Effect of Peptide and Guest Charge on the Structural, Mechanical

A. Markey , V. L. Workman , I. A. Bruce , T. J. Woolford , B. Derby , A. F. Miller , S. H. ... Hemar , Guang Mo , Alok K. Mitra , Jillian Cornish , Ma...
0 downloads 0 Views 2MB Size
Article pubs.acs.org/Langmuir

Effect of Peptide and Guest Charge on the Structural, Mechanical and Release Properties of β‑Sheet Forming Peptides D. Roberts,† C. Rochas,



A. Saiani,§ and A. F. Miller*,†



School of Chemical Engineering and Analytical Science & Manchester Institute of Biotechnology, The University of Manchester, 131 Princess Street, Manchester M1 7DN, United Kingdom ‡ CERMAV-CNRS, BP 53, 38041 Grenoble, France § School of Materials Science, The University of Manchester, Oxford Road, Manchester M13 9PL, United Kingdom S Supporting Information *

ABSTRACT: The effect of peptide charge on the selfassembly, gelation behavior, and model drug release profiles has been explored here for three octa-peptides, VEVKVEVK (VEK2), VKVKVEVK (VEK3), and VEVEVKVE (VEK1), that carry a net charge of 0, +2, and −2 at neutral pH, respectively. Transparent, self-supporting hydrogels were found to form above a critical concentration when the peptide charge modulus was >1 and this was independent of the sign of the charge. TEM, SAXS, and shear rheology revealed that there were no differences in hydrogel structure or mechanical properties when the peptides were at the same concentration and carried the same charge modulus. All peptides were found to form dense fibrillar networks formed by β-sheet rich single fibers where lateral aggregation of the fibers occurred and increased with decreasing charge modulus. Such behavior was found to correlate with an increase in hydrogel mechanical properties, demonstrating that fiber lateral aggregation is inextricably linked with the mechanical properties of these hydrogels. Two hydrophilic model drug molecules, namely napthol yellow (NY) and martius yellow (MY), were subsequently incorporated within the VEK1 and VEK3 hydrogels at pH 7 and although they did not effect the self-assembly of the peptide at a molecular level, they did effect the level of lateral fiber aggregation observed and, therefore, the mechanical properties of the hydrogels. The release of each molecule from the hydrogels was monitored over time and shown to be controlled by Fickian diffusion where the diffusion rate, D, was dependent on the ratio between the overall effective charges carried by the peptide, i.e., the fibrillar network, and the overall charges carried by the guest molecules, but independent from the hydrogel concentration and mechanical properties within the ranges investigated. This work highlights the possibility of controlling the rate of release of small drug molecules by manipulating the charges on the guest molecules as well as the charged state of the self-assembling peptide.



INTRODUCTION Self-assembling peptide hydrogels are currently attracting considerable attention as they have significant potential for use in a wide variety of applications including tissue engineering, 1−3 drug delivery, 4−7 biosensing, 8−10 electronics,11,12 and energy harvesting.13 A number of design rules are now beginning to emerge to control the self-assembly of peptides into fibers of defined thickness, which either branch or entangle to form 3D networks and hydrogels that have tunable mechanical properties.14 Moreover, peptide hydrogels can be designed to be responsive to a range of stimuli, including pH,15,16 salt concentration,17 light,18 temperature19,20 or the presence of enzymes.21−23 Such properties have led to several groups exploring the potential of peptide hydrogels to control the release of model drug and therapeutic molecules.24−33 The primary structure of peptides is easily programmable and this not only allows the structure and mechanical properties of the hydrogels to be manipulated, but also the specific peptide-drug molecule interactions. This programmability and versatility makes peptide-based hydrogels ideal candidates for use as controlled, topical, drug delivery vehicles. © 2012 American Chemical Society

It is well-known from the significant body of work existing on polymeric hydrogels that the size of the guest, or drug molecule, its degree of interaction with the polymer fibers, the quantity of water media, the average mesh size of the fibrillar network, and its shape are all key parameters that influence the rate of release of the guest molecule.5,34−38 The use and understanding of electrostatic interactions in these systems is of particular interest as a large number of therapeutic agents are indeed charged molecules. Here, we investigate the possibility of controlling the release rate of small guest molecules from peptide hydrogels by manipulating the nature and strength of the electrostatic interactions between the guest and the selfassembling peptide fibrillar network. This was done by varying the charged state of both the peptide building block and the guest molecule. To this end, three ionic-complementary octapeptides carrying net charges of 0, +2, and −2 at pH 7 were synthesized VEVKVEVK (VEK2), VKVKVEVK (VEK3), and Received: August 16, 2012 Revised: October 9, 2012 Published: October 22, 2012 16196

dx.doi.org/10.1021/la303328p | Langmuir 2012, 28, 16196−16206

Langmuir

Article

TOF mass spectroscopy to confirm the heating process did not degrade the peptides. Due to the presence of residual TFA, the pH of the peptide solutions after heating was ∼2.8. The pH of the samples was, therefore, adjusted to the desired values by adding the required quantity of 1 M NaOH. The total sample volume was then adjusted to 1 mL with water before leaving the sample to gel and equilibrate for 24 h at room temperature prior to characterization. Loaded samples were prepared using the same method, where the water was replaced by an aqueous solution of the guest molecule at the required concentration and the pH adjusted to 7 using a 1 M NaOH solution with the guest molecule at its required concentration. In all guest-loaded samples used in this work, the total number of effective charges carried by the peptide fibrillar network was always in excess in comparison to the total number of charges carried by the guest molecules. The composition and charge ratios for all samples studied at pH 7 are given in Table 1.

VEVEVKVE (VEK1) (V: valine; E: glutamic acid; K: lysine) and two small, model drug molecules napthol yellow (NY) and martius yellow (MY) carrying charges of −2 and −1 at pH 7 respectively, were encapsulated within the peptide hydrogels. The peptides were chosen as they are simple, short, and are known to self-assemble into β-sheet rich fibers which subsequently entangle to form hydrogels when above a critical concentration.39,40 The guest molecules were selected due to their similarity in structure and size, their spectroscopic characteristic signal that enabled their quantification using UV−vis spectroscopy and their agreement with the Lipinski rules for good drug candidates.41 The chemical structures of the peptides and guest molecules used in this work are given in Figure 1. Initially, the self-assembling properties of the pure

Table 1. Loaded Sample Compositions and Charge Ratios (Number of Effective Charges Carried by the Peptides/ Number of Charges Carried by the Guest Molecules) at pH 7a peptide concentration (mg mL−1)

guest loading (mM)

30 40 50

0.25 0.25 0.25

30 40 50

peptide to NY charge ratio (pH 7)

(a) constant drug loading 129 172 215 (b) constant charge ratio 7.18 7.5 5.74 7.5 4.31 7.5

peptide to MY charge ratio (pH 7) 258 344 430

Figure 1. Chemical structures at pH 7 of peptide and model drug molecules used.

(Peptides molar masses: MVEK1 = 930 g mol−1 and MVEK3 = 928 g mol−1).

peptides as well as the structure and properties of the resulting hydrogels, were investigated as a function of peptide concentration and media pH. Subsequently, any effect of incorporating the guest molecules on the structure and properties of the hydrogels at pH 7 was studied and identified. Finally, the effect of the nature and strength of the electrostatic interactions existing between the peptide fibrillar network and guest molecules on the release of the latter was explored. Hydrogel characterization was performed using a combination of Fourier transform infrared spectroscopy (FTIR), oscillatory rheology, transmission electron microscopy (TEM), and smallangle X-ray scattering (SAXS) while guest molecule release studies were undertaken using UV−vis spectroscopy.

Fourier Transform Infrared (FTIR) Spectroscopy. FTIR spectra were collected using a Nicolet 5700 spectrometer equipped with a multibounce attenuated total reflectance (ATR) zinc selenide crystal. Samples for FTIR were prepared as described above by replacing water with deuterated water. Each sample was scanned 256 times at a resolution of 4 wavenumbers. Scans were averaged to obtain a good signal-to-noise ratio and auto corrected for the ATR crystal. Data were analyzed using the OMNIC software provided with the instrument. Rheology. All rheological studies were undertaken using a stresscontrolled rheometer (Thermal Analysis AR-G2) equipped with 20 mm parallel plates. In each experiment, 1 mL of sample was loaded onto the stage and the upper plate lowered until a 0.25 mm gap was reached. Any excess material was then soaked away and the sample was left to equilibrate at 25 °C for several minutes before measurement. Frequency sweeps were subsequently undertaken between 0.1 and 10 rad s−1 using a strain of 1%, which is within the linear viscoelastic regime for all samples. All measurements were repeated at least 3 times to ensure reproducibility. Transmission Electron Microscopy (TEM). TEM micrographs were obtained using a Joel 1220 TEM operating at 100 keV and a Gatan Orius CCD camera at a specimen level increment of 2.9 Å pixel−1. Carbon-coated copper grids (no. 400, Agar Scientific) were glow-discharged for 30 s and then placed shiny side down on the surface of a 10 μL droplet of 20-fold diluted samples for ca. 10 s. Once adsorbed, the loaded grids were immediately placed on a 10 μL droplet of doubly distilled water for 10 s and blotted with Whatman 50 filter paper. Washed grids were then placed on a 10 μL droplet of freshly filtered 2% (w/v) uranyl acetate solution for 60 s for negative staining and then blotted again before being air-dried. Small Angle X-ray Scattering (SAXS). SAXS experiments were performed on beamline BM02 at the European Synchrotron Radiation Facility (ESRF) in Grenoble, France. The energy of the beam was 16.0



a

MATERIALS AND METHODS

Peptide Synthesis. All Fmoc-amino acids, activator (HCTU) and Wang resin were purchased from Novabiochem (Merck) and used as received. All other reagents and solvents were purchased from Sigma Aldrich and used without further purification. All peptides were synthesized using standard solid phase peptide synthesis (SPPS) protocols using a CEM Discover SPS microwave peptide synthesizer and standard Fmoc protection strategies. The peptides were purified by precipitating three times in cold ether before being lyophilized for three days. Peptides were characterized by HPLC and MALDI-TOF mass spectrometry and their purity estimated to be ≥90%.42 Samples Preparation. Samples were prepared by dissolving the desired quantity of peptide in 0.5 mL of HPLC grade water at 90 °C for 12 h. VEK1 was found to be less soluble than the other two peptides, and consequently was heated for 48 h. Small sample aliquots were then extracted and once again analyzed by HPLC and MALDI16197

dx.doi.org/10.1021/la303328p | Langmuir 2012, 28, 16196−16206

Langmuir

Article

Figure 2. (a) Charge modulus carried by each peptide and (b) critical gelation concentration (CGC) of the corresponding hydrogels as a function of medium pH:  VEK3, ----- VEK2, - - - - VEK1. (pKa Lysine (K) = 10.5; pKa Glutamic acid (E) = 4.25; pKa N−terminus = 9.69; pKa C−terminus = 2.34). keV, which corresponds to a wavelength of 7.75 × 10−2 nm. The scattered photons were collected onto a two-dimensional CCD detector with typical acquisition times of about 120 s. The sample− detector distance was fixed at 1.57 m resulting in a momentum transfer vector range of 0.1 < q (nm−1) < 2.5, with q = (4π/λ)sin(θ/2), where θ is the scattering angle and λ the wavelength. The samples were placed into cylindrical thin wall (0.2 mm) glass tubes with an outer diameter of 3 mm (Deutero GmbH, Germany). The 2D isotropic scattering data collected were corrected for the detector response, dark current, and sample transmission and were radially averaged to obtain one-dimensional scattering patterns. Intensity normalization was achieved using a lupolen secondary standard and calibration of the momentum transfer was done using silver behenate. Under these conditions, the coherent intensity scattered by the peptides is as follows:43−45

IA(q) =

1 [IN(q) − (1 − C p)IS(q) − Ib] K

saline solution (PBS) was prepared by dissolving NaCl (8 g), Na2HPO4 (1.38 g), KH2PO4 (0.19 g), and NaN3 (0.2 g) in distilled water (1 L). The pH was then adjusted to 7.4 using HCl if required.



RESULTS AND DISCUSSION Three octa-peptides, VEVKVEVK (VEK2), VKVKVEVK (VEK3), and VEVEVKVE (VEK1), with different overall charges were synthesized (Figure 1) to explore the effect of peptide charge on their self-assembly, on their subsequent gelation behavior and also the release kinetics of small guest molecules from their corresponding hydrogels. The overall charge modulus, |Z|, carried by each peptide at a set pH can be calculated, via:

(1)

|Z | =

where IN(q) is the normalized intensity scattered by the sample, IS(q) the normalized intensity scattered by the water, CP the peptide concentration in g cm−3, Ib the background scattering mainly due to the incoherent scattering of the peptides and K the contrast factor expressed as follows:43−45 K=

2 vp ⎞ 4.76 ⎛ − Z Z ⎜ ⎟ s p vs ⎠ m02 ⎝

Kp q4

+ Ib

⎞2 10 pH ⎟ ∑ Nj pH pK a j ⎟ 10 10 + ⎠ j (4)

where Ni/j are the numbers and pKai/j the pKa values of the basic (i - amine groups present on the N-terminus and lysine side chains) and acidic (j - carboxylic acid groups present on the on C-terminus and glutamic acid side chains) groups present on the peptide. The values calculated for |Z| for each peptide are presented in Figure 2a as a function of pH. At neutral pH, the three peptides VEK1, VEK2, and VEK3 carry charges of −2, 0, and +2 and their isoelectric points are at pH 4, 7, and 10.2 respectively. Samples with a range of concentrations were prepared at various pHs and their critical gelation concentration (CGC) determined using the tilting test tube method, i.e., a sample was classified as a liquid when the sample flowed freely and as a gel when the sample was self-supporting upon inversion of the vial. The CGC values obtained for each peptide are presented in Figure 2b as a function of sample pH. As can be seen from Figure 2, the CGC is directly dependent on the charge modulus carried by the peptide. As the samples approach their isoelectric points, the CGC drops below 5 mg mL−1 for all three peptides. Moving away from the isoelectric point leads to a significant increase in the CGC up to 35 mg mL−1 for VEK1 and VEK3 at pH 11 and 3, respectively. The stability of the hydrogels was also found to be dependent on the value of |Z|. When |Z| < 1 (gray area Figure 2a) samples became cloudy and although selfsupporting hydrogels were still obtained, over time they were found to phase separate. These results clearly show the key role played by the overall charge modulus in dictating the phase behavior of these peptides. Such behavior is no surprise as it is

(2)

where mp is the peptide molecular weight, Zp and Zs are the numbers of electrons in the peptide and the water molecules and vp and vs their molar volumes, respectively. The molar volumes of the peptide were estimated by adding the molar volume values reported by Jacrot and Zaccai for each amino acid in the sequence.46 The background scattering, Ib, was estimated using the Porod law which gives the scattered intensity of a two phase system at high q values:43−45

I(q) =

⎛ 10 pK ai ⎜∑ N − i ⎜ 10 pH + 10 pK ai ⎝ i

(3)

where KP is the Porod constant. Ib was estimated by fitting the last 10 data points of the scattering curves using a Porod representation (q4I(q) vs q4). Ultra Violet-Visible (UV−vis) Spectroscopy. Calibration curves (concentration versus absorption) were established for each guest molecule at their detection wavelength of 428 nm using an UV−vis spectrometer by placing the samples in HELLMA quartz cells of 5 mm path length. The rate of release of the guest molecules from the gels was then determined by placing 3 mL of PBS solution on top of the loaded samples. One milliliter of this layer was then collected every 30 min, placed in the UV−vis spectrometer and the concentration of guest molecules determined from the absorption measured at 428 nm using the calibration curves. After analysis, the sample was carefully placed back on top of the drug loaded hydrogel. Phosphate buffered 16198

dx.doi.org/10.1021/la303328p | Langmuir 2012, 28, 16196−16206

Langmuir

Article

Figure 3. (a) Storage (G′, closed symbols) and loss (G″, open symbols) moduli for hydrogels prepared at 40 mg mL−1 and |Z| = 2 (●, ○: VEK3 at Z = +2, pH 7; ⧫, ◊: VEK2 at Z = +2, pH 3; ■, □: VEK2 Z = −2, pH 3; ▰, ▱: VEK1 Z = −2, pH 7) as a function of frequency at 1% strain. (b) Storage modulus, G′, at 5 rad s−1 as a function of peptide concentration for VEK2 at Z = −2, pH 10.5 (solid fill); VEK1 at Z = −2, pH 7 (horizontal stripes); VEK2 at Z = +2, pH 3 (hollow fill) and VEK3 at Z = +2, pH 7 (vertical stripes).

Figure 4. (a) FTIR spectra of the amide I band region obtained for hydrogels prepared at 40 mg mL−1 and |Z| = 2 (··−··−: VEK3 at Z = +2, pH 7; − −: VEK1 at Z = −2, pH 7; ·····: VEK2 at Z = +2, pH 3; : VEK2 at Z = −2, pH 10.5). (b) TEM micrograph obtained for a 20-fold diluted VEK2 hydrogel prepared at 40 mg mL−1 and Z = +2, pH 3.

known that charge plays a key role in the stabilization of peptide fibres within solutions and gels.16,17,47−50 As discussed by Caplan and co-workers,50 the stability of these hydrogels depends on a balance between repulsive forces that prevent fiber aggregation, and attractive forces that drive the assembly of the fibers into percolated networks. When |Z| < 1, the electrostatic repulsive forces are no longer strong enough to overcome the attractive hydrophobic forces leading to fiber aggregation. As a result, large (>1 μm) aggregates form that scatter light, resulting in the sample becoming cloudy and eventually phase separating. When |Z| increases, the electrostatic repulsion between fibers also increases resulting in a decrease in the level of fiber aggregation observed, and therefore a decrease in the size of the aggregates formed, i.e., samples become transparent for |Z| > 1. It is evident from Figure 2 that the increase in the CGC is directly linked to the increase in |Z|. This suggests that fiber−fiber aggregation, which is controlled by the overall effective charge carried by the peptides, plays a key role in the formation of a percolated network and, therefore, in the formation of a self-supporting hydrogel in these systems. Eventually, for high |Z| values, the strength of electrostatic repulsion between the peptides at a molecular level will prevent their self-assembly into fibers and,

therefore, prevent gelation of the systems. The symmetry around pH 7 of the two graphs presented in Figure 2 also indicates that the self-assembly and gelation behavior of our peptides does not depend on the sign of the charge carried by the peptide, but only on its overall modulus. This was further confirmed by examining the phase behavior of each peptide as a function of temperature. The CGCs of each peptide were measured for a range of pHs from 20 to 80 °C and the same phase behavior was observed for all the peptides when the overall charge moduli were the same. A selection of these diagrams is presented Figures S1 and S2 (Supporting Information). The mechanical properties of the hydrogels at constant charge modulus |Z| = 2 (i.e., at pH 3 and 10.5 for VEK2 and at pH 7 for VEK3 and VEK1) were assessed using shear rheology. Typical frequency sweeps for hydrogels prepared at 40 mg mL−1 are given in Figure 3a. For all four hydrogels, the elastic modulus, G′, was approximately an order of magnitude higher than the loss modulus, G″, and showed little dependency on frequency which is typical of stable hydrogels with long relaxation times where the interactions/junctions between fibers are relatively permanent.14,51 The moduli obtained at 5 rad s−1 for hydrogels prepared at 20, 30, and 40 mg mL−1 are 16199

dx.doi.org/10.1021/la303328p | Langmuir 2012, 28, 16196−16206

Langmuir

Article

Figure 5. (a) log[IN(q)] vs log q and (b) ln[qIA(q)] vs q2 plots of the intensity scattered at room temperature by VEK3 hydrogels prepared at different concentrations and Z = +2, pH 7.

sheets.56,57 Unfortunately, this absorption band is masked by the TFA peak and therefore cannot be clearly identified here. The presence of an entangled/aggregated network of fibres in these systems was confirmed by TEM. A typical TEM micrograph obtained for a 20-fold diluted VEK2 sample at 40 mg mL−1 and Z = +2 (i.e., pH 3) is shown in Figure 4b. Both single fibers and bundles of fibers are observed, where the latter is a result of lateral aggregation of single fibers. These form a dense fibrillar network, which is typical of networks formed via noncovalent, physical interactions. This network topology clearly shows that fiber−fiber interactions and lateral aggregation play a key role in these systems. Such interactions and aggregations are typically dominated by electrostatic (repulsive) and hydrophobic (attractive) interactions and, therefore, keeping the balance between these interactions constant, results in gels with similar morphologies. For all three peptides, an average diameter of ∼2.6 ± 0.4 nm was obtained for the single fibers (smallest fibers observed on the TEM micrographs). This is in good agreement with the theoretical length of the peptide in a fully extended β-sheet conformation (2.8 nm). SAXS was used to confirm the network topology observed by TEM in the gel state. The SAXS patterns obtained for the VEK3 samples at 20, 30, and 40 mg mL−1 and Z = +2 (i.e., pH 7) are presented in Figure 5a. As can be seen, at low q, a q−1 behavior typical of rod-like structures is observed at all three concentrations, confirming the presence of fibres in our systems. For infinitely long rod-like structures in the q range investigated the scattered intensity can be written as follows:43,51,58

given in Figure 3b. As can be seen, the shear moduli of the hydrogels are the same at each concentration for all three peptides, but are found, as expected, to increase with increasing concentration: from 20 to 800 Pa when the concentration of peptide is increased from 20 to 40 mg mL−1. These values are typical of this family of peptides.52 More interestingly, these results confirm that the properties of these hydrogels do not depend on the peptide sequence as long as the overall charge modulus carried by the peptides is the same. The overall hydrophobic character of the peptide is not expected to change significantly when replacing glutamic acid residues by lysine residues, therefore, the hydrophobic interactions are expected to be similar for all three peptides. Similarly, when the charge modulus carried by the peptides is kept constant, the electrostatic interactions are also kept constant. As a result, the balance between these two dominating interactions is similar for all three peptides, resulting in hydrogels with similar properties. It is well-known that this family of ioniccomplementary octa-peptides form β-sheet rich fibers.52−55 FTIR spectra of the amide I band region (ca., 1750 − 1600 cm−1) for hydrogels prepared at 40 mg mL−1 and |Z| = 2 are presented in Figure 4a. A strong absorption peak is observed in the spectra centered around 1618 cm−1, which is indicative of the peptides adopting a β-sheet secondary structure.56,57 The intensity of this peak is similar for all samples suggesting that a similar quantity of βsheet structure is present in all of them. The spectrum for the VEK1 sample shows some absorption around ∼1650 cm−1. This region is usually associated with the presence of random coil or α helix structures.56,57 This suggests that for VEK1, a low level of unstructured peptide is present but it does not seem to affect it is overall self-assembling and gelation behavior. This observation correlates with the lower solubility observed for this peptide (see Materials and Method). A strong absorption peak is observed at ∼1680 cm−1. This peak is due to residual TFA from the synthesis that binds strongly to the amine groups present along the peptides (N-terminus and lysine residue side chains). The intensity of this peak, consequently, was found to increase with the number of lysine residues present on the different peptide sequences; from VEK1 to VEK2 and VEK3 where the possible binding sites for TFA are 2, 3, and 4, respectively. It should also be noted that a small shoulder seems to be present around ∼1695 cm−1 which is usually associated with the formation of antiparallel β-

q2IA(q) = πqCpμL f (qR σ ) + Cst

(5) −1

where μL is the mass per unit length of the rod in g mol nm−1, Cp the peptide concentration in g cm−3 and f(qRσ) represents the cross-section scattering, Rσ being the cross-section radius of gyration of the rod. Cst is a constant term taking into account interscattering effects. For qRσ < 1 Equation 5 reduces to the following:43,51,58 ⎛ q 2R 2 ⎞ σ ⎟ qIA(q) = πCpμL exp⎜ ⎝ 2 ⎠

(6)

If the scattering observed is of the form described by eq 6, then at low q a linear curve should be obtained in a ln[qIA(q)] 16200

dx.doi.org/10.1021/la303328p | Langmuir 2012, 28, 16196−16206

Langmuir

Article

Figure 6. (a) Comparison of FTIR spectra obtained for unloaded (:VEK3; − − −: VEK1) and loaded (------: VEK3; ········: VEK1; loading: 250 μM of NY) hydrogels prepared at 40 mg mL−1 and |Z| = 2, pH 7 in D2O. (b) Elastic moduli, G′, obtained for unloaded (solid fills) and loaded with NY (vertical stripes; loading: 250 μM) and MY (horizontal stripes; loading: 250 μM) VEK3 and VEK1 hydrogels prepared at varying concentrations and |Z| = 2, pH 7.

polydisperse population of fibers (i.e., with varying radii) is confirmed by the appearance of a roll-over at 30 and 40 mg mL−1 in the ln[qIA(q)] vs q2 curves (Figure 5b). The position of this roll-over is linked to the fiber aspect ratio (length/ diameter).59 Although the single fibers can be considered infinitely long, the thicker fibers resulting from the lateral aggregation of single fibers tend to be shorter as can be seen from the TEM micrograph in Figure 4b. Therefore, the effect of the fiber aspect ratio starts to be observed in the sample scattering patterns when their volume fraction increases. Overall, these SAXS results confirm the network topology observed by TEM. To investigate the effect of electrostatic interactions on the rate of release of small, guest molecules from the peptide hydrogels, two model drug molecules with similar chemical structures and carrying different charges at pH 7 were used: napthol yellow (NY) (Z = −2 at pH 7) and martius yellow (MY) (Z = −1 at pH 7) (Figure 1). These molecules were encapsulated at 250 μM within the two charged hydrogels VEK1 (Z = −2 at pH 7) and VEK3 (Z = +2 at pH 7) and their release studied. As discussed earlier, VEK2 produces unstable hydrogels at pH 7 and was therefore not considered for these release studies. Before undertaking the release studies the properties of the loaded hydrogels were explored to investigate whether the presences of the guest molecules had any effect on gel structure and mechanical properties. The FTIR spectra for the VEK1 and

vs q2 representation. This is indeed the case, as can be seen from Figure 5b, which again confirms the presence of fibres in our systems. The cross-section radius of gyration, Rσ, of these rods can be estimated from the slope of the linear section. If we assume that the fibers can be modeled by a plain cylinder then Rσ is related to the diameter of the fiber, d, through: Rσ =

d2 8

(7)

For the 20 mg mL−1 sample, an average fiber diameter of 2.80 ± 0.05 nm was obtained for the fibers. This is in very good agreement with the single fiber size obtained from the TEM micrographs. This result suggests that at 20 mg mL−1 the scattering is dominated by single fibres pointing toward a low level of fiber lateral aggregation (i.e., fiber thickening). When the hydrogel concentration is increased the average diameter of the fibers is found to increase to 2.98 ± 0.05 nm and 3.26 ± 0.05 nm for the 30 and 40 mg mL−1 samples, respectively. Keeping in mind that, in the q range investigated, the scattering observed can be assumed to be the average scattering resulting from all the fibers present in the sample (i.e., thin and thick) weighted by the volume fractions of each size population (i.e., interfibers scattering effects are neglected), the increase in average fiber size observed suggest that lateral fiber aggregation (i.e., fiber thickening) is occurring and is more prevalent at higher concentrations, as one would expect. The presence of a 16201

dx.doi.org/10.1021/la303328p | Langmuir 2012, 28, 16196−16206

Langmuir

Article

Figure 7. (a) TEM micrographs obtained for 20-fold diluted loaded VEK3 hydrogel prepared at 40 mg mL−1 and |Z| = 2, pH 7 (Loading: 1 mM of NY). (b) ln[qIA(q)] vs q2 plots of the intensity scattered at room temperature by the loaded and unloaded VEK3 and VEK1 hydrogels prepared at 30 mg mL−1 and |Z| = 2, pH 7 (Loading: 250 μM of NY).

the unloaded samples (Figure 4b). These observations were confirmed by SAXS in the gel state. The SAXS patterns obtained for the loaded (at 250 μM) and unloaded samples at 30 mg mL −1 are presented in the ln[qI A (q)] vs q 2 representation in Figure 7c. It is evident that linear curves are obtained, thus confirming the presence of fibres also in the loaded samples. The average fiber diameter was found to increase for the VEK3 sample from 1.5 ± 0.5 to 1.9 ± 0.5 nm in the presence of NY while for VEK1 sample it was found to decrease slightly from 1.4 ± 0.5 to 1.2 ± 0.5 nm. For the VEK3 system, the peptide and guest molecules carry charges of opposite sign, therefore, an attractive interaction is expected between the peptide fibers and guest molecules resulting in the shielding of the charge carried by the fibres. This is postulated to result in the promotion of lateral aggregation of the fibers through hydrophobic interactions and consequently the formation of thicker fibers. However, for the VEK1 system, the peptide and guest molecules carry charges of the same sign and repulsive interactions are expected between the peptide fibers and the guest molecules which will reduce the drive for fiber aggregation. The changes in the network topology observed for the VEK1 and VEK3 systems in the presence of the guest molecules tallies well with the change in mechanical properties. The increased propensity for the β-sheet fibers to aggregate laterally and form thicker fibers is indeed expected to result in a strengthening of the hydrogels. This is mainly due to thicker fibers being intrinsically more rigid and stiffer than single fibers and will also form strong and stable physical cross-links increasing the overall rigidity and strength of the fibrillar network and the hydrogel37,38 These results clearly show that fiber lateral aggregation and it's extent play a

VEK3 loaded and nonloaded hydrogels are compared in Figure 6a. No significant differences were observed, indicating that the presence of the guest molecules did not affect, at a molecular level, the self-assembly of the peptides into β-sheet rich fibers. Both guest molecules have highly polar side chains around the electron rich naphthalene core, hence they are hydrophilic in character. Since β-sheet formation is driven by hydrophobic interactions for these peptides, it is unlikely that the guest molecules would interfere with the formation of the fibers. The presence of the guest molecules was found, however, to alter the mechanical properties of the hydrogels (Figure 6b). The shear moduli of the VEK3 hydrogels were found to increase at all concentrations when the guest molecules were present, going from ∼800 to 3000 Pa for the 40 mg mL−1 sample in the presence of MY. Contrastingly for VEK1 (Z = −2 at pH 7) hydrogels the shear moduli were found to decrease slightly going from ∼800 to 600 Pa for the 40 mg mL−1 sample in the presence of MY. These changes in mechanical properties suggest that the presence of the guest molecules, affects the structure of the networks formed by the β-sheet fibers in different ways depending on the sign of the charge carried by the peptides, despite not affecting the self-assembly of the peptides at a molecular level. This was confirmed by TEM and SAXS. As can be seen from Figure 7a, TEM micrographs for a sample loaded with 1 mM of NY (loading was increased to exacerbate the effect of guest molecules on samples’ morphologies) suggest that for the VEK3 peptide the presence of the guest molecule promotes the formation of thick fibers, while for the VEK1 peptide the presence of the guest molecule seems to result in a smaller fraction of thick fibres being formed when compared to 16202

dx.doi.org/10.1021/la303328p | Langmuir 2012, 28, 16196−16206

Langmuir

Article

Figure 8. (a) Fraction of guest molecule released by loaded (loading: 250 μM; NY: closed symbols; MY: open symbols) VEK3 and VEK1 hydrogels prepared at 30 (■, □), 40 (●, ○) and 50 mg mL−1 (▰, ▱) and |Z| = 2, pH 7 vs square root of release time. For clarity purposes only data up to 4 h are presented. (b) Diffusion coefficients obtained for NY and MY from VEK3 (NY: ■ solid fill; MY: vertical stripes) and VEK1 (NY: □ no fill; MY: horizontal stripes) hydrogels prepared at different concentrations and |Z| = 2, pH 7. Diffusion coefficients were measured over 8 h.

time t = 0. Such behavior is indicative of the guest molecules being released via Fickian diffusion.60,61 The diffusion coefficients, D, were therefore calculated, from the slopes of the linear fits, using the nonsteady state diffusion equation:62

key role in determining the mechanical properties of these hydrogels. To monitor the kinetics of guest release from the peptide hydrogels, 3 mL of phosphate buffer solution (PBS) was placed on top of the loaded hydrogels. The rate of release, and hence the rate of diffusion of the guest molecules into the PBS supernatant was monitored over a period of 8 h using UV spectroscopy. No visible changes occurred to the surface or size of the gels, i.e., no surface roughening and no gel swelling were observed, during this time. However, after 7 days the VEK1 hydrogel samples were found to dissolve completely, while the VEK3 hydrogels remained stable and self-supporting for up to 5 weeks. Such behavior reflects differences in material properties. As discussed earlier the presence of the guest molecules with the VEK3 peptide tends to increase the strength and stability of the hydrogels by promoting fiber−fiber aggregation, while the presence of the guest molecules in the VEK1 system tends to weaken the hydrogels by reducing the drive for fiber lateral aggregation accelerating its dissolution. It should be noted that no significant degradation of the peptide through hydrolysis was observed under these conditions over 1 week. It is thought that the formation of β-sheet structures protects the peptide from hydrolytic degradation by preventing water molecules to access easily the amino bonds. Indeed β-sheet fibers will be relatively hydrophobic. It is clear from Figure 8a that linear relationships exist between Mt/M∞ and t1/2 for all the hydrogels where Mt is the number of molecules released at time t and M∞ is the number of guest molecules present in the hydrogels at

⎛ Dt ⎞1/2 Mt = 4⎜ 2 ⎟ ⎝ πλ ⎠ M∞

(8)

where λ is the gel thickness. The diffusion coefficients obtained (Figure 8b) are of the same order of magnitude as those obtained by other groups for similar systems11,12 Larger diffusion coefficients are obtained, as expected when repulsive electrostatic interactions are present between the fibres and the guest molecules, i.e., for VEK1 hydrogels. When attractive interactions are present, i.e., for VEK3 hydrogels, a marked reduction in D is observed. This is in particular the case when the hydrogels are loaded with MY which carries a −1 charge at pH 7. When replacing MY (Z = −1) by NY (Z = −2) an increase in D is observed as expected for VEK1 hydrogels as the overall repulsion forces between fibers and guest molecules is increased. In the case of VEK3 hydrogels, one would have expected D to decrease when replacing MY by NY as the attractive forces between fibres and guest molecules are increased. This is not the case as shown in Figure 8b and D is actually found to significantly increase. This highlights the importance of understanding guest−guest interactions. By increasing the charge of the guest molecules, the repulsive forces between guest molecules is also increased. This will limit 16203

dx.doi.org/10.1021/la303328p | Langmuir 2012, 28, 16196−16206

Langmuir

Article

As can be seen, the same D values are obtained at all concentrations, independently from the hydrogel mechanical properties, when the charge ratio is kept constant confirming the charge ratio is the key factor controlling the release rate of the guest molecule from the hydrogels. The use of a significantly higher guest molecule loading (Table 1) resulted in a 2 orders of magnitude increase in D for the VEK1 peptide (e.g., from 2.3 × 10−10 to 373 × 10−10 m2 s−1 for the 40 mg mL−1 sample) while for the VEK3 peptide a much more modest increase was observed (e.g., from 1.7 × 10−10 to 10 × 10−10 m2 s−1 for the 40 mg mL−1 sample). These results show that the release rates of drugs can be manipulated by controlling not only the charged state of the drug, but also by controlling the number and type of charged amino acid groups present on the self-assembling peptide molecules.

how close guest molecules can get and will, therefore, limit the number of guest molecules than can interact with the fibrillar network and/or destabilize the fiber-guest electrostatic molecular pairing. As a result, this strong guest−guest repulsion reduces the ability of the positively charged fiber network to trap/retain guest molecules through electrostatic interactions resulting in a larger diffusion coefficient. For all peptide−guest molecule combinations studied here, an inverse relationship between diffusion coefficient and peptide concentration is found, i.e., the rate of release becomes lower when the concentration of peptide is increased. In the experiments performed to this point, the loading of guest molecule was kept constant which means that when the peptide concentration increases the ratio between the overall effective charges carried by the peptides and overall charges carried by the guest molecules increases (Table 1). The increase in peptide concentration is also expected to result in an overall increase in fiber volume fraction and therefore in an increase of the fiber network density and a reduction in its overall mesh size. The decrease of D with increasing concentration could therefore be linked either to the change in overall electrostatic interactions in the gel due to the change in charge ratio, or to the physical trapping of the guest molecules in the fibrillar network. The latter explanation seems unlikely. SAXS scattering patterns obtained for these samples (Figure 7b) do not present any structural peak58 suggesting that the mesh size of the fibrillar network is larger than the size window explored in the experiments here, i.e., larger than 50 nm for all the samples. Considering the size of the guest molecules (∼ 1 nm), it is unlikely that in the concentration range investigated the change in network mesh size, i.e., porosity, results in the physical entrapment of the guest molecules. To demonstrate that it is the charge ratio that actually controls D here, new release experiments were undertaken using NY where the ratio between the number of effective charges carried by the peptides, i.e.: the fibrillar network, and number of charges carried by the guest molecules was kept constant at ∼7.5:1 (Table 1). A higher guest molecule loading was used to exacerbate the effect of peptide effective charge on guest molecule release. The diffusion coefficients obtained are presented in Figure 9.



CONCLUSIONS

We have investigated the effect of peptide charge on the selfassembly and gelation behavior of three octa-peptides: VEVKVEVK (VEK2), VKVKVEVK (VEK3), and VEVEVKVE (VEK1). The CGC of each peptide correlated with the charge modulus carried by the peptide and was found to be independent of the sign of the charge. Hydrogels formed were transparent and stable for |Z| > 1 and became cloudy and ultimately phase separated for |Z| < 1. No differences in hydrogel structure or mechanical properties were found when the peptides were at the same concentration and carried the same charge modulus. The shear modulus did increase, however, when the peptide concentration increased. These peptides were shown to form dense fibrillar networks formed by β-sheet rich single fibres and fiber lateral aggregation was controlled by |Z|. The increase in fiber lateral aggregation with decreasing |Z| was found to correlate with the increase in hydrogel mechanical properties, showing that fiber lateral aggregation pays a key role in controlling the mechanical properties of these hydrogels. The release of two hydrophilic model drug molecules, namely napthol yellow (NY) and martius yellow (MY), from VEK1 and VEK3 at pH 7 was subsequently explored. The incorporation of the guest molecules did not affect the selfassembly of the peptide at a molecular level but did affect the level of lateral fiber aggregation observed and, therefore, the mechanical properties of the hydrogels. The release of each of the model compounds was monitored over time and shown to be controlled by Fickian diffusion. The guest molecule diffusion rate, D, was dependent on the ratio between the overall effective charges carried by the peptide, i.e., the fibrillar network, and the overall charges carried by the guest molecules but independent from the hydrogel concetration and mechnical properties in the concetration and guest loading range investigated. This work shows that the rate of release of small drug molecules can be manipulated, not only by changing the charges on the guest molecules, but also by manipulating the charged state of the self-assembling peptide molecule and through it the charge state of the fribrillar network. Manipulation of these interactions will allow fine control over the diffusion rates which could have widespread implications and applications within biomedical sciences.

Figure 9. Elastic modulus (G′) (■) and diffusion coefficients (diagonal stripes) for loaded VEK3 and VEK1 hydrogels prepared at 30, 40, and 50 mg mL−1 and |Z| = 2, pH 7. Each sample had a constant peptide effective charge to guest molecule charge of ∼7.5:1 (Table 1). 16204

dx.doi.org/10.1021/la303328p | Langmuir 2012, 28, 16196−16206

Langmuir



Article

(27) Branco, M. C.; Pochan, D. J.; Wagner, N. J.; Schneider, J. P. Biomaterials 2010, 31 (36), 9527−9534. (28) Nagai, Y.; Unsworth, L. D.; Koutsopoulos, S.; Zhang, S. G. J. Control. Release 2006, 115 (1), 18−25. (29) Sutton, S.; Campbell, N. L.; Cooper, A. I.; Kirkland, M.; Frith, W. J.; Adams, D. J. Langmuir 2009, 25 (17), 10285−10291. (30) Matson, J. B.; Newcomb, C. J.; Bitton, R.; Stupp, S. I. Soft Matter 2012, 8 (13), 3586−3595. (31) Feng, Y.; Lee, M.; Taraban, M.; Yu, Y. B. Chem. Commun. 2011, 47 (37), 10455−10457. (32) Liang, G. L.; Yang, Z. M.; Zhang, R. J.; Li, L. H.; Fan, Y. J.; Kuang, Y.; Gao, Y.; Wang, T.; Lu, W. W.; Xu, B. Langmuir 2009, 25 (15), 8419−8422. (33) Kumada, Y.; Hammond, N. A.; Zhang, S. G. Soft Matter 2010, 6 (20), 5073−5079. (34) Hoare, T. R.; Kohane, D. S. Polymer 2008, 49 (8), 1993−2007. (35) Hamidi, M.; Azadi, A.; Rafiei, P. Adv. Drug Delivery Rev. 2008, 60 (15), 1638−1649. (36) Colombo, P.; Sonvico, F.; Colombo, G.; Bettini, R. Pharm. Res. 2009, 26 (3), 601−611. (37) McCoy, C. P.; Brady, C.; Cowley, J. F.; McGlinchey, S. M.; McGoldrick, N.; Kinnear, D. J.; Andrews, G. P.; Jones, D. S. Exp. Opin. Drug Deliv. 2010, 7 (5), 605−616. (38) Wolinsky, J. B.; Colson, Y. L.; Grinstaff, M. W. J. Control. Release 2012, 159 (1), 14−26. (39) Bowerman, C. J.; Liyanage, W.; Federation, A. J.; Nilsson, B. L. Biomacromolecules 2011, 12 (7), 2735−2745. (40) Bowerman, C. J.; Ryan, D. M.; Nissan, D. A.; Nilsson, B. L. Mol. Biosyst. 2009, 5 (9), 1058−1069. (41) Lipinski, C. A.; Lombardo, F.; Dominy, B. W.; Feeney, P. J. Adv. Drug Delivery Rev. 1997, 23 (1−3), 3−25. (42) Chan, W.; White, P. Fmoc Solid Phase Peptide Synthesis: A Practical Approach; Oxford University Press: New York, 2000. (43) Higgins, J. S.; Benoit, H. C. Polymer and Neutron Scattering; Clarendon Press: Oxford, 1994. (44) Roe, R.-J. Methods of X-Ray and Neutron Scattering in Polymer Science; Oxford University Press: New York, 2000. (45) Guinier, A.; Fournet, G. Small-Angle Scattering of X-rays; John Wiley & Sons, Inc.: New-York, 1955. (46) Jacrot, B.; Zaccai, G. Biopolymers 1981, 20 (11), 2413−2426. (47) Yokoi, H.; Kinoshita, T. J. Nanomater. 2012. (48) Carrick, L. M.; Aggeli, A.; Boden, N.; Fisher, J.; Ingham, E.; Waigh, T. A. Tetrahedron 2007, 63 (31), 7457−7467. (49) Aggeli, A.; Bell, M.; Carrick, L. M.; Fishwick, C. W. G.; Harding, R.; Mawer, P. J.; Radford, S. E.; Strong, A. E.; Boden, N. J. Am. Chem. Soc. 2003, 125 (32), 9619−9628. (50) Caplan, M. R.; Moore, P. N.; Zhang, S. G.; Kamm, R. D.; Lauffenburger, D. A. Biomacromolecules 2000, 1 (4), 627−631. (51) Guenet, J.-M. Thermoreversible Gelation of Polymers and Biopolymers; Academic Press: London, 1992. (52) Caplan, M. R.; Schwartzfarb, E. M.; Zhang, S. G.; Kamm, R. D.; Lauffenburger, D. A. J. Biomater. Sci.-Polym. Ed. 2002, 13 (3), 225− 236. (53) Saiani, A.; Mohammed, A.; Frielinghaus, H.; Collins, R.; Hodson, N.; Kielty, C. M.; Sherratt, M. J.; Miller, A. F. Soft Matter 2009, 5 (1), 193−202. (54) Caplan, M. R.; Lauffenburger, D. A. Ind. Eng. Chem. Res. 2002, 41 (3), 403−412. (55) Caplan, M. R.; Schwartzfarb, E. M.; Zhang, S. G.; Kamm, R. D.; Lauffenburger, D. A. Biomaterials 2002, 23 (1), 219−227. (56) Barth, A. Prog. Biophys. Mol. Biol. 2000, 74 (3−5), 141−173. (57) Barth, A.; Zscherp, C. Q. Rev. Biophys. 2002, 35 (4), 369−430. (58) Guilbaud, J. B.; Saiani, A. Chem. Soc. Rev. 2011, 40 (3), 1200− 1210. (59) Ramachandran, S.; Trewhella, J.; Tseng, Y.; Yu, Y. B. Chem. Mater. 2006, 18 (26), 6157−6162. (60) Ritger, P. L.; Peppas, N. A. J. Control. Release 1987, 5 (1), 23− 36.

ASSOCIATED CONTENT

S Supporting Information *

Temperature-concentration phase diagrams for the compounds investigated herein (Figures S1 and S2). This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*Tel: +44 (0)1613065781; E-mail: [email protected]. Notes

The authors declare no competing financial interest.



REFERENCES

(1) Jonker, A. M.; Lowik, D.; van Hest, J. C. M. Chem. Mater. 2012, 24 (5), 759−773. (2) Collier, J. H.; Rudra, J. S.; Gasiorowski, J. Z.; Jung, J. P. Chem. Soc. Rev. 2010, 39 (9), 3413−3424. (3) Loo, Y.; Zhang, S. G.; Hauser, C. A. E. Biotechnol. Adv. 2012, 30 (3), 593−603. (4) Guvendiren, M.; Lu, H. D.; Burdick, J. A. Soft Matter 2012, 8 (2), 260−272. (5) Ladewig, K. Exp. Opin. Drug Deliv. 2011, 8 (9), 1175−1188. (6) Overstreet, D. J.; Dutta, D.; Stabenfeldt, S. E.; Vernon, B. L. J. Polym. Sci., Part B: Polym. Phys. 2012, 50 (13), 881−903. (7) Ramachandran, S.; Yu, Y. B. Biodrugs 2006, 20 (5), 263−269. (8) Lowik, D.; Leunissen, E. H. P.; van den Heuvel, M.; Hansen, M. B.; van Hest, J. C. M. Chem. Soc. Rev. 2010, 39 (9), 3394−3412. (9) Ulijn, R. V.; Smith, A. M. Chem. Soc. Rev. 2008, 37 (4), 664−675. (10) Williams, R. J.; Mart, R. J.; Ulijn, R. V. Biopolymers 2010, 94 (1), 107−117. (11) Ashkenasy, N.; Horne, W. S.; Ghadiri, M. R. Small 2006, 2 (1), 99−102. (12) Xu, H. X.; Das, A. K.; Horie, M.; Shaik, M. S.; Smith, A. M.; Luo, Y.; Lu, X. F.; Collins, R.; Liem, S. Y.; Song, A. M.; Popelier, P. L. A.; Turner, M. L.; Xiao, P.; Kinloch, I. A.; Ulijn, R. V. Nanoscale 2010, 2 (6), 960−966. (13) Kameta, N.; Ishikawa, K.; Masuda, M.; Asakawa, M.; Shimizu, T. Chem. Mater. 2012, 24 (1), 209−214. (14) Yan, C. Q.; Pochan, D. J. Chem. Soc. Rev. 2010, 39 (9), 3528− 3540. (15) Fletcher, N. L.; Lockett, C. V.; Dexter, A. F. Soft Matter 2011, 7 (21), 10210−10218. (16) Tang, C.; Ulijn, R. V.; Saiani, A. Langmuir 2011, 27 (23), 14438−14449. (17) Ozbas, B.; Kretsinger, J.; Rajagopal, K.; Schneider, J. P.; Pochan, D. J. Macromolecules 2004, 37 (19), 7331−7337. (18) Jana, P.; Maity, S.; Maity, S. K.; Ghorai, P. K.; Haldar, D. Soft Matter 2012, 8 (20), 5621−5628. (19) Huang, Z.; Lee, H.; Lee, E.; Kang, S. K.; Nam, J. M.; Lee, M. Nat. Commun. 2011, 2. (20) Maslovskis, A.; Tirelli, N.; Saiani, A.; Miller, A. F. Soft Matter 2011, 7 (13), 6025−6033. (21) Guilbaud, J. B.; Vey, E.; Boothroyd, S.; Smith, A. M.; Ulijn, R. V.; Saiani, A.; Miller, A. F. Langmuir 2010, 26 (13), 11297−11303. (22) Bremmer, S. C.; Chen, J.; McNeil, A. J.; Soellner, M. B. Chem. Commun. 2012, 48 (44), 5482−5484. (23) Hughes, M.; Frederix, P.; Raeburn, J.; Birchall, L. S.; Sadownik, J.; Coomer, F. C.; Lin, I. H.; Cussen, E. J.; Hunt, N. T.; Tuttle, T.; Webb, S. J.; Adams, D. J.; Ulijn, R. V. Soft Matter 2012, 8 (20), 5595− 5602. (24) Zhao, Y.; Tanaka, M.; Kinoshita, T.; Higuchi, M.; Tan, T. W. J. Control. Release 2010, 147 (3), 392−399. (25) Koutsopoulos, S.; Unsworth, L. D.; Nagaia, Y.; Zhang, S. G. Proc. Natl. Acad. Sci. U.S.A. 2009, 106 (12), 4623−4628. (26) Branco, M. C.; Pochan, D. J.; Wagner, N. J.; Schneider, J. P. Biomaterials 2009, 30 (7), 1339−1347. 16205

dx.doi.org/10.1021/la303328p | Langmuir 2012, 28, 16196−16206

Langmuir

Article

(61) Zhao, Y.; Tanaka, M.; Kinoshita, T.; Higuchi, M.; Tan, T. J. Control. Release 2010, 147 (3), 392−399. (62) Baker, R. W.; Lonsdale, H. K. Controlled Release of Biologically Active Agents; Plenum: New York, 1974.

16206

dx.doi.org/10.1021/la303328p | Langmuir 2012, 28, 16196−16206