DTD 5
ARTICLE IN PRESS
Journal of Controlled Release xx (2004) xxx – xxx www.elsevier.com/locate/jconrel
Sustained interferon-g delivery from a photocrosslinked biodegradable elastomer Frank Gua, Husam M. Younesb,1, Ayman O.S. El-Kadib, Ronald J. Neufelda, Brian G. Amsdena,b,* a Department of Chemical Engineering, Queen’s University, Kingston, ON, Canada K7L 3N6 Faculty of Pharmacy and Pharmaceutical Sciences, University of Alberta, Edmonton, Alberta, Canada T6G 2N8
b
Received 5 July 2004; accepted 16 October 2004
Abstract The application of protein therapeutics for long-term, localized delivery has been hindered by a lack of a delivery device that releases active protein at a concentration within their therapeutic window. A protein delivery system that uses an osmotic pressure delivery mechanism and a photocrosslinked biodegradable elastomer has been designed in an attempt to overcome this limitation. The elastomer is prepared through the UV initiated crosslinking of end terminal acrylated star-poly(q-caprolactoneco-d,l-lactide). Interferon-g (IFN-g) was released from the optimum formulation at a constant rate of 23 ng/day over 21 days. A cell-based assay showed that over 83% of released IFN-g was bioactive. Furthermore, it was demonstrated that bovine serum albumin co-lyophilized with IFN-g was released at the same rate as IFN-g. This delivery formulation may be clinically useful for sustained, local protein drug delivery applications. D 2004 Elsevier B.V. All rights reserved. Keywords: Biodegradable elastomer; Cytokine delivery; Protein stability; Osmotic delivery
1. Introduction Over the past decades, there has been an increasing interest in the development of protein therapeutics.
* Corresponding author. Department of Chemical Engineering, Queen’s University, Kingston, Ontario, Canada K7L 3N6. Tel.: +1 613 533 3093; fax: +1 613 533 6637. E-mail address:
[email protected] (B.G. Amsden). 1 Present address : School of Pharmacy, Memorial University of Newfoundland, St. John’s, Newfoundland, Canada A1B 3V6.
Many of these proteins, e.g., cytokines, require longterm and site-localized delivery. Presently, most protein drugs are administered via multiple injection regimens. Storage stability and low patient compliance have hampered their translation to clinic [1]. There is thus a demand for a more effective means of administering protein drugs. An example of a clinically relevant protein that suffers from these drawbacks is interferon-g (IFN-g). IFN-g possesses a number of stability problems. Firstly, IFN-g contains an acid-labile group (Asp–Pro bond at positions 2
0168-3659/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.jconrel.2004.10.020
COREL-03686; No of Pages 11
ARTICLE IN PRESS 2
F. Gu et al. / Journal of Controlled Release xx (2004) xxx–xxx
and 3) that is broken at pH 2.3 with subsequent loss of activity. Furthermore, recombinant IFN-g, as are many recombinant proteins, is susceptible to aggregation when present in relatively high concentration in solution [2]. Various means of achieving localized delivery of IFN-g have previously been investigated and include the use of liposomes, polymer hydrogels and biodegradable microspheres. These formulation approaches for IFN-g have recently been reviewed [3], and are briefly summarized here. Liposomes have the advantage of easy administration by injection, however, they provide short release durations necessitating additional painful injections, and release rates are neither sustained nor controllable. The release rates of IFN-g from the one hydrogel formulation examined to date are short (less than 7 days) and the presence of a large degree of water in the device may lead to IFN-g stability issues during release. An extensive amount of research has gone into the development of polymeric microspheres that are capable of delivering a virtually constant amount of an encapsulated protein [1,4–6]. These formulations typically consist of biodegradable poly(lactide-co-glycolide) (PLG), throughout which the protein is distributed as discrete solid particles. Protein release is considered to begin with an initial burst, followed by a diffusion-controlled phase, and ultimately a polymer degradation-controlled phase. The initial burst is due to the presence of protein particles exposed at the surface, while the diffusioncontrolled release is a result of dissolved protein diffusing through the water-filled pores and channels within the microspheres. To obtain a constant release rate from PLG microspheres, the diffusion phase must overlap with the polymer erosion release phase such that new pores or channels are created. Polymeric microspheres are easily injected at the target site, provide a long-term release, consist of proven biocompatible materials, and have a reasonable shelf-life. However, there are numerous reports of destabilization of protein drugs during release from these systems (see review of Weert et al. [6]). The factor considered to be of primary importance is the local pH drop within the microspheres due to the production of acidic groups as a result of the polymer hydrolysis. Various techniques have measured the micro-environmental pH to be between 1.5 and 4.7 [6]. For example, Fu et al., using a pH sensitive dye
technique, have reported a pH drop to 1.5 within PLG microspheres after 15 days in phosphate buffered saline [7]. At these pH levels, it is likely that IFN-g will be denatured. Moreover, a recent study has implicated acylation of peptide and proteins by lactic acid and glycolic acid as a cause of inactivation [8]. There has been one report of IFN-g formulated in PLG microspheres [9]. Although release was obtained for greater than 15 days, the activity of the IFN-g after 7 days was only between 30% and 38% of its initial activity. Although continuous localized release of IFN-g is desirable goal, this goal has yet to be realized with the formulation approaches taken to date. A new formulation approach was therefore undertaken. Another method of obtaining a constant release rate of a protein from polymer devices is to distribute the protein as solid particles within a rubbery polymer and rely on either the drug’s or an incorporated excipient’s osmotic activity to drive it from the matrix [10]. This type of delivery proceeds as follows. Upon immersion into an aqueous medium, water vapor diffuses into the elastomer and dissolves the encapsulated solid particles, causing the capsule to swell and exert an osmotic pressure on the surrounding polymer (Fig. 1). At a pressure determined by the strength of the encapsulating elastomer, microcracks are formed and the internal capsule pressure forces the dissolved capsule contents out of the elastomer matrix through the microcrack network. This rupture-and-collapse process proceeds in a layer-by-layer fashion [11]. This
Fig. 1. Schematic of the osmotic pressure release mechanism. Elastomer encapsulated particles (black) draw water vapor (blue) into the elastomer. The water dissolves the solid at the polymer/solid interface, generating a saturated solution encapsulated by the surrounding elastomer (capsule). The imbibition of water into the capsule causes it to swell and exert pressure on the capsule wall. This pressure causes microcracks to form. The dissolved capsule contents are then released from the elastomer via the channels formed by the microcracks in the elastomer.
ARTICLE IN PRESS F. Gu et al. / Journal of Controlled Release xx (2004) xxx–xxx
mechanism has been confirmed by both Schirrer et al. [12] and Riggs et al. [13]. A constant release rate is achieved provided the volumetric loading of solid particles in the matrix is kept below a critical loading called the percolation limit. The percolation limit for slabs is approximately 30% (v/v) [14]. Above this limit, the release becomes non-constant and diffusionally controlled [15]. Presently, however, only nondegradable polymers have been used [10,16–18], which necessitate a further surgical procedure for their removal. It was hypothesized that continuous release of an active protein could be achieved by employing such an osmotic mechanism with a degradable elastomer. Provided that the majority of the protein was released before significant reduction in the mechanical properties of the elastomer occurs due to its hydrolysis, a period of constant release would occur. Further, the use of a slowly hydrolysable copolymer that has also been demonstrated to be biocompatible, such as poly(qcaprolactone-co-d,l-lactide), would reduce or eliminate acidic degradation of a cytokine incorporated within the device. Aggregation of the cytokine within the delivery device could be eliminated or reduced by incorporating the protein as a solid lyophilized with appropriate agents. The lyophilization agents would also serve as the driving force for the osmotic drug delivery mechanism. To facilitate production of protein-loaded elastomeric devices, the crosslinking could be introduced via a photo-initiated terminal end group. The advantages of photocuring are that it occurs rapidly, at room temperature and with minimal heat generation, and so minimal denaturation of the solid protein should occur during device fabrication.
2. Methods 2.1. Materials Recombinant murine IFN-g and its ELISA kit were purchased from Peprotech, Canada. BV-2 cells were provided by Dr. Joaquin Madrenas from the Robarts Institute (University of Western Ontario, Canada). d,l-Lactide and q-caprolactone were purchased from Purac (Canada) and Lancaster (Canada), respectively. All other reagents were purchased from SigmaAldrich (Canada).
3
2.2. Synthesis and characterization of the acrylated star copolymer (ASCP) The procedure followed is outlined in Fig. 2 [19]. A 7800-Da star-poly(q-caprolactone-co-d,l-lactide) was prepared by ring-opening melt polymerization of d,l-lactide and q-caprolactone initiated with glycerol. To make a 7800-Da SCP, glycerol, d,llactide and q-caprolactone were mixed in a flameddried glass ampoule at a molar ratio of 1:10:10 in an oven at 140 8C for 30 min. The mixture was removed from the oven and 1.4104 mol of stannous 2ethylhexanoate per mole of the monomer were added. The ampoule was then flame sealed under vacuum and replaced in the oven at 140 8C for 48 h. Monomer conversion and SCP composition were confirmed using 1HNMR in d6-DMSO using a Bruker Avance300 NMR spectrometer [19]. Thermal measurements of the polymers were carried out using a Seiko Instruments 5200 DSC. The samples were run using a cooling–heating–cooling cycle from ambient to 100 to 100 8C at a heating rate of 10 8C/min. The DSC was calibrated using indium and gallium standards. The glass transition temperature was determined from the inflection points of the second run endotherm using the internal DSC analysis program. A degree of error of F0.5 8C is associated with the reported glass transition temperature. The molecular weight of the SCP was measured using a Waters Alliance GPC system connected to a Wyatt Dawn EOS multi-angle laser light scattering (MALLS) detector. The mobile phase consisted of THF at a flow rate of 1 ml/min with the system at 35 8C. The concentration of the polymer used for the GPC measurements was 30 mg/ ml and the injection volume was 50 Al. The column configuration consisted of an HP Phenogel guard column attached to a Phenogel linear (2) 5 Am GPC column. The SCP was acrylated by reacting acryloyl chloride (ACl) with the terminal hydroxyl groups of the SCP in the presence of triethylamine (TEA) and a catalyst, 4-dimethylaminopyridine (DMAP). Briefly, the SCP was removed from the glass ampoule, and dissolved in anhydrous dichloromethane at a ratio of 1:10 (w/v) in a round-bottom flask. TEA and DMAP were added to the SCP solution at a molar ratio of 1.2 and 103 mol per mole of SCP hydroxyl group. The flask was sealed using a rubber septum and flushed
ARTICLE IN PRESS 4
F. Gu et al. / Journal of Controlled Release xx (2004) xxx–xxx
Fig. 2. Schematic of the synthesis and photocuring of acrylated star-poly(q-caprolactone-co-d,l-lactide) (ASCP). The ring opening polymerization of q-caprolactone and d,l-lactide was catalyzed by stannous-2-ethylhexanoate. The acrylated star copolymer was prepared by reacting the star copolymer with acryloyl chloride. To photocrosslink the material, acrylated polymer was mixed with the photoinitiator, 2,2dimethoxy-2-phenylacetophenone (DMPA), poured into a sealed glass mold of dimensions 10 mm (length) by 5 mm (width) by 1 mm (height), and then crosslinked under UV light (2 mW/cm2) for 2 min.
with nitrogen. ACl was diluted in anhydrous dichloromethane and was added slowly to the polymer solution at room temperature at a ratio of 1.2 mol per mole of SCP hydroxyl group. The reaction was continued at room temperature for 48 h under mild stirring. The final solution was dried at 45 8C under vacuum. The dried polymer was dissolved in ethyl acetate and filtered to remove the triethylamine hydrochloride salt formed during the reaction. The filtrate was dried in a fume hood for three days under air flow. To remove the impurities in the polymer, methanol was added to the ASCP at a volume ratio of 10:1 and the mixture was stirred under mild agitation for 1 h. The mixture was then cooled for 2 h at 20 8C and the methanol was decanted. The extraction process was repeated twice. The purified polymer was dried in the fume hood for several days. 2.3. Elastomer preparation and characterization In a glass scintillation vial, UV initiator solution (0.3% (w/v) of 2,2-dimethoxy-2-phenylacetophenone
dissolved in dichloromethane) was added to a solution of ASCP in anhydrous dichloromethane. The polymer solution was poured into a glass mould (1451 mm) and covered with a glass slide. The sample was then exposed to long-wave UV light at a distance of 1.5 cm at room temperature using an EXFO E3000 high-intensity lamp at a relative intensity of 2 mW/cm 2 for 2 min. A curing optimization study showed 0.015 mg of DMPA per mg of ASCP under UV-light at 2 mW/cm2 for 2 min was sufficient to obtain an elastomer with low sol content (b5 wt.%) [19]. The crosslinked polymer was dried in the fume hood for 2 days and kept in a desiccator under vacuum until required for analysis and characterization. The mechanical properties of the elastomer were carried out on slabs (8042 mm) using an Instron tensile tester model 4443. The crosshead speed was set at 500 mm/min according to ASTM D412. All specimens were tested at room temperature. Data analysis was carried out using a Merlin 4.11 Series IX software package. The sol content of the elastomer was determined, in triplicate, using a Soxhlet apparatus.
ARTICLE IN PRESS F. Gu et al. / Journal of Controlled Release xx (2004) xxx–xxx
2.4. Device preparation The fabrication of the biodegradable elastomer device loaded with IFN-g can be divided into two steps: (1) incorporation of solid protein/excipient particles in a solution of the ASCP and (2) photocuring of the ASCP into an elastomer. Murine recombinant IFN-g was co-lyophilized with trehalose and BSA at various ratios in 5.0 mM succinate buffer. Varying amounts of trehalose and BSA were used, while the IFN-g amount remained constant at 500 ng. The formulation solution was rapidly frozen in liquid nitrogen, and was then lyophilized on a Modulyo D freezer drier (Thermosavant, USA). Both the primary and secondary drying steps during lyophilization were 25 8C and 80 Abar for 48 h. The lyophilized product was ground into powder using a mortar and pestle and sieved through a Tyler 100 sieve to yield particles of less than 145 Am diameter, and mixed with ASCP/ dichloromethane solution. The resulting suspension was mixed using a vortex mixer and photocured in a rectangular glass mold (1051 mm) under the conditions described in the previous section. The volumetric loading of the solids was fixed at 10% of the elastomer volume to ensure that the loading was well below the polymer percolation threshold so that osmotic pressure induced elastomer rupturing was the dominant release mechanism [14]. 2.5. Release study The release study was carried out by immersing the elastomer slabs in 1 ml sterile pH 7.4 phosphate buffered saline containing 0.2% (w/v) sodium azide. Elastomer slabs without excipients were used as controls. The release medium was removed at predetermined time intervals and replaced with fresh buffer. IFN-g concentration in the release medium was measured using an IFN-g ELISA kit. The concentrations of BSA and trehalose in the release medium were measured by a Bradford assay [20] and a phenol sulfuric acid assay [21], respectively. 2.6. Measurement of microenvironmental pH The pH within elastomer slabs undergoing an in vitro degradation in PBS was measured using a pHsensitive dye technique described by Shenderova et al.
5
[22]. This technique relies on the fact that fluorescein no longer fluoresces below pH 5. Shortly after the photocrosslinking reaction, elastomer slabs were swollen at room temperature for 24 h in tetrahydrofuran solution containing 1 mg/mL of fluorescein. The fluorescein-loaded elastomer slabs were dried in vacuo for 3 days. These slabs were then incubated in 0.5 ml pH 7.4 PBS buffer containing 0.1 mg/mL of fluorescein. The fluorescein was incorporated into the degradation medium to reduce fluorescein loss from the slabs due to diffusion. At predetermined time intervals, elastomer slabs containing fluorescein were washed five times with PBS and placed on a glass coverslip. Images were obtained using a Leica TCS SP2 multi-photon confocal microscope. The excitation wavelength and the emission filter were 488 and 515 nm, respectively. Images were collected at 250 Am depth using a 10objective lens with the gain and the black level set at 7.5 and 5, respectively. The controls were generated by imaging a droplet of fluorescein solution on a glass coverslip at known pH values. 2.7. BV-2 cell culture BV-2 cells were cultured in RPMI-1690 media supplemented with 10% fetal bovine serum, penicillin–streptomycin, l-glutamine, sodium pyruvate, HEPES, nonessential amino acids and h-mercaptomethanol. Cells were incubated under standard conditions until they reached greater than 95% confluence. 2.8. IFN-c bioactivity assay The IFN-g activity was quantified by measuring the amount of nitric oxide (NO) produced by the BV-2 cells in response to IFN-g in the culturing medium. Briefly, BV-2 cells were cultured in a 24-well tissue culture plastic plate in serum free media overnight before the activity assay. The release study supernatant was mixed with serum free media, added to the BV-2 cells and incubated for 12 h. The controls were BV-2 cell culture medium injected with the release supernatant from blank elastomer slabs. The culture supernatant was then collected and the NO concentration was measured using a modified Griess assay provided by Sigma-Aldrich. The bioactivity is
ARTICLE IN PRESS 6
F. Gu et al. / Journal of Controlled Release xx (2004) xxx–xxx
reported as a percent of the activity expressed by the cells incubated with the release medium divided by the activity expressed by cells exposed to an equivalent concentration of freshly reconstituted IFN-g.
3. Results The SCP molecular weight and polydispersity index was measured to be 7910 Da and 1.05, respectively. There was no 1HNMR evidence of unreacted monomers in the SCP, and the q-caprolactone/d,l-lactide molar ratio was 52:48. The SCP exhibited only one glass transition temperature, at 19 8C, and no melting endotherms. The glass transition temperature of the ASCP was 14 8C, and that of the resulting elastomer was 6 8C. The elastomer had a sol content of 4F1%, indicative of a high degree of crosslinking. To determine whether IFN-g maintained its bioactivity during photocuring, the trehalose/IFN-g/BSA particles were suspended in a solution of tetrahydrofuran containing DMPA and exposed to UV radiation at an intensity of 10 mW/cm2 for 10 min. It should be noted that the exposure time for the device preparation was only 2 min. The suspension was filtered to collect the particles, which were then dissolved in PBS. This solution was added to a BV-2 cell culture to assess the bioactivity of IFN-g. BV-2 cells are an immortal murine microglial cell line, and secrete nitric oxide when stimulated by IFN-g [23]. In the absence of any protective excipient, the IFN-g was virtually nonbioactive. Although trehalose preserved IFN-g bioactivity during lyophilization, the recovery of IFN-g in the absence of BSA in the particles in the mock photocuring reaction was only 42.8% (Fig. 3). The addition of BSA in the formulation greatly increased the IFN-g recovery in active form, reaching levels statistically equivalent to the lyophilized form. Furthermore, BSA protected the IFN-g during photocrosslinking in a concentration-dependent manner, up to BSA/trehalose ratios of 1:2. Fig. 4a shows the release kinetics of IFN-g from the photocured elastomer as influenced by different BSA to trehalose ratios in the embedded particles. Without trehalose in the particles, IFN-g was released in a declining fashion, at an average rate of 14.1 ng/
Fig. 3. Effect of BSA and trehalose composition on IFN-g stability during lyophilization (clear bars) and during a simulated photocuring process (black bars). Total solids loading concentration=10% v/v. The data represent the mean of triplicate experiments.
day during the first week. In week 2 and week 3, the IFN-g release then dropped from 9.1 to 4.0 ng/day (1.7%/day to 0.8%/day). For the formulations containing trehalose, after the first day, the release became essentially constant, and the rate of release increased as the trehalose content in the embedded particles increased. The lines in Fig. 4a represent linear regressions to the data. For the case where trehalose constituted over half of the particle content (BSA/trehalose=1:1), IFN-g was released at a rate of 16.2 ng/day (R 2=0.994) over a 20-day period. When the trehalose concentration was increased to 2/3 of the particle content (BSA/trehalose of 1:2), the release rate of IFN-g over the same time period increased to 23 ng/day (R 2=0.987). The release profiles of BSA and trehalose from the device are shown in Fig. 4b, superimposed on the IFN-g release. Interestingly, the rates of BSA and trehalose release were essentially identical to the rate of IFN-g release. These results suggest that the mechanism of protein release from the elastomer matrix is controlled by the concentration of trehalose. A high concentration of trehalose accelerates the uptake of water vapor into the elastomer matrix and generates a greater solution osmotic pressure within a capsule (Fig. 1) and thereby increases the protein
ARTICLE IN PRESS F. Gu et al. / Journal of Controlled Release xx (2004) xxx–xxx
Fig. 4. (a) Release of IFN-g from the photocured elastomer prepared using different particle compositions. The solid lines represent linear regressions to the data for the 1:1 and 1:2 formulations, while the dotted line was drawn as a guide. Each point represents the mean of triplicate experiments, and the error bars represent one standard deviation about the mean. (b) The release kinetics of trehalose (triangles), BSA (circles) and IFN-g (squares). The 1:1 data was omitted for clarity. Each point represents the mean of triplicate experiments, and the error bars represent one standard deviation about the mean.
release rate. The total fraction of IFN-g released over the observed time period also increased as the trehalose content increased. In the absence of trehalose, the total fraction of IFN-g released over 25 days was 43% of the initial loading. As the trehalose composition in the elastomer increased, the total IFNg fraction released over 25 days increased from 75% to 94%. The bioactivity of the released IFN-g is shown in Fig. 5 for each formulation. The IFN-g release
7
supernatant was collected and stored at 20 8C until immediately prior to the bioactivity assay. The IFN-g bioactivity reported is normalized to account for the loss of protein activity during storage. The loss of IFN-g during the storage process was determined by measuring the bioactivity of a freshly reconstituted IFN-g solution after freezing and thawing in the same condition as for the samples in the release experiment. There was no influence of embedded particle trehalose content on IFN-g bioactivity. For each formulation, over 92% of the IFN-g released in week 1 remained biologically active. As release continued, the fraction of bioactive IFN-g in the release supernatant dropped to 83% in week 2 and 63% in week 3. The trehalose content in the particles did however influence the total amount of IFN-g released (Fig. 4a). The amount of bioactive IFN-g released from the three BSA/trehalose formulations over the 25-day release period were 38.2%, 60.2% and 83.4%, in order of increasing trehalose content. To confirm that release was driven predominantly by the osmotic mechanism and not elastomer degradation, an in vitro elastomer degradation study was done in sterile phosphate-buffered saline (PBS), and the mass and tensile properties followed with time over a period of 12 weeks (Fig. 6). Both the normalized Young’s modulus and the normalized stress at break were initially unaffected for the first 2 to 4 weeks, and decreased thereafter. The elastomer strain remained relatively constant and the elastomer
Fig. 5. The bioactivity of the IFN-g in the release medium during the release period. Each point represents the mean of triplicate experiments, and the error bars represent one standard deviation about the mean.
ARTICLE IN PRESS 8
F. Gu et al. / Journal of Controlled Release xx (2004) xxx–xxx
Fig. 6. Change in the elastomer normalized mass and normalized mechanical properties including Young’s modulus, stress at break (r b), and strain at break (e b) during in vitro degradation. Each point represents the mean of triplicate experiments, and the error bars represent one standard deviation about the mean.
slab exhibited no appreciable change in dimensions throughout the degradation study. By the end of week 8, the elastomer sample was too weak to undergo
tensile testing. There was no appreciable mass loss during the first 8 weeks of degradation, with a mass loss by week 12 of only 18%. The tensile test and mass loss data show that although the mechanical strength decreases with time, the elastomer still maintains its shape and extensibility over a 12-week time frame. The concentration of IFN-g in the release medium was too low to allow for the measurement of either aggregation or degradation products using standard techniques such as SDS–PAGE or light-scattering chromatography. However, the microenvironmental pH changes during degradation could be estimated using a pH sensitive fluorescent dye (fluorescein) that had been incorporated homogeneously throughout the elastomer. These results are displayed in Fig. 7. The controls are included to aid in the analysis. There is no change in fluorescent intensity within the elastomer during in vitro degradation for the first 15 days. By day 20, numerous small regions of black appear, indicating local areas of pH at or less than 5. By day 25, these regions have grown in size and there are few regions wherein the pH is greater than
Fig. 7. Alteration in elastomer microenvironmental pH during in vitro degradation in phosphate buffered saline, 37 8C. The top three figures show the influence of pH on fluorescein fluorescence intensity in buffered solutions. The bottom seven pictures show the fluorescent intensities of the interior section of the elastomer (depth 250 Am) obtained using a confocal microscope. The length of each of the seven bottom pictures is 0.5 mm.
ARTICLE IN PRESS F. Gu et al. / Journal of Controlled Release xx (2004) xxx–xxx
5, and by day 32, most of the viewable area is at a pH lower than 5.
4. Discussion Inside the swelling, polymer-encased capsule within the delivery device is a micro-environment consisting of a relatively high pressure and a concentrated excipient and protein solution. Proteins are not very sensitive to pressure and only at extremely large pressures will they exhibit denaturation [24]. Thus, the main problem to overcome is protein aggregation. The approach was to incorporate an excipient that would prevent this problem. Good candidates for preventing aggregation include polyols and small, neutral amino acids. Polyols were considered the best prospect because they can generate significant osmotic pressures and are highly effective at preventing protein aggregation. They accomplish this by re-ordering the water around the protein molecule, exerting pressure to reduce the surface contact between the protein and the solvent [25,26]. This pressure forces the hydrophobic sections of the protein to become further removed from the solvent, thus decreasing the likelihood of a hydrophobic–hydrophobic interaction leading to aggregation. Trehalose was chosen as a lead agent to provide the osmotic driving force because previous work on the lyophilization stability of proteins has demonstrated that it was an effective stabilizer [26,27], and it has been demonstrated to prevent denaturation of IFN-g solutions emulsified in dichloromethane [28]. BSA has also been shown to be an effective lyoprotectant for cytokines [27]. This effect has been attributed in part to an inhibition of the pH drop that occurs during lyophilization in a buffer and the inhibition of protein adsorption to surfaces [27]. The lyophilization results indicate that although BSA was effective at preventing denaturation of the IFN-g during lyophilization, trehalose provided more protection and the combination of trehalose and BSA provided the best protection. This may be explained as a result of their different mechanisms of action. The fact that trehalose and BSA are released at the same rate as IFN-g indicates that, not only do they protect the IFN-g during lyophi-
9
lization, but also continue their protective activity during release and in the release medium. This explains the prolonged period of time of high IFN-g activity. Although both excipients provided protection to the IFN-g during the simulated free radical photocrosslinking reaction, BSA was much more effective than trehalose. As neither excipient is appreciably soluble in the tetrahydrofuran solvent, the protective effect must be due to the nature of the arrangement of the molecules in the lyophilized crystals. Both excipients likely block exposure of the IFN-g to the free radicals generated in solution and to the UV radiation. The greater protecting capacity of the BSA may be due to complexation of the BSA to the IFN-g. This hypothesis is the subject of the current investigation. Provided trehalose was present in the particles, a constant release rate was observed, in accordance with both models describing this release mechanism and literature results [11,29]. Furthermore, increasing the trehalose content of the particles produced a faster release rate. This is explained in terms of the much greater osmotic activity of trehalose. The osmotic pressure of BSA in water at 37 8C can be estimated to be 8.6 atm, by extrapolating the data of Vilker et al. [30] to a saturated BSA solution concentration of 0.59 g/ml [31]. The osmotic pressure, P, of a saturated trehalose solution at 37 8C was calculated to be 98.3 atm using, P¼
DHf DTf T V1 Tf Tf 4
ð1Þ
wherein DH f is the latent heat of fusion of water (1436 cal/mol [32]), DT f is the freezing point depression of a saturated trehalose solution (7 8C [33]), T is the temperature of interest, V 1 is the molar volume of water, T f is the normal freezing point of water and T f* is the freezing point of the saturated solution. Eq. (1) was derived from the thermodynamic expressions for freezing point depression and osmotic pressure [34] by relating the activity of the solvent. Thus, increasing the content of trehalose in the embedded particles generated a greater driving force for release, leading to a faster release rate. The finding that the percent active IFN-g in the release medium was essentially independent of the
ARTICLE IN PRESS 10
F. Gu et al. / Journal of Controlled Release xx (2004) xxx–xxx
BSA and trehalose content in the lyophilized particles suggests that the decrease in activity that occurs after 15 days is due to the presence of degradation products of the elastomer. This explanation is supported by the microenvironmental pH analysis and the change in mechanical properties of the elastomer during degradation. Up until 2 weeks, the elastomer did not exhibit appreciable changes in mass or mechanical properties and the pH in the slabs was greater than 5. After this time, although the mass did not change, the mechanical properties began to deteriorate and the interior pH dropped to less than 5, suggesting that significant hydrolysis was taking place. Thus, the decrease in IFN-g activity is likely due to the production of acidic degradation products. The use of a trehalose content in the embedded particles of 67% (w/w) produced a release period that was almost 80% complete after two weeks, explaining the greater amount of bioactive IFN-g released for that formulation. These results represent an improvement over previously published results for IFN-g release from any type of formulation. For example, as noted above, Yang and Cleland [9], only achieved a retention of biological activity of IFN-g of 30–38% after 1 week from a PLGA microsphere formulation. The new formulation approach outlined in this paper can be adjusted relatively easily by altering the content of the more osmotic agent, and by manipulating the elastomer mechanical properties. The ability to release two different proteins of widely different molecular weights and physical properties from the same formulation at the same rate may also prove useful for applications such as tissue regeneration and indicates the potential of this approach to be used for the delivery of therapeutic proteins other than IFN-g.
Acknowledgements The authors thank Dr. Joaquin Madrenas from University of Western Ontario for providing BV-2 cells, and Dr. Gauri Misra, Dr. Stephen Pang and Dr. Yat Tse of Queen’s University for their assistance. This research was supported by grant MOP-53082 to B. Amsden from the Canadian Institutes of Health Research. F. Gu was supported by a CGS scholarship
from the Natural Sciences and Engineering Research Council of Canada.
References [1] V.R. Sinha, A. Trehan, Biodegradable microspheres for protein delivery, J. Control. Release 90 (2003) 261 – 280. [2] S. Gyorffy, K. Palmer, T.J. Podor, M. Hitt, J. Gauldie, Combined treatment of a murine breast cancer model with type 5 adenovirus vectors expressing murine angiostatin and IL-12: a role for combined anti-angiogenesis and immunotherapy, J. Immunol. 166 (10) (2001) 6212 – 6217. [3] H.M. Younes, B. Amsden, Interferon-g therapy: evaluation of routes of administration and delivery systems, J. Pharm. Sci. 91 (1) (2002) 2 – 17. [4] W.R. Gombotz, D.K. Pettit, Biodegradable polymers for protein and peptide drug delivery, Bioconjug. Chem. 6 (1995) 332 – 351. [5] G. Crotts, T.G. Park, Protein delivery from poly(lactic-coglycolic acid) microspheres: release kinetics and stability issues, J. Microencapsul 15 (1998) 699 – 713. [6] M. van de Weert, W.E. Hennink, W. Jiskoot, Protein instability in poly(lactic-co-glycolic acid) microparticles, Pharm. Res. 17 (10) (2000) 1159 – 1167. [7] K. Fu, D.W. Pack, A.M. Klibanov, R. Langer, Visual evidence of acidic environment within degrading poly(lactic-co-glycolic acid) (PLGA) microspheres, Pharm. Res. 17 (1) (2000) 100 – 106. [8] A. Lucke, J. Kiermaier, A. Gopferich, Peptide acylation by poly(alpha-hydroxy esters), Pharm. Res. 19 (2) (2002) 175 – 181. [9] J. Yang, J.L. Cleland, Factors affecting the in vitro release of recombinant human interferon-g from PLGA microspheres, J. Pharm. Sci. 86 (8) (1997) 908 – 914. [10] B. Amsden, Y.-L. Cheng, A generic protein delivery system based on osmotically rupturable monoliths, J. Control. Release 33 (1995) 99 – 105. [11] B. Amsden, A model for osmotic pressure driven release from cylindrical rubbery polymer matrices, J. Control. Release 93 (2003) 249 – 258. [12] R. Schirrer, P. Thepin, G. Torres, Water absorption, swelling, rupture and salt release in salt-silicone rubber compounds, J. Mater. Sci. 27 (1992) 3424 – 3434. [13] P.D. Riggs, P. Kinchesh, M. Braden, M.P. Patel, Nuclear magnetic imaging of an osmotic water uptake and delivery process, Biomaterials 22 (2001) 419 – 427. [14] B.G. Amsden, Y.L. Cheng, Enhanced fraction releasable above percolation threshold from monoliths containing osmotic excipients, J. Control. Release 31 (1) (1994) 21 – 32. [15] B.G. Amsden, Y.-L. Cheng, M.F.A. Goosen, A mechanistic study of the release of osmotic agents from polymeric monoliths, J. Control. Release 30 (1) (1994) 45 – 56. [16] R. Gale, S.K. Chandrasekaran, D. Swanson, J. Wright, Use of osmotically active therapeutic agents in monolithic systems, J. Membr. Sci. 7 (3) (1980) 319 – 331.
ARTICLE IN PRESS F. Gu et al. / Journal of Controlled Release xx (2004) xxx–xxx [17] V. Carelli, G. Di Colo, C. Guerrini, E. Nannipieri, Drug release from silicone elastomer through controlled polymer cracking: an extension to macromolecular drugs, Int. J. Pharm. 50 (3) (1989) 181 – 188. [18] M. Kajihara, T. Sugie, T. Hojo, H. Maeda, A. Sano, K. Fujioka, S. Sugawara, Y. Urabe, Development of a new drug delivery system for protein drugs using silicone (II), J. Control. Release 73 (2001) 279 – 291. [19] B. Amsden, G. Misra, F. Gu, H. Younes, Synthesis and characterization of a photocrosslinked biodegradable elastomer, Biomacromolecules (2004) (in press). [20] M.M. Bradford, A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein–dye binding, Anal. Biochem. 72 (1976) 248 – 254. [21] P. Gerhardt, R.G.E. Murrary, R.N. Costilow, E.W. Nester, W.A. Wood, N.R. Krieg, G.B. Phillips, Manual of Methods for General Bacteriology, Harcourt Press, Washington, DC, 1981. [22] A. Shenderova, T.G. Burke, S.P. Schwendeman, The acidic microclimate in poly(lactide-co-glycolide) microspheres stabilizes camptothecins, Pharm. Res. 16 (2) (1999) 241 – 248. [23] J. Kang, M. Yang, I. Jou, E. Joe, Identification of protein kinase C isoforms involved in interferon-gamma-induced expression of inducible nitric oxide synthase in murine BV2 microglia, Neurosci. Lett. 299 (2001) 205 – 208. [24] V.H.L. Lee, Changing needs in drug delivery in the era of peptide and protein drugs, in: V.H.L. Lee (Ed.), Peptide and Protein Drug Delivery, Marcel Dekker, New York, 1991, pp. 1 – 56.
View publication stats
11
[25] M.A. Hanson, S.K. Edmond Rouan, Introduction to formulation of protein pharmaceuticals, in: T.J. Ahern, M.C. Manning (Eds.), Stability of Protein Pharmaceuticals. Part B, Plenum Press, New York, 1992. [26] C.A.L.S. Colaco, C.J.S. Smith, S. Sen, D.H. Roser, Y. Newman, S. Ring, B.J. Roser, Chemistry of protein stabilization by trehalose, in: J.L. Cleland, R. Langer (Eds.), Formulation of Proteins and Peptides, ACS, Washington, 1994. [27] W. Wang, Lyophilization and development of solid protein pharmaceuticals, Int. J. Pharm. 203 (2000) 1 – 60. [28] J.L. Cleland, A.J.S. Jones, Stable formulations of recombinant human growth hormone amd interferon-g for microencapsulation in biodegradable microspheres, Pharm. Res. 13 (10) (1996) 1464 – 1475. [29] J. Wright, S.K. Chandrasekaran, R. Gale, D. Swanson, Model for the release of osmotically active agents from monolithic polymeric matrices, AIChE Symp. Ser. 77 (206) (1981). [30] V.L. Vilker, C.K. Colton, K.A. Smith, The osmotic pressure of concentrated protein solutions: effect of concentration and pH in saline solutions of Bovine Serum Albumin, J. Coll. Int. Sci. 79 (2) (1981) 548 – 566. [31] W. Saltzman, R. Langer, Transport rates of proteins in porous materials with known microgeometry, Biophys. J. 55 (1) (1989) 163 – 171. [32] R.C. Weast, M.J. Astle, CRC Handbook of Chemistry and Physics, CRC Press, Boca Raton, Florida, 1982. [33] D.P. Miller, J.J. de Pablo, H. Corti, Thermophysical properties of trehalose and its concentrated aqueous solutions, Pharm. Res. 14 (5) (1997) 578 – 590. [34] I.N. Levine, Physical Chemistry, McGraw-Hill Book, Toronto, 2002.
