cosurfactant. These results point at a possible direct connection between transport properties and size of water channels. Nonetheless, Received: September 24, 2012 Revised: November 1, 2012 Published: November 1, 2012 16455

dx.doi.org/10.1021/la303833s | Langmuir 2012, 28, 16455−16462

Langmuir

Article

commercial-grade form of monolinolein contains more than 98 wt % monoglyceride. Sucrose stearate S-1670 (referred to as S16 in what follows) was a gift of Mitsubishi-Kagaku Foods Corporation (Japan). This commercial grade of sugar ester consists of 75% of monoester and 25% of di-, tri-, and polyesters. Glucose and 10 000 g/mol Dextran were purchased from Sigma-Aldrich (Schnelldorf, Germany). Colloidal dispersions of gold nanoparticles of 2 and 5 nm were purchased from BB International (England). Milli-Q grade water was used for the formulation of the lyotropic liquid crystals. The water solutions were adjusted to 150 mM total ionic strength by HEPES buffer pH 7 prior to mixing with lipids. For the diffusion studies, glucose and 10 000 g/ mol Dextran were dissolved in the buffer solution at a concentration of 1 wt %. In the case of 2 and 5 nm gold nanoparticles, as these were used as received (0.0012 and 0.0063 wt %, respectively), the entire experiments were run in pure Milli-Q water conditions. To form the lipid blend with different sucrose stearate content (0, 5, 10, 15, 20, and 25 wt %), sugar ester and Dimodan U/J were weighed and mixed in ethanol. After the solvent was left to entirely evaporate, the mixture was further dried under vacuum overnight. The final mesophases of lipid:solution ratio of 70:30, were prepared by weighing the appropriate amounts of the lipid blend and water solution at pH 7. The mixture was homogenized in Pyrex tubes by repeated cycles of heating and vortex mixing; for samples containing sugar ester no heating was applied. The liquid crystal phase in the Pyrex tubes was then equilibrated in the oven at 37 °C for 24 h, prior to experiments. 2.2. Swelling and Diffusion Studies. The correlation between lattice parameter and diameter of the water channels was obtained by combining the results from small-angle X-ray scattering (SAXS) and gravimetric analysis; both techniques were necessary in order to evaluate the exact amount of water in the fully hydrated and swollen bicontinuous cubic phases and to apply triply periodic minimal surface (TMPS) topological models. Swelling and diffusion experiments were performed using the home-built setup described in a previous publication.3 Swelling studies were performed at 37 °C, and the excess solution was adjusted to 150 mM and pH 7 identical to the solution used to form the bulk mesophase. Samples were left for 1 week to allow total hydration and equilibration; afterward, the mesophase was extracted from the tubes, the excess water removed with filter paper, and the mesophase weighed to assess the final water intake. Diffusion studies were carried out in triplicates using the same method described earlier.3 In short, this setup is composed of two chambers, the delivering chamber and the receiving chamber, separated by the lipid mesophase.3 The time-dependent concentration of solute measured in the receiving chamber was normalized by the initial concentration present in the delivering chamber on the opposite side of the tube.3 To simulate perfect sink conditions, the solution in the receiving chamber containing the solute was periodically removed and replaced by fresh buffer solution.3 The cumulated total diffused solute concentration at time t was obtained by summing all relative concentrations collected over all times ≤t. The assumption of 100% drug diffusion at long times was made, thus neglecting possible specific interactions of the drug with the membrane, which could maintain an osmotic pressure imbalance between the solute contained within the mesophase and the receiving chamber. 2.3. Small-Angle X-ray Diffraction. Small-angle X-ray scattering (SAXS) measurements were used to identify the symmetry of the mesophases, to determine the lattice parameters of the liquid crystalline structures, and, therefore, to follow the swelling process. SAXS diffractograms were acquired using a Rigaku microfocused X-ray source of wavelength λ = 1.54 Å operating at 45 kV and 88 mA. Diffracted X-rays signal was collected on a gas filled two-dimensional detector. The scattering vector q = (4π/λ) sin θ, with 2θ the scattering angle, was calibrated using silver behenate. The q-range was assessed by the sample to detector distance (1 m) to be in the range from 0.01 to 0.45 Å−1. Data were collected and azimuthally averaged using the Saxsgui software to yield one-dimensional intensity versus scattering vector q. Samples were loaded in the Linkam hot-stage between two thin mica sheets and sealed by an O-ring, with a sample thickness of ca. 1 mm. Measurements were performed at 37 °C, and samples were

none of the previous studies have established in detail a quantitative correlation between diffusion properties and topological parameters of the swollen mesophase. In this work we study diffusion of molecules and colloidal particles through fully hydrated bicontinuous cubic bulk phases based on the monolinolein (MLO)/water system, but to which we have added sucrose stearate as cosurfactant to shift the cubic phase/excess water phase boundary and thus to increase the water channel diameter. We then correlate measured diffusion rates of molecules and colloidal particles directly with the diameter of the water channels of the swollen mesophase. Furthermore, while the system containing octyl glucoside (OG) used before18,21 allows enlarging the water channels up to a maximum of 7.3 nm,24 the sugar ester:monolinolein system used in the present work is capable to increase the water channel diameter up to 12 nm, which represents an outstanding, unprecedented topological feature of bicontinuous cubic phases and which could potentially open new doors to applications unattainable with traditional lipidic cubic phases. Specifically, in this new system, the maximum size of water channels could be controlled from 3.85 to 12 nm in a precise and continuous manner, simply by varying the amount of sugar ester in the system (see Figure 1 for a schematic of the swelling

Figure 1. Schematics of the swelling process of the bicontinuous cubic phase upon increase of sugar ester concentration.

process). By means of small-angle X-ray scattering (SAXS) and gravimetric analysis, we followed the structural changes of the fully hydrated bicontinuous cubic phase (BCP) upon increasing sugar ester content, and we determined the characteristic structural length scales by triply periodic minimal surface (TPMS) models. We then discuss the influence of the water channel dimensions on the diffusion behavior of solutes of varying size, ranging from glucose, to polysaccharides, and to colloidal particles. We finally show the possibility of using this new swollen mesophase either as a drug delivery system or as a membrane for size-selective separation processes with welldefined molecular cutoff.

2. MATERIALS AND METHODS 2.1. Materials and Bulk Mesophases Preparation. Dimodan U/J was a gift of Danisco (Denmark) and was used as received. This 16456

dx.doi.org/10.1021/la303833s | Langmuir 2012, 28, 16455−16462

Langmuir

Article

Figure 2. Swelling studies on the sucrose stearate:monolinolein system. (a) SAXS data on functionalized monolinolein system in excess of water solutions and upon increasing sugar ester content from 0 to 25 wt %. A transition from Pn3m to Im3m is observed at 10 wt % sugar ester. (b) Swelling behavior of the system obtained by combination of SAXS and gravimetric techniques: maximum water intake (▲), lattice parameter (⧫), and water channel diameter (●) profiles increasing sugar ester content. The red arrow indicates the conditions used for the diffusion studies. The blue dashed line indicates a transition from Pn3m to Im3m. equilibrated for 30 min prior to measurements, while scattered intensity was collected over 1 h. 2.4. Optical Rotation Dispersion (ORD) and Atomic Absorption Spectroscopy (AAS). The concentration of glucose and 10 000 g/mol Dextran was determined by an optical rotatory dispersion device (ORDE-Y02/15) mounted on a CD spectrometer (Jasco J-815); the measurements were performed at room temperature, and the ORD signal was acquired for each sample three times at a fixed wavelength of 436 nm; a series of solutions (six different concentrations) for both molecules were prepared to construct a calibration curve, and the drug concentration was determined by linear interpolation within the linear range (data not shown). The concentration of the gold nanoparticles was measured by a Varian AA240Z Zeeman graphite-furnace (GTA 120) atomic absorption spectrometer equipped with a PSD 120 programmable sample dispenser and pyrolytically coated graphite-furnace tubes. The light source was a hollow cathode lamp, which was used at 4 mA at 242.8 nm. 10 μL sample aliquots were used for the measurements, and each concentration was determined three times repeatedly. For a better detection, 5 μL of 2 mg Ni/mL was added to each sample as a matrix modifier to thermally stabilize gold. A separate calibration curve was established by measuring gold solutions with concentrations ranging from 4 to 20 μg/L.

critical packing parameter (CPP) to the surface-average Gaussian curvature (⟨K⟩)25 and from geometrical considerations: =

3 (1 − CPP) 2l 2

(1)

where CPP is expressed as v/a0l, which is the ratio between the volume of the hydrophobic lipid tail, v, and the product of the cross-sectional lipid head area, a0, and the lipid chain length, l. Infinite periodic minimal surface (IPMS), such as the bicontinuous cubic phases, is characterized by a zero mean curvature (⟨H⟩) and a negative Gaussian curvature (⟨K⟩) in all points of the bilayer−water interface. The previous eq 1 indicates that a decrease of the CPP causes an increase in the negative Gaussian curvature and vice versa, which is responsible for the observed order−order transition. The same approach was followed by Nakano et al.26 to explain the order−order transition in the monoolein−oleic acid system caused by the effect of electrostatic interactions on the surface of the lipid membrane. More in general, it is the progressive decrease of CPP upon hydration which imposes the changes in Gaussian curvature causing the order−order transitions sequence gyroid (Ia3d) → diamond (Pn3m) → primitive (Im3m) observed in lipidic cubic phases.12,27 In our case, as discussed above, the addition of a sugar ester with its highly hydrophilic sucrose polar head increases the average area of the head groups (a0) decreasing then the CPP value; subsequently, from the equation, the negative value of the Gaussian curvature (⟨K⟩) increases (e.g., the radii of curvature increase) and the spontaneous curvature of the lipid bilayer toward the water region decreases, inducing a change from the Pn3m to a Im3m, which possesses higher radii of curvature and thus a greater water intake capacity.28,29 To quantitatively evaluate the swelling of the water channels with respect to the sugar ester content, SAXS data were combined with gravimetric analysis to evaluate the exact water intake capacity of the fully hydrated mesophase. To calculate the diameter of the water channel for the two bicontinuous cubic phases (Pn3m and Im3m), TPMS arguments were used and the following equation from Tuner et al.30 was applied:

3. RESULTS AND DISCUSSION 3.1. Swelling Study of the Bicontinuous Cubic Phase Sucrose Ester:Monolinolein System. The swelling capacity of the system was studied by combining SAXS, gravimetric analysis, and structural models of the cubic phases. Figure 2a shows the results obtained by SAXS on mesophases with different sugar ester content. Up to a sugar ester:lipid ratio equal to 5%, the Pn3m mesophase (reflections spaced as √2, √3, √4, √6, √8, and √9) is retained and the lattice parameter increases, as can be observed by the decrease in q1 reflection from 0.104 to 0.096 Å−1; a further addition of sugar ester induces a phase transition from Pn3m to Im3m, the latter characterized by reflections spaced as √2, √4, √6, √10, √12, and √14. The presence of the sucrose stearate, characterized by a very large hydrophilic head, increases the hydration of the mesophase; the average area of the lipid polar head progressively increases, the mesophase swells, and an order−order phase transition occurs. This phenomenon can be well explained by eq 1, relating the 16457

dx.doi.org/10.1021/la303833s | Langmuir 2012, 28, 16455−16462

Langmuir

Article

3 ⎛l⎞ 4 ⎛l⎞ ϕ = 2A 0⎜ ⎟ + πx⎜ ⎟ ⎝a⎠ 3 ⎝a⎠

increase of sugar ester without any apparent discontinuity (see Figure 2b). This analysis focusing on the structural characterization of the swollen mesophases highlights the possibility to tune in a precise and continuous way the dimensions of the water channels by simply varying the sugar ester:monolinolein ratio, whose implications will be discussed in the following part of the article. 3.2. Diffusion Study. The size-selectivity properties of the sugar−ester:monolinolein liquid crystal membrane were investigated in the present work by studying the diffusion of model solutes within the swollen mesophase. For this purpose, we selected the system with a fixed sugar ester:lipid ratio of 20% and compared, for each class of solutes, the diffusion within this system with that occurring within the corresponding unperturbed monolinolein−water system. Under these conditions, transport occurs within water channels of 9.74 nm in diameter for the swollen mesophase and 3.85 nm for the normal, unperturbed mesophase (see Figure 2b and Table 1), respectively. To measure the diffusion of glucose, 10 000 g/mol Dextran and two colloidal gold nanoparticles, of 2 and 5 nm in diameter, we used the home-built setup described in the experimental part and more in details in our previous reports.3 Figure 3a shows the glucose diffusion profiles through the two liquid crystalline membranes, swollen (red squares) and standard (blue triangles), plotted as a percentage of the diffused drug versus diffusion time. For a better understanding of the transport process occurring within the lipid mesophase, we apply the following general diffusion power law equation:

(2)

where a is the lattice parameter as measured by SAXS, ϕ is the lipid volume fraction, which can be obtained knowing the maximum bulk water content (see Table 1) and the density of Table 1. Structural Data Obtained from the Swelling Study: Space Group, Maximum Water Intake, Lattice Parameter (a) and Water Channel Diameter (Dw) at the Excess Water Phase Boundary %S16

space group

water [wt %]

a [nm]

Dw [nm]

0 5 10 15 20 25

Pn3m Pn3m Im3m Im3m Im3m Im3m

37.00 41.33 42.73 46.03 47.00 49.52

8.57 9.23 13.14 14.60 17.50 21.00

3.85 4.39 7.05 8.06 9.74 11.94

the monolinolein (ρMLO = 1.05 g/cm3), and A0 and χ are respectively the ratio of the area of the minimal surface in a unit cell to (unit cell volume)2/3 and the Euler−Poincaire characteristic, which have the following values depending on the specific cubic phase: A0 = 1.919 and χ = −2 for Pn3m and A0 = 2.345 and χ = −4 for Im3m. Following Briggs et al.,31 one can derive then the radius of the water channels by using the following equivalences: (Pn3m)

r = 0.391a − l

(3a)

(Im3m)

r = 0.305a − l

(3b)

Mt = kt n M∞

(4)

where Mt and M∞ are respectively the amounts of drug released at time t and at equilibrium (in our case, due to the sink conditions used, M∞ = M0 where M0 is the initial quantity of drug), k is a proportionality constant, and n is the diffusional exponent which has the following values depending on the specific case: n = 0.5 for pure diffusion controlled drug release and n = 1 for swelling-controlled drug release or case II transport phenomena.32 Since transport in bicontinuous cubic phase structures has already been demonstrated to be diffusive,3,33,34 the percentage diffused versus the square root of time (n = 0.5) is plotted in Figure 3b to better quantify the changes in transport mechanism induced by the presence of

Figure 2b gives the variation of the maximum bulk water content, the lattice parameter, and the size of the water channel at the maximum hydration points as a function of sugar ester content. As can be observed, there is a linear increase tendency of the water channel dimension with increasing of the sugar ester content (the precise values are reported in Table 1). After the Pn3m−Im3m order−order transition appearing at 10 wt % sugar ester:lipid ratio, a further addition of sugar ester favors the hydration of the Im3m mesophase consistently with the similar trend reported previously by Angelov et al.18 with the hydration-modulating agent octyl glucoside (OG). Yet, the increase in water channel diameter progresses linearly with

Figure 3. Effects of 20% sugar ester addition on the diffusion behavior of glucose within the lipid mesophase (swollen and unperturbed). Glucose diffusion from the swollen-mesophase (⧫) containing 20 wt % sugar ester and the normal mesophase without sugar ester (▲) plotted against time (a) and square root of time (b). 16458

dx.doi.org/10.1021/la303833s | Langmuir 2012, 28, 16455−16462

Langmuir

Article

Figure 4. Effects of 20% sugar ester addition on the diffusion behavior of 2 nm gold nanoparticles within the lipid mesophase (swollen and unperturbed). 2 nm gold nanoparticles diffusion from the swollen mesophase containing 20 wt % sugar ester (⧫) and the normal mesophase without sugar ester (▲) plotted against time (a) and square root of time (b).

sugar ester. The characteristic lag time appearing at the early stages of the experiments is a clear evidence of the diffusion time required by the drug to diffuse through the mesophase from the donor to the receiving chamber. As it can be observed, in the swollen mesophase the slope of the glucose diffusion is twice larger than in the case of diffusion through the normal mesophase. Differently from the release experiments, in the diffusion experiments through the setup used for the study, the diffusion coefficients are not of straightforward extraction.3 Nonetheless, the linearity of both diffusion profiles indicates a purely Fickian diffusive process driven by concentration gradients in both the normal and swollen mesophases. Furthermore, because the thickness of the mesophase is maintained constant in the setup, it is possible to conclude that the ratio between the two diffusion coefficients is directly proportional to the square of the ratio of the diffusion slopes,3 which allows one to conclude that in the case of the swollen mesophase the diffusion coefficients of small molecules such as glucose has a 4-fold increase. Interestingly, the same trends are observed for the diffusion of the 2 nm gold nanoparticle, as shown in Figure 4. In both the cases of glucose and 2 nm gold nanoparticles the ratio between the diffusion slopes from the swollen (red squares) and the normal mesophase (blue triangles) is approximately two, inferring a direct correlation between the diffusion process and the size of the water channels; furthermore, as expected, it is noticeable that the diffusion of the 2 nm gold nanoparticles is much slower (about 10 times) compared to that of glucose, indicating that the diffusion of the solute is also size-dependent. Similar findings were reported by Caffrey et al.20 in a previous work where the transport properties of the monoolein/water bicontinuous cubic phase with respect to size, shape, and charge of the diffusing species were studied within the water channel range 4.7−6.6 nm, achieved in that case by changing the lipid type. Successively, we moved forward to study the diffusion behavior of polymeric species, taking 10 000 g/mol Dextran as a model system; the corresponding transport profiles through both lipid membranes are given in Figure 5. Also in this case, the polymer diffuses faster through the mesophase swollen by the sugar ester (red squares) than in the unperturbed mesophase (blue triangles). The inset of Figure 5, however, which gives the amount of diffused polymer versus square root

Figure 5. Effects of 20% sugar ester addition on the diffusion behavior of the 10 000 g/mol Dextran within the lipid mesophase (swollen and unperturbed). Dextran diffusion from the swollen mesophase containing 20 wt % sugar ester (⧫) and the normal mesophase (▲) plotted against time. The inset shows the percentage of diffused solute versus the square root of time (n = 0.5), indicating that the transport behavior is purely diffusive only in the case of the swollen mesophase.

of time (n = 0.5), indicates that only in the swollen mesophase transport is purely Fickian (linear behavior), whereas in the normal mesophase diffusion follows a more complex behavior. A rationale for the deviation from the purely Fickian behavior can be advanced by considering the dimensions of the polymer compared to the size of the water channels. The dextran polymer in water follows the self-avoiding random walk statistics of polymers in a good solvent, in which the end-toend distance and the radius of gyration are proportional to the molecular weight via the Flory exponent 3/5. In a previous work, Mezzenga et al.35 estimated by molecular dynamics simulations the average end-to-end distance of a Dextran of 1500 g/mol, ⟨S1500⟩, to be ∼2.5 nm. By taking ⟨S⟩ ≈ aM3/5, where a is a constant for a fixed polymer series and M is the molecular weight, we can estimate the end-to-end distance of the Dextran of 10 000 g/mol as ⟨S10000⟩ ≈ ⟨S1500⟩(10000/ 1500)3/5 ≈ 7.8 nm, which is intermediate to the diameters of the water channels in the two mesophases considered. More specifically, the polymer is smaller than the diameter of the 16459

dx.doi.org/10.1021/la303833s | Langmuir 2012, 28, 16455−16462

Langmuir

Article

Figure 6. Effects of 20% sugar ester addition on the diffusion behavior of 5 nm gold nanoparticles (a) and on the mesophase structure before and after the diffusion study (b). (a) Diffusion profiles versus time and square root of time (inset) of 5 nm gold nanoparticles from the swollenmesophase containing 20 wt % sugar ester (⧫) and the normal-mesophase (▲). (b) SAXS profiles of the mesophases before and after the diffusion study.

the 5 nm gold nanoparticles were retained by the normal mesophase, undergoing aggregation at the interface (as shown by visual inspection with the formation of a red layer in the monolinolein/water system) and at grain boundaries, while the swollen mesophase allowed free diffusion through the entire membrane. The SAXS curves for the swollen mesophase undergo a very moderate shift toward higher q possibly due to a minimal partition of sugar ester on the excess water, which however does not affect the diffusion study. This feature makes the present system very appealing for applications as sizeselective membranes, in which the cutoff level can be adjusted at will, drug delivery systems, or host mesophases for in meso crystallization of large proteins.

swollen mesophase (9.7 nm), and thus it can follow Fickian behavior across that system, but it is larger than the water channel of the normal mesophase (3.8 nm) and thus forced to follow a more complex transport behavior in which the confinement imposed by the tight channels plays a crucial role (Figure 5). This confinement effect is strong: while in the present case this is only reflected on the diffusing behavior of dextran due to the excess water conditions used, in bulk mesophases this effect has been also shown to be responsible for order-to-order transitions among different types of mesophases.35 We finally studied the diffusion of 5 nm gold nanoparticles to probe the efficiency of the swollen mesophase as a size-selective separation system. Figure 6a shows the diffusion profile of 5 nm gold nanoparticles through the swollen and normal mesophases, in which free unrestricted diffusion through the swollen mesophase is demonstrated (red squares), whereas essentially no diffusion is detected through the normal mesophase (blue triangles). The penalty of incorporating nanoparticles well exceeding in size the channel diameter of cubic mesophases can be estimated using the Helfrich theory,36 and this problem has been recently considered by Venugopal et al.37 Results indicate that the energy penalty reaches several dozens of KbT as soon as the mismatch between the nanoparticle diameter and the water domains becomes comparable with the size of the nanoparticles. In other words, this implies that the diffusion of 5 nm gold nanoparticle inside the water channel of 3.8 nm in diameter, as in the case of the normal, unperturbed mesophase considered in this work, is not possible. Yet, Figure 6a indicates that some diffusion of the order of 0.2−0.3 wt % is still detectable at very long diffusion times. This very low amount of nanoparticles is likely to diffuse through the grain boundaries of the mesophase, a result that is supported by the presence of red veins within the samples upon visual inspection. To substantiate these observations, SAXS measurements were performed on the samples at the beginning and the end of the diffusion study. As can it be observed in Figure 6b, a broad peak appeared at low q after the 5 nm gold nanoparticle diffusion occurred in the normal, nonswollen mesophase, while no noticeable difference could be observed for the swollen mesophase. This peak is indicative of particle aggregation, as previously shown by Venugopal et al.37 Notably,

4. CONCLUSIONS We have presented a new lyotropic liquid crystal membrane able to separate molecules with different dimensions and possessing a size selectivity which can be easily engineered a priori. In particular, the fine-tuning of the size of the water channels in the bicontinuous cubic phase was achieved by simply adding to the monolinolein base system a sugar ester as a cosurfactant. Because of the large hydrophilic head of the sugar ester, this cosurfactant enhances water intake and at 10% load induces an order-to-order transition from a Pn3m to an Im3m bicontinuous cubic phase. The swelling behavior of the mesophase was followed by SAXS measurements and gravimetric analysis. The lattice parameter, water intake, and diameter of the water channels all show a linear increase with the sugar ester content, without discontinuities and this, independently from the Pn3m → Im3m order-to-order transition, enables a fine control in the diameter of the water channels. The size-selectivity properties of the mesophase were investigated by diffusion studies on a model mesophase system containing 20% sugar ester (swollen mesophase, water channels diameter 9.7 nm) and compared to the monolinolein/water system in absence of the sugar ester (normal mesophase, water channels diameter 3.8 nm), using a series of different solutes differing in size and structure. A pure Fickian diffusive behavior was demonstrated for small molecules and colloidal particles whose size is well below the diameter of the water channels. Thus, glucose or 2 nm gold nanoparticles were found to have Fickian diffusion in all cases (swollen and normal mesophases), although in both solute classes, the diffusion coefficient in 16460

dx.doi.org/10.1021/la303833s | Langmuir 2012, 28, 16455−16462

Langmuir

Article

(11) Seddon, A. M.; Lotze, G.; Plivelic, T. S.; Squires, A. M. A highly oriented cubic phase formed by lipids under shear. J. Am. Chem. Soc. 2011, 133, 13860−13863. (12) Fong, C.; Le, T.; Drummond, C. J. Lyotropic liquid crystal engineering−ordered nanostructured small molecule amphiphile selfassembly materials by design. Chem. Soc. Rev. 2012, 41, 1297−1322. (13) Luzzatti, V.; Husson, F. The structure of the liquid-crystalline phases of lipid-water systems. J. Cell Biol. 1962, 12, 207−219. (14) Shah, J. C.; Sadhale, Y.; Chilukuri, D. M. Cubic phase gels as drug delivery systems. Adv. Drug Delivery Rev. 2001, 47, 229−250. (15) Lynch, M. L.; Ofori-Boateng, A.; Hippe, A.; Kochvar, K.; Spicer, P. T. Enhanced loading of water-soluble actives into bicontinuous cubic phase liquid crystals using cationic surfactants. J. Colloid Interface Sci. 2003, 260, 404−413. (16) Amar-Yuli, I.; Azulay, D.; Mishraki, T.; Aserin, A.; Garti, N. The role of glycerol and phosphatidylcholine in solubilizing and enhancing insulin stability in reverse hexagonal mesophases. J. Colloid Interface Sci. 2011, 364, 379−387. (17) Chang, C. M.; Bodmeier, R. Effect of dissolution media and additives on the drug release from cubic phase delivery systems. J. Controlled Release 1997, 46, 215−222. (18) Angelov, B.; Angelova, A.; Ollivon, M.; Bourgaux, C.; Campitelli, A. Diamond-type lipid cubic phase with large water channels. J. Am. Chem. Soc. 2003, 125, 7188−7189. (19) Engblom, J.; Miezis, Y.; Nylander, T.; Razumas, V.; Larsson, K. On the swelling of monoolein liquid-crystalline aqueous phase in the presence of distearoylphosphatidylglycerol. Prog. Colloid Polym. Sci. 2000, 116, 9−15. (20) Clogston, J.; Caffrey, M. Controlling release from the lipidic cubic phase. Amino acids, peptides, proteins and nucleic acids. J. Controlled Release 2005, 107, 97−111. (21) Zabara, A.; Mezzenga, R. Plenty of room to crystallize: swollen lipidic mesophases for improved and controlled in-meso protein crystallization. Soft Matter 2012, 8, 6535−6541. (22) Uehara, H.; Kakiage, M.; Sekiya, M.; Sakuma, D.; Yamonobe, T.; Takano, N.; Barraud, A.; Meurville, E.; Ryser, P. Size-selective diffusion in nanoporous but flexible membranes for glucose sensors. ACS Nano 2009, 4, 924−932. (23) Yaghmur, A.; de Campo, L.; Sagalowicz, L.; Leser, M. E.; Glatter, O. Control of the internal structure of MLO-based isasomes by the addition of diglycerol monooleate and soybean phosphatidylcholine. Langmuir 2006, 22, 9919−9927. (24) Angelov, B.; Angelova, A.; Garamus, V. M.; Lebas, G.; Lesieur, S.; Ollivon, M.; Funari, S. S.; Willumeit, R.; Couvreur, P. Small-angle neutron and x-ray scattering from amphiphilic stimuli-responsive diamond-type bicontinuous cubic phase. J. Am. Chem. Soc. 2007, 129, 13474−13479. (25) Hyde, S. T.; Andersson, S. A cubic structure consisting of a lipid bilayer forming an infinite periodic minimum surface of the gyroid type in the glycerolmonooleat-water system. Z. Kristallogr. 1984, 168, 213−219. (26) Aota-Nakano, Y.; Li, S. J.; Yamazaki, M. Effects of electrostatic interaction on the phase stability and structures of cubic phases of monoolein/oleic acid mixture membranes. Biochim. Biophys. Acta 1999, 1461, 96−102. (27) Templer, R. H.; Seddon, J. M.; Warrender, N. A.; Syrykh, A.; Huang, Z.; Winter, R.; Erbes, J. Inverse bicontinuous cubic phases in 2:1 fatty acid/phosphatidylcholine mixtures. the effects of chain length, hydration, and temperature. J. Phys. Chem. B 1998, 102, 7251−7261. (28) Hyde, S. T. Microstucture of bicontinuous surfactant aggregates. J. Phys. Chem. 1989, 93, 1458−1464. (29) Larsson, K. Cubic lipid-water phases: structures and biomembrane aspects. J. Phys. Chem. 1989, 93, 7304−7314. (30) Tuner, D. C.; Wang, Z. G.; Gruner, S. M.; Mannock, D. A.; McElhaney, R. N. Structural study of the inverted cubic phases of didodecyl alkyl-beta-D-glucopyranosyl-rac-glycerol. J. Phys. II 1992, 2, 2039−2063.

swollen mesophases was found to be 4 times higher than in the normal mesophases, highlighting the effect of water channels size in transport properties. Large polymers with end-to-end distance falling between the size of the normal and swollen mesophase water channels (10 000 g/mol dextran, 7.8 nm) were found to follow Fickian diffusion only in swollen mesophases, whereas strong confinement in the normal mesophases imposed a more complex and restricted (nonFickian) diffusion. In the case of colloidal nanoparticles whose diameter falls exactly between the size of the normal and swollen mesophase water channels (5 nm gold nanoparticles), an efficient cutoff of the diffusion in the normal mesophases was demonstrated, whereas swollen mesophases still enabled a selective diffusion of the nanoparticles. These swollen mesophase systems can be engineered to meet specific structural requirements and show promising features suitable in the area of size-selective membranes for molecular separation, drug release, or crystallization of large molecules, such as proteins and DNA.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: raff[email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors thank Christophe Zeder (ETH Zürich) for support on the atomic absorption spectroscopy (AAS) measurements and Horst Adelman (ETH Zürich) for the assistance during the setup design.

■

REFERENCES

(1) Amar-Yuli, I.; Libster, D.; Aserin, A.; Garti, N. Solubilization of food bioactives within lyotropic liquid crystalline mesophases. Curr. Opin. Colloid Interface Sci. 2009, 14, 21−32. (2) Fong, W. K.; Hanley, T.; Boyd, B. J. Stimuli responsive liquid crystals provide ‘on-demand’ drug delivery in vitro and in vivo. J. Controlled Release 2009, 135, 218−226. (3) Negrini, R.; Mezzenga, R. pH-responsive lyotropic liquid crystals for controlled drug delivery. Langmuir 2011, 27, 5296−5303. (4) Vallooran, J. J.; Handschin, S.; Bolisetty, S.; Mezzenga, R. Twofold light and magnetic responsive behavior in nanoparticlelyotropic liquid crystal systems. Langmuir 2012, 28, 5589−5595. (5) Guo, C. Y.; Wang, J.; Cao, F. L.; Lee, R. J.; Zhai, G. X. Lyotropic liquid crystal systems in drug delivery. Drug Discovery Today 2010, 15, 1032−1040. (6) Boyd, B. J.; Whittaker, D. V.; Khoo, S. M.; Davey, G. Lyotropic liquid crystalline phases formed from glycerate surfactants as sustained release drug delivery systems. Int. J. Pharm. 2006, 309, 218−226. (7) Masum, S. M.; Li, S. J.; Tamba, Y.; Yamashita, Y.; Tanaka, T.; Yamazaki, M. Effect of de novo designed peptides interacting with the lipid-membrane interface on the stability of the cubic phases of the monoolein membrane. Langmuir 2003, 19, 4745−4753. (8) Rummel, G.; Hardmeyer, A.; Widmer, C.; Chiu, M. L.; Nollert, P.; Locher, K. P.; Pedruzzi, I.; Landau, E. M.; Rosenbusch, J. P. Lipidic cubic phases: New matrices for the three-dimensional crystallization of membrane proteins. J. Struct. Biol. 1998, 121, 82−91. (9) Caffrey, M.; Cherezov, V. Crystallizing membrane proteins using lipidic mesophases. Nat. Protoc. 2009, 4, 706−731. (10) Zabara, A.; Mezzenga, R. Tuning in-meso-crystallized lysozyme polymorphism by lyotropic liquid crystal symmetry. Langmuir 2011, 27, 6418−6425. 16461

dx.doi.org/10.1021/la303833s | Langmuir 2012, 28, 16455−16462

Langmuir

Article

(31) Briggs, J.; Chung, H.; Caffrey, M. The temperature-composition phase diagram and mesophase structure characterization of the monoolein/water system. J. Phys. II 1996, 6, 723−751. (32) Ritger, P. L.; Peppas, N. A. A simple equation for description of solute release II. Fickian and anomalous release from swellable devices. J. Controlled Release 1987, 5, 37−42. (33) Boyd, B. J.; Whittaker, D. V.; Khoo, S. M.; Davey, G. Lyotropic liquid crystalline phases formed from glycerate surfactants as sustained release drug delivery systems. Int. J. Pharm. 2006, 309, 218−226. (34) Burrows, R.; Collett, J. H.; Attwood, D. The release of drugs from monoglyceride-water liquid crystalline phases. Int. J. Pharm. 1994, 111, 283−293. (35) Mezzenga, R.; Grigorov, M.; Zhang, Z.; Servais, C.; Sagalowicz, L.; Romoscanu, A. I.; Khanna, V.; Meyer, C. Polysaccharide-Induced Order-to-Order Transitions in Lyotropic Liquid Crystals. Langmuir 2005, 21, 6165−6169. (36) Helfrich, W. Elastic properties of lipid bilayers: theory and possible experiments. Z. Naturforsch. 1973, 28c, 693−703. (37) Venugopal, E.; Bhat, S. K.; Vallooran, J. J.; Mezzenga, R. Phase behavior of lipid-based lyotropic liquid crystals in presence of colloidal nanoparticles. Langmuir 2011, 27, 9792−9800.

16462

dx.doi.org/10.1021/la303833s | Langmuir 2012, 28, 16455−16462