J. Phys. Chem. B 2005, 109, 6257-6261
6257
Simultaneous Release of Hydrophobic and Cationic Solutes from Thin-Film “Plum-Pudding” Gels: A Multifunctional Platform for Surface Drug Delivery? Iseult Lynch,*,†,‡ Paolo de Gregorio,† and K. A. Dawson† Irish Centre for Colloid Science and Biomaterials, Department of Chemistry, UniVersity College Dublin, Belfield, Dublin 4, Ireland, and Physical Chemistry 1, Center for Chemistry and Chemical Engineering, Lund UniVersity, P. O. Box 124, 22100 Lund, Sweden ReceiVed: January 12, 2005
The release of two compositionally different solutes from a composite gel composed of two different populations of microgel particles embedded in a single bulk gel matrix is described, showing the potential of the “plumpudding gel” as a multifunctional platform for controlled surface release. One hydrophobic solute (pyrene) and one hydrophobic and charged solute (rhodamine 123) were chosen as the solutes to be released. Hydrophobic microgels composed of 50% N-isopropylacrylamide (NIPAM) and 50% N-tert-butylacrylamide (BAM) were loaded with pyrene, and anionic microgels composed of 30% acrylic acid (AAc), 20% NIPAM, and 50% BAM were loaded with rhodamine 123. The two solute-loaded microgel populations were incorporated into a single bulk gel network, from which the two solutes were released simultaneously and independently. Using this structural motif, solutes that are mutually incompatible can be incorporated into a single matrix with which they may also be incompatible. The electrostatically incorporated solute was released much more slowly than the hydrophobically attracted solute, indicating that the microgel composition can be tailored to the specific solute, and thus control its release rate. The choice of bulk matrix was also found to influence the release rate much more than expected, offering a further control element to the system.
Introduction Responsive polymers and gels have received much interest in the past decade due to their potential as “smart biomaterials”.1 In this case the smartness comes from the coil-globule transition of the responsive polymers,2 which is manifested as the swelling-shrinking transition in the gel form.3 This transition can lead to control of the release rate of solutes from gels via “on-off” switching.4 Recently, however, the concept of smart biomaterials has been expandedsthe idea is now that a single material should provide multiple functionalities while remaining processable and mechanically satisfactory in use. Examples of such smart biomaterials include wound dressings that aid the healing process by delivery of active cells or other woundhealing materials,5 or coating materials that not only prevent protein adherence but also deliver materials to the active site.6 Progress in the field of functional materials has been hampered by the poor mechanical properties of responsive polymers.7 In designing films and coatings for use as controlled delivery devices there are many mechanical issues to be considered. Typical requirements for such a material would include low drag and moderate compliance. Such properties are associated with polymers above their glass transition temperatures. However, controlled release is associated with polymers below their glass transition temperatures, leading to incompatibility of functionality and mechanical concepts. Thus, it would seem that the way forward is to separate the issues of functionality and mechanical strength as this would allow the mechanical properties to be optimized independent of the functional properties. * Corresponding author. E-mail:
[email protected]. † University College Dublin. ‡ Lund University.
The latest generation of coatings for medical devices is being designed to release drugs that inhibit the body’s immune response. However, the body’s defense mechanisms are complex and usually involve hundreds of steps that occur in a cascade, once triggered.8 Thus it is unlikely that a single drug can target the entire cascade. One requires a coating that can release a cocktail of drugs in a preprogrammed or even an adaptive manner, to target a range of problems. Thus, each of the drugs to be delivered will need to be released over a different time scale at the required therapeutic concentration. Furthermore, and crucially, the likelihood of several drugs having similar physicochemical properties and being compatible with a single matrix is negligible. It is likely that the variety of drugs to be released is infinite, some hydrophobic, some hydrophilic, some charged, and all essentially incompatible. Besides, conventional thin-film coatings do not have the capability to release drugs with different release profiles, since it is well-known that the release of a drug obeys a simple scaling law with time (D ∼ tR, where R ≈ 1/2). In this work we show that, using the “plum-pudding” gel motif, it is possible to release several active substances simultaneously, each with its own release rate, independent of the others, and that by tailoring the microgel compositions to the solutes to be released incompatible solutes can be incorporated into a single matrix. With the aim of separating the concepts of mechanical and multifunctional properties and simultaneously keeping the issue of solute incompatibility in mind, we recently prepared a composite gel material, composed of poly(N-isopropylacrylamide) (NIPAM) based microgel particles entrapped in a hydrogel matrix. We called this material the “plum-pudding gel” because the microgel particles resembled plums in a traditional plum pudding when viewed using confocal microscopy.9 If the microgel particles were entrapped in the bulk gel below their
10.1021/jp0502149 CCC: $30.25 © 2005 American Chemical Society Published on Web 03/15/2005
6258 J. Phys. Chem. B, Vol. 109, No. 13, 2005
Lynch et al.
Figure 1. Structures of the solutes: (1) pyrene; (2) rhodamine 123.
transition temperatures (i.e., while they were swollen), they imparted thermoresponsive behavior onto the entire network. This means that any polymeric material can now be made thermoresponsive (or pH, light, or enzyme responsive), enabling the bulk material to be chosen with other issues in mind, such as mechanical strength or bioadhesion. On the other hand, if the microgel particles were entrapped in the bulk gel matrix above their transition temperatures (i.e., in the collapsed state), the microgel particles acted as hydrophobic reservoirs in the material, from which hydrophobic solutes could be slowly released, with the release rate being related to the hydrophobicity of the solute.10 The release of pyrene from “plum-pudding” gels containing 50% NIPAM and 50% N-tert-butylacrylamide (BAM) microgel particles was found to be Fickian, whereas release of the more hydrophobic solute BODIPY (4,4-difluoro-1,3,5,7,8pentamethyl-4-bora-3a,4a-diaza-s-indacene) from similar gels was found to be slightly non-Fickian.10 However, the strength of the “plum” concept is its capacity as a platform with multiple functionality. Thus, in this work we demonstrate the release of two solutes of different chemical compositions from two distinct populations of plum particles embedded in a single gel matrix. Two solutes of different chemical nature were chosen to illustrate the versatility of the system: one a hydrophobic solute, pyrene (1), and the other a cationic solute, rhodamine 123 (2), as shown in Figure 1. Microgel particles were tailored to the solutesshydrophobic microgels for pyrene and anionic (i.e., oppositely charged) microgels for rhodamine 123. Experimental Section Materials. N-Isopropylacrylamide (NIPAM) monomer (purity >99%) supplied by Acros Organics Ltd. (Geel, Belgium) was recrystallized twice from hexane. Dimethylacrylamide (DAM) monomer (purity >99%) supplied by TCI Chemicals (Tokyo, Japan), N-tert-butylacrylamide (BAM) (purity g97.0%), and N,N′-methylenebisacrylamide (BisAM) (purity >99.5%) from Fluka (Dorset, England), ammonium peroxydisulfate (APS) (purity 99.99%) from Aldrich (Dorset, England), N,N,N′,N′tetramethylethylenediamine (TEMED) from Sigma (Dorset, England), pyrene from Aldrich (Steinheim, Germany), and rhodamine 123 from Molecular Probes (Leiden, The Netherlands) were all used as supplied. All water used was doubly deionized Milli-Q water. Synthesis of 50:50 NIPAM:BAM and 30:20:50 AAc: NIPAM:BAM Microgel Particles. Both types of microgel were synthesized by dispersion polymerization according to the method of Li and Bae.11 The monomers NIPAM (0.1 g), BAM (0.1 g), and BisAm (0.02 g) or AAc (0.06 g), NIPAM (0.04 g), BAM (0.1 g), and BisAm (0.02 g) were dissolved in 36 mL of water. A 1 mL volume of 0.1 wt % Triton 100 solution was added. The solution was heated to 70 °C, and degassed by bubbling with N2 for 30 min. APS (0.02 g) was dissolved in 4 mL of water, degassed, and added slowly to the stirring monomer solution, under an N2 atmosphere. The reaction was left for 12 h at 70 °C. The resulting 1 wt % microgel in water
Figure 2. Schematic representation of the preparation of “plumpudding” gel films.
dispersions were cleaned by dialysis and freeze-dried. The diameter of both particle types was determined to be ∼200 nm, using transmission electron microscopy. The particles were deposited onto a holey carbon grid from a dispersion in ethanol, and the ethanol was subsequently evaporated, leaving the dried particles. Thus, the diameters measured correspond approximately to microgel particles in the collapsed state. Loading of the Microgels with Hydrophobic Fluorescent Dyes. The fluorescent dyes used were pyrene and rhodamine 123. For pyrene, a 2 × 10-6 M aqueous standard solution was prepared. For rhodamine 123, a standard solution of 1 mg in 25 mL of ethanol was prepared. An aqueous standard solution was prepared from this by diluting 1 mL of rhodamine 123 in ethanol solution to 10 mL with water, giving a concentration of 1.0 × 10-5 mg/mL rhodamine 123. For pyrene, 5 mL of the standard solution was used to make a 0.5 wt % dispersion of the 50:50 BAM:NIPA microgel particles and shaken at room temperature for 2 h. For rhodamine 123, 5 mL of the standard solution was used to make a 0.5 wt % dispersion of the 30:20:50 AAc:NIPAM:BAM microgel particles and shaken at room temperature for 2 h. Solutions were then centrifuged at 3500 min-1 for 30 min, and the supernatant was removed and used to calculate the amount of the fluorophor absorbed into the microgel particles. Water was added to resuspend the microgels, and an aliquot of each microgel type was added to the monomers to make the pregel solution. Synthesis of a Flat Sheet of Gel Containing 0.5 wt % Each of the Two Populations of Fluorescent Microgels. Flat sheets of gel containing the two different dye-loaded microgels (physically absorbed pyrene or rhodamine 123) were prepared according to the method of Kokufuta and Nakaizumi.12 Pregel solutions were prepared as follows. The required amounts of monomer (700 mmol of dimethylacrylamide, DAM), crosslinker (8.6 mmol of BisAM), and promotor (0.001% v/v TEMED) were dissolved in 0.5 mL each of the two microgel solutions. The solution was degassed under vacuum, and initiator (3.5 mmol of ammonium persulfate) was added. The solution was injected between two glass slides separated by 300 µm and left to gel for 24 h at room temperature, i.e., above the transition temperature of the microgel particles, so that they were in collapsed state. Thus, the two different microgel populations are incorporated into a single bulk matrix, as shown schematically in Figure 2. Release of Fluorescent Dyes (Pyrene and Rhodamine 123) from a Flat Sheet of the Composite Gel. The release of hydrophobic and cationic solutes from a thin gel film containing the two populations of microgel particles, the hydrophobic microgels of 50:50 NIPAM:BAM loaded with pyrene and the anionic microgels of 30:20:50 AAc:NIPAM:BAM loaded with cationic rhodamine 123, was measured by fluorescence spectroscopy. The freshly prepared “plum-pudding” gel was placed
Simultaneous Release of Solutes from Composite Gel
J. Phys. Chem. B, Vol. 109, No. 13, 2005 6259
Figure 3. Emission spectra of standard solutions after excitation at two different excitation wavelengths (337 and 507 nm): (red curve) excitation at 337 nm of rhodamine 123 solution; (green curve) excitation at 337 nm of pyrene solution; (blue curve) excitation at 507 nm of rhodamine 123 solution; and (purple curve) excitation at 507 nm of pyrene solution.
in a beaker containing 45 mL of deionized water and thermostated at 25 °C, above the transition temperature (Tc) of both of the microgel particles (Tc for the 50:50 NIPAM:BAM particles is 12 °C,10 while Tc of the 30:20:50 AAc:NIPAM:BAM particles is ∼8 °C, as determined from the transmission as a function of temperature with Tc being taken as the point at which the transmission is reduced to 50% of the initial transmission). At regular intervals the water was removed and replaced by fresh water. This setup represents the release of solutes from a thin gel film in a well-stirred infinite bath. The amount of each of the two solutes in the water at each sampling time was determined by fluorescence spectroscopy using a Spex Fluorolog II equipped with a 450 W Xe lamp as the excitation source, and two 0.22 m monochromators (Spex 1680) for wavelength selection of the excitation and emission light, respectively. The emission is detected with a water-cooled 950V photomultiplier tube (PMT) and is plotted as the percent release as a function of time. The initial solute concentration in the different microgel populations was calculated by subtracting the concentration in the supernatant after shaking and centrifuging from the concentration in the solution before shaking (2 × 10-6 M for pyrene and 1.0 × 10-5 M for rhodamine 123), taking into account that only one-third of the loaded beads of each type were incorporated into the composite gel. Thus the actual concentrations of the two dyes in the composite gel were determined to be 3.7 × 10-7 M pyrene and 3.3 × 10-6 M rhodamine 123. This actually represents almost 99% uptake of rhodamine 123 by the 30:20:50 AAc:NIPAM:BAM particles and approximately 56% uptake of pyrene by the 50:50 NIPAM: BAM particles, indicating that the 2 h of loading was sufficient to get good uptake of the solutes into the beads. Additionally, the rate of diffusion of these two small molecules into the particles should be very fast as they have a high affinity for the particles, indicating that they would easily have penetrated to the center of the particles over this time period, even allowing for the torturous diffusion route necessitated by the porous structure of the particles. Results and Discussion The two dyes were chosen so that their emission spectra would not overlap, and thus their individual concentrations could be determined in the same solution. For pyrene, the excitation and emission wavelengths are 337 and 385 nm, respectively, and for rhodamine 123 they are 507 and 529 nm, respectively. The emission spectra of the two solutes after excitation at each of the two excitation wavelengths are shown in Figure 3, which clearly shows that the emission peaks do not overlap and thus
Figure 4. Simultaneous release of pyrene and rhodamine 123 from 300 µm thick DAM gel containing 0.5 wt % 50:50 NIPAM:BAM microgel particles loaded with pyrene and 0.5 wt % 30:20:50 AAc: NIPAM:BAM microgel particles loaded with rhodamine 123. (O) Pyrene; (0) rhodamine 123.
there is no effect of the presence of a second solute on the determination of the concentration of either of the solutes from their emission peak intensity. There is a small peak in the rhodamine 123 spectrum in the region of the pyrene emission, but it is very small compared to that of pyrene and is therefore not expected to affect the results. Figure 4 shows the simultaneous elution profiles of pyrene and rhodamine 123 from a 300 µm thick gel matrix, composed of 0.5 wt % pyrene loaded 50:50 NIPAM:BAM microgel particles and 0.5 wt % rhodamine 123 loaded 30:20:50 AAc:NIPAM:BAM microgel particles in a 700 mmol DAM bulk gel, as a function of time at 25 °C. The amount of pyrene released was just 28% of the loaded amount after 400 h. The amount of rhodamine 123 released from the anionic (and hydrophobic) microgel particles was not even 5% after 400 h, meaning that rhodamine 123 was released much more slowly than pyrene from the gel. The difference in the release rates is most probably due to the fact that the charged solute, rhodamine 123, is held in the mircogel particles by an additional electrostatic interaction as well as the hydrophobic interaction. Electrostatic interactions have been shown previously to have a large effect on the rate of release of solutes, with release of solutes from gels with similar charge being very rapid due to charge repulsion, while release of solutes from gels of opposite charge is extremely slow due to charge attraction.13 The low percent of solute released after such long times has also been observed in a study on the effect of microgel and/or matrix composition on the release of the antirestenosis agent fluvastatin from composite gel networks.14 Here, even after 2 months of release, up to 70% of the loaded fluvastatin remained in the microgel particles. The release mechanisms for the different solutes were determined by plotting log M(t)/M(∞) against log Kt and determining n, the slope of the line obtained, where M(t)/M(∞) is the fraction of solute released at time t, M(t) is the amount of solute released at time t, and M(∞) is the total amount of solute present in the matrix initially. The index n is the effective slope, and this determines the release mechanism.15 When n < 0.5, the release is by Fickian diffusion of the solute, when 0.5 < n < 1.0, the release is by non-Fickian diffusion, and when n )
6260 J. Phys. Chem. B, Vol. 109, No. 13, 2005
Figure 5. log-log plot of simultaneously released pyrene (O) and rhodamine 123 (0), with calculated (effective) slopes (n values).
1, there is continuous zero-order release. Figure 5 shows the log-log plots for the two different solutes, as well as the calculated slopes (n values) from which the release mechanisms were determined. Thus, from the present composite gels, both pyrene and rhodamine 123 are released by Fickian diffusion, and both show sustained release over very long times. Once the release kinetics for a particular drug from a polymeric film have been established, it is possible to determine the diffusion coefficient for the solute at the temperature used. Where the release kinetics conform to Fickian diffusion (n < 0.5) the diffusion coefficient can be obtained from the equation M(t)/M(∞) ) 2(Dt/πl2), where D is the diffusion constant of the solute, and l is the thickness of the gel film (300 µm). Since both solutes have n < 0.5, the diffusion coefficients can be calculated directly using this equation. Taking t to be 300 h, and M(t)/M(∞) to be the amount released after 300 h, the diffusion coefficient of pyrene was calculated to be 3.51 × 10-12 m2 s-1 while that of rhodamine 123 was calculated to be 6.28 × 10-13 m2 s-1. Thus, the diffusion of pyrene from the gel is over 5 times faster than the diffusion of rhodamine 123 from the gel, as indicated in Figure 4. The slower overall release of rhodamine 123 is most probably due to the additional attraction between the cationic moieties in rhodamine 123 and the anionic charges on the microgel particles. Due to the composite nature of the gels, the release kinetics will consist of two steps: first, release of the solutes from the microgel particles into the bulk gel, and then release from the bulk gel into the surrounding solvent. Due to the much looser network structure of the bulk gel compared to the microgel particles, and the lower hydrophobicity (and thus lower attractiveness to the solutes) of the bulk matrix, the rate of release from the bulk gel to the solvent was expected to be much faster than the rate of release from the microgels into the bulk gel. However, the release of pyrene from a “plum-pudding” gel where the bulk matrix was composed of NIPAM, at 38 °C (above the transition temperature of the NIPAM gel), was found to be significantly faster than the release of pyrene measured here.10 In that case, 50% of the loaded pyrene was released after 300 h, compared with just over 25% released here, as shown for comparison purposes in Figure 6. One other difference between the two systems was the thickness of the bulk matrix, with the matrix in the NIPAM case being significantly thinner than that used here (150 µm for the NIPAM matrix compared
Lynch et al.
Figure 6. Comparison of release profiles for pyrene from composite gels with (O) DAM as bulk matrix or (0) NIPAM as bulk matrix (data taken from ref 10).
to 300 µm for the DAM network). However, it is unlikely that this difference in thickness could account for the much retarded release rate of pyrene from the DAM bulk network, as it should simply increase the diffusion distance. Thus, the 2-fold decrease in the release rate upon changing the bulk matrix from NIPAM above the transition temperature to DAM indicates that the bulk matrix itself must affect the diffusion process. In the case of the NIPAM bulk gel matrix10 we suggested that release of hydrophobic compounds from the bulk matrix to the solvent may be via a “partition” type method of release, where the solute molecule can interact with the polymer chains of the bulk network, and thus diffuse along the chains to the bulk solution. However, we did not realize that this in fact may be the driving force for the initial release of the solutes from the hydrophobic microgel particles into the bulk gel, as the solutes are quite happy in their hydrophobic havens and have no particular affinity for the bulk solution, and thus, if the bulk gel is also hydrophobic it facilitates transport of the solutes to the bulk. In the case where the bulk network is composed of DAM chains instead, which are hydrophilic, there is thus no incentive or driving force for the solutes to leave the hydrophobic microgel particles, or nothing to overcome the hydrophobic attraction between the solute and the microgel particles. Thus the release from the microgel particles into the bulk gel is severely reduced, which in turn reduces the release into the surrounding medium. This is even more pronounced where there is an additional electrostatic attraction between the solute and the microgel particle to be overcome (i.e., in the rhodamine 123 case). Thus, in the absence of a “partition” type interaction between the solute and the bulk gel network, as in the case where the bulk network is composed of the hydrophilic DAM network, the overall release rate is much slower because the solute cannot diffuse along the network chains, but instead must diffuse through the pores of the gel, in a free diffusion process, for which there is no strong driving force. Further evidence that the retarded release from the microgel particles into the hydrophilic bulk gel is due to the absence of a driving force, such as an attraction between the solute and the bulk gel network, can be obtained by an analysis of the process of formation of the composite gel matrixes. In the first step, the particles are loaded with solute, and the excess solvent is removed, leaving the loaded particles which are then
Simultaneous Release of Solutes from Composite Gel resuspended in the aqueous monomer solution. The gelation reaction takes place over the next 24 h, during which time some release from the microgel particles into the bulk gel would be expected to occur. This should then lead to a “burst” of release in the first couple of hours of the release experiment, and indeed that was the case from the NIPAM bulk matrix, as shown in Figure 6, where 35% of the pyrene was released in the first 24 h of the experiment, compared with just 22% where the bulk matrix was DAM. This again indicates that, in the case of the DAM matrix, there is a greater barrier to the release of the solute from the microgel particles into the bulk matrix, but as the composition of the microgel particles was identical in both cases there can be no difference in the attraction between the solute and the microgel particles. Thus it would seem that the rate of release and the release mechanism are largely influenced by the choice of bulk matrix, as well as the attractiveness of the microgel particles for the solute. This effect of the bulk matrix on the release process was entirely unexpected, and further work is needed to fully understand it. We have begun some NMR self-diffusion measurements to systematically determine the effect of the hydrophobicity of the bulk matrix on the release of solute from the microgel particles, the results of which will be reported later.
J. Phys. Chem. B, Vol. 109, No. 13, 2005 6261 Since the release rate was controlled in large part by the solute structure and the microgel composition (hydrophobicity or presence of charged groups), it is clear that by further tailoring the composition of the microgel particles to the solute composition the release rate can be increased or decreased as desired. The core difficulty of mutually incompatible solutes and matrix may therefore be resolved, and the mechanical aspect of the system decoupled from functionality. Additionally, by incorporating microgel particles that are responsive to different environmental triggers, delivery devices can be prepared where one solute has continual release, another releases in response to, for example, pH changes, and another still is self-regulating and releases only when its concentration in the bulk medium drops below some cutoff point. Thus, the “plum-pudding” gel offers the possibility of real controlled release of several compounds simultaneously, each with its own release rate and/ or release trigger. Acknowledgment. This work was funded by grants from the Health Research Board, Ireland, and Enterprise Ireland. I.L. thanks the department of Chemical Physics at Lund University for the use of their spectrophotometer. References and Notes
Conclusions Using the “plum-pudding” gel motif, a single gel matrix with two different microgel populations loaded with two different solutes was prepared. The different solutes (with different physicochemical properties, such as hydrophobicity, and uncharged or cationic) were simultaneously released from a single gel matrix, each with its own distinct release profile. The release rate was much slower for the solute held in the microgel via hydrophobic and electrostatic attractions (rhodamine 123) than for the solute held in the microgels via hydrophobic attractions only (pyrene). Additionally, the hydrophobicity or hydrophilicity of the bulk matrix also affected the release, offering an additional degree of control over the release rate, as increasing the hydrophobicity of the bulk matrix resulted in faster release, probably due to diffusion of the hydrophobic solute along the hydrophobic network chains.
(1) Hoffman, A. S. MRS Bull. 1991, 16, 42. (2) Heskins, M. a. G., J. E. J. Macromol. Sci., Chem. 1968, A2, 1441. (3) Hirotsu, S. J. Chem. Phys. 1991, 94, 3949. (4) Bae, Y. H.; Okano, T.; Kim, S. W. Pharm. Res. 1991, 8, 531. (5) Rees, R. S.; Adamson, B. F.; Lindblad, W. J. Wound Repair Regen. 2001, 9, 297. (6) Gunn, J.; Cumberland, D. Eur. Heart J. 1999, 20, 1693. (7) Stammen, J. A.; Williams, S.; Ku, D. N.; Guldberg, R. E. Biomaterials 2001, 22, 799. (8) Suthanthiran, M.; Morris, R. E.; Strom, T. B. Am. J. Kidney Dis. 1996, 28, 159. (9) Lynch, I.; Dawson, K. A. J. Phys. Chem. B 2003, 107, 9629. (10) Lynch, I.; Dawson, K. A. J. Phys. Chem. B 2004, 108, 10893. (11) Li, Y. D. B.; Y. C. J. Appl. Polym. Sci. 1998, 67, 2088. (12) Kokufuta, E.; Nakaizumi, S. Macromolecules 1995, 28, 1704. (13) Lee, W.-F.; Chiu, R.-J. Mater. Sci. Eng., C 2002, 20, 161. (14) McGillicuddy, F.; Lynch, I.; Dawson, K. A.; Keenan, A. K. Am. J. Cardiol. 2004, 94 (Suppl. 1), 224E. (15) Yasuda, H.; Lamaze, C. E.; Ikenberry, L. D. Makromol. Chem. 1968, 118, 18.
