Application of supercritical fluid to preparation of powders of high-molecular weight drugs for inhalation

Available online at www.sciencedirect.com

Advanced Drug Delivery Reviews 60 (2008) 433 – 446 www.elsevier.com/locate/addr

Application of supercritical fluid to preparation of powders of high-molecular weight drugs for inhalation☆ Hirokazu Okamoto ⁎, Kazumi Danjo Faculty of Pharmacy, Meijo University, 150 Yagotoyama, Tempaku-ku, Nagoya 468-8503, Japan Received 17 September 2006; accepted 21 February 2007 Available online 9 October 2007

Abstract The application of supercritical carbon dioxide to particle design has recently emerged as a promising way to produce powders of macromolecules such as proteins and genes. Recently, an insulin powder for inhalation was approved by authorities in Europe and the USA. Other macromolecules for inhalation therapy will follow. In the 1990s proteins were precipitated with supercritical CO2 from solutions in an organic solvent such as dimethylsulfoxide, which caused significant unfolding of protein. Since 2000, aqueous solutions of proteins and genes have generally been used with a cosolvent such as ethanol to precipitate in CO2. Operating conditions such as temperature, pressure, flow rates, and concentration of ingredients affect the particle size and integrity of proteins or genes. By optimizing these conditions, the precipitation of proteins and genes with supercritical CO2 is a promising way to produce protein and gene particles for inhalation. © 2007 Elsevier B.V. All rights reserved. Keywords: Supercritical carbon dioxide; Dry powder; Protein delivery; Gene therapy; Inhalation therapy; Microspheres

Contents 1. 2.

3.

4.

5.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Drug absorption from the lung . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Features of the lung as a site of drug absorption . . . . . . . . . . . . . . . . . . . . . . 2.2. Particle size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Application of CO2 for preparation of powders . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Principles of particle precipitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Integrity of proteins in pressurized CO2 . . . . . . . . . . . . . . . . . . . . . . . . . . Precipitation of proteins in CO2 from organic solvent . . . . . . . . . . . . . . . . . . . . . . . 4.1. Stability of proteins in DMSO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2. Precipitation of proteins in CO2 from DMSO by GAS processing . . . . . . . . . . . . . 4.3. Precipitation of proteins in CO2 from DMSO by SAS, ASES, PCA, or SEDS processing . Precipitation of proteins in CO2 from aqueous solution . . . . . . . . . . . . . . . . . . . . . . 5.1. Precipitation of proteins in CO2 from aqueous solution . . . . . . . . . . . . . . . . . . 5.2. Improvement of stability of proteins . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3. Other methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

☆

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

This review is part of the Advanced Drug Delivery Reviews theme issue on “Drug Delivery Applications of Supercritical Fluid Technology”. ⁎ Corresponding author. Tel.: +81 52 832 1781; fax: +81 52 834 8090. E-mail address: [email protected] (H. Okamoto).

0169-409X/$ - see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.addr.2007.02.002

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

434 434 434 435 435 435 436 437 437 438 438 439 440 440 441

434

6.

H. Okamoto, K. Danjo / Advanced Drug Delivery Reviews 60 (2008) 433–446

Microspheres and microcapsules of 6.1. Microspheres of proteins . . 6.2. Microcapsules of proteins . 7. Gene powders . . . . . . . . . . . 8. Perspectives . . . . . . . . . . . . References . . . . . . . . . . . . . . .

proteins . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

1. Introduction Macromolecules such as proteins and genes having biological functions in the body are promising therapeutic agents if they can be delivered to target tissue or cells without losing their biological activity. Peroral administration is most convenient for patients; however, the peroral bioavailability of proteins and genes is generally low due to high-molecular weight and susceptibility to enzymes in the gastrointestinal tract. Intravenous, intramuscular, and subcutaneous injections are so far the most practical routes for administration of these macromolecules for systemic therapy. The lungs have been used as a site for drug administration for the local treatment of respiratory diseases, such as bronchial asthma. There have been a number of basic studies in which protein solutions were administered to animals via the lungs. These studies showed that proteins, which are little absorbed from the small intestine, could be absorbed after intratracheal administration [1,2]. Recently, the pulmonary route has attracted attention as a noninvasive route of administering proteins and/or genes for systemic therapy as well as local therapy. For effective inhalation therapy, tiny particles with an aerodynamic diameter of less than 7 μm are required to deposit deep in the lung [3,4]. Pressurized metered-dose inhalers (MDIs), nebulizers, and dry powder inhalers (DPIs) are the three major delivery systems that produce tiny droplets or particles for aerosol inhalation in humans [3]. Among these, DPIs appear to be the most promising for future use because the device is small and relatively inexpensive, no propellants are used, and breath-actuation can be used successfully by many patients with poor MDI technique [3,4]. Proteins and genes can be formulated in micron-sized particles by several methods [5,6]. Milling is a simple method; however, the mean particle size and size distribution are relatively large for use in inhalation therapy. This process tends to denature proteins. Fluid energy grinding can produce 1–10-μm particles; however, the particles tend to be charged electrostatically. The precipitation of proteins from an aqueous solution can be achieved by the addition of an organic solvent to reduce the solubility or an acid or base to move the solution's pH to its isoelectric point. These methods have the problem of residual solvents or salts. Lyophilization is one of the most practical ways to produce protein and gene particles suitable for inhalation. However, it is time consuming and the particles obtained have a broad size distribution. Spray drying is a useful and widely applied method of preparing powders for inhalation. However, it is likely that proteins are susceptible to degradation upon spray drying due to relatively high temperatures [7]. Spray drying of a 5-mg/mL aqueous insulin solution caused significant degradation of insulin at outlet temperatures above 120 °C [8]. β-Galactosidase

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

442 442 442 443 444 444

activity is susceptible to the spray drying temperature and only half of the activity remained after spray drying without additives at an outlet temperature of 50 °C [9]. Fluids at temperatures and pressures above the critical values are known as supercritical fluids (SCFs). SCFs have unique thermophysical properties, i.e. a liquid-like density, large compressibility, and viscosity intermediate between that of a gas and liquid. In addition, the mass transfer rate is higher in a supercritical fluid than in a liquid, and the dielectric constant and solubility can be changed markedly with a small change of pressure around the critical level [10]. Carbon dioxide is the most widely used supercritical solvent, because it is cheap and nontoxic and has an easily accessible critical point (31.1 °C and 73.8 bar). Recently, the precipitation of particles using supercritical carbon dioxide has attracted much attention as an alternative way to prepare protein and gene powders suitable for inhalation. By optimizing the operating conditions, protein and gene powders having a suitable size for inhalation can be produced with little loss of activity and a high yield. 2. Drug absorption from the lung 2.1. Features of the lung as a site of drug absorption The lung is an attractive site for drug absorption because of its wide surface area, thin epithelial membrane, large blood supply, and low levels of enzymatic activity. The total cross sectional area of respiratory bronchioles is about 10 m2 and that of the alveoli in human is more than 100 m2 as large as that of the small intestine [11]. The alveolar epithelium is so thin that drugs in alveoli only have to travel 0.5 to 1.0 μm to enter the blood stream. The total volume of fluid in the human lung is estimated to be approximately 10 mL [12]. A comparison of the pulmonary absorption of several electrolytes from buffered and unbuffered solutions in rats indicated the lung pH at the site of absorption to be about 6.6 [13]. The average weight of the human lung is as little as 0.6 kg; however, the blood flow is as rich as 5700 mL/ min because the lung receives the entire cardiac output. This flow is more than 5 times that of the portal system (1125 mL/ min) including the stomach and small and large intestines [14]. The systemic bioavailability of budesonide for the oral route is 11%, whereas that for the inhaled route is 73%. Fluticason propionate, which has 99% hepatic first-pass metabolism, has no pulmonary first-pass metabolism [15]. Although the metabolic activity of the lung is much weaker than that of the intestinal wall and liver, peptides such as insulin are subjected to enzymatic degradation in the lungs [16,17]. However, avoiding the hepatic first-pass effect by using the pulmonary route would overcome the disadvantage of pulmonary metabolism.

H. Okamoto, K. Danjo / Advanced Drug Delivery Reviews 60 (2008) 433–446

In the early 1970s, Schanker et al. examined the disposition of weak acids and bases from buffered aqueous solutions administered in rat lungs. In general, unionized species were absorbed rapidly from the lungs as in the case of gastrointestinal absorption, suggesting that the pulmonary absorption of small molecules basically obeys the pH-partition theory [18,19]. They reported that inulin, whose molecular weight is 5250 and too large to be absorbed from the gastrointestinal tract, was absorbed from rat lung according to first-order kinetics with a half-life of 225 min. The absorption constants of various saccharides (molecular weights, 122–75,000) including inulin were inversely related to molecular weights and directly related to the diffusion coefficients of the compounds [20]. A study of the transport of dextrans (4 to 150 kDa) across cultured rat alveolar epithelial cell monolayers in vitro suggested that macromolecules with a radius of b 5 μm traverse the alveolar epithelial barrier via paracellular pathways, while macromolecules ≥ 6 μm in radius cross the barrier via other pathways such as pinocytosis [21]. These results suggest that macromolecules can be absorbed from the lung. There are many reports on pulmonary absorption of proteins [22,23]. Clinical studies for inhaled DNase, insulin, interferon α, interferon γ, leuprolide acetate, and α-1-antitrypsin showed virtually no adverse lung reactions [24]. 2.2. Particle size Particle size and distribution affect retention in the lung. The optimum aerodynamic diameter to reach and deposit in the alveolar region is reportedly 1–5 μm [25,26]. The aerodynamic diameter (da) of individual particles can theoretically be related to the mass median particle diameter (dm), measured by light microscopy, by the equation da = (r / F)0.5dm [27]. Where r is particle density and F is the dynamic shape correction factor. The F values for a sphere and cube are 1.00 and 1.08, respectively [27]. This equation predicts that a large light particle

435

may have the same aerodynamic diameter as a small heavy particle. Large porous particles are preferable for inhalation over small nonporous particles. Large porous particles have a smaller surface-to-volume ratio, which results in less aggregation. Another benefit of using large particles is that phagocytic clearance from the lungs is avoided. 93% of the cells in the lavage of a normal adult lung are macrophages, 7% are lymphocytes, and less than 1% are neutrophils, eosinophils, or basophils. Alveolar macrophages interact with particles as well as microorganisms [28]. It has been reported that the clearance of particles larger than 10 μm by pulmonary macrophages is much less extensive than that of smaller particles [29]. Indeed, large porous particles increased the systemic bioavailability of insulin in rats compared to small nonporous particles having the same aerodynamic diameter [30]. 3. Application of CO2 for preparation of powders 3.1. Principles of particle precipitation There are two major principles for particle precipitation with supercritical fluids. In some cases, subcritical fluids at a temperature below the critical temperature are used instead of supercritical fluids. One utilizes SCF as a good solvent for the drugs to be precipitated and the other utilizes SCF as an antisolvent. The Rapid Expansion of Supercritical Solution (RESS) and Particles from Gas Saturated Solutions (PGSS) methods belong to the first category while the Gas Anti-Solvent (GAS), Aerosol Solvent Extraction System (ASES), Supercritical Fluid Antisolvent (SAS), Precipitation with Compressed Antisolvent (PCA), and Solution Enhanced Dispersion by Supercritical Fluids (SEDS) methods belong to the second category (Fig. 1). When a substance is soluble in a SCF, a powder can be obtained by spraying the SCF solution at atmospheric pressure through an adequate nozzle. This is the principle of the RESS

Fig. 1. Schematic diagrams of the apparatus for precipitation of powders with supercritical fluids. (A) RESS or PGSS, (B) GAS, (C) SAES, SAS, or PCA, (D) SEDS, and (E) SAA.

436

H. Okamoto, K. Danjo / Advanced Drug Delivery Reviews 60 (2008) 433–446

technique that utilizes a drastic decrease in the solvating power of the SCF as it expands rapidly into a gas [31]. However, many of the proteins that are candidates for therapeutic agents are water-soluble and have very limited solubility in SCFs. When a substance is insoluble in a SCF, the SCF can be used as an antisolvent to precipitate the substance dissolved in a good solvent. GAS processing introduces the SCF in a protein solution with a good solvent such as dimethylsulfoxide (DMSO). The SCF expands the volume of the good solvent with an increase in pressure to reduce its solvating power and to precipitate protein particles. ASES, SAS, and PCA disperse a protein solution at a relatively low flow rate in the SCF flowing in a vessel. The SCF that penetrates the protein solution droplets decreases the solubility of proteins to precipitate. After the precipitation process, a solvent such as DMSO can be removed by allowing the SCF only to flow. The SEDS technique is based on the same principle. It employs a coaxial nozzle to flow the SCF and protein solution to maximize admixing of the two fluids. Supercritical CO2-assisted aerosolization (SAA) or nebulization with bubble drying utilizes the expansion power of CO2 to disperse aqueous solutions. An aqueous solution and supercritical CO2 are delivered into a low dead volume mixing tee. The resultant emulsion is allowed to disperse from a nozzle equipped with the tee into a drying tube heated by flowing N2 gas. The explosive release of dissolved CO2 from the aqueous solution makes very fine aqueous droplets to be dried [32]. 3.2. Integrity of proteins in pressurized CO2 When a protein solution is processed in compressed CO2 to precipitate, the protein is exposed to a high pressure, normally 20 MPa (200 bar, 197 atm) or less. In general, pressure at this range does not cause irreversible unfolding of proteins. Many enzymes are biologically active in CO2 in this pressure range. Supercritical carbon dioxide has been used as a medium of enzymatic reactions. For instance, cholesterol oxidase, which is nearly insoluble (4.7 μM) in water but 50 times more soluble in CO2 pressurized at 123 bar and 308 K, catalyzes cholesterol in pressurized CO2. The enzymatic activity of cholesterol oxidase isolated from Streptomyces sp. was lost in CO2 at 100 bar and 308 K with a half-life of 45–60 min which was comparable to that in water at the same temperature, suggesting that the loss of activity was due not to the supercritical solvent but to thermal inactivation [33]. Electron paramagnetic resonance spectroscopy revealed that the confirmation of cholesterol oxidase isolated from G. chrysocreas, more stable than that isolated from Streptomyces sp., was not changed in CO2 at 100 bar and 308 K. The addition of a small amount of cosolvent such as methanol and tert-butanol did not affect the confirmation of the enzyme [33]. Thin films of lysozyme and albumin prepared by drying 2% aqueous solutions of them on a CaF2 window were exposed to compressed gas at 40 °C to characterize FT-IR absorbance. When the protein films were exposed to N2 (critical temperature = − 147 °C and critical pressure = 3.39 MPa) at 2 to 12 MPa, no changes in the FT-IR spectra were observed, suggesting that the secondary structure of the proteins was not affected by the pressure at up to 12 MPa. Although compressed

CO2 at 2 and 5 MPa had a minimal effect on the FT-IR spectra of the proteins, CO2 at pressures above 9 MPa affected the spectra depending on the pressure. The changes in spectra suggested a reaction between CO2 and the alcoholic or amine function of side groups of amino acid residues and formation of amide bonds between CO2 and amine groups in amino acid residues. It took about 20 min to obtain a stabilized FT-IR peak below 8–9 MPa. At higher pressures, it took a few hours for equilibration, suggesting that the secondary structural changes proceeded gradually. An important finding was that the FT-IR spectra obtained after reducing the pressure to atmospheric conditions were almost identical to those obtained before pressurization. This suggests that the change in the secondary structure of lysozyme and albumin exposed to CO2 up to 12 MPa was reversible [34]. When aqueous solutions of protein are treated in compressed CO2, conformational changes occur that are either reversible or irreversible. The α-helix content estimated from CD spectra of myoglobin in an aqueous solution bubbled with CO2 at 30 MPa and 35 °C for 30 min was irreversibly reduced from 74% to 12%. The α-helix content was also reduced by lowering the solution's pH to 2 or 3 by adding HCl. This change was reversible because the CD spectrum was restored by raising the pH to 7.0. Irreversible conformational changes in myoglobin occurred also when CO2 was bubbled at atmospheric pressure but at high temperature (70 °C) while the conformational change at 80 °C without exposure to CO2 was reversible [35]. When orange juice containing pectinesterase was treated with CO2 at 31 MPa and 35 °C or 27 MPa and 40 °C, pectinesterase was inactivated rapidly compared to that at atmospheric pressure, especially at higher temperature. The activation energy for inactivation of pectinesterase at atmospheric pressure was 166.6 kJ/mol, which was reduced to 97.4 kJ/mol at 31 MPa, suggesting that pectinesterase was more sensitive to temperature under compressed CO2 and denatured rapidly. Pectinesterase is inactivated at pH 2.4 or below. The pH of orange juice during treatment with CO2 dropped to 2.96 or 3.1. The reduction could not fully explain the inactivation in this case; however, it is likely that the decrease in pH affects the integrity of proteins and genes processed under compressed CO2 [36]. It should be noted that the pH of water in contact with compressed CO2 decreases. When water containing 0.00154% bromophenol blue as an indicator was placed under compressed CO2 at 70–200 atm and 25–70 °C, the light absorbance spectra suggested that pH values were 2.80–2.95. The effect of temperature and pressure on pH was small, although the pH decreased with the increase in pressure and decrease in temperature. The spectra obtained at 175 atm with N2 instead of CO2 was identical to that obtained at atmospheric pressure, suggesting that the pressure change itself had little or no effect on the pH of water [34]. As mentioned in Section 5, some techniques involve spraying aqueous protein solutions into compressed carbon dioxide. Even with a small amount of water in CO2, unfolding of protein may occur due to decreased pH. A conformational change of trypsin has been reported by monitoring the change in the emission center of gravity of the fluorescence of the excited tryptophan residue. The emission spectrum of trypsin in CO2 with

H. Okamoto, K. Danjo / Advanced Drug Delivery Reviews 60 (2008) 433–446

0.2% water pressurized at 50–250 atm at 35 °C red-shifted within 2–3 min after the pressure increase and approached the emission center of denatured trypsin in 8 M urea. When the pressure was reduced from 200 bar to 85 bar, the emission was identical at these two pressures, suggesting that the denaturation of the protein was irreversible in pressurized CO2 in this pressure range [37]. The sensitivity of enzyme activity to compressed CO2 with and without water depends on the protein. Taniguchi et al. examined the activity of nine enzymes treated in supercritical CO2 at 200 atm and 35 °C for 1 h. The activity of the enzymes, α-amylase, glucoamylase, β-galactosidase, glucose oxidase, glucose isomerase, lipase, thermolysin, alcohol dehydrogenase, and catalase, was more than 90% that of untreated enzymes. The addition of 3% ethanol to CO2 had almost no effect except for lipase and alcohol dehydrogenase, the activity of which was lowered by adding ethanol. The addition of 0.1% water to CO2 had minimal effect on the activity of the enzymes [38]. Aqueous solutions of 2.5 and 5% β-lactoglobulin were pressurized with CO2 at a high pressure 450 MPa and 25 °C for 15 min and analyzed by DSC using water as a reference. The residual enthalpy of denaturation (ΔH) was halved compared to that of unprocessed protein, suggesting disruption of the intermolecular hydrogen bonds. When 5% sucrose was added to the 2.5% β-lactoglobulin solution, the temperature of the endothermal peak (Tm) under atmospheric pressure increased compared to that in 2.5% β-lactoglobulin solution with no sucrose, suggesting some protective effect of sucrose against thermal denaturation. Pressure processing in the presence of 0.5– 5% sucrose also reduced the ΔH of the 2.5% β-lactoglobulin solution compared to that at atmospheric pressure. However, the ΔH of the pressurized 2.5% β-lactoglobulin solution with 0.5– 5% sucrose was significantly larger than that with no sucrose, suggesting some protective effect on β-lactoglobulin [39]. As mentioned above, the pH of water in contact with pressurized CO2 is reduced to 2–2.5. This can be applied to the sterilization of the protein product. Dillow et al. treated PLA and PLGA microsphere suspensions contaminated with grampositive bacteria (S. aureus, B. cereus, and L. innocua), and gram-negative bacteria (S. salford, P. vulgaris, L. dunnifii, P. aeruginosa, and E. coli) with pressurized CO2. The sterilization unit was partially depressurized and repressurized 5 cycles an hour to make the pressure difference more than 100 bar. This process enhances the mass transport of CO2 into bacteria. The bacteria were sterilized completely by CO2 treatment at 205 bar and 25–40 °C for 0.6–4 h except B. cereus that required treatment at 60 °C. The sterilization process did not affect microspheres according to SEM observation, DSC, and FT-IR. When the bacteria were treated with supercritical nitrogen at the same pressure and temperature at which CO2 successfully sterilized, no sterilizing effect was observed. The condition is far removed from the critical point of nitrogen (− 147 °C and 33.9 bar) and N2 may not exhibit special gas-like mass transport properties (liquid-like density) that make near-critical fluids ideal for extraction. In addition, dried E. coli cultures were more resistant to pressurized CO2 sterilization, suggesting that the existence of water and reduction of pH within the cell play an important role in the sterilization [40].

437

4. Precipitation of proteins in CO2 from organic solvent Many proteins for use as medicines are hydrophilic and not very soluble in SCFs. In such cases, the SCF can be applied as a poor solvent or an antisolvent. Proteins to be processed should be dissolved at first in a good solvent that is miscible with the supercritical fluid. The solvent should be expandable by the introduction of the supercritical fluid to reduce the protein solubility to precipitate. Because of the immiscibility of water, which is a good solvent for many proteins, with supercritical carbon dioxide, many of the studies appeared before 2000 employed organic solvents that are miscible with supercritical carbon dioxide to dissolve proteins. Thiering et al. surveyed organic solvents capable of dissolving lysozyme, insulin, and albumin. Anhydrous ethandiol, ethylenediamine, and DMSO were the most suitable solvents for protein dissolution. Ethanol is a commonly used organic solvent in the pharmaceutical field; however, the disadvantage of ethanol is that it is a poor solvent for proteins and the yield of precipitates is low with respect to the size of the precipitator or the volume of expanded solution. Glycerol and ethylene glycol are unsuitable for use in GAS precipitation since they are not expanded by common dense gas antisolvents. [41]. When a protein solution in DMSO comes into contact with compressed CO2, the volume of DMSO expands according to the pressure. The expansivity of DMSO and the solubility of vapor-phase CO2 in liquid DMSO at any given pressure decrease with increased temperature [42]. It was experimentally determined that dissolving proteins in DMSO, such as 1 mg/mL trypsin, 1 mg/mL ribonuclease, 4 mg/mL lysozyme, 3.4 mg/mL alkaline phosphatase, and 2.5 mg/mL insulin, had no significant effect on the expansivity of DMSO. The pressure at which nucleation of solid-phase precipitates begins depends on the protein dissolved in DMSO and its initial concentration. A protein solution in DMSO with a higher initial concentration precipitates at a lower CO2 pressure. There is a linear relationship between the logarithm of the initial concentration and the pressure at which a protein of given initial concentration precipitates [42]. For instance, the solubility of lysozyme in DMSO decreases rapidly from N6 mg/mL to b0.5 mg/mL and catastrophic precipitation of lysozyme occurs on the volumetric expansion of DMSO of approximately 100% when CO2 is introduced at 4.8 MPa [43]. 4.1. Stability of proteins in DMSO Dimethylsulfoxide is the organic solvent that has been most commonly used as a good solvent of proteins for precipitation in SCFs. However, it disrupts protein conformation and remains as a trace contaminant. Fourier transform infrared spectroscopy showed that significant perturbations of the secondary structure of myoglobin and concanavalin A are induced by DMSO and DMSO/2H2O mixtures. Intermolecular β-sheet formation and aggregation are induced until moderate DMSO concentrations (around 0.33-mol fraction) because of the disruption of the interaction between intramolecular peptide groups by DMSO (partial unfolding). At higher DMSO concentrations (above

438

H. Okamoto, K. Danjo / Advanced Drug Delivery Reviews 60 (2008) 433–446

0.75-mol fraction), such aggregates are dissociated by disruption of the intermolecular C_O…2H–N deuterium bonds. The proteins are completely unfolded, lacking any secondary structure in pure DMSO. Even at low DMSO concentrations showing no detectable effect on the gross secondary structure of myoglobin and concanavalin A, the thermal stability of both proteins was markedly reduced [44]. Lysozyme and trypsin were also unfolded in DMSO [5]. However, as mentioned below, some proteins unfolded in DMSO and precipitated powders may refold when the powders are rehydrated. 4.2. Precipitation of proteins in CO2 from DMSO by GAS processing According to reports available so far, GAS-operating at a low temperature with a high concentration of protein in solution is preferable to obtain tiny discrete particles with good biological activity. Muhrer and Mazzotti examined the effects of CO2 addition rate (3–50 g/min), initial solute concentration (2–8 mg/mL), and temperature (19, 25, and 35 °C) on the size of lysozyme particles GAS-precipitated from DMSO solutions. A 50-mL volume of lysozyme solution in a 400-mL precipitator was precipitated by introducing CO2. After full expansion the feed was shut-off, stirring was continued for 1 h, and the precipitates were rinsed with CO2 flowing at 20–30 g/min for at least 5 h. However, the obtained particles exhibited a residual DMSO content of a few percent in mass. The starting material was partly crystalline, whereas the GAS-processed protein powder was amorphous. Typically, the processed lysozyme had around 75% biological activity relative to the starting material. Although the mean particle size was 200–300 nm, the calculated volume-average diameter was typically 20–30 μm due to the presence of some larger agglomerates. No significant effect of CO2 addition rate on mean particle size or particle size distribution was observed. The lysozyme concentration exhibited no major effect on average particle size; however, there seemed to be a tendency for particle size to decrease according to the increase in lysozyme concentration from 2 to 5 mg/mL. The mean particle size seemed to decrease slightly with temperature. However, a clear effect of temperature was the increase in the degree of agglomeration of the articles [45]. Thiering et al. also examined the GAS precipitation of lysozyme from a DMSO solution and reported that agglomeration of the precipitate occurred with a low concentration of lysozyme in DMSO. A high protein concentration close to the gelation limit led to the formation of large particles with a porous structure. Precipitation from low concentration solutions requires high solution expansion. The agglomeration of lysozyme particles at high DMSO expansion was also observed during SEDS processing as mentioned in the next chapter [46]. The effect of operating temperature (18–45 °C) on the particle size of the precipitated lysozyme was minimal. On the other hand, the biological activity of GAS-precipitated lysozyme depended on the operating temperature. The particles precipitated at 25, 35, and 45 °C had 100%, 92%, and 60% activity compared to the unprocessed lysozyme [43].

The particle size of insulin precipitated from a DMSO solution by GAS processing at a high temperature (50 °C) increased, which was attributed to the thermal instability of insulin. Operating at a low temperature has the advantage of producing discrete particles without denaturation [43]. Significant changes to protein size or shape were achieved by modification of the solvent system. Typically the stronger the protein solvent, the larger the precipitate. For instance, the particle size of insulin obtained from a DMSO solution by GAS process was 1.4– 1.8 μm, while that precipitated from an ethanol solution was 0.05–0.3 μm [41]. 4.3. Precipitation of proteins in CO2 from DMSO by SAS, ASES, PCA, or SEDS processing Carbon dioxide is continuously provided in SAS, ASES, PCA, or SEDS processing. The faster flow of CO2 seems to be preferable to reduce particle size and the residual DMSO concentration. The residual DMSO level seems to be lower than that of GAS-processed particles because of the rapid removal of DMSO by the flowing CO2. Operation at a lower temperature may minimize the conformational change during processing. The first report on the preparation of microparticles of proteins using the SAS process appeared in 1993. Yeo et al. dissolved insulin in DMSO at 5 or 15 mg/mL to spray at 0.3 mL/ min into carbon dioxide flowing at 8.0 standard L/min in a 75mL precipitator kept at 86.2 bar. Thereafter, pure CO2 continued to flow for 2 h to remove any residual organic solvent. Operation at a higher temperature was preferable, both with respect to the shorter drying time, as well as with respect to the degree of removal of residual solvent from the particles. Spherical and spheroidal particles were continuously obtained. It was found that the morphology of the resulting powders was relatively insensitive to the temperatures and concentrations explored. Higher protein concentrations led to lower threshold pressures. A low CO2 injection rate (0.57 bar/min) resulted in the growth of larger particles. This suggests that the slow expansion of DMSO increases particle size and the pressure can control particle size [6]. In every case, 90% of the insulin particles formed were smaller than 4 μm. Blood glucose levels in rats after intravenous administration of a rehydrated insulin solution were examined. In every case, the processed insulin's activity was indistinguishable from that of the original material, irrespective of processing conditions [6]. Insulin exhibits two strong Raman band envelopes in the 1500–1800 cm− 1 region. The large shift in the amide I mode is indicative of a significant perturbation in secondary structure. The SAS-precipitated insulins generated at 25 and 35 °C gave amide I band spectra that were shifted roughly + 10 cm− 1 relative to the commercial powder. The corresponding secondary structural estimates had significantly increased β-sheet contents with concomitant decreases in α-helix contents relative to the commercial protein. However, the spectra of the commercial powder and the SAS precipitates dissolved in 0.01 M HCl were almost identical. This similarity reflects the reversible nature of the SAS precipitation-induced structural changes for insulin under these conditions [47].

H. Okamoto, K. Danjo / Advanced Drug Delivery Reviews 60 (2008) 433–446

Winters et al. examined the effect of SAS precipitation processing on the secondary structure of lysozyme and trypsin as well as insulin by using amide I band Raman spectroscopy. The protein solutions in DMSO were sprayed at 0.5–2.4 mL/min into CO2 flowing at 5–26 L/min using a 75-mL pressurized view cell operated at 26.6–46.5 °C and 73.4–142 bar. Exposure to DMSO alone extensively perturbed the secondary structure of lysozyme and trypsin. The SAS-precipitated proteins were greatly distorted from their native, commercial form. The perturbation in secondary structure suggests that the most significant event during supercritical fluid-induced precipitation involved the formation of β-sheet structures with concomitant decreases of αhelix. Amide I band Raman and FT-IR spectra indicated that higher operating temperatures and pressures led to more extensive β-sheet-mediated intermolecular interactions in the precipitates. Unlike the insulin SAS precipitates, lysozyme particles refolded during SAS precipitation almost to their starting conformation, even after the severe structural distortion caused by dissolution in DMSO. Upon hydration with 0.01 M HCl, SAS-precipitated insulin refolded to its native form. αHelix contents for redissolved SAS-precipitated lysozyme and trypsin were very similar to the values given for native structures form X-ray crystal data. The structural distortion of lysozyme SAS precipitates increased with the increase in operating temperature. Upon redissolution in water, SAS-precipitated lysozyme recovered between 88 and 100% of its biological activity relative to commercial lysozyme. Trypsin recovered between 69 and 94% of its biological activity [5]. When these SAS-processed protein microparticles were stored in sealed containers at − 25, − 15, 0, 3, 20, 22, and 60 °C, the perturbed secondary structure of the protein particles remained constant in time, regardless of the storage temperature. The recoverable biological activity upon reconstitution for the SAS-processed lysozyme and trypsin microparticles was also preserved and found to be independent of storage temperature [48]. The effects of the SEDS operating parameters such as pressure (80, 115, and 150 bar), temperature (40, 45, and 50 °C), flow rates of CO2 (18, 24, and 30 mL/min) and lysozyme solution in DMSO (0.2, 0.4, and 0.6 mL/min), lysozyme concentration (0.5, 0.75, and 1.0 w/v%), and the diameter of the opening of the nozzle (0.20 and 0.30 mm) on the biological activity and physicochemical properties of lysozyme particles were examined. The particles were partially bridged at 150 bar at 40 °C, while individual particles were obtained at 80 bar. The high volumetric expansion of DMSO at a high pressure may have resulted in too fast precipitation for the particles to precipitate individually. By the SEDS process, spherical lysozyme particles with diameters of 1–5 μm were obtained. No conformational changes in the precipitated lysozyme particles were detected by high-performance cation-exchange chromatography, high-sensitivity DSC, or X-ray powder diffraction. The typical residual amount of DMSO was less than 20 ppm. The biological activity of the processed lysozyme depended on the processing variables. A high level of biological activity close to 100% of that of the unprocessed lysozyme was obtained at 150 bar and the lowest level of activity was less than 50% of that of the initial material resulted at several conditions

439

operated at 80 bar. At 80 bar, the expansion of DMSO dispersed in CO2 is not sufficient and particles will precipitate in a DMSOrich phase compared to those precipitated at 150 bar. Concentrations greater than 2% of lysozyme in DMSO led to an increased risk of line blockage [46]. To improve the SAS process to prepare particles up to one order of magnitude smaller with a narrow size distribution, supercritical antisolvent with enhanced mass transfer (SAS-EM), a process that uses a surface vibrating at an ultrasonic frequency to atomize the jet into microdroplets, has been proposed [49]. The ultrasound field generated by the vibrating surface enhances turbulence and mixing within the supercritical phase, resulting in a high mass transfer between the solution and the antisolvent. The combined effect of a fast rate of mixing between the antisolvent and the solution, and reduction of solution droplet size due to atomization, provides particles up to ten-fold smaller than those obtained by the conventional SAS process. When a 5-mg/mL solution of lysozyme in DMSO was dispersed in CO2 at 96.5 bar and 37 °C from a capillary tube placed at an angle of 40° touching the ultrasound horn surface operated at 20 kHz, the mean particle size decreased to a minimum (190 nm) at the horn amplitude corresponding to a 90-W power supply. Increasing the power supply beyond this point tended to increase particle size by a small amount due to agglomeration. The lysozyme particles obtained by the SAS-EM technique at an ultrasound horn amplitude corresponding to a 60-W power supply retained about 87% of biological activity [49]. Whereas DMSO tends to increase the random structure of proteins, halogenated alcohols are helix formers, increasing the amount of ordered structure in proteins. In addition, halogenated alcohols are much more volatile than DMSO, and hence should leave little residual solvent in a PCA precipitate. Insulin dissolves in 1,1,1,3,3,3-hexafluoro-2-propanol (HFIP) at more than 38 mg/mL. Although HFIP affected the CD spectra of insulin dissolved in it compared to insulin in phosphate buffer, the PCA-precipitated insulin and unprocessed insulin in 5 mM phosphate buffer had a similar secondary to quaternary structure. The precipitated powder consisted of physical aggregates of 50nm spheres. Through deagglomeration of these aggregates, it may be possible to obtain discrete uniform particles (1–5 μm) suitable for pulmonary therapy [50]. 5. Precipitation of proteins in CO2 from aqueous solution Exposure of a protein to DMSO results in significant perturbation of the secondary structure. Furthermore, high boiling DMSO is less volatile and special attention has to be paid to a careful operating procedure allowing the residual solvent in the precipitates to be removed to a satisfactory degree. From this viewpoint, precipitation from an aqueous protein solution should be preferable. Although water is immiscible with CO2, a limited volume of water is miscible with compressed CO2 modified with organic solvents such as ethanol and isopropyl alcohol. α-Chymotrypsin was precipitated by the PCA process with ethanol. An acidic aqueous solution at pH 2.04 was used to minimize self-proteolysis. PCA processing was operated at subcritical conditions (25 °C and 68.9 bar) with a CO2:ethanol:

440

H. Okamoto, K. Danjo / Advanced Drug Delivery Reviews 60 (2008) 433–446

water molar ratio of 12:7:1 or supercritical conditions (36 °C and 136 bar) with a CO2:ethanol:water molar ratio of 14:5:1. The primary particles were spherical and approximately 0.2 μm in size; however, highly agglomerated particles of 250–400 μm were obtained. The protease activity of α-chymotrypsin precipitated at 36 °C was 59% of that of the unprocessed protein. This activity was greater than that of α-chymotrypsin precipitated by spraying DMSO solution into supercritical CO2 (6.5%) [51]. Even with ethanol in CO2, the volume of water that makes a single phase with CO2 and ethanol is very limited [52], which limits the production performance. GAS processing is inadequate for the precipitation of proteins from water. A recombinant immunoglobulin G1 (rIgG) can be precipitated by the SEDS process with CO2 modified by the addition of 0.44 v/v% ethanol. Without ethanol, hardly any dry powders were obtained because the walls of the vessel and the particle collection filter were wet after the SEDS processing [53]. The modifier is usually added to CO2. Recombinant human growth hormone can be precipitated from its 1–2% aqueous solution by SEDS processing with CO2 modified with isopropanol at 100– 200 bar and 40–50 °C. When isopropanol was added into the aqueous protein solution at 30%, the precipitation process in CO2 without isopropanol was incomplete [54]. Admixing the cosolvent with an aqueous solution is not recommended because only a limited amount of the cosolvent is available. 5.1. Precipitation of proteins in CO2 from aqueous solution To obtain protein particles from an aqueous solution adequate for inhalation, a lower temperature and higher protein concentration seem to be preferable as in the case of precipitation from DMSO solutions. A slower flow of aqueous solution may result in high yield and strong biological activity. When a 10–20-mg/mL aqueous solution (0.4 mL/min) of rhDNase was ASES-processed in CO2 (12 mL/min) modified with 0.2-mol fractions of ethanol at 155 bar, the rhDNase in powder obtained at 45 °C was totally aggregated and no monomer was detected. By reducing the temperature to 20 °C, about 40% monomer was recovered [55]. Bustami et al. sprayed an aqueous solution of protein, lysozyme, insulin, rhDNase, or albumin, through the inner tube of a coaxial nozzle into pressurized CO2 flowing from the outer tube. The mole fraction of ethanol in CO2 was 0.2 and the flow ratio of the aqueous solution to the antisolvent was set to 0.4/12. All proteins precipitated as spherical particles of 100 to 500 nm, which tended to agglomerate during the ASES process. Higher temperature was favorable to precipitate smaller particles with less agglomeration. At higher temperature, mass transfer of the antisolvent into the water droplet was enhanced. The solubility of water in CO2 increases at high temperature. They proposed that water is extracted more efficiently at a water/ethanol/CO2 composition close to the critical point (plate point). When a low concentration (5 mg/mL) of lysozyme solution was operated at 45 °C and 155 bar, only agglomerated particles were obtained at a low yield because supersaturation was not attained immediately. Higher concentrations of the aqueous solution resulted in smaller particles with less agglomeration. Lysozyme processed

at 35 and 45 °C retained all biological activity. rhDNase processed at 45 °C was denatured to leave no monomer. Lowering the temperature to 20 °C improved the integrity; however still two-thirds of all the protein was denatured. The operating temperature and the acidic environment in the ASES process affected the integrity of proteins such as rhDNase [56]. The optimization of SEDS conditions for lysozyme particle precipitation was examined with a pressure of 100, 150, and 200 bar, temperature of 40, 45, and 50 °C, ethanol flow rate of 1.5, 1.75, and 2 mL/min, and aqueous solution flow rate ranging 0.1, 0.15, and 0.2 mL/min. The flow rate of CO2 and lysozyme concentration were fixed to 30 mL/min and 3 w/v%, respectively. Operating the system at the highest water flow rate failed to give a homogeneous single phase and no particles were obtained. The biological activity of the precipitates obtained was 63–101% of the unprocessed protein. Particles obtained at the lowest temperature tended to show strong activity. Another important factor in obtaining particles with high level of biological activity was the ratio of the flow rates. A slower flow of the aqueous solution resulted in precipitates with strong activity. The mean particle size was 1–5 μm. FT-Raman spectroscopy suggested perturbation of the secondary structure of the solid samples. The lysozyme particles obtained at 200 bar and 40 °C had a small mean diameter and narrow particle size distribution, the least secondary structural change, and the greatest biological activity [57]. A recombinant immunoglobulin G1 (rIgG) was precipitated by the SEDS process at 175 bar and 45 °C with CO2 modified by the addition of 0.44 v/v% ethanol. The aqueous solution was composed of 5 mg/mL rIgG, 20 mM sodium citrate, 15 mM NaCl, 6.2% sucrose, and 0.02% polysorbate-80 and the pH was adjusted to 6.0. The flow rates of CO2, ethanol, and the aqueous protein solution were 9.0, 0.04, and 0.003 mL/min. Extremely deliquescent powder was obtained at a yield of 40–50%. The antigen-binding activity was less than half that of the unprocessed protein [53]. Todo et al. examined in rats the hypoglycemic activity of insulin powder produced from DMSO and aqueous solutions. They tried four different types of nozzles (Fig. 2). By dispersing an insulin/mannitol (1:20 by weight) solution in DMSO from a parallel nozzle, powder with an aerodynamic diameter of 6.4 μm was obtained at a yield of 89%. However, the other nozzles were unsuitable for the DMSO solution because the tips were clogged. On the other hand, an insulin/citric acid/mannitol (1:0.8:20) powder with an aerodynamic diameter of 3.2 μm was obtained with all except the parallel nozzle. It seemed that the mixing of water (0.035 mL/min) with the CO2 (14 mL/min)/ ethanol (0.665 mL/min) admixture was incomplete using the parallel nozzle. Intratracheal administration of the insulin powders successfully showed a hypoglycemic effect for 6 h with a rapid onset [58]. 5.2. Improvement of stability of proteins As mentioned earlier, the pH of water in contact with compressed CO2 decreases, which may affect the secondary structure of protein. The decrease in pH of water can be

H. Okamoto, K. Danjo / Advanced Drug Delivery Reviews 60 (2008) 433–446

441

more labile to SEDS processing. The undissolved protein was increased to more than 35% by SEDS processing without sugar. The addition of sucrose and trehalose partially decreased the amount of undissolved protein to 12.8% and 8.4%, respectively. Freeze-drying of myoglobin without sugar slightly increased the amount of undissolved protein (2.5%). The addition of sugars suppressed the increase in undissolved protein. The UV/VIS spectroscopy and CD spectra suggested SEDS-processed myoglobin, especially with trehalose, had protein–heme interaction and the ternary structure was significantly different from the unprocessed myoglobin. Although the sugars were effective at maintaining protein stability during SEDS processing, optimization should be required [61]. Using the SEDS process at 100 bar and 50 °C with a 1% rhGH aqueous solution with 5% sucrose, 1– 6-μm spherical amorphous particles were obtained. The flow rates of CO2, modifier (isopropanol), and the aqueous solution were 25, 1, and 0.1 mL/min, respectively, and 91.6% of the soluble protein was present in monomeric form [54]. Fig. 2. Schematic diagram of the nozzles used in supercritical antisolvent method. (A) V type, (B) U type, (C) parallel type, and (D) two-channeled coaxial nozzle.

suppressed by the addition of a buffer species. In the course of supercritical CO2-assisted aerosolization mentioned below, the addition of 100 mM phosphate buffer (pH 7.5) was more effective at suppressing the drop in pH than the same concentration of TRIS buffer (pH 8.0), citrate buffer (pH 6.0) and acetate buffer (pH 5.5) [32]. The integrity of the rhDNase precipitated with CO2 from an aqueous solution was lowered due to the low pH of the aqueous solution in contact with CO2. On the addition of 0.0012-mol fraction of triethylamine (TEA) in CO2 modified with 0.2-mol fraction of ethanol, the integrity of rhDNase was drastically improved by the suppression of the decrease in pH during ASES processing. Without TEA, more than 50% of the protein precipitated at 20 °C was recovered as insoluble aggregates, compared to just 2% with TEA. TEA is not classified as a Class I solvent that should be avoided in pharmaceutical processing. The residual level of TEA in the precipitated powder was less than 10 ppm [55]. It has been reported that the addition of a sugar is effective at preserving the native protein confirmation during freeze-drying process [59]. Notably, amorphous sugar matrices embedding protein molecules prevent conformational change during drying and storage [60]. Recently, Jovanovic et al. examined the effect of sucrose and trehalose on the stability and characteristics of SEDSprocessed lysozyme and myoglobin. They compared supercritical CO2 drying and freeze-drying using the same protein solutions. A 0.1-mg/mL protein solution with a 10-mg/mL sugar or a 2-mg/mL protein solution without sugar was dispersed from the inner tube of a coaxial nozzle into CO2 modified with ethanol at 37 °C and 100 bar. The flow rates of aqueous solution, CO2, and ethanol were 0.5 mL/min, 250 g/mL, and 25 mL/min, respectively. The biological activity of lysozyme was fully preserved by the SEDS processing and freeze-drying. However, undissolved protein was increased by SEDS processing with and without sugars (4–6%). Freeze-drying gave no undissolved lysozyme. Myoglobin was

5.3. Other methods Supercritical CO2-assisted aerosolization or nebulization with bubble drying utilizes the expansion power of CO2 to disperse aqueous solutions. An aqueous solution and supercritical CO2 are delivered into a low dead volume mixing tee. The resultant emulsion is allowed to disperse from a nozzle equipped with the tee into a drying tube the temperature of which is kept at below 70 °C by the flow of heated N2 gas. The explosive release of dissolved CO2 from the aqueous solution makes very fine aqueous droplets to be dried [32]. Although the biological activity of the reconstituted lysozyme powder was more than 90%, the secondary structure of lysozyme in the dry powder obtained by this process with phosphate-buffered solution was found to be perturbed. Mannitol that formed crystalline powders partially improved the integrity of lysozyme and sucrose that gave amorphous powders successfully suppressed the conformational change. Lactate dehydrogenase (LDH) is a more labile protein and the biological activity of LDH in the dry powder obtained by this process with phosphate-buffered solution was only 15% of the unprocessed LDH. Interestingly, the activity found for the aerosolized solution without drying had 87% activity, suggesting that the activity was lost during the drying process rather than the nebulization process. The addition of 10% mannitol and sucrose to the aqueous solution increased the biological activity of the powders to 40% and 70%, respectively. The combination of 10% sucrose and 0.01% Tween 20 was most effective at reducing activity loss and more than 90% of the activity was retained in the resulting powder [32,62]. The solubility of a protein decreases at its isoelectric point. Mineral acids such as sulfuric acid and hydrochloric acid are used to precipitate proteins under acidic conditions; however, neutralization should be done afterwards, which results in residual salts in the product. Carbon dioxide, even at a low pressure, can be used as a volatile electrolyte, instead of an antisolvent of the proteins. Depressurization can easily remove CO2 to obtain a product without residual salt. By placing a soy meal aqueous extract (pH 9.0) in contact with pressurized CO2,

442

H. Okamoto, K. Danjo / Advanced Drug Delivery Reviews 60 (2008) 433–446

soy proteins such as glycinin and β-conglycin with an isoelectric point of 4.8 can be precipitated. The higher the protein extract concentration, the higher the pH of the solution in contact with CO2 at a given pressure. The equilibrium pHs of 4, 20, and 40 g/ L extracts in contact with CO2 at 55 bar were 4.05, 4.55, and 4.75 at 25 °C. To reduce the pH of a concentrated extract, a higher pressure was required. By reducing the extract's pH to 4.8–5.0, about 75% protein was precipitated [63]. 6. Microspheres and microcapsules of proteins Microspheres and microcapsules of proteins and genes with biodegradable polymers are promising formulations for optimizing inhalation therapy. These microparticles may provide new functions such as the sustained release of proteins, protection of proteins against enzymatic degradation, increased retention due to bioadhesive properties, and so on. Microspheres have been prepared by emulsion-solvent extraction, spray drying, and phaseseparation techniques [64]. The application of supercritical CO2 to microencapsulation of proteins is a promising way to prepare microspheres with low residual solvent and strong protein activity. 6.1. Microspheres of proteins Biodegradable polymers such as poly(L-lactic acid) (L-PLA) and poly(DL-lactide-co-glycolide) (PLGA) are soluble in dichloromethane. However, the solubility of proteins in dichloromethane is low. To prepare microspheres with protein using compressed CO2, three methods have been proposed: spraying a protein suspension in dichloromethane with polymer, spraying a protein solution in DMSO and a polymer solution separately, and spraying a protein solution in a dichloromethane/DMSO admixture with a polymer. It should be noted that CO2 lowers the glass transition temperature of polymers [65]. Operating at lower temperature and higher pressure is preferable to obtain particles without agglomeration; however, too low a temperature may result in a high level of residual solvent. A faster CO2 flow is preferable to reduce residual solvent [66]. Lysozyme was formulated in PLGA microspheres by spraying lysozyme suspension with PLGA in dichloromethane into CO2 vapor phase equilibrated with liquid CO2 phase. Because the glass transition temperature of PLGA, 45–50 °C, was decreased to −40 °C or below, the operation at 0 °C or above caused extensive agglomeration of PLGA particles. Spraying 0.5% lysozyme suspension with 5% PLGA into static CO2 successfully produced 5–60-μm PLGA particles containing lysozyme. A low PLGA concentration such as 1% resulted in small particles 0.5–5 μm in diameter, which was too small to entrap solid lysozyme. When CO2 was flowed from the annular region in a coaxial nozzle, the increased mixing caused a higher degree of supersaturation, more nucleation, faster solvent removal and quenching, and less time for growth, which resulted in far less agglomeration compared to the case with static CO2 [64]. Dichloromethane has been used as a solvent for L-PLA; however, the solubility of protein in it is low. To precipitate microspheres from an L-PLA solution in a dichloromethane and protein solution in DMSO, a multiple coaxial nozzle was used.

A lysozyme solution, a polymer solution and pressurized CO2 were flowed through the inner, middle, and outer nozzle of the multiple coaxial nozzle at 76 bar and 25 °C. The encapsulation efficiency was 14.6% [67]. Insulin is more soluble in DMSO than in dichloromethane while PLA is more soluble in dichloromethane than in DMSO. By admixing the same volumes of 1 mg/mL insulin in DMSO and 20 mg/mL PLA in dichloromethane, a 1% PLA (Mw = 102,000) solution containing 5% insulin with respect to the polymer was prepared. Dichloromethane is more expandable than DMSO by CO2 at a given pressure. The expansion of the 1:1 admixture was closer to that of DMSO. Below 80 bar, microfibrils were produced. Smaller particles were obtained at 130 bar. Temperature affected the morphology of particles. The SAS process at 40 °C resulted in agglomeration and flocculation phenomena while a particulate powder was obtained at 20 °C. SAS processing at 130 bar and 20 °C produced 0.5–2-μm PLA particles without agglomeration. More than 80% of insulin was incorporated in the PLA particles without loss of hypoglycemic activity, which was determined by subcutaneous injection into diabetic mice of insulin extracted from the microspheres [68]. Polyethylene glycol (PEG) with high hydrophilicity was introduced into the PLA/insulin formulation to improve the drug release. The residual concentration of dichloromethane and DMSO was 8 ppm and 300 ppm. The entrapment of insulin was decreased with the increase in PEG 6000 content. The higher molecular weight PEG released insulin rapidly in the initial 80 h. However the amount released was less than 3% and no more release was observed thereafter. PEGs of smaller molecular weight (1900, 750, and 350) were favorable to incorporate insulin at high yield. Constant release was observed up to 1500 h without an initial burst, although the release was very slow (b 1.5%/ 1500 h) [69]. The increase in PEG 1900 content from 0 to 75% decreased insulin loading yield from 95 to 65%, while the release rate of insulin increased. These trends resulted in the same amount of insulin released from the same amount of microspheres regardless of the PEG content. The insulin release in 3 months from the microsphere with 23% PEG 1900 was 10% of that loaded, while that from the microsphere with 67% PEG 1900 was 100%. The increase in insulin release was due to the pore and channel formation by PEG dissolution. The melting points of PEGs 350, 750, and 1900 under normal conditions are 288, 298, and 323 K, respectively. Higher operating temperatures resulted in a sticky material and low product recovery, while too low a temperature causes low elimination of DMSO which may induce material plasticization and particle agglomeration. Operating a few degrees below the PEG melting point seemed to be preferable for the formation of particles and elimination of the solvent [70]. 6.2. Microcapsules of proteins In some cases, the admixture of compressed CO2 and an organic solvent, both of which act in pure form as a nonsolvent or poor solvent for a polymer, can dissolve the polymer. Protein microparticles coated with a polymer can be produced by RESSprocessing by spraying a suspension of protein in the admixture of CO2 and an organic solvent containing dissolved polymer at

H. Okamoto, K. Danjo / Advanced Drug Delivery Reviews 60 (2008) 433–446

atmospheric pressure. Because the organic solvent is a nonsolvent for the polymer, it is not expected to cause agglomeration of the particles after depressurization. Although the addition of ethanol in CO2 did not increase the solubility of polystyrene or PLA, the solubility of PEGs in the 1:1 admixture of CO2 and ethanol was much higher than that in pure ethanol. Lysozyme and lipase are insoluble in CO2, ethanol, and their admixtures at any ratios. The protein suspensions with PEG 6000 in compressed CO2 containing 27.1 wt.% of ethanol at 20 MPa and 308 K were sprayed through a capillary nozzle to a target plate placed under atmospheric pressure. Discrete protein particles coated with PEG 6000 were successfully obtained. By the increase in the weight ratio of PEG 6000 to protein, the mean particle diameter increased. When toluene or acetone, which is a good solvent for PEG 6000, was employed instead of ethanol, the particles obtained were agglomerated because these solvents dissolved the polymer after spraying. The residual ethanol in the particles was less than 1 wt.% [71]. By selecting proper coating materials that are soluble in compressed CO2, protein particles can be coated without using an organic solvent, which may denature the proteins or have adverse effects on the environment, at a low temperature. For example, 1.5 g of Gelucire 50-02, which is a mixture of glycerides and fatty acid esters, and 1.0 g of Dynasan 114 (trimyristin) dissolve completely in 1 L of CO2 at 45 °C and 200 bar and 35 °C and 200 bar, respectively. CO2 was introduced in an autoclave containing the coating materials and bovine serum albumin (BSA) particles and heated at 35–45 °C and 200 bar for 1 h to dissolve the coating materials. Cooling the autoclave decreases the solubility of the coating materials to precipitate on the surface of BSA particles. The coated BSA particles were harvested from the autoclave after reducing the pressure inside to atmospheric level. No adverse effect of this process on BSA was detected by electrophoresis. Coating with Gelucire 50-02 successfully prolonged the BSA release with time in pH 7.4 phosphatebuffered solution. Uncoated BSA dissolved immediately while the Gelucire particles released 50% of BSA in 180–360 min. BSA content was 36–67%. When 0.4 g of the 125–250-μm fraction of BSA was treated with 1.5 g of Gelucire 50-02, two BSA particles on average were incorporated in each particle [72]. 7. Gene powders It has been reported that the intravenous injection of gene with a proper vector can deliver the gene to the lung. The intravenous administration of radiolabelled plasmid DNA (pDNA) encoding chloramphenicol acetyltransferase (CAT) with a vector DOTIM in mice increased the radioactivity in the lung. However, the radioactivity in the lung decreased rapidly; in turn, that in the liver increased. The majority of pDNA was found to degrade in blood in 2 h [73]. The respiratory system, from the nasal cavity to alveoli, is a site suitable for gene therapy because direct access of a gene delivery system via the airway is possible. By the direct administration of genes into the lungs for the treatment of pulmonary diseases, there are many advantages such that the distribution of the gene into organs other than the lungs will be minimized, that the degradation of

443

the gene in plasma will be avoided, that the total dose of the gene will be reduced, and that the patients will be relieved from the pain of injections. Nebulization is one of the practical systems for the administration of non-viral gene delivery systems. The aerosol administration of a pDNA encoding CAT complexed to cationic liposomes produces high-level, lung-specific CAT gene expression in mice in vivo with no apparent treatment-related damage [74]. However, the pDNA is degraded rapidly during aerosolization or jet nebulization, although complexed pDNA is largely protected [75,76]. In addition, the genes are not stable in solution [77,78]. Therefore, dry gene powders are a promising formulation for inhalation therapy for pulmonary diseases. Tservistas et al. described the first application of the SEDS process involving supercritical CO2 for the production of pDNAloaded particles. They employed a three-channeled coaxial nozzle, through which supercritical CO2, isopropanol (a cosolvent), and an aqueous solution of the 6.9-kb plasmid pSVβ with mannitol as an excipient were flowed. A high degradation of the pDNA during powder formation was observed. The addition of sodium acetate to the aqueous-feed solution successfully increased the recovered supercoiled proportion from 7% to 80%. This suggests that the degradation of pDNAwas due to a pH drop by the formation of carboxylic acid in the CO2/isopropanol/ water system, which was encountered by sodium acetate [79]. There are very few reports on gene expression after intratracheal administration of powders for inhalation. Freeman and Niven administered 200 μg of pCMV-Luc as a spray-dried pCMV-Luc:trehalose (1:9) powder into rat lung through the trachea to examine the effect of insufflated pDNA on the transfection of lung tissue. Unfortunately, no response to the insufflated pDNA powder could be obtained [80]. Okamoto et al. prepared pCMV-Luc powder with chitosan as a cationic vector and mannitol as an excipient. An admixture of CO2 (5.7 g/min) and ethanol (0.665 mL/min) was flowed into the particle formation vessel through one end of a specially designed V-shaped nozzle (Fig. 2) at 35 °C and 15 MPa. The aqueous solution of gene/chitosan/mannitol flowed into the particle formation vessel at a rate of 0.035 mL/min through the other end of the V-shaped nozzle to precipitate them as tiny particles. Without chitosan, an intensive degradation of the gene was detected, which was suppressed by the addition of sodium acetate as reported by Tservistas et al. Even without sodium acetate, the addition of chitosan successfully suppressed the degradation of the gene during processing [81]. Chitosan, a weak base with a pKb value reportedly of 7.7 [82], may reduce the pH drop through the formation of carboxylic acid in the present CO2/ethanol/water system as sodium acetate does. In addition, the complexation of pDNAwith chitosan may protect the pDNA against chemical and/ or physical degradation processes. The optimized powder having a N/P ratio (the ratio of number of nitrogen in chitosan to number of phosphorus in gene) of 5 increased gene expression in the mouse lung after insufflation 27 times that of the instillation of the same dose of a gene solution [81]. When the pCMV-Luc/chitosan/ mannitol powder placed in a stability chamber at 40 °C for 4 weeks, it still kept gene expression activity, which was several times higher than that of the same dose of a gene solution that was

444

H. Okamoto, K. Danjo / Advanced Drug Delivery Reviews 60 (2008) 433–446

freshly prepared [83]. The electrophoresis detected a marginal amount of supercoiled DNA but a strong band of open circular DNA that reportedly has more than 90% of the transfer efficiency compared to the supercoiled DNA [84], while the genes in solutions with or without chitosan were completely degraded in 4 weeks at 25 and 40 °C [83]. 8. Perspectives Inhalation therapy has been limited to the treatment of local pulmonary diseases such as asthma and infection with small molecular weight drugs. Recombinant human deoxyribonuclease (rhDNase) was the first protein approved for inhalation therapy: 2.5 mL of a 1-mg/mL solution of rhDNase is administered by nebulization to patients with cystic fibrosis [85]. In January 2006, a new inhalation system of dry insulin powder was approved for use in Europe and the USA. This is the very first inhalation of protein medicine for systemic therapy. Because of the difficulty of administering many proteins and genes orally, inhalation systems will be extended to other macromolecular medicines. There have been a lot of improvements on supercritical carbon dioxide processing to maximize integrity and inhalation performance of proteins and genes. Although there is still room for optimizing operating conditions, such as temperature, pressure, flow rates, and concentration of ingredients, for each formulation, precipitation of proteins and genes with supercritical CO2 is a promising way to produce protein and gene particles for inhalation. References [1] H. Yoshida, K. Okumura, R. Hori, T. Anmo, H. Yamaguchi, Absorption of insulin delivered to rabbit trachea using aerosol dosage form, J. Pharm. Sci. 68 (1979) 670–671. [2] J.S. Patton, P. Trinchero, R.M. Platz, Bioavailability of pulmonary delivered peptide and proteins: α-interferon, calcitonins and parathyroid hormones, J. Contr. Rel. 28 (1994) 79–85. [3] M.P. Timsina, G.P. Martin, C. Marriott, D. Ganderton, M. Yianneskis, Drug delivery to the respiratory tract using dry powder inhalers, Int. J. Pharm. 101 (1994) 1–13. [4] S.P. Newman, A. Hollingworth, R. Clark, Effect of different modes of inhalation on drug delivery from a dry powder inhaler, Int. J. Pharm. 102 (1994) 127–132. [5] M.A. Winters, B.L. Knutson, P.G. Debenedetti, H.G. Sparks, T.M. Przybycien, C.L. Stevenson, S.J. Prestrelski, Precipitation of proteins in supercritical carbon dioxide, J. Pharm. Sci. 85 (1996) 586–594. [6] S.-D. Yeo, G.-B. Lim, P.G. Debenedetti, H. Bernstein, Formation of microparticulate protein powders using a supercritical fluid antisolvent, Biotechnol. Bioeng. 41 (1993) 341–346. [7] H.R. Costantino, J.D. Andya, P.A. Nguyen, N. Dasovich, T.D. Sweeney, S.J. Shire, C.C. Hsu, Y.F. Maa, Effect of mannitol crystallization on the stability and aerosol performance of a spray-dried pharmaceutical protein, recombinant humanized anti-IgE monoclonal antibody, J. Pharm. Sci. 87 (1998) 1406–1411. [8] K. Stahl, M. Claesson, P. Lilliehorn, H. Linden, K. Backstrom, The effect of process variables on the degradation and physical properties of spray dried insulin intended for inhalation, Int. J. Pharm. 233 (2002) 227–237. [9] J. Broadhead, S.K. Rouan, C.T. Rhodes, The effect of process and formulation variables on the properties of spray-dried β-galactosidase, J. Pharm. Pharmacol. 46 (1994) 458–467. [10] G. Donsi, E. Reverchon, Micronization by means of supercritical fluids: possibility of application to pharmaceutical field, Pharm. Acta Helv. 66 (1991) 170–173.

[11] S. Suarez, A.J. Hickey, Drug properties affecting aerosol behavior, Respir. Care 45 (2000) 652–666. [12] R. Vanbever, J.D. Mintzes, J. Wang, J. Nice, D.H. Chen, R. Batycky, R. Langer, D.A. Edwards, Formulation and physical characterization of large porous particles for inhalation, Pharm. Res. 16 (1999) 1735–1742. [13] L.S. Schanker, M.J. Less, Lung pH and pulmonary absorption of nonvolatile drugs in the rat, Drug Metabol. Dispos. 5 (1977) 174–178. [14] N. Benowitz, R.P. Forsyth, K.L. Melmon, M. Rowland, Lidocaine disposition kinetics in monkey and man I. Prediction by a perfusion model, Clin. Pharmacol. Ther. 16 (1974) 87–98. [15] B.J. Lipworth, Pharmacokinetics of inhaled drugs, Br. J. Clin. Pharmacol. 42 (1996) 697–705. [16] Z. Shen, Q. Zhang, S. Wei, T. Nagai, Proteolytic enzymes as a limitation for pulmonary absorption of insulin: in vitro and in vivo investigations, Int. J. Pharm. 192 (1999) 115–121. [17] M.C.P. Hsu, J.P.F. Bai, Investigation into the presence of insulin-degrading enzyme in cultured type II alveolar cells and the effects of enzyme inhibitors on pulmonary bioavailability of insulin in rats, J. Pharm. Pharmacol. 50 (1998) 507–514. [18] S.J. Enna, L.S. Schanker, Absorption of drugs from the rat lung, Am. J. Physiol. 223 (1972) 1227–1231. [19] R.C. Lanman, R.M. Gillilan, L.S. Schanker, Absorption of cardiac glycosides from the rat respiratory tract, J. Pharmacol. Exp. Ther. 187 (1973) 105–111. [20] S.J. Enna, L.S. Schanker, Absorption of saccharides and urea from the rat lung, Am. J. Physiol. 222 (1972) 409–414. [21] Y. Matsukawa, V.H.L. Lee, E.D. Crandall, K.J. Kim, Effect of absorption enhancers on ciliated epithelium: a novel in vivo toxicity screening method using rotifers, Pharm. Res. 14 (1997) 541–545. [22] F. Komada, S. Iwakawa, N. Yamamoto, H. Sakakibara, K. Okumura, Intratracheal delivery of peptide and protein agents: absorption from solution and dry powder by rat lung, J. Pharm. Sci. 83 (1994) 863–867. [23] A. Yamamoto, S. Umemori, S. Muranishi, Absorption enhancement of intrapulmonary administered insulin by various absorption enhancers and protease inhibitors in rats, J. Pharm. Pharmacol. 46 (1993) 14–18. [24] R.K. Wolff, Safety of inhaled proteins for therapeutic use, J. Aerosol Med. 11 (1998) 197–219. [25] L.J. Kanig, Pharmaceutical aerosols, J. Pharm. Sci. 52 (1963) 513–535. [26] C. Bosquillon, C. Lombry, V. Preat, R. Vanbever, Comparison of particle sizing techniques in the case of inhalation dry powders, J. Pharm. Sci. 90 (2001) 2032–2041. [27] T.M. Crowder, J.A. Rosati, J.D. Schroeter, A.J. Hickey, T.B. Martonen, Fundamental effects of particle morphology on lung delivery: predictions of Stokes' law and the particular relevance to dry powder inhaler formulation and development, Pharm. Res. 19 (2002) 239–245. [28] G.W. Hunninghake, J.E. Gadek, O. Kawanami, V.J. Ferrans, R.G. Crystal, Inflammatory and immune processes in the human lung in health and disease: evaluation by bronchoalveolar lavage, Am. J. Pathol. 97 (1979) 149–206. [29] M.B. Snipes, M.F. Clem, Retention of microspheres in the rat lung after intratracheal instillation, Environ. Res. 24 (1981) 33–41. [30] D.A. Edwards, J. Hanes, G. Caponetti, J. Hrkach, A. Ben-Jebria, M.L. Eskew, J. Mintzes, D. Deaver, N. Lotan, R. Langer, Large porous particles for pulmonary drug delivery, Science 276 (1997) 1868–1871. [31] D.W. Matson, J.L. Fulton, R.C. Petersen, R.D. Smith, Rapid expansion of supercritical fluid solutions: solute formation of powders, thin films, and fibers, Ind. Eng. Chem. Res. 26 (1987) 2298–2306. [32] S.P. Sellers, G.S. Clark, R.E. Sievers, J.F. Carpenter, Dry powders of stable protein formulations from aqueous solutions prepared using supercritical CO2-assisted aerosolization, J. Pharm. Sci. 90 (2001) 785–797. [33] T.W. Randplph, D.S. Clark, H.W. Blanch, J.M. Prausnitz, Enzymatic oxidation of cholesterol aggregates in supercritical carbon dioxide, Science 238 (1988) 387–390. [34] K.L. Toews, R.M. Shroll, C.M. Wai, N.G. Smart, pH-defining equilibrium between water and supercritical CO2. Influence on SFE of organics and metal chelates, Anal. Chem. 67 (1995) 4040–4043. [35] H. Ishikawa, M. Shimoda, A. Yonekura, K. Mishima, K. Matsumoto, Y. Osajima, Irreversible unfolding of myoglobin in an aqueous solution by supercritical carbon dioxide, J. Agric. Food Chem. 48 (2000) 4535–4539.

H. Okamoto, K. Danjo / Advanced Drug Delivery Reviews 60 (2008) 433–446 [36] M.O. Balaban, A.G. Arreola, M. Marshall, A. Peplow, C.I. Wei, J. Cornell, Inactivation of pectinesterasein orange juice by supercritical carbon dioxide, J. Food Sci. 56 (1991) 743–746. [37] J.A. Zagrobelny, F.V. Bright, In situ studies of protein conformation in supercritical fluids: trypsin in carbon dioxide, Biotechnol. Prog. 8 (1992) 421–423. [38] M. Taniguchi, M. Kamihira, T. Kobayashi, Effect of treatment with supercritical carbon dioxide on enzymatic activity, Agric. Biol. Chem. 51 (1987) 593–594. [39] E.M. Dumay, M.T. Kalichevsky, J.C. Cheftel, High-pressure unfolding and aggregation of b-lactoglobulin and the baroprotective effects of sucrose, J. Agric. Food Chem. 42 (1994) 1861–1868. [40] A.K. Dillow, F. Dehghani, J.S. Hrkach, N.R. Foster, R. Langer, Bacterial inactivation by using near- and supercritical carbon dioxide, Proc. Natl. Acad. Sci. U. S. A. 96 (1999) 10344–10348. [41] R. Thiering, F. Dehghani, A. Dillow, N.R. Foster, Solvent effects on the controlled dense gas precipitation of model proteins, J. Chem. Technol. Biotechnol. 75 (2000) 42–53. [42] M.A. Winters, D.Z. Frankel, P.G. Debenedetti, J. Carey, M. Devaney, T.M. Przybycien, Protein purification with vapor-phase carbon dioxide, Biotechnol. Bioeng. 62 (1999) 247–258. [43] R. Thiering, F. Dehghani, A. Dillow, N.R. Foster, The influence of operating conditions on the dense gas precipitation of model proteins, J. Chem. Technol. Biotechnol. 75 (2000) 29–41. [44] M. Jackson, H.H. Mantsch, Beware of proteins in DMSO, Biochim. Biophys. Acta 1078 (1991) 231–235. [45] G. Muhrer, M. Mazzotti, Precipitation of lysozyme nanoparticles from dimethyl sulfoxide using carbon dioxide as antisolvent, Biotechnol. Prog. 19 (2003) 549–556. [46] S. Moshashaee, M. Bisrat, R.T. Forbes, H. Nyqvist, P. York, Supercritical fluid processing of proteins. I: lysozyme precipitation from organic solution, Eur. J. Pharm. Sci. 11 (2000) 239–245. [47] S.D. Yeo, P.G. Debenedetti, S.Y. Patro, T.M. Przybycien, Secondary structure characterization of microparticulate insulin powders, J. Pharm. Sci. 83 (1994) 1651–1656. [48] M.A. Winters, P.G. Debenedetti, J. Carey, H.G. Sparks, S.U. Sane, T.M. Przybycien, Long-term and high-temperature storage of supercriticallyprocessed microparticulate protein powders, Pharm. Res. 14 (1997) 1370–1378. [49] P. Chattopadhyay, R.B. Gupta, Protein nanoparticles formation by supercritical antisolvent with enhanced mass transfer, Aiche J. 48 (2002) 235–244. [50] W.K. Snavely, B. Subramaniam, R.A. Rajewski, M.R. Defelippis, Micronization of insulin from halogenated alcohol solution using supercritical carbon dioxide as an antisolvent, J. Pharm. Sci. 91 (2002) 2026–2039. [51] M. Sarkari, I. Darrat, B.L. Knutson, CO2 and fluorinated solvent-based technologies for protein microparticle precipitation from aqueous solutions, Biotechnol. Prog. 19 (2003) 448–454. [52] M.L. Gilbert, M.E. Paulaitis, Gas–liquid equilibrium for ethanol–water– carbon dioxide mixtures at elevated pressures, J. Chem. Eng. Data 31 (1986) 296–298. [53] D.P. Nesta, J.S. Elliott, J.P. Warr, Supercritical fluid precipitation of recombinant human immunoglobulin from aqueous solutions, Biotechnol. Bioeng. 67 (2000) 457–464. [54] S.P. Velaga, J. Carlfors, Supercritical fluids processing of recombinant human growth hormone, Dru. Dev. Ind. Pharm. 31 (2005) 135–149. [55] R.T. Bustami, H.K. Chan, T. Sweeney, F. Dehghani, N.R. Foster, Generation of fine powders of recombinant human deoxyribonuclease using the aerosol solvent extraction system, Pharm. Res. 20 (2003) 2028–2035. [56] R.T. Bustami, H.K. Chan, F. Dehghani, N.R. Foster, Generation of microparticles of proteins for aerosol delivery using high pressure modified carbon dioxide, Pharm. Res. 17 (2000) 1360–1366. [57] S. Moshashaee, M. Bisrat, R.T. Forbes, E.A. Quinn, H. Nyqvist, P. York, Supercritical fluid processing of proteins: lysozyme precipitation from aqueous solution, J. Pharm. Pharmacol. 55 (2003) 185–192. [58] H. Todo, K. Iida, H. Okamoto, K. Danjo, Improvement of insulin absorption from intratracheally administrated dry powder prepared by supercritical carbon dioxide process, J. Pharm. Sci. 92 (2003) 2475–2486.

445

[59] K. Imamura, T. Ogawa, T. Sakiyama, K. Nakanishi, Effects of types of sugar on the stabilization of protein in the dried state, J. Pharm. Sci. 92 (2003) 266–274. [60] K. Izutsu, S. Yoshioka, T. Terao, Effect of mannitol crystallinity on the stabilization of enzymes during freeze-drying, Chem. Pharm. Bull. (Tokyo) 42 (1994) 5–8. [61] N. Jovanovic, A. Bouchard, G.W. Hofland, G.J. Witkamp, D.J. Crommelin, W. Jiskoot, Distinct effects of sucrose and trehalose on protein stability during supercritical fluid drying and freeze-drying, Eur. J. Pharm. Sci. 27 (2006) 336–345. [62] R.E. Sievers, E.T.S. Huang, J.A. Villa, J.K. Kawamoto, M.M. Evans, P.R. Brauer, Low-temperature manufacturing of fine pharmaceutical powders with supercritical fluid aerosolization in a Bubble DryerR, Pure Appl. Chem. 73 (2001) 1299–1303. [63] G.W. Hofland, A. de Rijke, R. Thiering, L.A. van der Wielen, G.J. Witkamp, Isoelectric precipitation of soybean protein using carbon dioxide as a volatile acid, J. Chromatogr. B 743 (2000) 357–368. [64] T.J. Young, K.P. Johnston, K. Mishima, H. Tanaka, Encapsulation of lysozyme in a biodegradable polymer by precipitation with a vapor-overliquid antisolvent, J. Pharm. Sci. 88 (1999) 640–650. [65] D.J. Dixon, K.P. Johnston, R.A. Bodmeier, Polymeric materials formed by precipitation with a compressed fluid antisolvent, Aiche J. 39 (1993) 127–139. [66] F. Ruchatz, P. Kleinebudde, B.W. Muller, Residual solvents in biodegradable microparticles. Influence of process parameters on the residual solvent in microparticles produced by the aerosol solvent extraction system (ASES) process, J. Pharm. Sci. 86 (1997) 101–105. [67] L. Sze Tu, F. Dehghani, N.R. Foster, Micronisation and microencapsulation of pharmaceuticals using a carbon dioxide antisolvent, Powder Technol. 126 (2002) 134–149. [68] N. Elvassore, A. Bertucco, P. Caliceti, Production of protein-loaded polymeric microcapsules by compressed CO2 in a mixed solvent, Ind. Engineer. Chem. Res. 40 (2001) 795–800. [69] N. Elvassore, A. Bertucco, P. Caliceti, Production of insulin-loaded poly (ethylene glycol)/poly(L-lactide) (PEG/PLA) nanoparticles by gas antisolvent techniques, J. Pharm. Sci. 90 (2001) 1628–1636. [70] P. Caliceti, S. Salmaso, N. Elvassore, A. Bertucco, Effective protein release from PEG/PLA nano-particles produced by compressed gas anti-solvent precipitation techniques, J. Contr. Rel. 94 (2004) 195–205. [71] K. Mishima, K. Matsuyama, D. Tanabe, S. Yamauchi, T.J. Young, K.P. Johnston, Microencapsulation of proteins by rapid expansion of supercritical solution with a nonsolvent, Aiche J. 46 (2000) 857–865. [72] I. Ribeiro Dos Santos, J. Richard, B. Pech, C. Thies, J.P. Benoit, Microencapsulation of protein particles within lipids using a novel supercritical fluid process, Int. J. Pharm. 242 (2002) 69–78. [73] R. Niven, R. Pearlman, T. Wedeking, J. Mackeigan, P. Noker, L. SimpsonHerren, J.G. Smith, Biodistribution of radiolabeled lipid–DNA complexes and DNA in mice, J. Pharm. Sci. 87 (1998) 1292–1299. [74] R. Stribling, E. Brunette, D. Liggitt, K. Gaensler, R. Debs, Aerosol gene delivery in vivo, Proc. Natl. Acad. Sci. U. S. A. 89 (1992) 11277–11281. [75] J.C. Birchall, I.W. Kellaway, M. Gumbleton, Physical stability and in-vitro gene expression efficiency of nebulised lipid–peptide–DNA complexes, Int. J. Pharm. 197 (2000) 221–231. [76] S.J. Eastman, J.D. Tousignant, M.J. Lukason, H. Murray, C.S. Siegel, P. Constantino, D.J. Harris, S.H. Cheng, R.K. Scheule, Optimization of formulations and conditions for the aerosol delivery of functional cationic lipid:DNA complexes, Hum. Gene Ther. 8 (1997) 313–322. [77] J.Y. Cherng, H. Talsma, D.J.A. Crommelin, W.E. Hennink, Long term stability of poly((2-dimethylamino)ethyl methacrylate)-based gene delivery systems, Pharm. Res. 16 (1999) 1417–1423. [78] B. Li, S. Li, Y.D. Tan, D.B. Stolz, S.C. Watkins, L.H. Block, L. Huang, Lyophilization of cationic lipid–protamine–DNA (LPD) complexes, J. Pharm. Sci. 89 (2000) 355–364. [79] M. Tservistas, M.S. Levy, M.Y. Lo-Yim, R.D. O'Kennedy, P. York, G.O. Humphrey, M. Hoare, The formation of plasmid DNA loaded pharmaceutical powders using supercritical fluid technology, Biotechnol. Bioeng. 72 (2001) 12–18.

446

H. Okamoto, K. Danjo / Advanced Drug Delivery Reviews 60 (2008) 433–446

[80] D.J. Freeman, R.W. Niven, The influence of sodium glycocholate and other additives on the in vivo transfection of plasmid DNA in the lungs, Pharm. Res. 13 (1996) 202–209. [81] H. Okamoto, S. Nishida, H. Todo, Y. Sakakura, K. Iida, K. Danjo, Pulmonary gene delivery by chitosan–pDNA complex powder prepared by supercritical carbon dioxide process, J. Pharm. Sci. 92 (2003) 371–380. [82] K.Y. Lee, I.C. Kwon, Y.H. Kim, W.H. Jo, S.Y. Jeong, Preparation of chitosan self-aggregates as a gene delivery system, J. Contr. Rel. 51 (1998) 213–220.

[83] H. Okamoto, Y. Sakakura, K. Shiraki, K. Oka, S. Nishida, H. Todo, K. Iida, K. Danjo, Stability of chitosan–pDNA complex powder prepared by supercritical carbon dioxide process, Int. J. Pharm. 290 (2005) 73–81. [84] R.C. Adami, W.T. Collard, S.A. Gupta, K.Y. Kwok, J. Bonadio, K.G. Rice, Stability of peptide-condensed plasmid DNA formulations, J. Pharm. Sci. 87 (1998) 678–683. [85] H.K. Chan, A. Clark, I. Gonda, M. Mumenthaler, C. Hsu, Spray dried powders and powder blends of recombinant human deoxyribonuclease (rhDNase) for aerosol delivery, Pharm. Res. 14 (1997) 431–437.