Active Loading and Tunable Release of Doxorubicin from Block

Vesicles are spherical bilayers that offer a hydrophilic reservoir, suitable for the incorporation of water-soluble molecules, as well as a hydrophobi...
1 downloads 4 Views 193KB Size
Download PDF
9308

Langmuir 2005, 21, 9308-9313

Active Loading and Tunable Release of Doxorubicin from Block Copolymer Vesicles Amira Choucair, Patrick Lim Soo, and Adi Eisenberg* Department of Chemistry, McGill University, 801 Sherbrooke Street West, Montreal, Qc., Canada H3A 2K6 Received March 16, 2005. In Final Form: June 10, 2005 Vesicles are spherical bilayers that offer a hydrophilic reservoir, suitable for the incorporation of watersoluble molecules, as well as a hydrophobic wall that protects the loaded molecules from the external solution. The permeability of a vesicle wall made from polystyrene can be enhanced by adding a plasticizer such as dioxane. Tuning the wall permeability allows loading and release of molecules from vesicles to be controlled. In this study, vesicles are prepared from polystyrene310-b-poly(acrylic acid)36 and used as model carriers for doxorubicin (DXR), a weak amine and a widely used anticancer drug. To increase the wall permeability, different amounts of dioxane are added to the vesicle solution. A pH gradient is created across the vesicle wall (inside acidic) and used as an active loading method to concentrate the drug inside the vesicles. The results show that a pH gradient of ca. 3.8 units can enhance the loading level up to 10-fold relative to loading in the absence of the gradient. After loading, the release of DXR from vesicles is followed as a function of the wall permeability. The diffusion coefficient of doxorubicin through polystyrene (D) is evaluated from the initial slope of the release curves; the value of D ranges from 8 × 10-17 to 6 × 10-16 cm2/s, depending on the degree of plasticization of the vesicle wall.

1. Introduction Vesicles prepared from amphiphilic block copolymers are spherical bilayers in which the insoluble polymer chains constitute the vesicle wall (which is a bilayer), whereas the chains of the soluble block extend from the inner and outer surfaces of the vesicle into the solvent system. The interest in polymeric vesicles is motivated both by the versatility of these structures, and by the variety of their potential applications. For instance, vesicles can be prepared from several diblock1-9 and triblock copolymers,10-12 in different solvents1,2,10,11,13-18 and with sizes ranging between tens of nanometers11,17 to * Corresponding author. (1) Zhang, L.; Eisenberg, A. Science 1995, 268, 1728-1731. (2) Ding, J.; Liu, G. Macromolecules 1997, 30, 655-657. (3) Holder, S. J.; Sommerdijk, N. A. J. M.; Williams, S. J.; Nolte, R. J. M.; Hiorns, R. C.; Jones, R. G. Chem. Commun. 1998, 1445-1446. (4) Yu, K.; Bartels, C.; Eisenberg, A. Macromolecules 1998, 31, 93999402. (5) Maskos, M.; Harris, J. R. Macromol. Rapid Commun. 2001, 22, 271-273. (6) Luo, L.; Eisenberg, A. J. Am. Chem. Soc. 2001, 123, 1012-1013. (7) Harris, J. K.; Rose, G. D.; Bruening, M. L. Langmuir 2002, 18, 5337-5342. (8) Kukula, H.; Schlaad, H.; Antonietti, M.; Foerster, S. J. Am. Chem. Soc. 2002, 124, 1658-1663. (9) Vriezema, D. M.; Hoogboom, J.; Velonia, K.; Takazawa, K.; Christianen, P. C. M.; Maan, J. C.; Rowan, A. E.; Nolte, R. J. M. Angew. Chem., Int. Ed. 2003, 42, 772-776. (10) Schillen, K.; Bryskhe, K.; Mel’nikova, Y. S. Macromolecules 1999, 32, 6885-6888. (11) Nardin, C.; Hirt, T.; Leukel, J.; Meier, W. Langmuir 2000, 16, 1035-1041. (12) Chen, X. L.; Jenekhe, S. A. Macromolecules 2000, 33, 46104612. (13) Discher, B. M.; Won, Y.-Y.; Ege, D. S.; Lee, J. C. M.; Bates, F. S.; Discher, D. E.; Hammer, D. A. Science 1999, 284, 1143-1146. (14) Ding, J.; Liu, G. J. Phys. Chem. B 1998, 102, 6107-6113. (15) Cornelissen, J. J. L. M.; Fischer, M.; Sommerdijk, N. A. J. M.; Nolte, R. J. M. Science 1998, 280, 1427-1430. (16) Zhang, L.; Eisenberg, A. Polym. Adv. Technol. 1998, 9, 677699. (17) Luo, L.; Eisenberg, A. Langmuir 2001, 17, 6804-6811. (18) Gravano, S. M.; Borden, M.; von Werne, T.; Doerffler, E. M.; Salazar, G.; Chen, A.; Kisak, E.; Zasadzinski, J. A.; Patten, T. E.; Longo, M. L. Langmuir 2002, 18, 1938-1941.

few microns.18-21 Potential applications of vesicles include their use as mimics of biological membranes,22 as synthetic models for biological cells,23 as microreactors,22 and as carriers for hydrophilic molecules.21,24 In this latter regard, an interesting property of vesicles is their ability to offer a hydrophilic reservoir suitable for the incorporation of water-soluble molecules and, at the same time, a hydrophobic wall, or a barrier, that protects the loaded molecules from the external solution. The wall can also act as a “gate” that controls the diffusion of molecules in and out of the vesicle. By tuning the permeability of this “gate”, the extent of loading and release from vesicles can be controlled. The present study exploits this property of polymeric vesicles in order to control the loading and release of doxorubicin from the aqueous cavity of polystyrene310-b-poly(acrylic acid)36 vesicles. Doxorubicin (DXR) is a potent antibiotic used in the treatment of a wide range of solid tumors and leukemias.25-28 However, the cardiotoxicity of this drug is a major drawback, limiting its direct administration and cumulative dosage.25,26 In an effort to reduce the accumulation of DXR in the heart muscle, several studies (19) Jenekhe, S. A.; Chen, X. L. Science 1998, 279, 1903-1907. (20) Ilhan, F.; Galow, T. H.; Gray, M.; Clavier, G.; Rotello, V. M. J. Am. Chem. Soc. 2000, 122, 5895-5896. (21) Lee, J. C. M.; Bermudez, H.; Discher, B. M.; Sheehan, M. A.; Won, Y.-Y.; Bates, F. S.; Discher, D. E. Biotechnol. Bioeng. 2001, 73, 135-145. (22) Nardin, C.; Widmer, J.; Winterhalter, M.; Meier, W. Eur. Phys. J. E 2001, 4, 403-410. (23) Hammer, D. A.; Discher, D. E. Annu. Rev. Mater. Res. 2001, 31, 387-404. (24) Brown, M. D.; Schaetzlein, A.; Brownlie, A.; Jack, V.; Wang, W.; Tetley, L.; Gray, A. I.; Uchegbu, I. F. Bioconjugate Chem. 2000, 11, 880-891. (25) Barenholz, Y.; Amselem, S.; Goren, D.; Cohen, R.; Gelvan, D.; Samuni, A.; Golden, E. B.; Gabizon, A. Med. Res. Rev. 1993, 13, 449491. (26) Gabizon, A.; Dagan, A.; Goren, D.; Barenholz, Y.; Fuks, Z. Cancer Res. 1982, 42, 4734-4739. (27) Son, Y. J.; Jang, J.-S.; Cho, Y. W.; Chung, H.; Park, R.-W.; Kwon, I. C.; Kim, I.-S.; Park, J. Y.; Seo, S. B.; Park, C. R.; Jeong, S. Y. J. Controlled Release 2003, 91, 135-145. (28) Kabanov, A. V.; Batrakova, E. V.; Alakhov, V. Y. J. Controlled Release 2003, 91, 75-83.

10.1021/la050710o CCC: $30.25 © 2005 American Chemical Society Published on Web 08/20/2005

Loading and Release of Doxorubicin

have focused on using liposomes as drug delivery vehicles.25,26,29-31 In vivo studies showed that, depending on the stability of the liposomes and their ability to retain the entrapped drug molecules (i.e., minimum leakage), administering liposomal doxorubicin reduces the accumulation of the drug in the heart tissues by decreasing the concentration of free drug in circulation.26,30 Obtaining high levels of drug loading was also of interest. In 1976, Nichols and Deamer32 reported that, by preparing liposomes under acidic conditions and then raising the pH of the solution in which the liposomes are suspended, that is, by creating a pH gradient across the liposome membrane (pHinside ) 5 and pHoutside ) 8), the level of incorporation of weak amines (such as dopamine, norepinephrine, and epinephrine) increases by 10- to 20fold, in comparison to loading in the absence of a pH gradient. Later, this concept of active loading was used by Bally et al.33,34 and Mayer et al. to concentrate doxorubicin (a weak amine, pKa ) 8.25),35 and other types of drugs (all of which are weak bases)36 into liposomes. High loading levels were achieved (internal drug concentration ranging between 50 and 130 mM); however, the permeability of the liposome membrane to protons made the pH gradient difficult to maintain, and in many instances, the resulting leakage of the incorporated molecules was significant.36 In light of these findings, it was interesting to investigate the applicability of such an active loading method to polymeric vesicles, which offer a tougher and thicker wall than liposomes do.13,37 In this study, unilamellar vesicles of polystyrene310-b-poly(acrylic acid)36 are used as model carriers for doxorubicin. The active loading method is applied by creating a pH gradient across the vesicle membrane (pHinside ) 2.5 and pHoutside ) 6.3). For comparison, DXR is added to another series of vesicle solutions in which no pH gradient was established. The extent of incorporation, using both methods, is determined as a function of the permeability of the polystyrene wall of the vesicle. To increase the permeability of polystyrene, different concentrations of dioxane, a plasticizer, are added to each solution prior to loading. The release of DXR from vesicles is also followed as a function of the wall permeability (i.e., in the presence of different concentrations of dioxane). The diffusion coefficient of DXR though polystyrene is evaluated from the release profile. It is shown that by varying the degree of plasticization of the vesicle wall, the extent of loading and release can be controlled, independently. 2. Experimental Section 2.1. Materials. The vesicles used in this study were prepared from polystyrene310-b-poly(acrylic acid)36 (PS310-b-PAA36) copoly(29) Sadzuka, Y.; Nakade, A.; Tsuruda, T.; Sonobe, T. J. Controlled Release 2003, 91, 271-280. (30) Mayer, L. D.; Cullis, P. R.; Bally, M. B. J. Liposome Res. 1994, 529-553. (31) Abraham, S.; McKenzie, C.; Masin, D.; Ng, R.; Harasym, T.; Mayer, L.; Bally, M. Clin. Cancer Res. 2004, 10, 728-738. (32) Nichols, J. W.; Deamer, D. W. Biochim. Biophys. Acta 1976, 455, 269-271. (33) Dos Santos, N.; Cox, K.; McKenzie, C.; van Baarda, F.; Gallagher, R.; Karlsson, G.; Edwards, K.; Mayer, L.; C, A.; Bally, M. Biochim. Biophys. Acta 2004, 1661, 47-60. (34) Chiu, G.; Abraham, S.; Ickenstein, L.; Ng, R.; Karlsson, G.; Edwards, K.; Wasan, E.; Bally, M. J. Controlled Release 2005, 104, 271-288. (35) Mayer, L. D.; Bally, M. B.; Cullis, P. R. Biochim. Biophys. Acta 1986, 857, 123-126. (36) Madden, T. D.; Harrigan, P. R.; Tai, L. C. L.; Bally, M. B.; Mayer, L. D.; Redelmeier, T. E.; Loughrey, H. C.; Tilcock, C. P. S.; Reinish, L. W.; Cullis, P. R. Chem. Phys. Lipids 1990, 53, 37-46. (37) Discher, B. M.; Hammer, D. A.; Bates, F. S.; Discher, D. E. Curr. Opin. Colloid Interface Sci. 2000, 5, 125-131.

Langmuir, Vol. 21, No. 20, 2005 9309 Table 1. Chemical Structure, Molecular Weight, and pKa40 of Doxorubicin Hydrochloride

molecular weight (g/mol) pKa

580 8.25

mer. The numbers 310 and 36 represent the average degree of polymerization of the hydrophobic and the hydrophilic block, respectively. The copolymer was synthesized by anionic polymerization and has a polydispersity of 1.03, as determined by size exclusion chromatography using polystyrene standards. A detailed description of the synthesis and characterization procedures is given in previous publications.17,38,39 Dioxane (HPLC grade) and doxorubicin hydrochloride were purchased from Sigma-Aldrich and were used as received. Table 1 gives the chemical structure of doxorubicin hydrochloride, along with its molecular weight and pKa. Dialysis chambers (Slide-A-Lyzer Mini Dialysis Unit) used for the loading and release experiments were purchased from MJS BioLynx Inc. (Brockville ON, Canada) and have a molecular weight cutoff of 3500 g/mol. 2.2. Sample Preparation. Vesicle solutions were prepared in the following way: the PS310-b-PAA36 copolymer was first dissolved in dioxane, a good solvent for both blocks. The initial polymer concentration was 0.5% (w/w). pH 2.5 water was then added dropwise at a rate of 0.2% (w/w) per minute to induce self-assembly. Water addition was continued until a final water concentration of 15% (w/w) was reached. The presence of vesicles at this water content was confirmed using transmission electron microscopy (TEM). Polystyrene constitutes the wall of the vesicle, and poly(acrylic acid) chains cover the internal and external surfaces. The average diameter of the vesicles is ≈213 ( 80 nm (as determined using TEM). The vesicle solution was stirred overnight and then quenched in excess water (4-fold dilution) in order to extract much of the dioxane from the polystyrene wall and raise its glass transition temperature (Tg) well above room temperature. This procedure preserves the vesicle morphology during the subsequent dialysis against pH 2.5 water, which is done in order to remove dioxane from the vesicle solution. The aqueous solution obtained after dialysis has a pH of ca. 2.5 and contains vesicles with an internal aqueous cavity of the same pH. This solution was divided into a series of vials, to each of which a given amount of dioxane and doxorubicin (as an aqueous solution) was added. The dioxane content ranged between 0 and 60% (w/w), whereas the doxorubicin concentration was kept constant in all solutions at ca. 5.6 × 10-4 M. The solutions were allowed to stir for about 4 h, and then the pH of each solution was measured and adjusted to a value of ca. 6.3 pH units (AR10 pH meter, Fisher Scientific), using small volumes of NaOH aqueous solution (0.1004 N). A pH difference between the interior of the vesicles (pHinside ) 2.5) and the outside aqueous solution (pHoutside ) 6.3) was thus created. The vesicle solutions in which a pH gradient is created will be referred to as samples. For comparison, DXR was added to another set of solutions in which no pH gradient was established (i.e., pHinside ) pHoutside ) 2.5). These solutions will be referred to as controls. Solutions containing doxorubicin were wrapped in aluminum foil and kept in the dark. After allowing the solutions to stir for 3 days, during which the loading of the drug took place, an aliquot of each solution (both samples and controls) was quenched in excess water (ca. 7-fold dilution) and dialyzed against water for 3 days in order to remove unloaded drug molecules. Quenching in water prior (38) Zhong, X. F.; Varshney, S. K.; Eisenberg, A. Macromolecules 1992, 25, 7160-7167. (39) Yu, K.; Eisenberg, A. Macromolecules 1998, 31, 3509-3518.

9310

Langmuir, Vol. 21, No. 20, 2005

Choucair et al.

to dialysis decreases the dioxane percentage in the solution and, therefore, its content in the polystyrene wall of the vesicles.41 This, in turn, reduces the permeability of the wall and minimizes the diffusion and loss of the incorporated molecules during dialysis. The concentration of doxorubicin in each vesicle solution after dialysis was determined using fluorescence spectroscopy (SPEX Fluoromax-2; at λex ) 488 nm). 2.3. Determining the Ethylbenzene-Water Partition Coefficient. The partition coefficient of DXR between ethylbenzene and water was calculated in order to determine the relative affinity of the drug for the polystyrene wall of the vesicle versus the aqueous solution. A known concentration of doxorubicin was added to two solutions: the first is a mixture of 1.5 mL of ethylbenzene and 2.0 mL of pH ) 2.5 water, and the second is a mixture of 1.5 mL of ethylbenzene and 2.0 mL of pH ) 6.3 water. Each solution, which consists of two immiscible phases, was allowed to mix for two weeks. The aqueous phase was then separated from the ethylbenzene phase, and the concentration of DXR in each phase was determined using fluorescence spectroscopy. The ethylbenzene-water partition coefficient, KEB/H2O was then calculated using the following equation:

KEB/H2O )

molarity of DXR in EB phase molarity of DXR in water phase

(1)

It is important to note that polystyrene, which constitutes the vesicle wall, has different solubility properties than ethylbenzene especially that at room temperature polystyrene is below its glass transition temperature and is, therefore, in a glassy state. Nevertheless, ethylbenzene is chosen as a model solvent due to its hydrophobicity and structural similarity to the styrene monomer. The solubility of doxorubicin in ethylbenzene should be considered as an upper limit for the solubility of the drug in the polystyrene wall. 2.4. Determining the Interaction of Doxorubicin with Poly(acrylic acid). Interaction of doxorubicin with poly(acrylic acid) was determined in order to estimate the extent of interaction between the drug and the surfaces of the vesicles. Two aqueous solutions of homo-poly(acrylic acid) (average molecular weight ) 450 000 g/mol) containing the same concentration of DXR were prepared. After adjusting the pH of the first solution to ca. 2.5, and that of the second to ca. 6.3, the solutions were allowed to stir for 3 days. A 400 µL aliquot of each solution was then placed in a mini-dialysis chamber and dialyzed against water for 3 days, which is the same period of time over which DXR-loaded vesicles were dialyzed to remove free, unincorporated drug molecules. The concentration of doxorubicin present in each solution after dialysis was determined using fluorescence spectroscopy. The obtained fluorescence signal is assumed to be due to doxorubicin molecules interacting with poly(acrylic acid). 2.5. Release of Doxorubicin from Vesicles. The vesicle samples used for the release experiment were dialyzed repeatedly for 3 days to remove any unincorporated DXR. A 200 µL solution of DXR loaded vesicles was placed into a dialysis chamber, and the composition of the solution was adjusted to give a ratio of 50% water to 50% dioxane, 75% water to 25% dioxane, or 100% water. Several dialysis chambers were prepared at the same DXR concentration (120 µM) and placed into a dialysis float device. The device was then placed into a large reservoir, which contained twenty liters of solvent. The solvent composition in the reservoir was adjusted to match that inside the dialysis chambers (i.e., 50%, 75%, or 100% water). The release of DXR is assumed to be under sink conditions, since the volume of the reservoir (20 L) is 50 000 times larger than the volume of the vesicle solution inside the dialysis chamber (400 µL). The drug release was assumed to start as soon as the dialysis float device containing the dialysis chambers was placed into the reservoir. The large reservoir was kept under constant stirring, and at various time points, one of the dialysis chambers was removed. The concentration of DXR in the dialysis chamber was quantified using fluorescence spectroscopy. (40) Lasic, D. D.; Frederik, P. M.; Stuart, M. C. A.; Barenholz, Y.; McIntosh, T. J. FEBS Lett. 1992, 312, 255-258. (41) Yu, Y.; Zhang, L.; Eisenberg, A. Macromolecules 1998, 31, 11441154.

Figure 1. TEM picture of PS310-b-PAA36 vesicles loaded with doxorubicin via a pH gradient. Average vesicle diameter = 210 nm and average wall thickness = 30 nm. 2.6. Transmission Electron Microscopy. Samples for transmission electron microscopy were prepared by placing a drop (ca. 10 µL) of a vesicle solution on a copper grid, previously coated with a thin layer of carbon. The grid was then left to air dry. All TEM measurements were carried out on a JEOL JEM2000 FX electron microscope, operating at an acceleration voltage of 80 keV. The pictures were taken using a multiscan CCD camera. Figure 1 shows a representative TEM picture of PS310b-PAA36 vesicles loaded with DXR.

3. Results and Discussion 3.1. Active Loading of Doxorubicin into Vesicles. The loading of doxorubicin (DXR) into PS310-b-PAA36 vesicles was carried out for a series of samples and controls. The samples represent vesicle solutions in which a transmembrane pH gradient was created (pHinside ) 2.5, and pHoutside ) 6.3). In the control solutions, no pH gradient was established (pHinside ) pHoutside ) 2.5). Since in both cases (samples and controls) the drug was added to solutions of preformed vesicles, incorporation into the vesicle cavity must involve diffusion though the polystyrene wall. Therefore, it is of interest to examine the effect of wall permeability on the extent of loading, both in the presence and absence of a pH gradient. To increase the permeability of polystyrene, different amounts of dioxane, a plasticizer, were added to each solution. Dioxane is a better solvent for polystyrene than water. This can be seen from the values of the solubility parameters of polystyrene, dioxane and water, which are δPS ) 8.1-9.9, δdioxane ) 10.0, and δwater ) 23.4 [cal/cm3]1/2.42 In dioxane/ water mixtures, dioxane partitions between polystyrene and water, and its content in polystyrene increases with its overall concentration in the solution. For example, Yu et al. showed that the dioxane content in polystyrene richphase increases from ca. 0.4 (v/v) to ca. 0.65 (v/v) as the dioxane content in solution increases from 83% to 91% (w/w).41 The results of DXR loading in samples and in controls are shown in Figure 2. It is important to recall that all solutions were allowed to stir for 3 days (72 ( 2 h), before the extent of incorporation was determined. In the presence of dioxane, the loading level in the samples is (42) Brandrup, J.; Immergut, E. H. Polymer Handbook, 3rd ed.; John Wiley and Sons: New York, 1989.

Loading and Release of Doxorubicin

Langmuir, Vol. 21, No. 20, 2005 9311

Figure 2. Loading of doxorubicin in PS310-b-PAA36 vesicles as a function of the dioxane content in solution.

higher than that in the controls, reflecting the ability of the pH gradient to enhance loading. At 0% dioxane, however, the sample and the control show a similar degree of loading. This indicates that, in the absence of a plasticizer, the given 3 days are not sufficient for a significant amount of the drug molecules to diffuse and, therefore, for a concentration difference between the sample and the control to develop. Creating a transmembrane pH gradient (in which the inside of the vesicle is more acidic than the external solution) has been used previously in order to enhance the loading of doxorubicin (and other weak bases) into liposomes.32,35,36,43-46 The proposed loading mechanism assumes that the neutral (non protonated) form of the drug is membrane permeable, whereas the charged form is not. At equilibrium, the concentration of the neutral form is equal on both sides of the membrane. The distribution of the protonated molecules, on the other hand, is governed by the pH difference on the opposite sides of the membrane, and can be expressed using the following relation:32,36

[XNH3+]in [XNH3+]out

)

[H3O+]in [H3O+]out

(2)

where [XNH3+] is the concentration of the protonated form of the drug, [H3O+] is the proton concentration, and the subscripts “in” and “out” refer to the inside and outside of the vesicle membrane, respectively. The derivation of eq 2 and a description of the equilibria involved in the loading process are given in the Supporting Information. Figure 2 also shows that the degree of loading is strongly dependent on the dioxane content. It is important to note that such dependence is only a kinetic effect. As the dioxane concentration increases, permeability through the polystyrene wall is improved, and therefore, diffusion of the drug molecules becomes faster. After a given period of time (before equilibrium is reached) and for a given external concentration of doxorubicin, the amount loaded is proportional to the speed of diffusion, and hence, to the (43) Mayer, L. D.; Bally, M. B.; Hope, M. J.; Cullis, P. R. Biochim. Biophys. Acta 1985, 816, 294-302. (44) Haran, G.; Cohen, R.; Bar, L. K.; Barenholz, Y. Biochim. Biophys. Acta 1993, 1151, 201-215. (45) Li, X.; Hirsh, D. J.; Cabral-Lilly, D.; Zirkel, A.; Gruner, S. M.; Janoff, A. S.; Perkins, W. R. Biochim. Biophys. Acta 1998, 1415, 23-40. (46) Mayer, L. D.; Reamer, J.; Bally, M. B. J. Pharm. Sci. 1999, 88, 96-102.

dioxane content. However, when the system is given sufficient time and equilibrium is reached, the degree of loading should be independent of the dioxane content. The increase in the extent of encapsulation with dioxane concentration shown in Figure 2 indicates, therefore, that equilibrium has not been reached after the given 3 days of mixing. However, the higher the dioxane content, the faster is the diffusion, and the closer the system is to equilibrium. Loading of doxorubicin in liposomes was also time dependent, although in general, shorter times were required to reach equilibrium.32,33,35,44 For example, Mayer et al.35 showed that loading of DXR into liposomes increases over time and levels off after ca. 80 min. The liposome wall (2-5 nm) is thinner than that of polymeric vesicles (30 nm in the present system) and is generally more fluid. This leads to faster diffusion through the membrane and explains why in the case of liposomes equilibrium is reached in shorter times. In contrast to the trend observed between a dioxane content of 0 and 48% (w/w), the extent of loading in the samples drops when the dioxane concentration increases from 48 to 60% (w/w). It is possible that the enhanced permeability of the polystyrene wall at such high dioxane contents allows a significant diffusion of protons. The resulting decrease in the pH gradient would lead to a decrease in the extent of incorporation. In the absence of a pH gradient (i.e., for the control solutions), one would expect the amount of DXR accumulated into the vesicles versus that preset in the external aqueous solution to be proportional to the volume ratio between the vesicle interior and the solution in which the vesicles are suspended. For the present system, the ratio between the total cavity volume of vesicles and the external solution does not exceed 1.7 × 10-3 (the details of the calculations are given in the Supporting Information). However, the ratio between the amount of drug loaded into control vesicles and that present in the external aqueous solution ranges between 1.4 × 103 and 4.7 × 102 (mol/mol). It is clear then that enrichment occurs even in the absence of a pH gradient. The mechanism through which the drug molecules accumulate into the vesicles is not clear in this case, however, it might be related to the fact that the concentration of poly(acrylic acid) on the inside of the vesicle is higher than that on the outside. 3.2. Determining the Internal Concentration of Doxorubicin. In the present system, comparing the extent of loading in samples to that in controls shows that establishing a transmembrane pH gradient (inside acidic) enhances the degree of doxorubicin loading. The internal concentration of doxorubicin in the vesicles can be calculated using the amount incorporated (moles of DXR per gram of polymer), and the total internal volume of the vesicles, Total Vint, estimated to be 9.6 × 10-4 L/g of polymer. The internal concentration of the drug, calculated for the data shown in Figure 2, ranges between 0.12 and 0.80 M in the presence of a pH gradient. However, in the control solutions, the internal concentration of doxorubicin varies between 0.14 and 0.060 M (details of the calculations are given in the Supporting Information). Therefore, the pH mediated enhancement in the internal drug concentration ranges between 1- and 10-fold. Refer to Table 2 in the Supporting Information for more details. This enhancement is to be distinguished from the ratio of DXR inside the vesicles to that in the external aqueous solution, which is in the order of 104 (mol/mol) in samples and 103 (mol/mol) in controls. The incorporation of doxorubicin in liposomes using a pH gradient resulted in similar high internal drug concentrations (0.12,44 0.13,36 0.17,40 and up to 0.30 M 45).

9312

Langmuir, Vol. 21, No. 20, 2005

Upon examining the physical state of doxorubicin inside the liposomes, (using a number of techniques, such as absorption spectroscopy, fluorescence, and X-ray diffraction), it was shown that, at such high internal concentrations, the drug molecules do not remain dissolved in solution, but instead, they aggregate inside the liposomes.40,44,45 Li et al. showed that, at internal concentrations between 0.2 and 0.3 M, doxorubicin arranges into bundles of fibers, each of which is comprised of several stacked molecules.45 In the present study, no independent experiments were conducted to examine the physical state of doxorubicin inside the vesicles. Recalling that doxorubicin is water-soluble, and that under acidic conditions similar to those present in the vesicle interior (pH ) 2.5) it is positively charged, the most likely location of the incorporated DXR molecules is the aqueous cavity of the vesicle. However, interactions of DXR with the polystyrene wall or with the poly(acrylic acid) surfaces are also possible and might contribute to the extent of loading. In the following section, the extent of such interactions is estimated using model systems. 3.3. Estimating the Interactions of DXR with Polystyrene and with Poly(acrylic acid). The solubility of doxorubicin in polystyrene was estimated using ethylbenzene as a model solvent. The partitioning of DXR between ethylbenzene and water was determined at pH ) 2.5 and 6.3. The partition coefficient, KEB/H2O, calculated using eq 1, is equal to ca. 0.19 and is independent of the pH. Using the above value of KEB/H2O, along with the definition of eq 1, the number of moles of DXR possibly present in the polystyrene wall of the vesicles (out of the total moles of DXR present in solution) ranges between 7.0 × 10-3 and 2.0 × 10-2 % (mol/mol). The details of the calculations are given in the Supporting Information. Therefore, the contribution of the polystyrene wall towards the extent of loading is insignificant. Interactions of doxorubicin with the poly(acrylic acid) chains covering the internal surface of the vesicle were estimated by determining the amount of DXR bound to homo-poly(acrylic acid) at pH ) 2.5. A similar determination was carried out at pH ) 6.3 in order to evaluate possible interactions between DXR and the acrylic acid chains on the external surface of the vesicle. The results show that out of the total number of moles of doxorubicin present in solution, the number of moles possibly interacting with poly(acrylic acid) is negligible and ranges between 1.8 × 10-3 and 5.4 × 10-3 % (mol/mol) at pH ) 2.5 and 2.0 × 10-1 and 4.0 × 10-1 % (mol/mol) at pH ) 6.3 (refer to the Supporting Information for details). At pH ) 2.5, molecules of doxorubicin are mainly positively charged (pKa of DXR ≈ 8.2), whereas those of poly(acrylic acid) are neutral (pKa of PAA ≈ 5.2); therefore, electrostatic interactions are expected to be weak. The increase in association between the drug and PAA at pH ) 6.3 can be attributed to the fact that under these conditions PAA is above its pKa and negatively charged, whereas doxorubicin is still below its pKa and mainly positively charged. It is important to note that calculations based on model systems might not reflect the exact content of DXR in the different parts of the vesicle; however, they provide a useful estimate of the extent of interaction between the drug and the vesicle wall and surfaces. Based on the above estimates, one can conclude that the aqueous cavity of the vesicle is the main incorporation site. 3.4. Release of Doxorubicin from the Vesicles. The release of DXR from PS310-b-PAA36 vesicles was studied in dioxane/water mixtures of different compositions (dioxane content ) 0%, 25%, and 50% (w/w)). The results, summarized in Figure 3, show that the rate of release

Choucair et al.

Figure 3. Release of DXR from PS310-b-PAA36 vesicles present in dioxane/water mixtures of different ratios: 0%/100% (b), 25%/75% (2), and 50%/50% (9). The straight lines are meant to guide the eye.

increases with the dioxane content. For example, after 200 h, ca. 20% of the loaded DXR is released when the vesicles are present in 100% water, whereas ca. 30% and almost 90% of the drug is released when the vesicle solution contained 25% and 50% (w/w) dioxane, respectively. This shows that the release of DXR can be tuned by adding different amounts of dioxane to the solution. The release of DXR from different types of carriers was previously reported. Mayer et al. followed the release of DXR from egg PC and egg PC/cholesterol (1:1) liposomes, and showed that ca. 50% of DXR was released after 16 and 30 h, respectively.43 Liu et al. showed that the time at which 80% of DXR is released from sulfopropyl dextrans (ion exchange) microspheres ranges from 10 to 120 min, depending on the amount of DXR initially present in the carriers.47 Khopade et al. showed that almost 90% of DXR loaded into poly(styrenesulfonate) and fourth generation poly(amidoamine) microcapsules is released into 0.15 M NaCl solution within 4-5 h, with an initial burst release of ca. 22%.48 For the present system, the release profile does not show a burst release, (i.e., a large amount of drug released initially over a very short time period). Typically burst release occurs when the drug molecules are surface absorbed48 or loosely bound to the delivery vehicle.49 The absence of burst release in the current system is most probably due to the fact that, prior to release measurements, solutions of the PS-b-PAA vesicles were dialyzed extensively to remove unloaded DXR molecules and, possibly, those bound to the vesicles external surface. The release of DXR molecules from vesicles involves their diffusion through the polystyrene wall, which is a unilamellar bilayer, into a large volume of liquid. The diffusion coefficient of doxorubicin through the wall of this reservoir system can be calculated by fitting the release data to the following equation:50,51

M)

DAKC t L

(3)

where M is the amount released (mol) at time t (s), D is (47) Liu, Z.; Cheung, R.; Wu, X. Y.; Ballinger, J. R.; Bendayan, R.; Rauth, A. M. J. Controlled Release 2001, 77, 213-224. (48) Khopade, A. J.; Caruso, F. Biomacromolecules 2002, 3, 11541162. (49) Jones, C.; Burton, M. A.; Gray, B. N. J. Pharm. Pharmacol. 1989, 41, 813-816. (50) Lee, V. H.-L.; Robinson, J. R. In Sustained Controlled Release Drug Delivery Systems; Robinson, J. R., Ed.; Marcel Dekker: New York, 1978; Vol. 6, pp 123-209. (51) Park, K.; Wood, R. W.; Robinson, J. R. In Medical Applications of Controlled Release; Wise, D. L., Ed.; CRC Press Inc.: Boca Raton, FL, 1984; Vol. 1, pp 159-201.

Loading and Release of Doxorubicin

Langmuir, Vol. 21, No. 20, 2005 9313

the diffusion coefficient of the drug through the polystyrene wall (cm2/s), A is the external surface area of the vesicle (cm2), K is the partition coefficient (taken as the ethylbenzene-water partition coefficient), C is the internal concentration of the drug (mol/cm3), and L is the wall thickness (cm). The values of D (calculated from the initial slope of the straight lines obtained by plotting M versus t) increase from ca. 7.8 × 10-17, to 8.7 × 10-17, and then to 6.0 ×10-16 cm2/s as the dioxane content in the solvent mixture increases from 0, to 25, and then to 50% (w/w), respectively. Refer to the Supporting Information for more details. In an aqueous solution of vesicles at room temperature, polystyrene is in a glassy state (Tg of polystyrene ) 100 °C).52 By adding dioxane, which is a plasticizer for polystyrene, the viscosity of the vesicle wall is reduced and diffusion is enhanced. This is reflected by the increase in the value of the diffusion coefficient (and the extent of release) with the dioxane content in the vesicle solution. To the best of our knowledge, these are the first reported diffusion coefficients for DXR through a polystyrene wall of a vesicle or through polystyrene in general. The diffusion coefficient of DXR through albumin-heparin conjugate microspheres was calculated to be ca. 2.5 × 10-10 cm2/s.53 The release from such microspheres is expected to occur more quickly than through PS-b-PAA vesicles since the highly water swollen albumin has a much lower local viscosity than polystyrene. Controlling the rate of release of DXR from model carriers has been investigated by several groups. For DXR loaded into ion exchange microspheres, the rate of drug release was varied by changing the ionic strength of the release medium. The release mechanism in this case occurs mainly via ion exchange between the DXR cations bound to anionic sites of the microspheres and cations (such as Na+ or Ca2+) present in solution. For example, Cremers et al.53 showed that 90% of DXR loaded into albuminheparin microspheres is released in ca. 45 min into an ion-containing release medium. However, over the same time period, only 30% is released into distilled water.53 Sawaya et al. also showed that release of DXR from crosslinked albumin microspheres in water was less than 5% in 24 h.54 However, in the presence of 0.9% NaCl, more than 70% of DXR was released in the same time period. Similar results were reported by Khopade et al.48 and Liu et al. using sulfopropyl dextran microspheres.47 Discher et al. showed that the release of encapsulants from polybutadiene-b-poly(ethylene glycol) (PEO-PBD) vesicles can be controlled by blending a degradable copolymer, such as poly(ethylene glycol)-poly-L-lactic acid, into the membranes of these PEO-PBD polymersomes.55

(DXR). To control the loading and release of the drug, the permeability of the vesicle wall was enhanced by adding increasing amounts of dioxane, which is a plasticizer for polystyrene. An active loading method that involves creating a pH gradient across the vesicle membrane was used to concentrate the drug molecules into the aqueous cavity of the vesicles. The extent of incorporation obtained using this active loading method was compared to that obtained in vesicle solutions in which no pH gradient was established (i.e., pHinside ) pHoutside). The results show that a transmembrane pH gradient of ca. 3.8 units (inside acidic) can enhance the loading level of doxorubicin by up to 10-fold, compared to the extent of loading obtained in the absence of the gradient. The contribution of the polystyrene wall and poly(acrylic acid) surfaces to the extent of loading was estimated using model systems and was found to be only minor, indicating that the vesicle aqueous cavity is the main incorporation site. After loading, the plasticizer can be removed from the vesicle solution (by dialysis), and the loaded vesicles become stable with the drug molecules inside. The release of doxorubicin from the vesicles was also followed. Different amounts of dioxane were added to the vesicle solutions in order to vary the permeability of the vesicle wall and, therefore, tune the extent of release. The diffusion coefficient of DXR through the polystyrene vesicle wall, D, was evaluated from the release profiles. The value of D ranges from 7.8 × 10-17 to 6.0 ×10-16 cm2/s, depending on the amount of plasticizer used. The present system offers the possibility of preparing empty polystyrene-b-poly(acrylic acid) vesicles in pure water, in which polystyrene is glassy and the rate of loading is extremely slow. Subsequently, the vesicle wall can be plasticized to different extents, allowing the incorporation of drug molecules to take place over a range of rates. Once the desired amount of the drug is loaded, the vesicles can be dialyzed against water to remove the plasticizer, thus, hardening the wall and locking the drug inside the carrier. When release is desired, the hard walls of the vesicles can be plasticized once again, and to different extents, allowing the rate of drug release to be tuned.

4. Summary and Conclusions

Supporting Information Available: Details for evaluating the internal drug concentration, the solubility of doxorubicin in polystyrene and poly(acrylic acid) and the diffusion coefficient of doxorubicin through polystyrene are available free of charge via the Internet at http://pubs.acs.org.

Polystyrene310-b-poly(acrylic acid)36 vesicles were used as model carriers for the antitumor drug doxorubicin (52) Zhengcai, P. Polymer Data Handbook; Oxford University Press: New York, 1999. (53) Cremers, H. F. M.; Verrijk, R.; Noteborn, H. P. J. M.; Kwon, G.; Bae, Y. H.; Kim, S. W.; Feijen, J. J. Controlled Release 1994, 29, 143155. (54) Sawaya, A.; Fickat, R.; Benoit, J. P.; Puisieux, F.; Benita, S. J. Microencapsulation 1988, 5, 255-267.

Acknowledgment. We thank Professor C. Allen for useful discussions and for the use of her laboratory during preliminary experiments, as well as the Natural Sciences and Engineering Research Council (NSERC) for financially supporting this work.

LA050710O (55) Ahmed, F.; Discher, D. E. J. Controlled Release 2004, 96, 3753.