Formation of Microcapsules from Polyelectrolyte and Covalent

Three kinds of interactions were generated in the same membrane: (1) ... they influence the density of the covalent bonds in the polymeric network. ...
0 downloads 0 Views 236KB Size
Download PDF
9764

Langmuir 2005, 21, 9764-9772

Formation of Microcapsules from Polyelectrolyte and Covalent Interactions Ve´ronique Breguet,† Raphae¨l Gugerli,† Mimma Pernetti,‡ Urs von Stockar,† and Ian W. Marison*,† Laboratory of Chemical and Biochemical Engineering, Ecole Polytechnique Fe´ de´ rale de Lausanne (EPFL), CH-1015 Lausanne, Switzerland, and Centre for Environmental Technology and Chemistry, Department of Chemical Engineering, University of Rome “La Sapienza”, I-00184 Rome, Italy Received May 13, 2005. In Final Form: July 29, 2005 A new approach combining electrostatic and covalent bonds was established for the formation of resistant capsules with long-term stability under physiological conditions. Three kinds of interactions were generated in the same membrane: (1) electrostatic bonds between alginate and poly-L-lysine (PLL), (2) covalent bonds (amides) between propylene-glycol-alginate (PGA) and PLL, and (3) covalent bonds (amides) between BSA and PGA. Down-scaling of the capsules size (e1 mm diameter) with a jet break-up technology was achieved by modifying the rheological properties of the polymer solution. Viscosity of the PGA solution was reduced by 95% with four successive pH stabilizations (pH 7), while filtration (0.2 µm) and sterilization was possible. Covalent bond formation was initiated by addition of NaOH (pH 11) using a transacylation reaction. Kinetics of the chemical reaction (pH 11) were simulated by two mathematical models and adapted in order to preserve immobilization of animal cells. It was demonstrated that diffusion of NaOH in the absence of BSA resulted in gelation of 94% of the bead and death of 94% of the cells after 10 s reaction. By addition of BSA only 46% of the cells were killed within the same reaction time (10 s). Mechanical resistance of this new type of capsule could be increased 5-fold over the standard polyelectrolytic system (PLL-alginate). Encapsulated CHO cells were successfully cultivated for 1 month in a repetitive batch mode, with the mechanical resistance of the capsules decreasing by only 10% during this period. The combination of a synthetic and natural protein resulted in enhanced stability toward culture medium and proteolytic enzymes (250%).

Introduction Microencapsulation is a promising technique for the immobilization of cells, finding applications in many different fields, such as agro-alimentary, production of cell derived biomolecules, gene therapy, and bioartificial organs.1-3 Production of recombinant proteins using encapsulated animal cell cultures presents several advantages over suspension cell cultures, such as protection of cells from shear stress, reduction of toxic metabolite accumulation, facilitated cell retention, and simplified downstream processing.4,5 For optimal cell growth,6 microcapsules should have a liquid core, in which cells can grow freely surrounded by a semipermeable membrane, which ensures * Corresponding author. E-mail: [email protected]. Tel.: +41 21 693 31 94. Fax +41 21 693 36 80. † Ecole Polytechnique Fe ´ de´rale de Lausanne (EPFL). ‡ University of Rome “La Sapienza”. (1) Le´vy, M.-C.; Poncelet, D. Bioencapsulation, the technology. Biofutur 1994, 132, 16-21. (2) Lysaght, M.; Aebischer, P. Encapsulated cells as therapy. Sci. Am. 1999, April, 52-58. (3) Uludag, H., et al. Viability and Protein Secretion from Human Hepatoma (Hepg2) Cells Encapsulated in 400-Mu-M Polyacrylate Microcapsules by Submerged Nozzle Liquid Jet Extrusion. Biotechnol. Bioeng. 1994, 44 (10), 1199-1204. (4) Posillico, E. G.; Kallelis, M.; George, J. M. Large-Scale Production and Purification of Monoclonal Antibodies Using Cellular Microencapsulation. In Commercial Production of Monoclonal Antibodies: a Guide for Scale-Up; Seaver, S. S., Ed.; M. Dekker: New York, 1987; p 139-157. (5) Rupp, R. G. Use of Cellular Microencapsulation in Large-Scale Production of Monoclonal Antibodies. In Large-Scale Mammalian Cell Culture; Tolbert, W. R., Feder, J., Eds.; Academic Press: New York, 1986; p 19. (6) Gugerli, R., et al. Quantitative study of the growth and stability of encapsulated CHO cells. Biotechnol. Bioeng. 2004, submitted.

mechanical and chemical resistance, and a small size that minimizes diffusional limitations. Permeability and molecular weight cutoff of the membrane should be adjustable in order to allow the controlled transport of cellular products while simplifying the recovery process. Such capsules may be produced by extrusion of a polymer solution into a gelation bath, followed by formation of an external membrane and re-liquefaction of the core.7 Polymers used should be biocompatible, soluble in physiological solutions, and filterable (0.2 µm) in order to carry out sterilization and extrusion. In addition, the polymers should be stable over long periods with gelation occurring under mild conditions. Microcapsules are frequently formed through polyelectrolyte interactions involving complexation of alginate with a polycation.6,8-12 Complexation is enhanced by the (7) Wong, H.; Chang, T. M. S. The Microencapsulation of Cells within Alginate Poly-L-Lysine Microcapsules Prepared with the Standard Single Step Drop Technique - Histologically Identified Membrane Imperfections and the Associated Graft-Rejection. Biomater. Artif. Cells Immobilization Biotechnol. 1991, 19 (4), 675-686. (8) Bartkowiak, A.; Hunkeler, D. Alginate-Oligochitosan Microcapsules: A Mechanistic Study Relating Membrane and Capsule Properties to Reaction Conditions. Chem. Mater. 1999, 11, 2486-2492. (9) Peirone, M., et al. Encapsulation of various recombinant mammalian cell types in different alginate microcapsules. Biomed. Mater. Res. 1998, 42 (4), 587-596. (10) Prokop, A.; et al. Water soluble polymers for immunoisolation II: Evaluation of multicomponent microencapsulation systems. In Microencapsulation - Microgels - Iniferters; Springer: New York, 1998; pp 53-73. (11) Prokop, A.; et al. Water soluble polymers for immunoisolation I: Complex coacervation and cytotoxicity. In Microencapsulation Microgels - Iniferters. Springer: New York, 1998; pp 1-51. (12) Thu, B.; et al. Alginate Polycation Microcapsules 0.1. Interaction Between Alginate and Polycation. Biomaterials 1996, 17 (10), 10311040.

10.1021/la0512796 CCC: $30.25 © 2005 American Chemical Society Published on Web 09/10/2005

Capsules with Polyelectrolytic/Covalent Membranes

charge density on the polyanion and polycation, which depends strongly on the chemical structure and pKa of the polyelectrolytes, as well as the pH of the solution. Alginate has a pKa of 3.813 and therefore is negatively charged at neutral pH. The optimal pH of the incubation bath is influenced by the pKa of the polycation, which generally varies between 7 and 11. Numerous polyion combinations have been tested to produce resistant polyelectrolytic membranes, including poly-L-lysine-alginate, cellulose sulfate-PDADMAC, chitosan-alginate, PMCG-alginate, and chitosan-carrageenan.10,11,14 The alginate-polylysine system is often adopted for microencapsulation,15-18 due to its relative simplicity and mild conditions of gelation. Nevertheless, the resulting microcapsules exhibit low mechanical resistance and poor stability, which explains the limited success in industrial and medical applications.6,19 To increase the mechanical resistance and long-term stability of microcapsules, various reactions have been used to covalently link the membrane polymers.20-24 These include coating of alginate beads by using transacylation reactions.25 Such reactions involve nucleophilic additionelimination between an esterified carboxylic group and an amino group, yielding an amide and an alcohol. The reaction takes place when amino groups become uncharged under alkaline conditions (-NH2). Membranes formed by the reaction of propylene-glycol-alginate (PGA) with different natural proteins (human serum albumin, ovalbumin, hemoglobin) have been extensively investigated,26 with the reaction being initiated by the addition of NaOH. It has been shown that the key parameters influencing membrane characteristics are the pH of reaction, reaction time, and protein concentration,27,28 since they influence (13) Draget, K. I.; Braek, G. S.; Smidsrod, O. Alginic Acid Gels - The Effect of Alginate Chemical-Composition and Molecular-Weight. Carbohydr. Polym. 1994, 25 (1), 31-38. (14) Bartkowiak, A.; Hunkeler, D. Carrageenan-oligochitosan microcapsules: optimization of the formation process. Colloids Surf. B-Biointerfaces 2001, 21 (4), 285-298. (15) Lacik, I.; et al. New Capsule with Tailored Properties for the Encapsulation of Living Cells. J. Biomed. Mater. Res. 1998, 39 (1), 52-60. (16) Quong, D.; Neufeld, R. J. DNA protection from extracapsular nucleases, within chitosan- or poly-L-lysine-coated alginate beads. Biotechnol. Bioeng. 1998, 60 (1), 124-134. (17) Thu, B.; et al. Alginate Polycation Microcapsules 0.2. Some Functional-Properties. Biomaterials 1996, 17 (11), 1069-1079. (18) Vandenbossche, G. M. R.; et al. Host-Reaction Against Empty Alginate-Polylysine Microcapsules - Influence of Preparation Procedure. J. Pharm. Pharmacol. 1993, 45 (2), 115-120. (19) Duff, R. Microencapsulation technology: a novel method for monoclonal antibody production. Trends Biotechnol. 1985, 3, 167-170. (20) Birnbaum, S., et al. Covalent Stabilization of Alginate Gel for the Entrapment of Living Whole Cells. Biotechnol. Lett. 1981, 3 (8), 393-400. (21) Chang, S. J.; Lee, C. H.; Wang, Y. J. Microcapsules prepared from alginate and a photosensitive poly(L-lysine). J. Biomater. Sci.Polym. Ed. 1999, 10 (5), 531-542. (22) Dupuy, B., et al. In situ polimerization of a microencapsulating medium round living cells. J. Biomed. Mater. Res. 1988, 22, 10611070. (23) Lu, M. Z., et al. Cell encapsulation with alginate and alphaPhenolxyxinnamylidene-Acetylated Poly(allylamine). Biotechnol. Bioeng. 2000, 70 (5), 479-483. (24) Moe, S.; Skjak-Braek, G.; Smidsrod, O. Covalently cross-linked sodium alginate beads. Food Hydrocolloids 1991, 5, 119-123. (25) Levy, M.-C.; Edwards-Levy, F. Coating alginate beads with crosslinked biopolymers: a novel method based on a transacylation reaction. J. Microencapsulation 1996, 13 (2), 169-183. (26) Le´vy, M.-C.; Edwards-Le´vy, F.; Particles prepared by transacylation reaction between an esterified polysaccarude and a polyamine, methods of preparation therefor and compositions containing same. U.S.A. patent, 5,635,609, 1997. (27) Edwards-Le´vy, F.; Le´vy, M.-C. Serum albumin-alginate coated beads: mechanical properties and stability + Erratum. Biomaterials 1999, 20, 2069-2084. (28) Guney, H.; Gugerli, R. Encapsulation of animal cells: study of a new polymeric system (diploma thesis). 2000, EPFL: Lausanne.

Langmuir, Vol. 21, No. 21, 2005 9765

the density of the covalent bonds in the polymeric network. A higher concentration of PGA does not increase the mechanical resistance, since the ester groups are in excess. HSA-PGA beads (HSA: human serum albumin), generated using an air-jet device, have been used for the encapsulation of porcine hepatocytes in order to create bioartificial livers.29 Such covalent membranes have been shown to be more stable than polyelectrolytic ones.27 However, even when proteins were irreversibly crosslinked to polysaccharides, the gels rapidly degraded when exposed to hydrolytic enzymes.25 In the present work, a combination of polyelectrolytic and covalent interactions was investigated in order to produce stable capsules that are also resistant toward proteolytic enzymes. Two polyamines were used simultaneously, a natural one (BSA), which is included in the polymer solution to be extruded, and a synthetic one (PLL), which is added to the incubation bath. At neutral pH, PLL is positively charged and diffuses through the bead (50-100 µm), reacting with alginate (AG) and forming polyelectrolytic bonds.6 Increasing the pH by the addition of NaOH results in a transacylation reaction between ester groups of PGA and amino groups of PLL and BSA, giving covalently bonded amides. A thin outer membrane is thus formed by PLL and a thicker one by BSA. Since BSA is present throughout the whole bead, it hinders the diffusion of NaOH, while protecting the cells in the core. This study focuses on the formation of such PLL-BSA-PGA covalent membranes and characterization of their mechanical resistance and stability. Since a majority of reported studies involves capsules with a diameter of 2-3 mm,25,27 a second aim of this work was to reduce capsule size (e1 mm) in order to avoid any mass transfer limitations.30 Extrusion with a jet-breakup encapsulator31,32 was investigated since this device enables formation of small monodisperse droplets (200-800 µm) at larger flow rates (100-1000 mL/h) than air-jet devices. Microencapsulation using the PLL-BSA-PGA system has not been previously reported due to practical problems resulting from the high viscosity of the polymer solution, rapidity of the transacylation reaction, and leaching of BSA from the beads. Optimization of polymer rheology is of primary importance to successful microencapsulation (polymer viscosity must be in the range of 10-250 mPa s) in order to avoid unstable jet break-up resulting in heterodisperse bead production. In an earlier work,28 it was shown that BSA leaches from the beads during the encapsulation procedure, thus decreasing the density of covalent bonds. Increasing the S/V ratio (surface to volume), by decreasing the bead size, results in a higher rate of BSA leaching, thus it is essential to modify the encapsulation procedure to minimize protein leaching. As bead size is reduced, diffusion of NaOH becomes faster and the transacylation time decreases dramatically.33 Therefore, it is necessary to develop a simple procedure to control the reaction time to prevent the total gelation of the beads. Since NaOH catalyses the trans(29) Joly, A.; et al. Survival, proliferation, and functions of porcine hepatocytes encapsulated in coated alginate beads: A step toward a reliable bioartificial liver. Transplantation 1997, 63 (6), 795-803. (30) Shuler, M. L.; Kargi, F. Bioprocess engineering: basic concepts. Prentice Hall PTR: Englewood Cliffs, NJ, 1992; pp 254-259. (31) Brandenberger, H.; et al. Monodisperse Particle Production: A New Method to Prevent Drop Coalescence Using Electrostatic Forces. J. Electrostatics 1999, 45, 227-238. (32) Serp, D.; et al. Characterization of a encapsulation device for the production of monodisperse alginate beads for cell immobilization. Biotechnol. Bioeng. 2000, 70 (1), 41-53. (33) Crank, J. The Mathematics of Diffusion, 2nd ed.; Oxford: Clarendon Press, 1975.

9766

Langmuir, Vol. 21, No. 21, 2005

acylation reaction, membrane thickness depends on the NaOH diffusion rate and the time of diffusion (reaction time). Therefore, modeling of NaOH diffusion as a function of reaction time is essential in order to predict the membrane thickness and to demonstrate whether the presence of BSA in the core hinders diffusion of NaOH throughout the whole capsule. Finally, the microencapsulation procedure was carried out under sterile conditions, in the presence of CHO cells, to verify nontoxicity of the transacylation reaction and long-term stability of the capsule structure. Materials and Methods Materials and Solutions. The biopolymers used in this study were alginate (Alginic acid, A-0682, Sigma, St-Louis, USA), propylene-glycol-alginate (PGA, Kelcoı¨d S, DE: 80-85%, Kelco International, Tadworth, England), bovine-serum-albumin (BSA, 05488, Fluka AG, Buchs, Switzerland), polyethylene-glycol 4000 (Fluka AG, Buchs, Switzerland), and poly-L-lysine (PLL, Sigma, 30kDa, St-Louis, USA). Alginate stock solutions were prepared by dissolving 4% (w/ w) alginate in aqueous MOPS buffer (10 mM) at pH 7 containing 0.92% (w/w) NaCl. The PGA stock solutions were prepared by dissolving 3% (w/w) PGA in aqueous MOPS buffer (20 mM) at pH 7. The BSA stock solutions were prepared by dissolving 4% (w/w) BSA in aqueous MOPS buffer (10 mM) at pH 7 containing 0.92% (w/w) NaCl. The PLL stock solution was prepared by dissolving 0.05% (w/w) PLL in aqueous MOPS buffer (10 mM) at pH 7 containing 110 mM CaCl2. All solutions were filter sterilized (0.2 µm, Pall Gelman, Ann Arbor, USA) prior to use. Polymer Characterization. Molecular weights of PGA and alginate have been measured by HPLC using a gel permeation column (Ohpak SB-804 HQ, Shodex, Kawasaki-city Kanagawa, Japan), 0.92% NaCl as solvent, and a RI detector (Nerma, ERC7515A, Ercatech AG, Bern, Switzerland). Calibration was made with pullulan standards (Shodex, Kawasaki-city Kanagawa, Japan). The viscosity of polymer solutions was measured using a rheometer (model VT 500, Haake, Karlsruhe, Germany). The surface tension was measured using a Dunoy ring (Sigma 7.03, KSV Instruments LTD, Helsinki, Finland). PGA Stabilization. PGA 3% (w/w) was first dissolved in MOPS buffer (20 mM) at pH 7. After 24 h incubation at room temperature, the pH was readjusted to pH 7 by the addition of NaOH (1 M). This operation was repeated 4 times until the pH remained stable, before sterile filtration (0.2 µm, Pall Gelman, Ann Arbor, USA). Encapsulation Procedure. Microcapsules were prepared using an extrusion device (Encapsulator, Inotech, Dottikon, CH), based on laminar jet break-up technology.31,32 Polymer beads were generated by extrusion of a polymer solution (1.2% alginate, 1.8% propylene glycol alginate, 4% bovine serum albumin and 1% poly(ethylene glycol)) through a vibrating nozzle (400 Hz, nozzle diameter 400 µm). Droplets were collected in a gelation bath containing 110 mM CaCl2 and 1% PEG. After 10 min incubation in CaCl2, beads were transferred to a 0.05% poly-Llysine solution for 20 min. The ratio of bead volume to PLL solution was 10. Varying amounts of NaOH (2 M) were then added to regulate the transacylation reaction. MOPS (2 M) was subsequently added to neutralize the pH (pH 7) and stop the reaction, followed by washing with saline buffer (0.92% NaCl containing 10 mM MOPS). The transacylation time or reaction time is the time interval between addition of NaOH and neutralization. Beads were subsequently incubated for 20 min in citrate solution (50 mM tri-sodium citrate in 10 mM MOPS) in order to liquify the core. The capsules were then washed twice with saline buffer and stored at 4 °C or used immediately. Capsule Characterization. Capsule size determination and microscopic observation were carried out with a laboratory microscope (Axiolab, Carl Zeiss Jena GmbH, Jena, Germany) connected to a digital video camera (Sony CCD Iris Camera, Sony Corporation, Tokyo, Japan) and software for image analysis (Cyberview, Cervus International, Courtaboeuf, France). Thirty capsules were measured individually in order to calculate the average diameter and the standard deviation. The mechanical

Breguet et al. resistance of capsules was determined using a texture analyzer (TA-2xi, Stable Micro Systems, Goldaming, U.K.) in which capsules were deformed by a probe moving at a constant speed of 0.05 mm s-1 until bursting. The probe, having a contact area of 36.3 mm2, allowed the deformation of several capsules simultaneously. For each batch of capsules, the average mechanical resistance and standard deviation, based on three hundred capsules, was determined. Stability tests were performed by storing 5 mL of capsules in vials containing 30 mL of a physiological solution (0.92% NaCl) or cell culture medium (ChoMaster HP-1, Cell Culture Technologies, Zu¨rich, Switzerland) at 37 °C in a thermostated water bath. Similar tests were performed using spent cell culture medium, which had been recovered from the stationary phase of batch cultures of CHO cells. The solution was changed every 2 days, over a period of 30 days. Leaching of BSA from the beads was determined spectrophotometrically at 280 nm by analyzing samples taken from the solution in which beads were produced (Spectrophotometer Uvikon 930, Kontron Instruments, Milan, Italy). Cell Encapsulation and Culture. Cells (CHO SSF3 cells), from a working cell bank, stored at -196 °C, were rapidly thawed at 37 °C and used to inoculate a T-flask (75 mL, Falcon, Beckton Dickinson, Fraga, S) containing 15 mL of a protein- free culture medium (ChoMaster HP-1, Cell Culture Technologies, Zu¨rich, CH) to an initial cell density of 2 × 105 cell mL-1. After incubation at 37 °C, under a humidified atmosphere of air containing 5% CO2, cells were harvested upon reaching a density of 106 cell mL-1 and used to inoculate a 1 L spinner flask (Integra Biosciences GmbH, Fernwald, Germany) containing 500 mL of culture medium. Upon reaching a cell density of 106 cell mL-1, cells were harvested by centrifugation and mixed with the polymer solution to yield a density of 106 cell mL-1, before extrusion under sterile conditions, to form capsules. Cellcontaining capsules were inoculated into spinner-flasks (Integra Biosciences GmbH, Fernwald, Germany) containing ChoMaster HP-1 medium to a starting cell density of 4 × 105 cell mL-1. Cells were enumerated using the Trypan blue exclusion method after destruction of the capsule membrane with a capillary (300 µm diameter) and liberation of the cells. After mixing with Trypan blue (0.4%, Sigma T-8154, 80 µL sample + 20 µL Trypan-blue), total and viable cells were counted microscopically using a Neubauer haemocytometer.

Results and Discussion Modification of the Fluid Rheological Properties for Jet Break-Up Extrusion. The first step of microcapsule formation is the extrusion of the polymer solution with a jet break-up droplet generator. Droplets are formed by extrusion of the solution through a vibrating nozzle, which breaks the laminar jet, forming regular and monodisperse droplets. Depending on the nozzle diameter and the rheological properties of the fluid (viscosity, density, surface tension), there is a range of frequencies and jet velocities, which gives uniform droplets.32 Moreover, to allow extrusion of a stable jet through a 400 µm nozzle, the viscosity must be lower than 0.25 mPa s (with a shear stress of 5 Pa) and the polymer solution must be filterable at 0.2 µm. The optimal polymer composition for large capsules (diameter 3 mm) produced using a syringe pump was 2% (w/w) propylene-glycol-alginate (PGA), 1.2-1.5% (w/w) alginate (AG) and 4% (w/w) bovine serum albumin (BSA). However, this polymer solution cannot be applied to the jet break-up extrusion technique due to the high viscosity (>0.25 mPa s; Figure 1). By measuring the viscosity of the single components, PGA and alginate viscosities were found to be approximately 150 and 100 mPa. Although the viscosity of a solution comprising multiple components is not simply the sum of each component, it is interesting to note that the viscosity of PGA and alginate represent 60% and 30% of the final solution viscosity (300 mPa), respectively. Similarly, the viscosity of a 4% BSA solution

Capsules with Polyelectrolytic/Covalent Membranes

Figure 1. Rheological behavior of 2% PGA solution (faint black line, left y axis), 1% alginate (AG) solution (bold dotted black line, left y axis), PGA 2%, AG 1%, and BSA 4% mixture (faint dotted black line, left y axis) and PGA 2%, AG 1.5%, and BSA 4% mixture (bold gray line, right y axis).

is only 3 mPa s, which increases to approximately 40 mPa s when mixed with the other polymers (Figure 1). The viscosity of the standard polymer solution cannot be modified by reducing the concentration of either alginate or BSA, since a lower alginate concentration would prevent the instantaneous gelation of the droplets, whereas a lower BSA concentration would greatly reduce the mechanical properties of the beads (data not shown). A further problem arises from the inability to perform sterile filtration of the PGA solution, which appears turbid when dissolved in demineralized water. Since PGA is composed of ionic units (∼10% carboxylic groups) and mainly of hydrophobic segments, intramolecular and intermolecular hydrophobic interactions promote chain aggregation at low concentration, which may form insoluble particles.34 Furthermore, the molecular weight distribution of PGA clearly showed two peaks (data not shown), indicating a great heterogeneity of molecular structures. The difficulty in filtering is probably due more to the presence of these compact aggregates than to viscosity, since a 4% alginate solution (300 mPa s) is easily filterable. As a result, it is essential to modify PGA in order to dissolve these aggregates and to decrease its viscosity, to enable extrusion with the jet break-up technique. This was realized by four repetitive pH stabilizations (pH 7) with NaOH 1 M. If PGA was simply dissolved in buffer at pH 7, the pH constantly decreased to reach pH 6-6.5. After four successive stabilizations, the pH remained stable for 24 h and the solution appeared very clear and could be filtered at 0.2 µm (up to 4% w/w). The repetitive pH adjustments decreased the initial viscosity by 95% and the molar mass, Mn, by 51% (Figure 2). Similar decreases in viscosities and molar mass were measured when PGA solutions were stabilized at different pH values (3-10). The higher the pH, the lower the molecular weight and the lower the viscosity (data not shown). As the pH is increased, basic hydrolysis probably occurs, breaking internal bonds that could have been previously formed by intramolecular and intermolecular transesterification, while splitting off some ester groups. The PGA chains separate from each other and become more linear and fewer aggregates are visible; therefore, the solution becomes clear, less viscous (Figure 2), almost Newtonian (data not shown), and filterable. The treatment of dilute PGA solutions with NaOH results in a partial saponification of esters. In addition there is a side reaction involving β-elimination which leads to some degradation (34) Sinquin, A.; et al. Rheological properties of semidilute aqueous solutions of hydrophobically modified propylen glycol alginate derivatives. Colloids Surf. A-Physicochem. Eng. Aspects 1996, 112, 193-200.

Langmuir, Vol. 21, No. 21, 2005 9767

Figure 2. Effect of pH stabilization on the viscosity and the molecular weight of a 3% PGA solution. The control solution was 3% PGA in 0.9% NaCl. To enable extrusion and filtration, PGA was dissolved in buffer (20 mM MOPS) and subsequently stabilized at pH 7 until pH was stable (4 stab). Viscosities (Dots) are given for a shear rate of 100 s-1. The average molecular weight is calculated by number, Mn (dashed box), and by weight, Mw (open box).

Figure 3. Viscosity (open box) at 100 s-1 and surface tension (dots) for different solutions of alginate-PGA-BSA-PEG. The standard solution was composed of 1.8% PGA and 1.2% alginate. The second solution contained 1.8% PGA, 1.2% AG, and 4% BSA, the third 1.8% PGA, 1.2% AG, 4% BSA, and 1% PEG, and the final solution had the same composition as the second although PGA but was not previously stabilized.

of alginate.35 Moreover, separation of aggregate coils occurs. As a consequence, a very marked drop in viscosity is observed. This hypothesis of aggregate separation is confirmed by the change in molecular weight. Indeed, if only complete de-esterification (saponification) occurred during the pH adjustments, the initial molecular weight would vary by 20% (according to a degree of esterification of 80% and assuming all ester groups are transformed in carboxyl) instead of 51%, suggesting that large aggregates separated into two single PGA chains. Further evidence that total de-esterification is not occurring is given by measuring mechanical properties of capsules composed of standard PGA and capsules composed of stabilized PGA. If all propylene glycol groups dissociated, the transacylation reaction would no longer be feasible. For these reasons, it was necessary to stabilize the pH (pH 7) by addition of the minimum amount of NaOH, such that filtration could be achieved. The new polymer solution, prepared with stabilized PGA, BSA, and alginate, had a viscosity that enabled stable jet-formation using a 400 µm nozzle (Figure 3). However, the gelled droplets presented pointed extremities, probably due to the high surface tension and adhesive properties of BSA. The addition of 1% PEG 4 kDa to the polymer solution resulted in an 8% decrease in surface tension and a 25% decrease in viscosity (Figure 3), such that the extrusion speed and adherence of the solution to the nozzle was reduced. PEG was also added to the gelation solution (CaCl2) resulting in the formation of spherical beads with a smooth homogeneous surface. (35) McDowell, R. H. New Reactions of Propylene Glycol Alginate. J. Soc. Cosmetic Chem. 1970, 21 (7), 441-&.

9768

Langmuir, Vol. 21, No. 21, 2005

Figure 4. Effect of PGA stabilization on the mechanical resistance of capsules. The capsules were made of 1.2% alginate, 4% BSA, 1.8% PGA, and 1% PEG and presented an average radius of 1.5 mm. The standard solution contained PGA in 0.9% NaCl (saline), the second one PGA in 20 mM MOPS buffer, the third solution contained PGA stabilized once at pH 7 and the fourth solution contained PGA stabilized 4 times at pH 7. Error bars represent triplicate measurements of an average of 50 capsules.

Effect of Reduction in Size on Capsule Composition. As previously explained, the polymer solution was optimized by reducing the PGA viscosity and by reducing the surface tension. To verify that the capsule mechanical properties were unaffected by PGA stabilization, the effect on the capsule structure was tested for large capsules (diameter 3 mm). The stabilization probably helped to break the PGA aggregates, making more ester groups available for the transacylation reaction. When PGA was further stabilized, the mechanical resistance slightly decreased (20%), due to partial saponification,35 which consequently reduced the transacylation reactivity. However, after the second stabilization, the mechanical resistance remained constant (60 g/capsule) even though a higher amount of NaOH was added (Figure 4). These results show that complete de-esterification did not occur during pH stabilization. In conclusion, the procedure for PGA stabilization at pH 7 enabled filtration and extrusion of the polymer solution, while ensuring a high mechanical resistance of the capsules. Reduction in size strongly affected the leaching of BSA from beads immediately after extrusion in the gelation bath. Such protein leaching was already observed with large beads (diameter 3 mm); however, at the end of the gelation period (10 min), the amount that diffused out corresponded to only 15% of the initial amount of BSA (data not shown). With microbeads (diameter 9.

coefficient of NaOH through the gel. The diffusion coefficient can be approximated to the one in water (1.51 × 10-5 cm2/s).38 The concentration of NaOH at the bead is constant (CS) and equal to the bulk concentration (it is assumed that the bulk solution is perfectly mixed). The differential equation [eq 1] can be integrated as follows,33 giving the concentration of NaOH as a function of radius and time (non steady state):

[

C ) CS 1 +

2R

∞

∑

(-1)n

πr n)1 n

sin

nπr R

(

exp -

)]

Dn2π2t R2

(2)

where R is the total radius of the bead and concentration of NaOH is 0 when t ) 0. Equation 2 has been solved as a function of time for beads (diameter 1.16 mm) composed of 1.2% AG, 1.8% PGA, coated with 0.05% PLL and incubated in a solution of 0.02 M NaOH. As illustrated in Figure 6, the NaOH concentration reaches 0.002 M (equivalent to pH 9 in our buffered conditions) at a distance of 62% of the bead radius after only 10 s of reaction time. After exposure to a solution at pH 9, cells were shown to be irreversibly damaged. As a result, cells would be killed in 94% of the bead volume after 10 s of the transacylation reaction. For larger beads (diameter 3.6 mm) the volume containing dead cells decreases to 36% for the same 10 s reaction time. These calculations indicate that diffusion of NaOH becomes difficult to control when the surface/volume ratio is increased, making this encapsulation method not suitable for animal cells. We should however minimize the consequences of this conclusion, as reaction was neglected. Then it can be deduced that the effective time to reach such a concentration is higher, since NaOH is consumed during the transacylation reaction. Diffusion of NaOH through BSA-containing capsules was simulated using the shrinking core model, which is used in chemical reaction engineering for noncatalytic reactions with particles of constant size in a surrounding fluid.39 In this model, the transacylation reaction takes place first in the external part of the particle. The reaction surface then progresses through the bead, leaving an external part completely converted (the membrane) and a nonreacted internal part (core). This diffusion-reaction process is controlled by NaOH which (1) diffuses through(38) Bennet, C. O.; Myers, J. E. Momentum, Heat and Mass Transfert, 3rd ed; McGraw-Hill chemical engineering series; McGraw-Hill: New York, 1982. (39) Levenspiel, O. Chemical reaction engineering, 3rd ed.; Wiley: New York, 1999.

9770

Langmuir, Vol. 21, No. 21, 2005

Breguet et al.

out the bead and (2) reacts by forming a membrane on the surface of the core, resulting in shrinking of the core. Mass transfer through the membrane is assumed to be the controlling step, the reaction being instantaneous. The moving NaOH boundary can be expressed by a mass balance [eq 3]

4 dN ) Fcored πrc3 3

(

)

(3)

where N(r,t) is the number of moles of NaOH diffusing, rc is the nonreacting core radius, and F is the density of the reacting groups in the core (mol/m3). As the reaction surface moves forward due to the diffusion of NaOH, both N and rc change with time. However, the pseudo-stationary-state (PSS) approximation can be applied, by considering the shrinkage of the nonreacting core, drc/dt, to be slower than the diffusion of NaOH through the membrane, dN/dt. It has been shown40 that if the ratio of solute concentration to reacting solid concentration is less than 0.1, and rc/R is greater than 0.5, all results (from the PSS approximation and from rigorous solution) give nearly the same values. Since the NaOH concentration is much lower than the PGA, PLL, and alginate concentrations, the PSS assumption can be applied to simulate the transacylation reaction. As a first approximation, the nonreacting core can be considered to be stationary and the rate of diffusion constant at any value of r [eq 4]

dC dN ) -4πr2Deff ) const dt dr

(4)

where C(r,t) is the concentration of NaOH, r the distance from the center and Deff is the effective diffusion coefficient. Second, the nonreacting core shrinks as soon as NaOH disappears [eq 5]

drc dN ) F4πrc2 dt dt

(5)

When eq 4 is integrated and substituted into eq 5, the final expression for reaction time, t, as a function of membrane thickness, s, can be calculated [eq 6]

t)

R2 R-s2 R-s 1-3 +2 DeffCs 6 R R F

(

(

)

(

3

) ))

R-s2 R-s 1 R2 1-3 +2 A 6Cs R R

(

(

)

(

3

) ) (6)

This expression has been correlated to experimental data (Figure 7) in order to determine the coefficient A (Deff/F). Experimental data were obtained by measuring the membrane thickness as a function of transacylation time for large capsules (3.6 mm) containing 2% PGA, 1%AG, and 2% BSA.28 It was not possible to collect sufficiently reliable data using small capsules (diameter 1 mm) since the reaction is very short (100 mL beads), such a rapid pH shift (10 s) under sterile conditions might be difficult to carry out since homogeneous mixing becomes less efficient and the dynamics of neutralization is slowed. Capsules were maintained for 1 month in spinner flask cultures, during which cells colonized the core up to a value of 1.4 × 107 cell mL-1 (Figure 9). This value is 8-fold lower than that obtained in PLL-alginate capsules6 and is probably due to the high viscosity of the internal core and/or a partial networking of the core. After one month in repetitive fed-batch mode, cell viability was 30%. During the culture, capsules were exposed to shear stresses due to stirring and to destabilization by five volume changes of medium. Furthermore, capsules were in close contact with proteolytic enzymes released by the cells, which may also degrade the BSA-PGA network.25 However, the mechanical resistance of the capsules decreased by only 10%, showing that long-term stability could be reached by using a combination of natural and synthetic peptides. The synthetic polyamine provided the thin rigid outer membrane and BSA the thick soft membrane. Long-term stability under physiological conditions is a key parameter for application of encapsulated cells in extended bioreactor cultures. To compare the stability under working conditions, different types of capsules formed from electrostatic or covalent bonds were stored in three solutions: (1) 0.9% NaCl, (2) fresh CHO culture medium, and (3) spent CHO culture medium which contained lactate, cell debris, various proteins, and enzymes. The effect of medium removal was subsequently studied to simulate continuous culture conditions. The results (Figure 10) show that storage during 30 days in a saline solution without medium removal already affects the mechanical properties. The standard electrostatic system PLL-AG and the original covalent system composed of BSA-PGA-AG both presented an approximate 20% decrease in resistance, whereas the resistance of the PLLPGA-AG system decreased by 10%. However, when saline medium was removed and replaced repetitively 15 times, the mechanical resistance of all types of capsule dropped by 20-30% (Figure 10). It is assumed that successive medium replacements displaced the equilibrium between the surrounding medium and the membrane composition,

Langmuir, Vol. 21, No. 21, 2005 9771

Figure 10. Mechanical stability as function of storage conditions for three types of capsules: (1) standard covalent capsules with BSA, (2) covalent capsules with a synthetic polyamine (PLL), and (3) the classical polyelectrolytic capsules composed of alginate and PLL. The three physiological media were: saline solution (0.9% NaCl); fresh cell culture medium and spent culture medium. Capsules stored for 30 days in the same medium were compared to capsules stored in a medium that was replaced 15 times over a period of 30 days.

thus extracting one or more constituents of the outer surface. In the case of AG-PLL capsules, the polycations are probably replaced by sodium ions, thereby rendering the alginate complex less mechanically resistant. It is also possible that the osmotic pressure introduced during medium removal also play a role. As a consequence, medium removal is a way to amplify the de-stabilization process. Storage in sterile culture medium (with removal) resulted in up to a 70% decrease in resistance for the AG-PLL system, whereas for the two covalent systems (Figure 10), the stability was similar to that obtained for saline solution ((15%). These results indicate that saline solutions cannot systematically be used as a reference medium for testing the mechanical stability under physiological conditions (300 mOs mol).41 Indeed, culture media are composed of buffers, salts, amino acids, vitamins, and carbon sources. Ions that are generally present at a high concentration include Na+, Cl-, CO32-, L-glutamine, and L-lysine. All of these components may destabilize the membrane if they have strong affinities for polycations or polyanions. The aim of the stability test in spent culture medium (Figure 10) was to investigate if byproducts or secreted enzymes had an effect on mechanical properties of the capsules. The electrostatic system AG-PLL showed the lowest resistance (65% destabilization), followed by the AG-PGA-BSA system with a 43% decrease in mechanical stability. Spent medium reduced the resistance of the AGPGA-PLL system by only 17%. The sensitivity of AGPGA-HSA (human serum albumin) capsules toward proteases (trypsin and pepsin) has been reported25 with rapid digestion of the gel (2-4 h) even if proteins were cross-linked to polysaccharides. However, the hydrolytic action of trypsin is known to be slowed when amino groups of lysine are substituted;42 therefore, it was expected that the replacement of BSA by PLL would result in a higher stability toward enzymatic degradation. The destabilization of the AG-PGA-BSA system (Figure 10) confirmed that the membrane could still be damaged by enzymes acting on a single membrane component, even if the latter is covalently cross-linked to other polymers. For a higher long-term stability in cell cultures, it is therefore essential (41) Dos Santos, V.; Leenen, E.; Ripoll, M. Relevance of rheological properties of gel beads for their mechanical stability in bioreactors. Biotechnol. Bioeng. 1997, 56, 517-529. (42) Desnuelle, P. The Enzymes of lipid metabolism. in Sixth International conference on the biochemistry of lipids; Pergamon Press: Oxford, U.K., 1960.

9772

Langmuir, Vol. 21, No. 21, 2005

to use synthetic polyamino acids or synthetic cationic polymers that are not susceptible to be attacked by proteolytic enzymes. Conclusions The combination of polyelectrolytic and covalent bonds for the formation of resistant and stable microcapsules was investigated and applied to a commercial vibrating nozzle encapsulation device. Reducing the size of microcapsules imposed modifications on the chemical composition of the extruded solution containing poly-ester (PGA) in order to reduce viscosity and to facilitate filtration, sterilization, and extrusion. The optimal composition of the polymer solution was 1.8% PGA, 1.2% AG, 4% BSA, and 1% PEG. Furthermore, BSA leaching increased with decreasing capsule size. Performing the encapsulation process in a single reaction vessel already containing BSA could simplify the production process while reducing BSA loss by up to 50%. A potential solution to this problem would be to use higher molecular weight proteins that diffuse through the bead at a lower rate. Transacylation was applied to covalently bind ester groups of PGA to amino groups of PLL and BSA. Although the reaction between PLL and alginate proved to be successful for electrostatic bonds, a subsequent transacylation induced a 5-fold increase in mechanical resis-

Breguet et al.

tance. These results clearly indicated that transacylation could be applied both to polymers (PLL) and proteins and that only part of the amino groups of PLL was electrostatically bound to alginate. Reducing the size of microcapsules resulted in a reduction of the transacylation reaction time. By modeling the reaction, it was demonstrated that, in the absence of BSA, the reaction would not be controllable and result in the death of most of the immobilized cells. A protein present in the whole bead volume could effectively slow the diffusion of NaOH, although the reaction time remained very short (10 s for a 100 µm membrane). In conclusion, reaction time and pH must be controlled very precisely during the process in order to avoid complete gelation of the bead and total cell death. The optimized process conditions enabled encapsulation of CHO cells, without affecting viability of cells within the core. This experiment shows that it is possible to immobilize fragile cells in harsh conditions, while generating covalent membranes with long-term stability. The principle of increasing the stability by the combination of a natural protein and a synthetic polyamino acid proved to be effective under physiological conditions and makes these capsules potentially interesting for long-term cultures in bioreactors or for use as implants. LA0512796