Celecoxib Encapsulation in β-Casein Micelles: Structure, Interactions

Jun 11, 2015 - Whitney C. Blocher , Sarah L. Perry ... Irina Portnaya , Sharon Avni , Ellina Kesselman , Yoav Boyarski , Shahar Sukenik , Daniel Harri...
0 downloads 0 Views 6MB Size
Article pubs.acs.org/Langmuir

Celecoxib Encapsulation in β‑Casein Micelles: Structure, Interactions, and Conformation Tanya Turovsky,† Rafail Khalfin,‡ Shifi Kababya,§ Asher Schmidt,§ Yechezkel Barenholz,∥ and Dganit Danino*,†,⊥ †

Russell Berrie Nanotechnology Institute, Technion − Israel Institute of Technology, Haifa 3200003, Israel Department of Chemical Engineering, Technion − Israel Institute of Technology, Haifa 3200003, Israel § Shulich Faculty of Chemistry, Technion − Israel Institute of Technology, Haifa 3200003, Israel ∥ Laboratory of Membrane and Liposome Research, IMRIC, the Hebrew University−Hadassah Medical School, Jerusalem 91120, Israel ⊥ Department of Biotechnology and Food Engineering, Technion − Israel Institute of Technology, Haifa 3200003, Israel ‡

S Supporting Information *

ABSTRACT: β-Casein is a 24 kDa natural protein that has an open conformation and almost no folded or secondary structure, and thus is classified as an intrinsically unstructured protein. At neutral pH, β-casein has an amphiphilic character. Therefore, in contrast to most unstructured proteins that remain monomeric in solution, β-casein self-assembles into well-defined core−shell micelles. We recently developed these micelles as potential carriers for oral administration of poorly water-soluble pharmaceuticals, using celecoxib as a model drug. Herein we present deep and precise insight into the physicochemical characteristics of the protein-drug formulation, both in bulk solution and in dry form, emphasizing drug conformation, packing properties and aggregation state. In addition, the formulation is extensively studied in terms of structure and morphology, protein/drug interactions and physical stability. Particularly, NMR measurements indicated strong drug− protein interactions and noncrystalline drug conformation, which is expected to improve drug solubility and bioavailability. Small-angle X-ray scattering (SAXS) and cryogenic transmission electron microscopy (cryo-TEM) were combined for nanostructural characterization, proving that drug−protein interactions lead to well-defined spheroidal micelles that become puffier and denser upon drug loading. Dynamice light scattering (DLS), turbidity measurements, and visual observations complemented the analysis for determining formulation structure, interactions, and stability. Additionally, it was shown that the loaded micelles retain their properties through freeze-drying and rehydration, providing long-term physical and chemical stability. Altogether, the formulation seems greatly promising for oral drug delivery.



suggestion builds upon some unique properties: β-CN is a calcium-sensitive phosphoprotein, consisting of a 24 kDa polypeptide chain with 209 amino acids.9 It has been classified as an intrinsically unstructured protein (IUP), i.e., it has an open conformation and no tertiary structure.10 However, while typical IUPs remain monomeric in solution, β-CN selfassembles and forms stable nanosized micellar structures at neutral pH.11 The spontaneous self-organization of β-CN in physiological conditions is associated with the distribution of charges that gives rise to the amphiphilic character of the protein; at neutral pH β-CN carries a moderate net charge of ∼14 e, concentrated mainly within the first 50 amino acid groups of the N-terminus, whereas the C terminus has only a few charged groups and a small net charge, and is rich in hydrophobic groups.9,11 Like small amphiphiles, β-CN has a

INTRODUCTION The research of protein micelles and nanoparticles as drug carriers has substantially expanded in the past decade owing to nanotechnology contributions to the world of nanomedicine.1 Such systems are formed by natural self-assembly of protein subunits into organized structures, where the subunits may be identical or distinct.2 The proteins are considered ideal for drug delivery applications due to their exceptional properties, e.g., biodegradability, biocompatibility, low toxicity, abundant renewable sources, ease of scaling up during manufacturing, and their being metabolizable.2,3 Common examples include albumin, collagen, gelatin, whey protein, and casein.4−6 Food proteins are of particular interest due to their high nutritional value and outstanding functional features, namely emulsification, gelation, water binding and foaming ability, and their application in the food industry.4 β-Casein (β-CN) is one of the four main caseins in bovine milk.7,8 In this study β-CN is suggested as an attractive natural alternative of synthetic block copolymers to construct an oral drug delivery system. This © 2015 American Chemical Society

Received: November 19, 2014 Revised: June 4, 2015 Published: June 11, 2015 7183

DOI: 10.1021/acs.langmuir.5b01397 Langmuir 2015, 31, 7183−7192

Article

Langmuir

to explore the aggregation state of Cx in the protein micelles. The effect of powder rehydration on the structural properties and stability was studied as well.

well-defined critical micellization concentration (CMC) that is found to vary between 0.05 and 0.2% w/v as a function of temperature, pH, solvent composition and ionic strength.9 Above the CMC, the protein self-organizes into classical micelles constructed of a hydrophobic core and a hydrophilic corona.11,12 The micelles are characterized by an aggregation number of 15−60, Rg values of 7.3−13.5 nm, diameter of about 20−25 nm, and stability over a wide pH and temperature range.9,13 In addition, β-CN has all the aforementioned benefits of natural food proteins. We took advantage of the self-aggregation behavior of β-CN and developed its micelles as drug nanocapsules for potential oral delivery applications. Oral drug delivery is the most common route of drug administration; it is convenient, costeffective, easily administered, and flexible in the design of dosage form.14 Nevertheless, poor water solubility constitutes an enormous problem in the development of new, as well as generic dosage forms, especially for drugs applied via the oral route. Poor drug solubility often leads to serious drawbacks, such as limited absorption, gastrointestinal mucosal toxicity, high fed/fasted variation, low and erratic bioavailability, poor stability, and more. This problem is especially relevant in class II and IV drugs of the biopharmaceutics classification system (BSC), where absorption is limited by dissolution rate. Solubility enhancement remains the dominant challenge in oral-dosage development.14,15 As a proof of concept, we utilized Celecoxib (Cx). Cx is a nonsteroidal anti-inflammatory drug (NSAID) that selectively inhibits cyclooxygenase-2 (COX-2) enzymes. It is a BSC class II drug (characterized by low solubility and high permeability) used for the treatment of rheumatoid arthritis and osteoarthritis, now also evaluated as a potent anticancer drug.16 However, its low aqueous solubility (≅5 μg/mL) contributes to high variability in absorption after oral administration17 and requires giving high doses. This might lead to severe toxicity, which may be reduced by drug encapsulation into β-CN micelles, as it substantially enhances drug dispersibility. Further, to carrying the drug, β-CN loaded micelles are expected to ease drug release through natural digestion of the protein in the stomach or by interaction of the micelles with the GI tract walls.18 We recently designed a simple, fast, reproducible, and efficient formulation that allows encapsulation of very high Cx loads,18 several orders of magnitude higher than other β-CN/ drug preparations.19 The formulation also does not require additives, as do other systems that have high loadings.20 Our previous research focused on finding the required conditions for stable formation of structures and improved drug loading.18 The present study aims to achieve detailed, quantitative information on the characteristics of the drug-loaded micelles, both in the bulk solution and in dry form, emphasizing the investigation of drug conformation, packing properties, and aggregation state. In addition, the formulation is extensively studied in terms of structure and morphology, protein/drug interactions and physical stability. To fulfill these purposes, the formulation was characterized by various techniques: shape and size were precisely determined by two complementary highresolution methods: small-angle X-ray scattering (SAXS) and cryogenic transmission electron microscopy (cryo-TEM); solution 19F nuclear magnetic resonance (19F-NMR), light microscopy (LM), dynamic light scattering (DLS), turbidity measurements, and visual observations complemented this analysis in determining the structural properties, interactions, and stability of the formulation; solid-state 13C NMR was used



EXPERIMENTAL SECTION

Dispersions Preparation. Bovine β-CN (>90%, Sigma-Aldrich) was dissolved in 20 mM Hepes buffer (pH 6.8, MP Biomedicals) containing 1 mM MgCl2 (Sigma-Aldrich), 2 mM ethylene glycol tetraacetic acid (EGTA; Sigma-Aldrich) and 10 mM NaCl (Merck), and was stirred overnight at 4 °C. The protein dispersions were prepared at a concentration of 10 mg/mL (1 wt %, 0.42 mM) or 20 mg/mL (2 wt %, 0.84 mM), at least an order of magnitude above the CMC (0.5 to 2 mg/mL, 0.021 to 0.083 mM respectively), where stable protein micelles exist.9 The dispersions were transparent. Cx (MW of 381.373 g/mol, TEVA, Israel) was dissolved in absolute ethanol (Bio Lab, Israel) and a known amount of that solution was added dropwise to the protein micellar dispersion under stirring for 30 min at 25 °C, to a final protein:drug mole ratio of 1:8, 5% v/v ethanol. For the solution NMR experiments, drug-loaded micelles were prepared as described above, but β-CN was dissolved in deuterium oxide (Sigma-Aldrich) containing 80 mM NaCl, 5.65 mM Na2HPO4, and 3.05 mM NaH2PO4 (all compounds from Merck) with ionic strength of 0.1, pH 7.0. Cx was dissolved in DMSO-d6 (CIL). Solid-state 13C NMR measurements were performed utilizing phosphate buffer saline, pH 7.4 (SigmaAldrich) for protein solubilization. In all formulations the solutions remained transparent upon drug loading. As a control, Cx solution in absolute ethanol was added dropwise under the same conditions into the buffer. All control drug dispersions were cloudy, indicating the poor dispersibility and solubility of Cx in aqueous solution without βCN. Lyophilization. Protein and protein−drug micellar dispersions were lyophilized by freezing in liquid nitrogen followed by drying in a Christ Alpha 1−4 lyophilizer for 24 h. The specimens were stored at 4 °C, and then suspended in aqueous solution back to the original concentration (10 mg/mL protein). Resuspension was performed by weighing the powder, adding a measured amount of double distilled water, and stirring for 30 min at room temperature. The suspensions obtained were transparent and stable for at least 3 weeks. Small Angle X-ray Scattering. SAXS experiments were performed using a small-angle diffractometer (Molecular Metrology SAXS system with Cu Kα radiation from a sealed microfocus tube (MicroMax-002+S), two Göbel mirrors, and three-pinhole slits. Generator powered at 45 kV and 0.9 mA). The scattering patterns were recorded by a 20 × 20 cm two-dimensional position sensitive wire detector that was positioned 150 cm behind the sample. The resolution of the SAXS system is greater than π/hmax = π/2.7, ∼ 1.16 nm.21 The scattered intensity I(h) was recorded in the interval 0.07 < h < 2.7 nm−1, where h is the scattering vector defined as h = (4π / λ) sin(θ), where 2θ is the scattering angle, and λ is the radiation wavelength (0.1542 nm). The solutions under study were sealed in a thin-walled glass capillary of about 2 mm in diameter and 0.01 mm wall thickness, and measured at 25 °C under vacuum. I(h) was normalized to time, solid angle, primary beam intensity, capillary diameter, transmission, and the Thompson factor.21 Scattering of the solvent (buffer), empty capillary, and electronic noise were subtracted. SAXS patterns of β-CN micelles in Hepes buffer, either with or without the drug, display an interference peak resulting from intermicellar interactions.12 To avoid this, the NaCl concentration in the buffer was increased from 10 mM to 50 mM. This eliminated the interparticle interference peak and allowed SAXS patterns to be analyzed more readily. Quantitative analysis was performed considering the model of an ellipsoid shape with dimensions a, a, b and a model of polydispersed spheres. All calculations were performed using the Mathcad software. Solution 19F NMR. For the solution NMR experiments, a specimen of 20 mg/mL β-CN with 1:8 mol ratio β-CN:Cx (Cx concentration of 2.54 mg/mL) was analyzed. This specimen was further used for cryo-TEM and DLS analysis. Due to the low Cx 7184

DOI: 10.1021/acs.langmuir.5b01397 Langmuir 2015, 31, 7183−7192

Langmuir



solubility in water, Cx was dissolved in DMSO-d6 to a concentration of 2.54 mg/mL as a control. Solution NMR spectra were acquired on AV-III 600 Bruker spectrometer with resonance frequencies for 19F, 565.06 MHz. Bruker Topspin 2.1 was used on a PC (windows-XP) for spectral acquisition and processing. Samples were at ambient temperature. 19F chemical shift was referenced according to IUPAC procedure.22 Solid-State 13C NMR. Solid state 125.81 MHz 13C MAS NMR spectroscopy measurements were carried out on a 500 MHz AVANCE III (Bruker) solid-state NMR spectrometer equipped with tripleresonance probes using 4 mm (OD) Zirconia rotors. 13C Crosspolarization (CP) MAS echo experiments (indirect excitation) were carried with a 10.0 μs π pulse widths, an echo interval τ (96 μs) identical to the rotational period TR (rotor spinning rate 10,416 Hz), with a 1H decoupling level of 100 kHz and 5.0 μs π/2 pulse. Hartmann−Hahn RF levels were matched at 50 kHz, with a 2 ms contact time. Relaxation delay was 3 s and up to 8000 transients were collected. 13C CPMAS experiments with Interrupted Decoupling (CPid)23 used 74 μs interval without 1H decoupling; this interval was centered about the rotor-synchronized refocusing π-pulse. CPid experiment is capable of spectral editing keeping only carbon with reduce dipolar coupling to hydrogens, therefore keeping peaks of, e.g., quaternary and methyl carbons. In all experiments, the data points of the free induction decay signals were left-shifted to the first rotational echo position prior to the Fourier transform. 13C chemical shifts were reported relative to tetramethylsilane. Solid-state 13C NMR measurements were performed on lyophilized β-CN solution (1 wt %), lyophilized β-CN/Cx solution (1 wt % β-CN with 1:8 mol ratio β-CN:Cx), and polycrystalline Cx powder. Direct-Imaging Cryogenic Transmission Electron Microscopy. Specimens were prepared either in an automated vitrification device (Vitrobot, FEI, The Netherlands) or in a homemade controlled-environment vitrification system (CEVS), at controlled temperature and humidity conditions (25 °C, 100% RH) to avoid loss of volatiles. Five microliters of the suspension was placed on a TEM copper grid covered with a perforated carbon film, and then blotted with a filter paper to form a thin liquid film of the specimen. The grid was plunged into liquid ethane at its freezing temperature (−183 °C) to form a vitrified specimen, and stored in liquid nitrogen (−196 °C) until examination. Some specimens were examined in a Philips CM120 transmission electron microscope (Philips, The Netherlands), others in a Tecnai 12 G2 TEM (FEI), always at an accelerating voltage of 120 kV. Gatan 626 cryo-holder was used. Specimens were studied in a lowdose imaging mode to minimize electron beam exposure and radiation damage.24 Images were recorded digitally in the CM120 on a cooled Gatan MultiScan 791 CCD camera (Gatan, UK), and in the Tecnai on a high-resolution 2k × 2k Ultrascan 1000 cooled CCD camera (Gatan), using the Digital Micrograph 3.6 software (Gatan), applying imaging procedures we developed.25 Light Microscopy. An Olympus BX51 light microscope was operated at Nomarski differential interference contrast (DIC) optics. A 5 μL drop was placed on a glass slide and covered with a cover slide. Images were recorded digitally at magnifications of 10- to 60-fold, with an Olympus DP71 digital camera connected to the light microscope. Image processing was done using Cell A (Olympus). Turbidity. Turbidity measurements were performed using an Amersham Biosciences Ultrospec 2100 pro spectrophotometer at a wavelength of 600 nm (far from the protein absorbance range) with a light path of 1 cm. Dynamic Light Scattering. Size distribution was determined using a combined DLS and zeta-potential analyzer (Nicomp 380 ZLS zeta potential/particle sizer, USA), at 25 °C. The scattered light intensity was detected with an Avalanche Photo Diode (APD) detector, used at a fixed angle of 90°. Laser wavelength was 658 nm, operating at 90 mW. Size distributions were calculated from the scattered light intensity fluctuations by using the intensity-weighted Nicomp bimodal distribution (solid particles) analysis method.

Article

RESULTS AND DISCUSSION Cx (Scheme 1) is a selective cyclooxygenase-2 inhibitor widely used in the treatment of osteoarthritis, rheumatoid arthritis, and Scheme 1. Structure of Celecoxib

acute pain. It is poorly water-soluble NSAID with low oral bioavailability. It was found that Cx possesses high permeability; however, when absorbed from solid dosage forms, its dissolution may constitute a rate-limiting factor.26 Therefore, keeping the drug in a disordered state is expected to assist in gaining enhanced pharmacokinetic parameters and in reducing the drug dosage, with subsequent benefits from economic and toxicological perspectives. Generally, in aqueous solution without β-CN, Cx has poor dispersibility and solubility; in buffer it exists as large crystals resulting in the formation of a cloudy dispersion. By contrast, the drug-loaded suspension in the presence of β-CN is completely transparent, and hardly any excess of drug crystals is observed by light microscopy (Supporting Information (SI) Figure S1). These results indicate on considerable enhancement in drug solubility and on strong interactions of the drug with the protein micelles, and suggest that β-CN/Cx complexes, like β-CN micelles, are of nanometric size.18 We previously found optimal protein and drug concentrations for our formulation and achieved complete drug solubilization in βCN micelles up to a very high loading (1:15 protein:drug mole ratio in 20 mg/mL β-CN).18 Herein, the protein-drug complexes were thoroughly characterized, providing detailed information regarding their structure, conformation, and interactions. Protein−Drug Interactions and Drug Conformation. First, we combined two complementary techniques, solid-state 13 C NMR and solution 19F-NMR spectroscopy, in order to characterize protein−drug interactions and the aggregation state of encapsulated drug in the dried formulation, as well as in the hydrated dosage form. Solid-State 13C NMR Spectroscopy. Solid-state 13C NMR spectroscopy is an essential and highly effective analytical technique nowadays considered as an integral tool in pharmaceutical sciences and drug delivery.27 It is well established now that pharmaceutical solids may exhibit several polymorphic forms, which may be translated to considerable variations in pharmaceutically significant characteristics as bioavailability, solubility, physical/chemical stability, and processability. The studies of order/disorder and polymorphism appear to be among the most widespread utilizations of this tool for pharmaceutical compounds.27 We have used solid-state 13C{1H} CPMAS NMR to determine the conformation of encapsulated Cx in the dried powder state of the dosage form. Measurements were performed on a lyophilized specimen of 1 wt % β-CN/Cx micelles of a 1:8 mol ratio, as well as on lyophilized 1 wt % βCN sample and Cx powder as controls. The spectrum of Cx 7185

DOI: 10.1021/acs.langmuir.5b01397 Langmuir 2015, 31, 7183−7192

Article

Langmuir

Figure 1. 13C 125.81 MHz NMR spectra of Cx powder (top); red trace represents lyophilized 1 wt % β-CN, blue trace represents lyophilized 1 wt % β-CN/Cx micelles of a 1:8 mol ratio (bottom). Left 13C{1H} CPMAS; right CPid (CPMAS with 74 μs interrupted decoupling). Focusing on a chemical shift range where peaks are exclusively from Cx (120−150 ppm), their pronounced broadening (bottom traces) compared to the crystalline peaks (top traces) is evident. This broadening is indicative that Cx is in a disordered from.

powder exhibits a set of narrow peaks indicative of its crystalline state (Figure 1; top traces). The two protein samples spectra show heterogeneously broadened peaks characteristic of lyophilized proteins (Figure 1; red trace βCN, blue trace β-CN/Cx). The β-CN/Cx spectrum additionally clearly shows Cx contributions between 120 and 150 ppm, indicating the existence of the drug in the sample. This observation is further substantiated by applying CPid, editing the 13C{1H} CPMAS spectrum, preserving primarily the quaternary carbon peaks of the drug in this spectral range (Figure 1; bottom traces). The presence of drug peaks considerably broader than the reference crystalline Cx peaks indicate that Cx in the lyophilized β-CN/Cx sample are not crystalline. These results suggest the drug is present either as a molecular dispersion that interacts with the protein, or alternatively, in the form of disordered (amorphous) molecular aggregates. Solution 19F-NMR Spectroscopy. The 19F-NMR spectra of Cx (in DMSO-d6) and Cx-loaded β-CN (in D2O-based buffer) are shown in Figure 2. The Cx spectrum shows a single narrow peak with a characteristic chemical shift at −63.4 ppm.28 The peak from Cx dissolved in β-CN micelles exhibits a pronounced broadening. We attribute this broadening to the much slower tumbling of Cx as a result of its incorporation inside the much larger micelles. Thus, the drug in micelles is not crystalline; as in the dried formulation, it may exist as a molecular dispersion interacting with the protein, or as amorphous molecular entities. To increase the NMR signal-to-noise, we doubled encapsulated Cx concentration, while keeping the same protein to drug mole ratio in the formulation. This was done by increasing βCN concentration up to 2 wt %. Drug encapsulation at a high protein concentration of 2 wt % resulted in slightly bluish dispersion with larger, more spherical, and clearly more swollen micelles (Figure 3 and Figure S2) in comparison with the lowconcentration drug-loaded micelles (as further discussed). Still, in both cases, the micelles remain flat oblates. DLS measure-

Figure 2. Solution 19F-NMR spectra of (a) Cx in DMSO-d6 (6.7 mM Cx), and (b) 2 wt % β-CN loaded with Cx (6.7 mM Cx, 1:8 β-CN/Cx mole ratio). Cx peak broadening indicates restricted motion of the molecule as a result of the drug−(micellar protein) interactions.

ments (Figure S2) indicated diameters of 31 ± 4 nm and 40 ± 7 nm for the empty micelles and the drug-loaded micelles, respectively. No larger structures were formed, confirming that the loading capacity of β-CN micelles is retained when protein concentration is raised to 2 wt %. An especially essential property from a pharmaceutical perspective is the aggregation state of the drug in the protein micelles; it may exist as amorphous or crystalline entities, or as molecular dispersion. The results obtained by solid-state NMR and solution NMR are consistent with our previously reported findings,18 where wide-angle X-ray diffraction (XRD) was applied. Both NMR and XRD suggest the existence of strong drug−protein interactions and their preservation in the dried formulation, and indicate that the broad drug peaks represent molecular dispersion of the drug that is tightly associated with the protein micellar scaffold/matrix. These results rule out the possibility that crystalline drug entities are encapsulated and 7186

DOI: 10.1021/acs.langmuir.5b01397 Langmuir 2015, 31, 7183−7192

Article

Langmuir

Figure 4. Photographs (upper panel) and turbidity measurements (lower panel) of empty 1 wt % β-CN micelles and micelles loaded with cx to 1:8 mol ratio (3.33 mM), fresh (A and B, respectively), and after drying and resuspension (C and D, respectively); Cx in buffer (3.33 mM) is shown in (E).

Figure 3. Cryo-TEM image of 2 wt % β-CN micelles loaded with cx to 1:8 mol ratio. In the image, micelles are positioned in different orientations; since the micelles are oblate-shaped, some of them seem round (inside black circles) and others very narrow (inside white circles).

Here, SAXS provided information on the micelle shape, size and density. These parameters were extracted from fitting the experimental data to a model shape. The building blocks for modeling and quantitative analysis were provided by cryoTEM, which showed shapes ranging from spheroids to spheres (Figure 5). Accordingly, these two extremes were considered as models. Specifically, the utilized models were monodisperse ellipsoids of dimensions a, a, b and polydisperse solid spheres of radius r (Figure 6). Three different parameters were derived from fitting to each of these model shapes: in the ellipsoid model, the parameters were the large semi axis a, the small semi axis b, and the micelle density, which was considered as constant; in the polydispersed spheres model the parameters were the sphere radius r, radius standard deviation, and the micelles density (number of micelles per unit volume). Initially, protein monomer radius of gyration and amplitude of monomer were considered as variables too; this yielded slightly different values for each of the samples in both models. However, since these parameters are not expected to change between our samples and models, average values of 4.6 and 1500 nm were chosen for protein monomer radius of gyration and amplitude of monomer, respectively (see Table 1 and Table 2). This gave consistency in initial conditions between the different models and was in excellent agreement with previously reported value of 4.6 nm for the protein monomer radius of gyration.32 Experimental SAXS curves and fitting curves to ellipsoid and sphere models are shown in Figures S3 and S4. The flat tails in the Kratky plots [I(h)·h2 versus h] indicate that intensity is proportional to 1/h2 for high values of h (Figure S5), thus correspond to scattering from coils of protein monomers in the dispersion, whereas the scattering in smaller h values imply on the presence of bigger particles−the micelles. The SAXS data (I(h) versus h) after normalization is presented in Figure 7A. One can observe that the large h values (tails) of all experimental SAXS curves (with and without Cx) coincide. This indicates that the number of protein monomers per volume unit is the same in all solutions (most likely representing the CMC); thus, the total number of casein molecules arranged in micelles is the same too. The micelles’ parameters of β-CN and β-CN loaded with Cx, of fresh samples and samples after drying and resuspension, obtained from both the experimental SAXS patterns and from fitting to ellipsoid and sphere models are summarized in Table 1 and Table 2,

indicate drug presence in micelles either as a molecular dispersion that interacts with the protein, or alternatively, in the form of disordered (amorphous) molecular aggregates. Altogether, NMR and XRD combined provide a powerful strategy for complete structural characterization: data obtained by solution NMR represents an average over molecules that undergo random reorientation in solution, while XRD data averages molecules that are arranged in long-range periodic crystal lattices. Like XRD, solid-state NMR gives structural information from samples in the solid state; however, the latter represents integrative description of the local environment.29,30 Most importantly, encapsulating the drug in a disordered form, as suggested by our results, is expected to improve Cx water-solubility in the protein/drug dry formulation, which in turn may potentially increase intestinal absorption. Indeed, recent examination of our formulation in vivo showed significant improvement in Cx bioavailability compared to the commercial drug oral formulation (Celebra).31 Turbidity. Turbidity analysis was performed on 1 wt % β-CN micellar solutions, empty and loaded with 1:8 mol ratio Cx. Samples that were lyophilized and then suspended were analyzed as well. Although to the naked eye all samples are transparent, a slight increase in turbidity was measured when Cx is loaded into the carrier (Figure 4). This observation is coherent with our finding (see next section) of slight increase in micelle size upon drug loading, yet no formation of large aggregates. Structural Analysis. We used two complementary structural high-resolution methods for precise shape and size determination of the drug-loaded micelles: cryo-TEM and SAXS. DLS constituted an additional method for micelle size estimation. Cryo-Transmission Electron Microscopy. Direct imaging cryo-TEM was utilized to study the native structure of the drugloaded assemblies at the nanometric scale. Images of both empty and drug-containing β-CN micelles show the existence of a uniform population of small oblate-shaped micelles (Figure 5, panels A1 and B1). Small-Angle X-ray Scattering. SAXS is a powerful technique for studying structural properties of colloidal size by analyzing scattering of X-rays, which occur as a result of electron density inhomogeneity in the sample.21 7187

DOI: 10.1021/acs.langmuir.5b01397 Langmuir 2015, 31, 7183−7192

Article

Langmuir

Figure 5. Cryo-TEM images (A1−D1) and schematic representative shapes (A2−D2). Empty 1 wt % β-CN micelles and micelles loaded with Cx to 1:8 mol ratio, fresh (A and B, respectively); and after drying and resuspension (C and D, respectively). Drug encapsulation results in micelle growth, but they remain oblate and uniform in size. Micelle morphologies shown on the right were drown according to the parameters a and b obtained from fitting of the experimental SAXS patterns to an ellipsoid shape.

Table 1. Micelle Structure Parameters Obtained from the Experimental SAXS Data and from Fitting of the Experimental Patterns to an Ellipsoid Model β-CN parameter 2a axis (from p(r)) [nm] 2a axis (from fitting) [nm] 2b axis (from fitting) [nm] axes ratio Porod integral [nm−6] micelle volume [nm3] micellar density [g/cm3] Rg of micelle from the spheroid axes (eq 8, SI) [nm] Rg of micelle from Guinier law (eq 1, SI) [nm] Rg of monomer [nm] amplitude of monomer [nm3]

Figure 6. (A) Ellipsoid (axes a, a, b) and (B) sphere (radius r).

respectively. The equations utilized for calculations are found in the SI section. Ellipsoid. The maximum dimension of the micelle large axis, 2a, was obtained from the experimental SAXS patterns and from the fitting procedures. To acquire 2a from the experimental data, the pair distance distribution function p(r) was calculated using eq 7 and eq 8 (SI), (see Figure 7B); the value of 2a is received from the point where p(r) drops to zero at the maximum dimension.33 Evidently, the large axis increases when drug is loaded to the β-CN micelles, in agreement with the cryo-TEM findings (Figure 5) and DLS (shown next). The experimental values found are ∼24 nm for empty β-CN micelles, and ∼27.5 nm for drug-loaded β-CN micelles, both in fresh solution and after rehydration. These are in excellent

β-CN/Cx

fresh

rehydrated

fresh

rehydrated

24.0 22.5 10.4 0.46 254 2757.5 1.23 7.5

24.0 24.4 11.1 0.46 249 3464.7 1.23 8.1

27.5 27.5 12.9 0.47 469 5114.7 1.27 9.1

27.5 27.7 13.3 0.48 408 5311.9 1.27 9.2

7.8

8.1

9.0

9.0

4.6 1500

4.6 1500

4.6 1500

4.6 1500

agreement with the dimensions found for the fitting to an ellipsoid shape (Table 1). Importantly, as pointed out in the SAXS profiles (Figure 7B and Figure S3) and the values in Table 1, lyophilization does not have a significant effect on the micelle parameters: the plots 7188

DOI: 10.1021/acs.langmuir.5b01397 Langmuir 2015, 31, 7183−7192

Article

Langmuir

When comparing Porod integral values of empty and drugloaded micelles, it is clear that drug incorporation leads to larger integral values, thus to an increase not only in micellar volume, but in micellar density and/or in their number (Table 1; eq 11 and 12 in SI). Indeed, Table 1 shows a slight increase in micellar density and significant growth in micellar volume after Cx loading, confirming the formation of drug-encapsulated micelles. The function f(r) = p(r)/r was utilized for the identification of flat particles and for rough estimation of their thickness, which is indicated by the transition point from the round curve to the linear area.12 According to Figure 7D, the thicknesses of empty β-CN micelles, fresh and rehydrated, are about 10 to 11 nm, and the thicknesses of drug-loaded β-CN micelles, fresh and rehydrated, are about 13 nm. These values agree very well with the small spheroid axis dimensions (2b) obtained by fitting (Table 1). Furthermore, Table 1 shows that the gyration radii from the experimental SAXS profiles (from Guinier approximation21) and from the fitting procedure (from a,a,b) almost coincide, showing again the good compliance between the fitting procedure and the experimental data. For illustration, the shapes of empty and drug-loaded β-CN micelles, fresh and following drying and resuspension, were drawn according to the parameters a and b obtained from the fitting (Figure 5, panels A2−D2). One can observe the growth in micelle dimensions following drug loading, as observed by cryo-TEM too (Figure 5, panels A1−D1). Polydisperse Spheres. When comparing between fresh and rehydrated samples, good agreement between all micelle parameters is shown in Table 2 and in the SAXS profiles

Table 2. Micelle Structure Parameters Obtained from the Experimental Saxs Data and from Fitting of the Experimental Patterns to Polydisperse Spheres Model β-CN

β-CN/Cx

parameter

fresh

rehydrated

fresh

rehydrated

average diameter [nm] diameter standard deviation [nm] mean micelle volume [nm3] amplitude [nm−9] Porod integral [nm−6] Rg of monomer [nm] amplitude of monomer [nm3]

16.22 7.35

16.38 8.70

16.47 14.57

16.46 14.37

2232 0.0056 219 4.6 1500

2302 0.0056 230 4.6 1500

2341 0.010 475 4.6 1500

2334 0.0092 436 4.6 1500

of fresh and rehydrated samples almost coincide with each other; further, there is very good matching between the values of the semiaxes obtained from fresh and rehydrated samples (Table 1). This is true for both empty β-CN micelles and β-CN micelles loaded with Cx. Information on the micelles’ shape can be obtained from the p(r) curve. The experimental curves were compared to the curve of a theoretical homogeneous sphere, and the curves were normalized by dividing p(r) values by p(r)max and r values by rmax (Figure 7C). The experimental curves are characteristic to a spheroid shape: overlapping with the theoretical sphere on the left side of the maximum, and deviating from it on the right side.12 This, again, confirms the selected micelle form of spheroid for the fitting. One can observe that the curves of all samples are very similar, as also evident from the nearly constant axes ratio value of about 0.47 in all samples (Table 1).

Figure 7. Normalized SAXS data (A) and analysis of SAXS curves (B−D) of 1 wt % β-CN and 1 wt % β-CN loaded with 1:8 β-CN:Cx mole ratio, fresh and rehydrated. (B) p(r) as obtained from the SAXS curves; (C) Normalized p(r) to show the deviations in shape from a homogeneous sphere. rmax is the value of r where the p(r) = p(r)max has its maximum; (D) The function f(r) = p(r)/r, to show the presence of oblate particles and to estimate their mean thickness. 7189

DOI: 10.1021/acs.langmuir.5b01397 Langmuir 2015, 31, 7183−7192

Article

Langmuir (Figure S4), both for pure β-CN micelles and drug-loaded βCN micelles. That is, the drying and rehydration procedure did not have a significant impact on micelle parameters. Therefore, the rest of discussion regarding sphere model does not separate between the fresh and rehydrated samples and treats both at once for convenience. Average sphere diameter and diameter standard deviation were calculated using eqs 19 and 20 in the SI, respectively. It can be seen from Table 2 that the average micelle diameter does not undergo a significant change as a result of drug loading and remains about 16 nm. As evident from the diameter standard deviation values, the protein micelles have broad size distribution; when the micelles are loaded with Cx, the distribution in their size is even higher. In accordance with the similarity in average diameter values, the mean micelle volumes show the same trend of close values of around 2300 nm3 in all analyzed samples (Table 2). Then, how does drug presence reflects on SAXS parameters of the micelles? The answer lies upon the dramatic change in amplitude (B), which is nearly doubled for β-CN/Cx micelles in comparison to pure β-CN micelles. Considering the fact that micellar density (same as for the ellipsoid model, see Table 1) increases only slightly upon drug loading, the significant growth in amplitude is attributed mainly to a substantial increase in the number of micelles per unit volume (see eq 17 in SI). That is, Cx loading results in almost 2-fold increase in the total number of micelles comparing to pure protein micelles, while the mean micellar volume is virtually unchanged. Ellipsoid model vs polydisperse spheres model. As discussed above, the fitting curves of both models agree well with the experimental plots. Also, both models show good fitting between fresh and rehydrated samples, in agreement with other methods used. Nevertheless, the models differ in the impact on micelle parameters they suggest as a result of drug incorporation. According to the ellipsoid model, Cx loading leads mainly to increase in the micellar volume, whereas the sphere distribution model suggests that drug loading results primarily in increased number of micelles per unit volume, without significantly changing micelle volumes, and in a broader micelle size distribution. If so, cryo-TEM can help to decide which model fits better in our case; while SAXS can provide an average of the feature size based on a relatively large sample volume, microscopy gives the local structural details.34 As cryo-TEM indicates (Figure 5), the micelles, at least partially, are clearly oblate-shaped, both when empty and when loaded with drug. Furthermore, there is sure evidence of growth in micelle volume and no implications of increased polydispersity upon drug loading. Therefore, the ellipsoid/ oblate shape model is more appropriate in the present system. This conclusion is further supported by previous extensive studies of the self-organization behavior of β-CN performed in our group.9,11,12 It has been shown that β-CN organizes into uniformly packed and dynamic micelles that respond to temperature change. Detailed SAXS analysis of 2 wt % β-CN dispersion showed that the micelles undergo structural change from flat, disk-like micelles at low temperatures to more spherical micelles at high temperature. Importantly, it was found that at room temperature, β-CN arranges to flat, oblateshaped micelles.12 Dynamic Light Scattering. DLS is an approved method to determine average size and size distribution of small particles in a suspension.35 Since the structures under analysis possess an oblate character, whereas DLS data was analyzed under the

assumption of spherical particles, the information obtained by this method is not expected to be highly accurate. Still, it may be useful as a complementary method to estimate micellar average size and size distribution. DLS data of 1 wt % β-CN and 1 wt % β-CN loaded with 1:8 β-CN:Cx mole ratio, fresh and rehydrated, is shown in Figure S6. Analysis of empty β-CN dispersions and dispersions of βCN loaded with Cx showed bimodal particle size distribution, where the lower-size peak (about 1−3 nm) represents β-CN monomers in the dispersion and the higher-size peak (around 20−30 nm) is attributed to β-CN micelles. Mean diameter of the fresh empty micelles obtained by DLS is ∼22 ± 3 nm, similar to the results attained by SAXS when fitting to an ellipsoid shape is applied, which showed maximum dimension measured of 24 nm. After drug loading, fresh micelle diameter evaluated by DLS increased to ∼29 ± 3 nm, again in agreement with SAXS fitting to an ellipsoid shape (∼27.5 nm). Similarly, DLS analysis of 2 wt % protein empty and drugloaded (1:8 β-CN:Cx mole ratio) dispersions showed a bimodal particle size distribution, where again the 1−3 nm peak is attributed to protein monomers and the higher-size peak is attributed to the micelles (Figure S2). As expected, micelle dimensions are somewhat larger in comparison to suspensions of 1 wt % protein, and the mean diameter of the drug-loaded micelles (∼41 ± 7 nm) is higher than of the pure protein micelles (∼31 ± 2 nm), again indicating drug incorporation inside the micelles. Previous SAXS study of micelle morphology at 2 wt % protein found a micellar diameter of ∼32 nm using fitting to an oblate spheroid model.12 This is in excellent agreement with the dimensions presented here for solution of 2 wt % empty βCN micelles, as measured by DLS, confirming that this model matches our system. Another confirmation of the chosen model is the cryo-TEM imaging of the 2 wt % β-CN micelles loaded with Cx (Figure 3) that exhibits micelles of different shapes ranging from narrow (white circles) to round (black circles), which is characteristic for oblate-shaped objects positioned in different orientations in the vitrified film. Interestingly, although size measurement in cryo-TEM is somewhat complex and is slightly affected by focus and film thickness, here the results from this method are in excellent agreement with DLS and SAXS; all the techniques displayed very similar dimensions and showed an increase in micelle size upon drug loading. Freeze-Drying. Detailed characterization of the freeze-dried powder in its dried and rehydrated form is significant for establishing β-CN functionality as a potential platform for oral drug delivery. Freeze-dried β-CN micelles and β-CN/Cx micelles were easily suspended in double distilled water to the original concentration, forming transparent solutions that resemble very much the initial micellar dispersions (Figure 4). No structures were seen by light microscopy, indicating that all aggregates are small and nanometric in size, below the detection resolution of the microscope (Figure S1). An observation at much higher magnification by cryo-TEM revealed that the dimensions and morphology of both empty and drug-containing β-CN micelles did not change after rehydration; the micelles remained uniform, small, and oblate spheroids (Figure 5, panels C1 and D1), as was further quantitatively indicated by SAXS that showed good agreement between the values of the semi axes (2a and ab) of fresh and rehydrated micelles (Table 1). 7190

DOI: 10.1021/acs.langmuir.5b01397 Langmuir 2015, 31, 7183−7192

Article

Langmuir



SUMMARY This study presents detailed, quantitative characterization in terms of structure, interactions, and conformation, of a potential oral drug-delivery system based on β-CN micelles, utilizing the important drug celecoxib as an example. We found that the drug interacts strongly with the protein and is located in the micelle’s hydrophobic core, where it is either molecularly dispersed, or exists as disordered (amorphous) small molecular clusters. The micelles are small, uniform and oblate. It was shown that drug loading swells the micelles and results in increased dimensions, volume, and density. We also illustrate the absolutely reversible restoration of our system toward freeze-drying by detailed characterization in terms of aggregation state, structure and morphology. We previously demonstrated that this formulation can be lyophilized without the need of a cryoprotectant, and can be kept as a dry powder, providing long-term physical and chemical stability, which is a great advantage for clinical application.18 Importantly, we achieved very high loading efficiency (>96%), thus eventually could meet the high daily dose of this drug, in 1 g of the dry dosage that potentially can be administered as a capsule. Alternatively, fully reversible reconstitution of the structures is possible by simple rehydration of the powder, potentially making drug administration in liquid form available.18 Furthermore, we scaled up the procedure to the level that ensures fast and reproducible manufacturing of the formulation and conducted bioavailability tests in pigs, which showed a higher rate and extent of drug dissolution from our formulation in comparison to the commercial product Celebra, leading to a 1.76-fold increase in Cx bioavailability,31 demonstrating the feasibility of our formulation for oral drug-delivery. In application, an enteric coating can be used to protect the formulation from the harsh acidic conditions in the stomach. In summary, the characterization techniques utilized in this study complement previous research of the formulation18,31 and extend our understanding of the unique properties of β-CN micelles, further demonstrating the highly promising potential of this protein for drug delivery applications.



author is grateful to Dr. Irina Portnaya, Dr. Inbal IonitaAbutbul, Dr. Ellina Kesselman, and Mrs. Luba Kolik of the Department of Biotechnology and Food Engineering, Technion, for their help and support of this research.



(1) Husseini, G. A.; Pitt, W. G. Micelles and nanoparticles for ultrasonic drug and gene delivery. Adv. Drug Delivery Rev. 2008, 60, 1137−1152. (2) MaHam, A.; Tang, Z.; Wu, H.; Wang, J.; Lin, Y. Protein-based nanomedicine platforms for drug delivery. Small 2009, 5, 1706−1721. (3) Elzoghby, A. O.; Helmy, M. W.; Samy, W. M.; Elgindy, N. A. Novel ionically crosslinked casein nanoparticles for flutamide delivery: Formulation, characterization, and in vivo pharmacokinetics. Int. J. Nanomed. 2013, 8, 1721. (4) Chen, L.; Remondetto, G. E.; Subirade, M. Food protein-based materials as nutraceutical delivery systems. Trends Food Sci. Technol. 2006, 17, 272−283. (5) Elzoghby, A. O.; Samy, W. M.; Elgindy, N. A. Protein-based nanocarriers as promising drug and gene delivery systems. J. Controlled Release 2012, 161, 38−49. (6) Menéndez-Aguirre, O.; Kessler, A.; Stuetz, W.; Grune, T.; Weiss, J.; Hinrichs, J. Increased loading of vitamin D2 in reassembled casein micelles with temperature-modulated high pressure treatment. Food Res. Int. 2014, 64, 74−80. (7) Swaisgood, H. E. In Chemistry of the Caseins; Fox, P. F.; McSweeney, P. L. H., Ed.; Advanced Dairy Chemistry; Elsiver: London, 2003; Vol. 1a, pp 139−201. (8) Holt, C. Structure and stability of bovine casein micelles. Adv. Protein Chem. 1992, 43, 63−151. (9) Portnaya, I.; Cogan, U.; Livney, Y. D.; Ramon, O.; Shimoni, K.; Rosenberg, M.; Danino, D. Micellization of bovine β-casein studied by isothermal titration microcalorimetry and cryogenic transmission electron microscopy. J. Agric. Food Chem. 2006, 54, 5555−5561. (10) Tompa, P. Intrinsically unstructured proteins. Trends Biochem. Sci. 2002, 27, 527−533. (11) Portnaya, I.; Ben-Shoshan, E.; Cogan, U.; Khalfin, R.; Fass, D.; Ramon, O.; Danino, D. Self-assembly of bovine β-casein below the isoelectric pH. J. Agric. Food Chem. 2008, 56, 2192−2198. (12) Moitzi, C.; Portnaya, I.; Glatter, O.; Ramon, O.; Danino, D. Effect of temperature on self-assembly of bovine ß-casein above and below isoelectric pH. Structural analysis by cryogenic-transmission electron microscopy and small-angle X-ray scattering. Langmuir 2008, 24, 3020−3029. (13) Horne, D. S. Casein interactions: Casting light on the black boxes, the structure in dairy products. Int. Dairy J. 1998, 8, 171−177. (14) Sastry, S. V.; Nyshadham, J. R.; Fix, J. A. Recent technological advances in oral drug delivery−A review. Pharm. Sci. Technol. Today 2000, 3, 138−145. (15) Gao, L.; Liu, G.; Ma, J.; Wang, X.; Zhou, L.; Li, X.; Wang, F. Application of drug nanocrystal technologies on oral drug delivery of poorly soluble drugs. Pharm. Res. 2013, 30, 307−324. (16) Bertagnolli, M. M.; Eagle, C. J.; Zauber, A. G.; Redston, M.; Solomon, S. D.; Kim, K.; Tang, J.; Rosenstein, R. B.; Wittes, J.; Corle, D. Celecoxib for the prevention of sporadic colorectal adenomas. N. Engl. J. Med. 2006, 355, 873−884. (17) Paulson, S. K.; Zhang, J. Y.; Breau, A. P.; Hribar, J. D.; Liu, N. W.; Jessen, S. M.; Lawal, Y. M.; Cogburn, J. N.; Gresk, C. J.; Markos, C. S.; Maziasz, T. J.; Schoenhard, G. L.; Burton, E. G. Pharmacokinetics, tissue distribution, metabolism, and excretion of celecoxib in rats. Drug Metab. Dispos. 2000, 28, 514−521. (18) Bachar, M.; Mandelbaum, A.; Portnaya, I.; Perlstein, H.; EvenChen, S.; Barenholz, Y.; Danino, D. Development and characterization of a novel drug nanocarrier for oral delivery, based on self-assembled β-casein micelles. J. Controlled Release 2012, 160, 164−171. (19) Shapira, A.; Assaraf, Y. G.; Livney, Y. D. Beta-casein nanovehicles for oral delivery of chemotherapeutic drugs. Nanomed.: Nanotechnol., Biol. Med. 2010, 6, 119−126.

ASSOCIATED CONTENT

* Supporting Information S

Additional figures, SAXS theory, and full DLS data of 1, wt %, 2 wt% β-CN and 1 and 2 wt% β-CN loaded with 1:8 β-CN:Cx mole ratio. The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/ acs.langmuir.5b01397.



REFERENCES

AUTHOR INFORMATION

Corresponding Author

*Address: Department of Biotechnology and Food Engineering Technion, Haifa 3200003, Israel. Phone: 972-4-829-2143; fax: +972-4-829-3399; e-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS The study was partially supported by the RBNI (Russell Berrie Nanotechnology Institute) at the Technion, by the ISF (Israeli Science Foundation), and by NOFAR from Israel OCS (Office of the Chief Scientist) to Y.B. and D.D. We would like to thank TEVA pharmaceuticals (Israel) for supplying celecoxib. The 7191

DOI: 10.1021/acs.langmuir.5b01397 Langmuir 2015, 31, 7183−7192

Article

Langmuir (20) Morgen, M.; Bloom, C.; Beyerinck, R.; Bello, A.; Song, W.; Wilkinson, K.; Steenwyk, R.; Shamblin, S. Polymeric nanoparticles for increased oral bioavailability and rapid absorption using celecoxib as a model of a low-solubility, high-permeability drug. Pharm. Res. 2012, 29, 427−440. (21) Glatter, O.; Kratky, O. Small Angle X-ray Scattering; Academic Press: London, 1982; Vol. 102. (22) Harris, R. K.; Becker, E. D.; Cabral de Menezes; Sonia, M.; Goodfellow, R.; Granger, P. NMR nomenclature. Nuclear spin properties and conventions for chemical shifts (IUPAC recommendations 2001). Pure Appl. Chem. 2001, 73, 1795−1818. (23) Sluszny, A.; Silverstein, M. S.; Kababya, S.; Schmidt, A.; Narkis, M. Novel semi-IPN through vinyl silane polymerization and crosslinking within PVC films. J. Polym. Sci., Part A: Polym. Chem. 2001, 39, 8−22. (24) Danino, D. Cryo-TEM of soft molecular assemblies. Curr. Opin. Colloid Interface Sci. 2012, 17, 316−329. (25) Danino, D.; Bernheim-Groswasser, A.; Talmon, Y. Digital cryogenic transmission electron microscopy: An advanced tool for direct imaging of complex fluids. Colloids Surf. Physicochem. Eng. Aspects 2001, 183, 113−122. (26) Paulson, S. K.; Vaughn, M. B.; Jessen, S. M.; Lawal, Y.; Gresk, C. J.; Yan, B.; Maziasz, T. J.; Cook, C. S.; Karim, A. Pharmacokinetics of celecoxib after oral administration in dogs and humans: Effect of food and site of absorption. J. Pharmacol. Exp. Ther. 2001, 297, 638−645. (27) Tishmack, P. A.; Bugay, D. E.; Byrn, S. R. Solid-state nuclear magnetic resonance spectroscopypharmaceutical applications. J. Pharm. Sci. 2003, 92, 441−474. (28) Edwards, M.; Ratho, S. Synthesis of p-chlorocelecoxib from 4methylacetophenone and its spectral characterization. Indo Global J. Pharm. Sci. 2012, 2, 239. (29) Brunger, A. T. X-ray crystallography and NMR reveal complementary views of structure and dynamics. Nat. Struct. Biol. 1997, 4 (Suppl), 862−865. (30) Helliwell, J. R. NMR crystallography; Harris, R. K., Wasylishen, R. E., Duer, M. J., Eds.; Crystallography Reviews; Wiley: Chichester, U.K., 2012; Vol. 18, pp 241−245. (31) Perlstein, H.; Bavli, Y.; Turovsky, T.; Rubinstein, A.; Danino, D.; Stepensky, D.; Barenholz, Y. Beta-casein nanocarriers of celecoxib for improved oral bioavailability. Eur. J. Nanomed. 2014, 6, 217−226. (32) Andrews, A.; Atkinson, D.; Evans, M.; Finer, E.; Green, J.; Phillips, M.; Robertson, R. The conformation and aggregation of bovine β-casein AI Molecular aspects of thermal aggregation. Biopolymers 1979, 18, 1105−1121. (33) Glatter, O. The interpretation of real-space information from small-angle scattering experiments. J. Appl. Crystallogr. 1979, 12, 166− 175. (34) Sawyer, L.; Grubb, D.; Meyers, G. F. Polymer Microscopy; Springer Science & Business Media: New York, 2008. (35) Hallett, F. R. Particle size analysis by dynamic light scattering. Food Res. Int. 1994, 27, 195−198.

7192

DOI: 10.1021/acs.langmuir.5b01397 Langmuir 2015, 31, 7183−7192