Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

High density animal cell culture systems using porous supports Lee, Daniel W. 1993

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_1993_spring_phd_lee_daniel.pdf [ 11.72MB ]
Metadata
JSON: 831-1.0058547.json
JSON-LD: 831-1.0058547-ld.json
RDF/XML (Pretty): 831-1.0058547-rdf.xml
RDF/JSON: 831-1.0058547-rdf.json
Turtle: 831-1.0058547-turtle.txt
N-Triples: 831-1.0058547-rdf-ntriples.txt
Original Record: 831-1.0058547-source.json
Full Text
831-1.0058547-fulltext.txt
Citation
831-1.0058547.ris

Full Text

HIGH DENSITY ANIMAL CELL CULTURE SYSTEMS USING POROUS SUPPORTS by  Daniel W. Lee B.A.Sc., The University of British Columbia, 1987 M.A.Sc., The University of Waterloo, 1989 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES DEPARTMENT OF CHEMICAL ENGINEERING We accept this thesis as conforming to the required standard  THE UNIVERSITY OF BRITISH COLUMBIA October, 1992 © Daniel W. Lee, 1992  In presenting the thesis in partial fulfillment of the requirements for an advanced degree at the university of British Columbia. I agree that the Library shall make freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission  Department of Chemical Engineering The University of British Columbia, Vancouver, Canada Dec. 28th, 1992  DE-6 (2/88)  ii  ABSTRACT Monolithic porous ceramic and porous polystyrene microcarriers were examined as supports for large scale mammalian cell culture. Vero cells and transformed baby hamster kidney (BHK) cells which produced human transferrin were grown in three different reactor configurations: a fixed bed ceramic perfusion system, an airlift system with draft tube made of porous ceramic and a stirred tank configuration which used porous polystyrene microcarriers. The porous matrices provide an increased surface area for cell attachment and growth and protect the entrapped or immobilized cells from shear stress in the bulk fluid. Steady state cell concentrations in all three systems were found to be in excess of 10 8 cells per mL porous matrix. All three systems intensified the culture process by increased cell mass per unit volume — a minimum five—fold increase in reactor volumetric cell density over simple suspension cultures. The airlift and microcarrier stirred tank system offer the potential to scale—up by increasing reactor volume. The fixed bed ceramic perfusion system can be used as a cell propagator to produce the required inoculum for other large scale bioreactors. Unlike the Opticore of the Opticell systems, the ceramic foam element can be reused. Its multiple interconnected channel structure greatly reduces the possibility of channel blockage due to over—grown cells. The biologically inert porous polystyrene microcarriers tested have distinct advantages, in terms of product purification, over the collagen based porous microcarriers such as Cultispher—G and Informatrix microcarriers. Cell attachment rates onto the porous polystyrene microcarriers treated with sulphuric acid were comparable to those of Cytodex-1 and Cultispher—G particles while the cell growth and productivity per unit carrier volume were 20% superior to the two tested commercial microcarriers. The use of airlift eliminates the need for a separate oxygenator or spin filter for gas exchange. The input gas created a differential pressure drop across the porous draft tube and forced the medium to perfuse through the porous draft tube. A simple  111  mathematical model was formulated to describe the hydrodynamic behaviour of the porous draft tube airlift system. The model allows gas holdup, liquid superficial velocity, and liquid perfusion rate through the porous draft tube to be predicted for a given gas input. For the cases examined, the predictions are in satisfactory agreement with the overall trend of experimental measurements.  iv  TABLE OF CONTENTS ABSTRACT ^  ii  TABLE OF CONTENTS ^  iv  LIST OF TABLES ^  ix  LIST OF FIGURES ^  x  ACKNOWLEDGMENTS ^  xvi  1.0 INTRODUCTION ^  1  1.1 Current Trend to Intensify Cultures ^  2  1.2 Objectives ^  3  2.0 LITERATURE REVIEW ^  5  2.1 Bioreactor Design Criteria ^  5  2.1.1 Shear effect ^  5  2.1.2 Oxygen supply ^  7  2.1.2.1 Bubble—free oxygenation ^  7  2.1.2.2 Oxygen transfer using external oxygenators ^ 8 2.1.2.3 Direct sparging ^  8  2.2 High Density Immobilized Cell Culture Systems ^ 9 2.2.1 Microcarrier culture systems ^ 2.2.1.1 Conventional systems ^  10 10  2.2.1.2 Systems utilizing porous microcarriers ^ 12 (a) Verax ^  12  (b) Cultispher—G microcarrier ^ 13 (c) Siran porous glass beads ^ 14 (d) Informatrix porous microcarriers ^ 14 2.2.2 Microcapsulation ^  15  2.2.3 Fixed bed bioreactor ^  15  V 2.2.4 Hollow fiber systems ^ 2.3 Airlift Bioreactors ^ 2.3.1 Classification of airlift reactors ^  17 18 19  2.3.2 Applications of airlift in animal cell culture ^ 19 2.3.2.1 Cell retention in airlift reactors ^ 20 2.3.3 Flow patterns ^  21  2.3.4 Power input ^  23  2.3.5 Gas velocity ^  23  2.3.6 Hydrodynamics ^  24  2.3.6.1 Effect of sparger^  24  2.3.6.2 Dispersion characteristics ^  25  2.3.6.3 Liquid circulating velocity, UL ^ 26 2.3.6.4 Gas holdup, 6g ^  29  2.3.6.5 Mass transfer ^  31  3.0 EXPERIMENTAL MATERIALS AND METHODS ^  35  3.1 Cells and Cell Maintenance ^  35  3.2 Cell Culture Systems ^  35  3.2.1 Ceramic foam ^  35  3.2.1.1 Fixed bed perfusion system ^ 36 3.2.1.2 Airlift system ^ 3.2.2 Microcarriers ^  38 42  3.2.2.1 Pretreatment of microcarrier particles. ^ 43 3.2.2.2 Microcarrier culture ^  45  (a) Roller culture ^  45  (b) Spinner cultures ^  46  3.2.3 Suspension cell culture ^ 3.3 Analytical Methods ^  47 47  vi 3.3.1 Cell numeration ^  47  3.3.1.1 Porous matrix ^  47  3.3.1.2 Microcarriers ^  48  3.3.1.3 Suspension cells ^  48  3.3.2 Assays ^  48  3.3.2.1 Glucose ^  48  3.3.2.2 Lactate ^  48  3.3.2.3 Lactic dehydrogenase (LDH) ^ 49 3.3.2.4 Transferrin ^  50  (a) Enzyme—linked immunosorbent assay (ELISA) ^ 50 (b) Particle concentration fluorescence immunoassay (PCFIA) ^ 3.3.3 Microscopy ^  50 51  3.3.3.1 Scanning electron microscopy (SEM) ^ 51 3.3.3.2 PEG embedding technique ^ 51 3.3.3.3 Thin—sectioning microscopy ^ 52 3.3.3.4 Confocal microscopy ^  53  4.0 RESULTS AND DISCUSSION —Fixed Bed Ceramic Foam Perfusion System ^  54  4.1 Effect of Foam Porosity ^  54  4.2 Cell Growth on the Ceramic Surface ^  55  4.3 Estimation of Total Cell Number ^  57  4.4 Lactate Production ^  60  4.5 Stability and Viability of Cells on the Matrix ^  62  4.6 Effect of Serum during Stationary Growth Phase in the Perfusion Reactor^ 4.7 Effect of Zinc on Transferrin Production ^  64 65  vii 4.8 Perfusion Propagator ^ 5.0 RESULTS AND DISCUSSION — Airlift System ^  66 69  5.1 Pressure Drop ^  70  5.2 Mathematical Modeling ^  76  5.3 Mass Transfer Coefficient ^  83  5.4 Scale—up Potential Assessed by Proposed Model ^ 85 5.5 Long Term Culture ^  88  6.0 RESULTS AND DISCUSSION — Porous Microcarriers ^ 91 6.1 Cell Attachment Rate ^  92  6.1.1 Effect of surface chemistry group modifications ^ 92 6.1.2 Cell attachment/entrapment rate for various microcarriers ^ 97 6.1.3 Sulphuric acid treatment ^  98  6.1.4 Influence of inoculation procedure ^  99  6.1.5 Effect of particle diameter ^  101  6.1.6 Influence of the inoculum cell concentration ^ 102 6.1.7 Effect of medium composition on cell attachment rate ^ 103 6.2 Long Term Microcarrier Cultures ^ 6.2.1 Biomass evaluation ^  104 105  6.2.2 Cell growth on Polyhipe in roller bottles ^ 108 6.2.2.1 Vero cell growth, effect of carrier surface modification ^  108  6.2.2.2 BHK cell growth, effect of carrier surface modification ^  111  6.2.2.3 Transferrin production ^  114  6.2.3 Particle pore size effect ^  115  6.2.4 Minimum cell inoculum requirement ^ 123 6.2.5 Cell growth on S40 Polyhipe in spinners ^ 127  viii 6.2.6 Cell penetration depth ^  129  6.2.7 Hybridoma cell growth in Polyhipe particles ^ 133 7.0 RESULTS AND DISCUSSION —Effect of Culture Systems on Cell Growth and Cell Productivity ^  137  7.1 Comparison of Cell Specific Transferrin Productivity of Different Culture Systems ^  137  7.1 Comparison of Cell Loading and Large Scale Cell Culture Suitablity ^ 143 8.0 CONCLUSION & RECOMMENDATIONS  ^  147  NOMENCLATURE ^  151  REFERENCES ^  154  APPENDICES ^  169  Appendix 1. Model PC-61 A/D board data logging program ^ 169 Appendix 2. Anglican controller data logging program listing ^ 173 Appendix 3. MathcadTM airlift model program listing ^ 174 Appendix 4 Sample Calculation ^  177  Appendix 5 Raw data ^  180  ix  LIST OF TABLES Table 1. Typical desirable conditions for mammalian cell growth ^ 2 Table 2. Commercially available microcarriers  ^ 11  Table 3. Properties of microcarrier beads tested ^  44  Table 4. Surface characteristics of the Polyhipe particles ^ 45 Table 5. Surface characteristics of S microcarriers ^  98  Table 6. Comparison of cell number (cells / mL beads) by alternative methods ^ 107  x  LIST OF FIGURES Figure 1. Schematic illustration of Opticell bioreactor ^ 16 Figure 2. Typical airlift bioreactor configuration ^  18  Figure 3. Airlift bioreactor configurations — (a) external loop (b) internal loop airlift ^  19  Figure 4. Ceramic foam cylinders of 30,50 and 100 PPI ^ 36 Figure 5. Schematic showing perfusion system ^  37  Figure 6. Schematic showing airlift system ^  39  Figure 7. Open structure of a macroporous microcarrier P40 particle ^ 42 Figure 8. Glucose utilization of BHK cells in batch culture on ceramic foam cylinders of various porosity ^  55  Figure 9. SEM photographs of BHK cells grown on ceramic surface. ^ 57 Figure 10. Oxygen uptake rate and estimated cell number based on the oxygen uptake rate ^  59  Figure 11. Glucose utilization of BHK cells in the perfusion system with 5% FCS and a perfusion rate of 128 mL,/min ^  60  Figure 12. Lactate production based on glucose utilized for BHK cells in the perfusion system. ^  61  Figure 13. Correlation between measured LDH activity and disrupted BHK cell concentration ^  62  Figure 14. Effect of fetal calf serum (FCS) concentration on steady—state transferrin production rate at a perfusion rate of 128 mL/min in DMEM medium ^ Figure 15. Induction of transferrin production. ^  64 66  xi Figure 16. Vero cell growth on ceramic perfusion system with repeated ^  harvesting.  68  Figure 17. Measured pressure drops across the various draft tubes ^ 72 Figure 18. Effect of porous draft tube length on the pressure drop across the ^ 73  thick wall 30 PPI porous draft tube Figure 19. Effect of porous draft tube length on the pressure drop across the thick wall 100 PPI porous draft tube ^  74  Figure 20. Effect of serum concentration on differential pressure drops across the thin wall 100 PPI porous draft tube, measured at the base of the reactor, at various riser superficial velocities ^ 74 Figure 21. Effect of serum concentration on differential pressure drops across the thin wall 30 PPI porous draft tube, measured at the base of the reactor, at various riser superficial velocities ^ 75 Figure 22. Effect of serum concentration on differential pressure drops across the thick wall 100 PPI porous draft tube, measured at the base of the reactor, at various riser superficial velocities ^ 75 Figure 23. Effect of serum concentration on differential pressure drops across the thick wall 30 PPI porous draft tube, measured at the base of the reactor, at various riser superficial velocities ^ 76 Figure 24. Pressure drop across the thick—wall porous draft tube, measured at the base of the reactor, versus riser superficial air velocity for different pore spacings  ^  80  Figure 25. Pressure drops across the thin—wall porous draft tube at the base of the reactor versus riser superficial air velocity for different pore spacings  ^  81  Figure 26. Average downcomer liquid superficial velocity at various riser gas superficial velocities ^  82  xii  Figure 27. Volumetric mass transfer coefficient at various riser gas superficial velocities ^  83  Figure 28. Effect of 5% serum addition on mass transfer coefficient ^ 84 Figure 29. Comparison of predicted mass transfer coefficients with measured values for airlift with 100 PPI thick—wall draft tube of 360 mm ^ 85 Figure 30. Calculated average perfusion velocity through the porous matrix versus draft tube wall thickness for the tested 5 L airlift with various riser area, AR ^  87  Figure 31. Cumulative glucose used and lactate produced by BHK cells in the porous draft—tube airlift bioreactor ^  90  Figure 32. Effect of surface modifications on BHK cell attachment rate ^ 94 Figure 33. Effect of surface modifications on Vero cell attachment rate ^ 94 Figure 34. Effect of surface modification on Vero cell attachment to polystyrene microcarriers ^  96  Figure 35. Attachment of BHK cells to treated polystyrene microcarriers ^ 96 Figure 36. Vero cell attachment to various microcarriers. Equal masses of each type of microcarriers (9 g/L) were used ^ Figure 37. BHK cell attachment to microcarriers ^  97 98  Figure 38. Attachment of BHK cells to sulphuric acid treated Polyhipe microcarriers as a function of time following treatment length of 15, 60 and 240 min ^  99  Figure 39. Attachment of Vero cells to sulphuric acid treated Polyhipe microcarriers as a function of time following treatment length of 15, 60 and 240 min. ^ Figure 40. Cell attachment to dry particles ^  100 101  Figure 41. Effect of particle diameter on the attachment rate of Vero cells to Polyhipe particles treated with sulphuric acid ^ 102  ^  Figure 42. Inoculum concentration effect on the Vero cell attachment rate ^ 103 Figure 43. Effect of medium composition on Vero cell attachment/entrapment rates ^  104  Figure 44. Effect of medium glucose concentration on BHK cell glucose uptake rate. ^  106  Figure 45. Glucose concentration of BHK CM80 roller culture in fed—batch operation ^  107  Figure 46. Glucose concentrations in semi—continuous perfusion of Vero cells on CM80 ^  108  Figure 47. Vero cells on DEA80 particles ^  109  Figure 48. Cumulative glucose utilization by Vero on particles with different modifications ^  110  Figure 49. Calculated Vero cell densities on various Polyhipe particles based on glucose utilization rates ^  110  Figure 50. Cumulative glucose used by BHK cells grown on 5 g/L of microcarriers ^  111  Figure 51. Cumulative glucose uptake of BHK cells on various types of microcarriers with a concentration of 2 g/L ^  112  Figure 52. Microcarrier (5g/L) cumulative glucose consumption of BHK roller cultures in DMEM medium ^  114  Figure 53. Cumulative transferrin produced in BHK roller cultures containing 2 g microcarrier/L DMEM medium, supplemented with 10 ,uM zinc ^ 115 Figure 54. Total glucose used by BHK cells on polystyrene particles of different chamber sizes  116  Figure 55. Cumulative glucose used for BHK polystyrene microcarrier spinner culture ^  117  Figure 56. Cumulative glucose consumed by Vero cells on P microcarriers ^ 118  xiv Figure 57. Cumulative lactate produced by Vero cells on P microcarriers ^ 119 Figure 58. Confocal images of a P80 particle ^  120  Figure 59. Thin sectioned microscopy images of Vero cells on polystyrene particles ^  122  Figure 60. Cumulative glucose used by Vero cells on Cytodex-1 microcarriers ^ 124 Figure 61. Cumulative glucose used by Vero cells on S40 microcarriers ^ 124 Figure 62. Cumulative lactate produced by Vero cells on Cytodex-1 microcarriers ^  125  Figure 63. Cumulative lactate produced by Vero cells on S40 microcarriers ^ 125 Figure 64. Scanning electron micrographs of Vero cells on S40 carrier after 600 h culture time ^  126  Figure 65. Cumulative glucose used and lactate produced by BHK cells on microcarriers (2 g/L) in 200 mL medium ^  127  Figure 66. Glucose utilization rates of various BHK microcarrier spinner cultures ^  128  Figure 67. Unit carrier cumulative glucose used by BHK cells on microcarriers ^ 129 Figure 68. Scanning electron micrograph of S40 particles ^ 132 Figure 69. Hybridoma cells growing within the pores of chloromethyl Polyhipe particle ^  135  Figure 70. Enlarged view of hybridoma cells growing in a pore of a chloromethyl Polyhipe particle ^  135  Figure 71. Growth of hybridoma cells with chloromethyl Polyhipe (5 g/L) in a roller bottle ^  136  Figure 72. Cumulative glucose utilization and antibody production rate of hybridoma cells entrapped in 5 g/L of CM80 macroporous beads ^ 136 Figure 73. Transferrin production by BHK cells in spinner suspension culture ^ 137  XV  Figure 74. Total transferrin produced by BHK on microcarriers (2 g/L) in 200 mL medium ^  138  Figure 75. Transferrin production per unit carrier volume by BHK cells on various microcarriers ^ Figure 76. Cumulative glucose used by BHK cells ^  139 140  Figure 77. Cumulative transferrin produced by BHK cells from various culture systems ^  142  Figure 78. Glucose to lactate conversion ratios for various BHK cultures ^ 142 Figure 79. Transferrin produced per g glucose utilized for various BHK culture systems ^  146  xvi  ACKNOWLEDGMENTS First, I would like to thank my supervisors, Dr. Doug Kilburn and Dr. John Grace for their useful suggestions, support and encouragement throughout this work. I would wish to thank my committee members, Dr. Norm Epstein and Dr. Ross MacGillivray for useful discussions.  I would also like to express my appreciation to all my friends in the two departments (Chemical Engineering and Microbiology) and in the Biotechnology Laboratory, especially to Dr. James Piret and Eric Jervis, for their help and useful suggestions.  The exceptional technical assistance provided by the technicians and staff in the Department of Chemical Engineering, Department of Microbiology, and Biotechnology Laboratory, especially by Doug Haddow, Randy Dean and Gary Lesnicki, is also gratefully acknowledged.  Financial support from the B.C. Science Council under a Science and Technology Development Fund core grant and the British Columbia Foundation for Non—animal Research is gratefully appreciated.  1.0 INTRODUCTION Many medically important pharmaceutical proteins, such as tissue plasminogen activator (tPA), cannot be produced in microbial systems by recombinant DNA technology and can only be expressed using animal cells as hosts. Such proteins made by bacteria might not be folded in the proper configuration; addition of sugar, phosphate or alkyl groups to the basic amino acid backbone might not be done properly. Although cell culture has a number of similarities to microbial fermentations used for brewing and antibiotics manufacture, there are important differences. Animal cells are much more sensitive to conditions in their environment; animal cells are characterized by their fragility due to their relatively large size (typical diameter about 15 ,um) and their lack of a protective cell wall. The mammalian cell bilayer membrane provides little protection against external disturbances such as turbulence in the external fluid. In addition to physical fragility, several other limitations of animal cells restrict their productivity in conventional large scale fermenters: (1) ill—defined nutritional requirements; (2) relatively long doubling time (typically 10 to 24 h); (3) growth to relatively low cell concentration (typically 10 6 to 10 7 cells/cm 3) in conventional cultures; (4) lack of quantitative information on the kinetics of product formation, nutrient utilization and formation of inhibitory metabolites. These limitations make animal cells more difficult to grow and the design of bioreactors for their growth more critical. Some of these limitations can be overcome by appropriate bioreactor design.  ^  A well designed bioreactor should expose the cells to low levels of shear and provide sufficient mass transfer and mixing so that there is an adequate chemical cell culture environment throughout. To successfully grow animal cells in bioreactors requires that the key chemical and physical aspects of their native environment be reproduced faithfully in the bioreactors. Table 1 lists the typical environmental requirements for mammalian cell growth. Table 1. Typical Environmental Conditions for Mammalian Cell Growth Temperature^32 to 40, usually 37°C 7.0 to7.5 pH^ Fluid shear stress^< 2 N/m 2 Nutrients —dissolved oxygen^30 to 80% air saturation —glucose^0.1 to 4.0 g/L Metabolic products —ammonium ion^< 4 mM —lactate^not critical if pH controlled between 7.2-7.5 Satisfying these key requirements becomes increasingly difficult as the cell concentration increases. Process intensification to improve reactor productivity involves increasing the cell concentration which requires higher rates of mass transfer. In conventional bioreactor systems, the improvement in mass transfer characteristics is achieved by increasing the agitation and gas sparging rate. Both changes increase the shear stress on the cells in the reactor. 1.1 Current Trend to Intensify Cultures  Less than a decade ago, the primary means of producing large amounts of biologicals using animal cells was the roller bottle. In a few cases, simple stirred vessels were used. With the increasing number of valuable therapeutic proteins being produced by genetically manipulated animal cells, various bioreactors have been developed. Cell immobilization has proven to be one of the most effective  ways of increasing cell concentration and prolonging the protein production period. It allows continuous or semi—continuous perfusion of medium without washing the cells out of the reactor. During the stationary growth phase, the need for serum and other complex medium components is reduced due to increased concentration of cell—derived products (growth factors,  etc.),  while the  concentration of product is generally increased at high cell concentration (Lydersen, 1987; Croughan et al., 1988). Some current immobilized cell culture systems provide good mass transfer characteristics and provide physical protection to the fragile animal cells  (e.g.  by encapsulation). However, most of these current  systems suffer from overly complex operating procedures, and expensive cell immobilization materials; most are not readily scaled up. Bioreactor efficiency must be increased further in order to produce the therapeutic proteins economically.  1.2 Objectives The overall objective of this work was to investigate new methods for large— scale culture of animal cells at high cell concentration. Such cultures require some method of immobilizing the cells within the culture. The use of porous media, either as monolithic blocks or as particles, was investigated. These porous materials offer significant advantages over many conventional non—porous supporting substrates. The specific objectives of the study were: (1) To design a new, high cell density bioreactor for the growth of anchorage dependent cells using fixed ceramic foam matrix. (2) To investigate the use of a new porous chemically modified polystyrene material as a microcarrier for cell culture.  The bioreactors were designed to provide high surface to volume ratios for cell growth, an adequate oxygen supply without direct contact between the cells and gas—liquid interfaces and minimum shear. Three different bioreactor designs (packed bed, airlift and stirred tank configurations) using two different inorganic porous substrates (ceramic and polystyrene) for cell culture were examined in this study. A further objective was to formulate a mathematical model to facilitate scale—up of the airlift bioreactor.  2.0 LITERATURE REVIEW 2.1 Bioreactor Design Criteria A well designed bioreactor for animal cell culture must satisfy many requirements. Two of the key design criteria for large—scale bioreactors are: (1) minimal shear induced stresses, and (2) sufficient supply of oxygen to maintain high cellular productivity and yield.  2.1.1 Shear effect It has been realized that shear can influence cell culture processes in various ways — fluid mixing, cell suspension, mass transfer, productivity, cell viability, cell growth and cell to cell or cell—to—substrata adhesion (Bliem and Katinger, 1988; Nollrty et al., 1991; Ludwig et al., 1992; Tramper and Vlak, 1988). Shear stress can be associated with either liquid motion or gas—liquid interfaces. In the absence of gas sparging, critical stirring speeds from 60 to 400 rpm have been reported beyond which cell death occurred (Hirtenstein et al. , 1980; de St. Groth, 1983; Telling and Radlett, 1971). The wide range of stirring speeds reflects differences in reactor geometry and cell lines and emphasizes the fact that agitation rate itself, without reference to vessel volume, impeller size and other geometric factors, does not quantify the hydrodynamic stress. Stathopoulos and Hellums (1985) found that liquid—induced shear stress above 2.6 N/m 2 caused a marked reduction in the viability of human embryonic kidney cells. At a lower shear stress (0.65 N/m 2) urokinase production was actually stimulated compared to no shear stress controls; similar findings were reported by Frangos et al. (1985, 1988). Shear stress greater than 1 to 5 N/m 2 was reported to be detrimental to animal cells in suspensions. A critical shear stress of 0.75 to 1.0 N/m 2 reported for adherent BHK cells (Ludwig et al., 1992).  was  Handa et al. (1987) studied the gas—liquid interfacial effect on the viability of hybridoma cells in bubble columns. They concluded that the survival of hybridomas depended on (1) cell type, (2) bubble size (smaller bubbles being more harmful), and (3) superficial gas velocity and bubble frequency. Emery et al. (1987) extended the investigation to include myeloma and baby hamster kidney (BHK) cells in various bubble columns of different aspect ratio (i.e. height—todiameter ratio). The findings of Emery et al. (1987) and Katinger and Scheirer (1982) support the hypothesis that cell death in sparged systems occurs mainly in the region of bubble disengagement from the free liquid surface. Overall cell viability was found to increase with increasing reactor volume due to reduction in the cell exposure time to disengaging bubbles. More recently, Handa—Corrigan et al. (1989) proposed two possible cell damage mechanisms associated with bubble disengagement from the liquid surface in sparged systems: damage due to rapid oscillations caused by bursting bubbles, and damage due to shearing in draining liquid film (or lamellae) in foams. Tramper et al. (1988) believe that, for cells in -suspension, shear stresses associated with liquid motion and rising of air bubbles are less than those associated with injection of air bubbles into the medium and their bursting at the surface. Kunas and Papoutsakis (1990) showed that in the absence of a vortex and bubble entrainment, hybridoma CRL-8018 cell damage due to stresses in the bulk turbulent liquid occurred only at very high agitation rates (above 700 rpm in a 2 L reactor with a 70 mm diameter impeller). They further showed that the entrainment and motion of very fine bubbles in the absence of a vortex did not cause growth retardation, even at an agitation rate of 600 rpm. This indicates that cell death in bioreactors, either with or without sparging, is primarily the result of air entrainment, bubble break—up and surface bursting. The elimination of direct  contact between cells and unstable gas—liquid interfaces should decrease cell damage in a bioreactor. 2.1.2 Oxygen supply  Dissolved oxygen is one of the most rapidly metabolized nutrients of mammalian cells. Initial concentrations of other nutrients such as glucose and amino acids are usually greater than ten—fold higher than that of oxygen. To sustain optimum cell growth, the dissolved oxygen concentration must be kept above some critical value. Reported values of critical dissolved oxygen concentration vary from 8 to 70% of saturation (Kilburn and Webb, 1968; Van Wezel and van der Velden de Groot, 1978; Radlett et al., 1972; Sinskey et al., 1981; Boraston et aL, 1984). The critical dissolved oxygen concentration depends on the cell line and is probably a function of the oxygen consumption rate. Reported oxygen consumption rates for animal cells vary from 0.04 to 0.5 ,umole 02 per 10 6 cells per hour (Spier and Griffiths, 1984; de Bruyne, 1988). A vast variety of systems have been developed to provide- oxygen in tissue culture bioreactors. 2.1.2.1 Bubble—free oxygenation  Given the damaging effects of gas bubbles, bubble—free aeration systems should offer significant advantages. Surface aerators which increase the turbulence of air—liquid interface can enhance oxygen mass transfer. However, scale—up would be problematic for surface—aerated systems of high aspect ratio. Surface aeration alone could not provide sufficient mass transfer for systems with high cell loading. An alternative method for aeration involves the use of gas permeable microporous membranes (Miltenburger and David, 1980). The immersed membrane increases the total surface area for oxygen transfer. The membrane is  often mounted onto a rotating shaft to enhance the mass transfer rate further. The feasibility of scale—up using multiple membranes in parallel has been demonstrated for reactors up to 150 L in volume (Lehmann et al., 1988). However, repeated autoclavings may cause pinholes in the membrane, hence increasing downtime and the maintenance costs. More exotic methods, such as the use of perfluorocarbon as oxygen carrier (Cho and Wang, 1989), have also been investigated. Although the costly perfluorocarbon particles can be recycled, the additional downstream recovery cost of the perfluorocarbon means that the technique is not feasible on an industrial scale (Yamaji et al., 1989).  2.1.2.2 Oxygen transfer using external oxygenators  Aeration systems in which cell free medium is continuously recirculated through an oxygenator are often used for cell cultivation (for examples in hollow fiber systems, OpticellTM, and Verax systems). High shear conditions for supplying oxygen can be used in the oxygenator (e.g. high fluid velocity, high speed stirring and direct sparging) without danger of cell damage. Such systems are relatively complex and require the use of a mechanical pump to circulate medium from the oxygenator to the bioreactor. This increases the risk of contamination and system breakdown.  2.1.23 Direct sparging  Direct sparging is the simplest method for providing oxygen to cell cultures. Foaming is inevitable due to the presence of proteins or serum in the culture medium. However, foaming can be decreased if the cells are grown in serum—free medium or in improved bioreactors with reduced serum requirements (e.g. Verax system). The addition of an antifoam agent can also control foaming (HandaCorrigan et al., 1989; Bently et al., 1989). However, the presence of an antifoam  agent complicates downstream processing. At the low levels of agitation usually employed in animal cell culture vessels, the entrapment and break—up of air bubbles is minimal. The oxygen transfer rate is limited by the short residence time and small interfacial area of the bubbles. A higher air flow rate or finer bubble size is required to achieve a higher oxygen transfer rate. Either of these measures may increase cell damage. Scale—up by increasing the reactor aspect ratio can increase the rate of oxygen transfer and the residence time of the bubbles without detrimental effects on cell viability (Boraston et al., 1984). Aspect ratios as high as 6:1 and 12:1 are sometimes used. Bubble columns and airlift systems have been used for animal cell cultivation with considerable success. However, most reported applications of direct sparging in these cell culture systems have been restricted to cells dispersed throughout the suspension at relatively low cell concentrations, i.e. 0.5-3 x106 cells/mL (Handa—Corrigan, 1988).  2.2 High Density Immobilized Cell Culture Systems The use of high cell density immobilized cultuie systems for large—scale growth of animal cells is becoming increasingly important. Immobilization of the cells provides an intrinsic separation of cells from medium which facilitates downstream processing. This together with reduced nutrient requirements, leads to significant economies of operation. Ideal immobilized cell culture systems should provide a high surface area for cell attachment, good mass transfer characteristics between the cells and the surrounding medium and a low liquid shear stress environment for cell growth. Current techniques for immobilizing cells usually utilize one of the following configurations: (A) Microcarriers in stirred tanks, e.g. Cytodex-1 (Levine et al., 1979), or macroporous gelatin beads (Nilsson et al., 1986; Reiter et al., 1990) or in fluidized beds, e.g. Verax system (Dean et al.,  10 1987) or Siran glass beads (Kratje et al., 1992). (B) Encapsulation in stirred tanks or airlift bioreactors (Lim and Sun, 1980; Bugarski et al., 1989; Kwong et al., 1989). (C) Fixed bed bioreactors — with ceramic matrices, e.g. OpticellTM system (Bognar et al., 1983), packed non—porous glass beads (Whiteside and Spier, 1981), porous glass beads (Kratje and Wagner, 1992), packed glass fibers (Perry and Wang, 1989), stainless steel matrices (Familletti and Fredericks, 1988), or polyurethane foam (Matsushita et al., 1990). (D) Hollow—fiber systems (Tharakan and Chau, 1986).  2.2.1 Microcarrier culture systems 2.2.1.1 Conventional systems Microcarrier particles are widely used for growing anchorage—dependent cells. The provision of sufficient culture surface area for cell attachment and growth is no longer a limiting factor with increased microcarrier loadings. Hence, relatively high cell density can be achieved under "homogenous" conditions. Diethylaminoethyl (DEAE) — Dextran beads with an optimized charge density (i.e. Cytodex-1) have been well characterized (Levine et al., 1979; Hu et al., 1985; Himes and Hu, 1987) and are among the most commonly used microcarriers in industry since the introduction of Cytodex-1. Many other types of microcarriers have been developed. Table 2 lists some commercially available microcarriers. Conventional microcarriers are generally used in stirred tank bioreactors. Aeration can be provided directly by sparging into a spin filter (Van Wezel, 1982; Tolbert et al. 1981). The cells on microcarriers are separated from the sparged medium by a rotating filter which is often fixed onto the agitator shaft stirrer to form a cage. The centrifugal force resulting from the rotational motion enhances mass transfer and also delays fouling of the filter membrane. The spin filter system is particularly effective for microcarrier systems (Cho and Wang, 1988). In  11 addition to isolating cells from gas sparging, it facilitates medium removal in continuous perfusion culture systems. For microcarrier cultures, direct sparging poses a more challenging problem. Microcarriers tend to rise to the liquid—gas interface carried by the air bubbles to remain there due to the small density difference between the microcarrier and the growth medium. This problem can be avoided by using a denser microcarrier. However, an increase in agitation level, resulting in an increase in the shear force, may be needed to compensate for the increased microcarrier density.  Table 2. Commercially available microcarriers Commercial Manufacturer Designation Cytodex 1 Non porous Pharmacia Cytodex 3 Pharmacia  Porous  Nunc Lux Solo Hill Solo Hill  Biosilon Cytosphere Bioplas Collagen  Solo Hill IBF BioRad BioRad Ventrex Verax Percell Biolytica Percell Biolytica Kirin Schott Biomat  Bioglas Micarcel G Bio—Carriers PS Ventregel Verax Cultispher—G Cultispher—H Cellsnow Siran Informatrix  Chemical composition DEAE—Dextran collagen coated DEAE—Dextran Polystyrene Polystyrene Polystyrene collagen coated _ polystyrene Glass Polyacrylamide Polyacrylamide Polystyrene Gelatin Collagen Gelatin Gelatin cellulose Glass Collagengl ycosamino—glycan  12 2.2.1.2 Systems utilizing porous microcarriers  Porous microcarrier systems (see Table 2) offer significant advantages over solid bead microcarriers. These systems are distinct from the conventional surface microcarrier culture system in that the cells are immobilized at high densities inside the matrix pores where they are protected from the fluid shear. Problems in long term cultures due to shear damage are reduced. Although primarily conceived for adherent cell applications, some porous carriers can also be used with nonadherent cells such as hybridomas (Almgren et al., 1991). Most of the systems involve the use of proprietary macroporous matrices. In an attempt to mimic the cell culture environment in mammals, most of these macroporous beads are collagen based (collagen, gelatin, or collagen-glycosaminoglycan). Macroporous beads can be inoculated directly from the bulk medium in the same fashion as conventional microcarriers. Suspended bead immobilization systems can be used in a number of different reactor configurations including fluidized beds or stirred tank bioreactors. Porous microcarrier systems can be scaled-up easily in both process intensity (cell density) and volume. -  (a) Verax  The Verax bioreactor system (Lebanon, New Hampshire) uses a proprietary weighted collagen sponge matrix (specific gravity 1.2 to 2.5). The porous microbeads, fluidized in the bioreactor, form a thick slurry (55% solids by volume), and the reactor is operated under continuous culture conditions for extended periods of time. Oxygen is supplied by a hollow fiber external gas exchanger which is connected to the fluidized bed reactor. Oxygen-rich medium from the gas exchanger is pumped into the base of the reactor chamber to fluidize the small microbeads. Oxygen-depleted medium from the top of the reactor chamber is then returned to the gas exchanger. Under conditions of perfusion  13 (medium replacement), a high cell density is developed within the fluidized bioreactor and microbeads (typically 4 x 10 7 cell /mL of reactor volume and 2 to 3 x 10 8 cells/mL inside the sponge bead matrix). Over 85 different cell lines, both suspension and anchorage—dependent cells, have been tested and cultured successfully using this system (Griffiths, 1990). The 90% void volume sponge— matrix microbeads can have diameters ranging from 200-600 gm and pore sizes from 30-100 gm (Dean et al., 1987). This system has been scaled up to 2000 liters. The microbeads are coated with collagen both on the interior and exterior carrier surfaces (Vournakis and Runstadler, 1989). Conventional heat sterilization cannot be used because it causes denaturation of the collagen. The cost of the sterile microbeads somewhat offsets the advantages of the system, but it is primarily its complexity that has limited its acceptance by industry.  (b) Cultispher-G microcarrier  Cultispher—G (CG) microcarriers (Biolytica, Lund, Sweden) are made of cross—linked gelatin with particle diameters of 170-270 gm (approximately 50% are smaller than 220 gm), pore size of 50 ptm, 50% void volume and density of 1.04 g/mL Cell densities up to 3 x 10 8 cells/ mL carrier have been reported (Nikolai and Hu, 1992; Mignot et al., 1990). Since gelatin is already heat denatured, CG particles can be sterilized by autoclaving. The CG microcarrier was originally designed for use in stirred—tank bioreactors. Recently, a new type of CG microcarriers with higher density (achieved by inclusion of a proprietary titanium compound) and large particle diameter (430-600 ptm) has been synthesized for use in fluidized bioreactors (Reiter et al., 1990).  14 (c) Siran porous glass beads Siran porous glass spheres (Schott Glaswerke, Mainz, Germany) have wide ranges of particle sizes (30-5000 gm), internal pore sizes (10-400 gm), dry densities (0.7-1.2 g/mL), and internal void fraction (up to 70%). In cell culture these glass spheres have been used in both packed bed (Looby and Griffiths, 1988) and two—phase fluidized bed bioreactors with separate oxygenator (Kratje and Wagner, 1992; Keller et aL, 1991). Cell densities up to 2.6 x 10 8 cells/mL carrier have been reported (Kratje and Wagner, 1992). The effect of bioreactor configuration on cell loading has been investigated (Kratje et al., 1991). Cell loading on the glass microbeads was reduced four—fold when the glass beads were utilized in a stirred reactor instead of a fluidized bed reactor. In a fixed bed configuration the biomass loading of Chinese Hamster Ovary (CHO) was three fold higher than in a fluidized bed reactor (Griffiths, 1990). Present evidence indicates that the porous glass spheres are unsuitable for cell cultivation in stirred tank bioreactors (Kratje et al., 1991; Griffiths, 1990).  (d) Informatrix porous microcarriers Informatrix microcarriers supplied by Biomat Corporation (Belmount, Massachusetts) are made of a collagen—glycosaminoglycan copolymer with mean diameters of 0.5 mm, pore sizes of 20-60 gm and densities of about 1 g/mL (Adema et al., 1990; Foran et al., 1991). The collagen—glycosaminoglycan copolymers are more resistant to collagenase degradation and have better mechanical properties than native collagen. The growth and differentiation of cells can be influenced by the composition of collagen—glycosaminoglycan (Cahn, 1990).  15 2.2.2 Microcapsulation Cells may be protected from the adverse effects of gas sparging by entrapment inside beads or microcapsules (Lim and Sun, 1980). Recently, Bugarski et al. (1989) and Kwong et al. (1989) were able to immobilize hybridoma cells in alginate microcapsules. A two—fold increase in maximum cell concentration was observed for cells immobilized in the alginate microcapsules (i.e. 3.5 x 10 6 cells/mL—alginate) compared to suspension cells in a conventional airlift system (Kwong et al. 1989). The secreted product can also be entrapped in the microcapsules facilitating downstream processing (Bugarski et al., 1989). However, microencapsulation techniques are both complicated and cannot currently be scaled up for large scale industrial processes (Yamaji et al., 1989).  2.2.3 Fixed bed bioreactor One of the best examples of a fixed bed bioreactor is the OpticellTM system (Charles River Biotechnical Services Inc., Wilmington, Massachusetts). This consists of a ceramic matrix, pump, oxygen and pH probes, gas permeator (oxygenator), medium reservoir, and feedback controller arranged as shown in Figure 1. The cylindrical ceramic matrix cartridge contains multiple channels running the length of the cylinder. Each channel has a square cross—section with sides of approximately 1 mm and wall thickness of about 0.15 mm. The ceramic matrix provides a surface area of 25-40 cm 2 per cm 3 volume. The most commonly used ceramic element has a nominal surface area of 4.25 m 2 . Two types of ceramic have been used. One provides a relatively smooth surface; the other is highly irregular, its rough surface covered with many small cavities. The size of the cavities can be manipulated to occupy up to 40% of the total surface area, with a mean diameter up to 50 Jim. The cavities allow cell entrapment to take place and provide added surface area for cell attachment.  16 Vero cell densities up to 5.7 x 10 5 cells/cm 2 ceramic surface ( 1.4 x 10 7 cells/mL of ceramic cartridge) have been reported (Lydersen et al., 1985). Seeding the Opticell reactor is problematic due to unevenly distributed cell inoculum population. A complex procedure is used to ensure uniform seeding of the matrix. Sufficient medium flow through the ceramic is needed to prevent growth— limiting gradients from developing along the channel of the ceramic matrix. Oxygen electrodes are provided to allow the consumption rate of dissolved oxygen to be monitored continuously based on the inlet and outlet dissolved oxygen concentrations and the medium flow rate. The oxygen supply is maintained by varying the medium flow rate. However, at high medium flows cells are exposed to excessive shear, causing detachment and blockage. The complex pumping loop between the external gas permeator and the ceramic cartridge is fragile and prone to contamination. Channel blockage due to over—grown cells is another common problem associated with the OpticellTM system. The high cost of the single use Ceramic Culture Chamber  Pump  Figure 1. Schematic illustration of Opticell bioreactor.  17 ceramic matrix and difficulties in scale—up have also hindered its acceptance.  2.2.4 Hollow fiber systems The use of hollow fiber bioreactors is a well established technique for cell culture (Feder, 1988; Tharakan and Chau, 1986). In these reactors, the medium flows through the fibers, and nutrients and wastes diffuse radially through the fiber wall to/from the cells in the shell space. The fibers are packed parallel to each other inside the shell space and are connected to a manifold at each end. The overall configuration is similar to that of a single—pass shell and tube heat exchanger. Cells grow in the extracapillary space (ECS). Cell densities in excess of 108 cells/mL can be established in the ECS (Tharakan and Chau, 1986). It is also possible to concentrate a high molecular weight product in the ECS. Concentration gradients pose major problems because of the heterogeneous culture growth conditions (Piret and Cooney, 1990). Scale—up of the culture volume is limited by the number and the length of fibers that can be accommodated within the shell.  18 2.3 Airlift Bioreactors A typical airlift reactor (see Figure 2) consists of 4 distinct sections — riser,  downcomer, gas separator and base (i.e. bottom zone below the riser). Only the riser section is usually sparged with gas. The different gas—holdups in the gassparged riser and the unsparged downcomer cause different fluid bulk densities in three flow regions which in turn induce fluid circulation in the reactor. The behavior of an airlift reactor is influenced by the interaction of these sections. Since the first development of the airlift fermenter by Lefrancois and his collaborators (1955), the device has been used for most types of fermentation ranging from single cell protein (SCP) production by microbial cells to more recent monoclonal antibody production by hybridoma cells. The simplicity of the design and construction, the well defined flow pattern, and the relatively low power input make the airlift reactor an attractive alternative to conventional stirred tank fermenters. Pneumatic agitation in airlift fermenters eliminates shaft seals and bearings associated with stirred tank reactors, hence reducing the possibility of contamination and mechanical failure. Gas separator--_____ 0 0  o  o  0  0 0 0  Downcomer  0 0  o  o  Riser  Draft Tube  00 0 0  0  Base  Figure 2. Typical airlift bioreactor configuration  19 2.3.1 Classification of airlift reactors  Various criteria have been proposed to classify airlift reactors (Onken and Weiland, 1983; Blenke, 1979). The two basic classes are: (1) external loop when the riser and downcomer are two separate conduits, usually round vertical pipes connected by horizontal sections near the top and the bottom (see Figure 3a). (2) Internal loop or baffled vessels — a bubble column with a draft tube or baffle which divides the column into a riser and a downcomer (see Figure 3b). Internal loop reactors can be further subdivided into two sub—classes, draft—tube (or concentric tube) and split—column. Simplicity of construction is the most appealing characteristic of the internal loop airlift reactor. Special attention is paid here to the draft—tube type internal loop airlift reactor with Newtonian fluid since these are the features of the system examined in this study. 2.3.2 Applications of airlift in animal cell culture  The use of airlift bioreactors to grow animal cells in suspension was first reported by Katinger et al. (1979). Airlift reactors can transfer sufficient oxygen  (a)^ (b) ^ O  0  C  0  o  0  Riser  0  0 0  0  0  0 0  0  0  0 0  00  O  O 0 0 F  \\  0  0  0 O  Downcomer Figure 3.  Airlift bioreactor configurations: (a) external loop (b) internal loop airlift  20 for conventional batch cultures of animal cells (maximum cell concentration < 10 7 cells/mL) at a relatively low shear rate (Wood and Thompson, 1986). Most data on airlift cell culture in the literature have been provided by the research group of Celltech Ltd (Berkshire, U.K.), primarily for suspension cell culture. Celltech has successfully grown more than 35 different cell lines, primarily hybridoma cells, in airlift fermenters ranging from 5 to 2,000 L to produce monoclonal antibodies (e.g. Lambert et al., 1987). They did not detect any significant effect of reactor type (i.e. stirred tank vs. airlift) per se on growth kinetics or specific antibody production rate. The effect of dissolved oxygen concentration (DO2) over the range of 8 to 100% was examined by Boraston et al. (1984). They found little effect on growth rate, glucose utilization rate, or maximum cell density of mouse hybridoma NB1. However, they did acknowledge that the optimum DO2 varies and must be determined for individual cell lines. Hiilscher and Onken (1988) found that the concentration of bovine serum albumin (BSA) in the serum—free medium significantly influences the growth of mouse hybridoma XR6—G10—B3. Hybridoma cell death rate was found to decrease with increasing BSA concentration.  2.3.2.1 Cell retention in airlift reactors  Hiilscher et al. (1992) suggested the use of an external cell settler and cell recycle to increase cell concentration in airlift reactors. Cells suspended in the effluent were collected in the inclined settler and returned to the reactor. However, the cell density was only doubled (up to a maximum of 6 x 106 cells/mL). The incorporation of a packed—bed for cell immobilization into airlift bioreactor configurations has also been proposed to increase cell loading (Murdin et al., 1989; Femilletti, and Fredericks, 1988; Lazar et. al., 1987; Chiou et al., 1991). The use of an airlift, where the liquid circulation is driven by the sparged bubbles in  21 the riser, eliminates the need for auxiliary mechanical pumping. The downcomer of the airlift is packed with a suitable support material for cell attachment and growth. Relatively high cell densities ( 6.8 x 10 7 CHO cells/mL of packed—bed volume; Chiou et al., 1991) have been reported. Advantages include: (1) There is a high surface to volume ratio in the packed—bed section. (2) In situ oxygenation is achieved without the use of an external loop.  (3) Cells are not directly exposed to gas—liquid interfaces. (4) Cell damage from excessive shear forces is reduced, due to the low liquid velocity in the packed—bed. (4) Nutrient is supplied to the immobilized cells by convection. (5) Mechanical agitation and external pumps are eliminated. (6) Reactors can be operated in a batch, fed—batch (i.e. semi—batch) or continuous/perfusion mode. However, the near plug flow of medium through the packed bed creates longitudinal gradients of nutrient and waste products making scale—up of such systems difficult. The application of airlift reactors in commercial biotechnology processes remains limited, due in part to the lack of basic design parameters in the literature. Unfortunately, the available information in the literature is often conflicting and shows large variations. Precise comparisons of different studies are often difficult since the operating conditions and reactor geometry have not been clearly reported. 23.3 Flow patterns The riser section of a concentric tube (CT) airlift can sometimes be considered as a conventional bubble column (BC). Shah et. al (1982) identified three flow regimes for BC:  22 (1) Homogeneous bubble flow, usually for superficial gas velocity, U gr, less than 0.05 m/s (2) Slug flow — bubbles occupy the entire column cross—section; the frequency and the length of spherical cap slug increases with increasing gas flow rate; slug flow often occurs in reactors with diameters less than 0.15 m. (3) Churn— turbulent flow — transitional region where the gas does not form a continuous phase Airlift and bubble column (BC) reactors for animal cell cultures should be operated in the homogeneous bubble flow regime because of the requirement for low shear. The main difference between a BC and a CT airlift is the liquid velocity (UL). Unlike airlift reactors, the net UL is nil for a BC operated in the batch liquid mode. As observed by Onken and Weiland (1983) and Merchuk (1986b), non— hindered liquid circulation in CT's tends to delay the transitions between the flow regimes discussed above. A priori determination of the flow regime is difficult, since it is affected by many parameters, such as the physical configuration of the reactor and the properties and velocity of the liquid phase. Shah et. al. (1982) suggested an approximate guideline of flow regime dependency on gas velocity and reactor diameter for bubble columns with no net liquid circulation (i.e. U L = 0). This is obviously of limited applicability to airlift reactors. The downcomer flow pattern in a CT is quite different from that observed in a BC. As the gas flow rate is increased, causing a parallel increase in the liquid velocity, bubbles are entrained in the flow stream to the downcomer. Oscillation between swirling flow, wavy flow and straight flow is observed at liquid velocities near the terminal rising velocity of the bubbles.  23 23.4 Power input The two contributors to the total power input, (Ei) to a pneumatic device are (1) isothermal gas expansion as the sparged gas moves from the higher hydrostatic pressure region at the bottom to the top of the liquid—gas interface, and (2) kinetic energy transferred to the fluid by the jet of gas entering the reactor, i.e.  P -Fp1 gh =QRTIn h +q)U20 Q^ •E 1^P 2  (1)  h^  where Ph = pressure at the top Q = flow rate of the gas T = temperature h = height of liquid column above the sparger (p = sparger efficiency (typically around 0.06) to account for the fact that the gas velocity just above the sparger is lower than the velocity in the orifice. U0 = velocity difference between the gas velocity inside the orifice of the sparger and the gas velocity just above the sparger.  pi = liquid density The second term of Eq. (1) is usually negligible for airlift fermenters (Robinson, 1986). However, Guy et al. (1986) pointed out that discrepancies (especially for pressurized vessels) have arisen in some cases as a result of neglecting the second term in Eq. (1).  23.5 Gas velocity Two factors dealing with the gas velocity may explain variations in the open literature. The first is the term, "superficial velocity" U g , which may be based on  24 either the entire cross—sectional area of the vessel or the riser cross—sectional area. The superficial velocity based on the entire cross—sectional area should be used only when comparing the power input to a BC and a CT airlift. In all other situations, the superficial velocity should be based on the riser cross—sectional area. The second factor is the effect of the axial variation of the dispersion height on the superficial gas velocity. Chisti and Moo—Young (1987) showed that the hydrostatic pressure can affect the superficial gas velocity significantly. They proposed that only the "true" superficial velocity should be used, i.e.  u _^ Q m R T . in [Ph +Pi g A h p g^P R L l h  g1  (2)  where AR = cross—sectional area of riser hi, = unaerated liquid height  2.3.6 Hydrodynamics 23.6.1 Effect of sparger: The type of sparger used can have a profound effect on the reactor behavior (Blenke, 1979; Deckwer et ell., 1974). Static spargers, e.g. porous plates, single nozzles, and perforated plates or pipes, are most commonly used in fermentation processes. Porous plates are more expensive and have higher operating costs due to greater pressure drops. Spargers coupled with a liquid jet nozzle, driven by an external pump, are rarely used in fermentation processes due to their need for associated pumping machinery and the higher shear force imposed on cells. The location of the sparger in the airlift device also has a strong influence on the overall hydrodynamic behavior. Chisti and Moo—Young (1987) found better gas distribution when the sparger was placed just inside the riser section (i.e.  25 above the base connecting zone). However, very little is known about the effect of the position of the sparger on the overall behavior of airlift devices.  2.3.6.2 Dispersion characteristics The gas bubble size in a typical turbulent bioreactor is generally controlled by the equilibrium between the dynamic forces which tend to break up the bubbles and the surface tension forces which preserve their size and shape. The average size of the bubbles is usually independent of its size at formation. Dussap and Gros (1982) proposed the following expression for predicting gas-liquid interfacial area, a D , in aqueous sodium sulphite in a CT (volume = 0.015 m 3 , DC = 0.11 m, Dci =0.0756 m, h=1.8 m). E• 0 77 aD = 3.66(-1-) VD  (3)  for 200 W/m 3 Ei/VD 1500 W/m 3 where VD = dispersion volume If one assumes spherical bubbles, the average bubble diameter, dB can be calculated knowing that  6£ d = g B aD  (4)  when £ g = gas hold-up so that d =1 . 64 • E B  E.  j-0.77  ^  g .V D  (5)  Chisti et al. (1987) proposed the following equation for estimating the bubble rise velocity (Ub) in a BC (air-water dispersion):  26  Ub = 0.284 + 2.7 U g (m/s)^  (6)  More accurate prediction of Ub in a BC can be made using the following equations (Chisti et al. , 1987): For U g < 0.05 m/s in an air-water dispersion, Ub = 0.284 + 1.1 E g (m/s)^  (7a)  for U g > 0.05 m/s, Ub = 0.284 + 11.2 E gg (m/s)^  (7b)  Hills (1976) correlated the bubble rise velocity in the riser for UL r > 0.3 m/s and an air-water dispersion as follows: Ub = 0.24 + 1.35 (U g + UL) (m/s)^  (8)  where UL = superficial liquid circulation velocity in the riser (m/s). Lee et. al (1986) showed that the terminal rise velocity of bubbles in the range of interest for airlift columns is independent of bubble size and approximately equal to 0.23 m/s. This is true for a CT containing bubbles of volume equivalent diameters from 0.002 m to 0.01 m in an air-water system which does not contain significant amounts of impurities (i.e. surface tension not less than 0.055 N/m).  2.3.63 Liquid circulating velocity, UL  As illustrated by Equation (8), the upflow of liquid in airlift devices increases the velocity of gas bubbles in the riser section, thereby lowering the riser gas hold-up, E g , and hydrodynamic pressure difference between the riser and downcomer. Airlifts have a much more uniformly distributed gas phase across the column cross-section, with a maximum E g near the column wall, compared to BC  27 (Onken and Weiland, 1983). UL affects all parameters which characterize the reactor behavior, such as E g , mixing time, and mass transfer. Unlike U g, UL cannot be varied independently. Several investigators (e.g. Onken and Weiland, 1983; Merchuk, 1986a; Bello et al., 1984) showed that UL varies with (U g) 13 , where j3 was approximately 0.33 to 0.4, depending on the reactor geometry and flow regime (Onken and Weiland, 1983).  p was found to decrease as the flow  regime changed from bubble flow to churn turbulent flow. Bello et al. (1984) found the ULr in CT with water or salt solution (0.15 kmol m -3 NaC1) can be predicted by ULr = 0.66 (AD/AR)0.78 ugr0.33 (9) Equation (9) indicates that UL r increases with increasing (AD/AR) ratio. However, both Weiland (1984) and Blenke (1979) found that UL r (for two phase flow) reached a maximum for (AD/AR) = 0.59. In fact, Equation (9) agrees reasonably well with literature values for (AD/AR) ratios up to 0.89. Significant deviation was observed for higher values of (AD/AR) (e.g. AD/AR = 1.12 in the work of Hatch, 1973 and 1.23 for the work of Blenke, 1979). Bello et al. (1984) attributed these differences to measurement errors arising from "insensitive" techniques used by other investigator. One should note that the distance between the lower end of the draft tube and the bottom of the reactor H u , was fairly large (100 and 260 mm, respectively) for the cases of Bello et al. (1984) and of Chakravarty et al. (1974). The devices used by Blenke (1979) and Hatch (1973) had smaller H u values of 70 mm and 33 mm, respectively. Two general theoretical approaches have been used to predict UL r . One is based on a momentum balance, while the second is based on an overall macroscopic energy balance. Bello (1981) found that the predicted ULr , when  28 based on a momentum balance, was insensitive to the flow regime and corresponded more closely to measured values than values obtained from an energy balance. However, the model proposed by Bello (1981) could not be used without a priori knowledge of gas holdups in the riser and downcomer. Lee et al. (1986) found that the energy balance could adequately predict values of UL r measured by Hatch (1973) and Jones (1985) in CT for both the bubbly and slug flow regimes. More recently, Chisti et al., (1988) and Calvo (1989) have demonstrated that predicted values of Uj based on an overall energy balance can satisfactorily describe most of the available liquid circulation data. They utilized Ei = ER + ED + EB + ET + EF^  (10)  where Ei = energy input as per Eq. (1) ER = energy dissipation due to wakes behind bubbles in the riser; ED = energy loss in downcomer due to upflow of bubbles; EB(T) = energy loss due to fluid turn around at the bottom (at the top) of the reactor; and EF = energy loss due to friction in the riser and the downcomer. Both Lee et al. (1986) and Chisti et al. (1988) found that for CT, EF and ET are negligible compared to other factors in Equation (10). Equation (10) then reduces to Ei = ER + ED + EB where, according to Chisti et al., (1988), one can write EB = 0.5 pi (VLd) 3 KB AD (1 — E gd);^  (10a)  ER = Ei —pi g hD Ufr AR E gr;^(10b) ED = pi g hp ULd AD Egd^  (10c)  29 For CT, Chisti et a/. (1988) proposed  U  Lr  =  2.g h D (£ gr —£ ) gd  1).5  . (i_ Egd ) K . (A / A )2 ^  B R D  KB is the dimensionless frictional loss coefficient and can be estimated from the correlation (Chisti et al. 1988), KB = 11.402 (AD/AB)  °389^(12)  where AB = free area for liquid flow between riser and downcomer at the bottom of the CT airlift reactor. Note that AB is affected by the clearance between the base and the bottom of the draft tube. AB in Equation (12) depends on both the internal draft—tube diameter, Da, and the clearance between the base and the draft—tube, H u , for CT airlifts. This may explain the dependence of U L on (AD/AR) observed by Weiland (1984), Blenke (1979) and Bello et a/. (1984). Calvo (1989) further simplified Equation (10) by excluding ED and obtained reasonable agreement with the data of Jones (1985). However, KB values obtained by Calvo (1989) were generally much larger than those used by Chisti et a/. (1988) and were found to depend on (AR/AD). The larger KB values used by Calvo (1989) seemed to compensate for the error introduced by neglecting ED which, according to Lee et a/. (1986), accounted for 50 — 60% of the total energy loss in the CT.  2.3.6.4 Gas holdup, eg  In order to predict ULr , the overall gas holdup, e g , is needed. e g is influenced not only by UL but also by the gas residence time, oxygen transfer, and  30 liquid mixing. Merchuk (1986b) found that the gas hold-up was lower for an airlift than for a BC. In addition, a maximum hold-up, which characterizes the behavior of the BC near the critical superficial velocity, does not appear in the corresponding experimental data obtained in the airlift device. On the other hand, many investigators (e.g. Koide et al. , 1983a, 1983b; Kawase and Moo-Young, 1986; Chisti  et  A, 1987) found the same overall e g (i.e.  E gr + e gd) in the CT and the corresponding BC when U g , based on the entire  cross-section, is used for comparison. This occurs even though the net effect of UL and Ub in the riser section tended to reduce the gas holdup. The downcomer gas holdup is increased by the downward liquid flow. Increasing the liquid velocity increases the_number of the bubbles dragged down the downcomer and the distance traveled by the large bubbles inside the downcomer. The gas-liquid separation ability near the top of reactors can influence the overall gas holdup substantially. External loop airlifts generally have a much lower e gd (i.e. 300 to 700%) than CT airlifts for the same unit volume energy input (Bello Many empirical correlations for estimating (e.g.  6g  et al.  1985).  in CT airlifts have been proposed  Chakravarty et al. 1973; Miyahara et al., 1986; Koide et a/., 1983a and b).  For example, Bello et al. (1985) suggest two equivalent empirical correlations: E gr = 3.4 x 10  -3  (1 + AD/AR) -1 (Ei/VD) (2/3)  E gr = 0.16 (1 + AD/AR) (U gr / ULr) a  (13)  (14)  with a = 0.56 (water); 0.58 (salt solution) and E gd = 0.89 E gr for air-water only (15) The suggested correlations were for an air-water or a salt solution ( 0.15 kmol m 3 NaCl) with bubbly flow (i.e. U g = 0.0137 to 0.086 ms -1 ), (AD/AR) = 0.13, 0.35, and 0.56, DC = 0.152 m, hp = 1.8 m, and the annulus sparged. The properties of the liquid such as surface tension, density, viscosity and ionic strength may also affect gas holdup. The downcomer area or diameter is often  31 incorporated in expressions for predicting eg. It should be noted that most gas holdup expressions have been based on U g for fresh gas input only. The recirculated gas is generally not considered. Merchuk (1986a, 1986b), Freedman and Davidson (1969) and Schiigerl et al. (1977) have all reported that the use of multiple—orifice spargers gives higher eg (up to 30% higher) than single orifice spargers for all UL and Ug tested. However, the differences reported by Merchuk (1986) were generally insignificant (<5%). Conflicting results were also reported by Siegel et al. (1986) who found no appreciable difference in the gas holdup or the gas recirculation rate caused by changing the sparger configuration and sparger orifice size. One can predict ULr , E gr , E gd in CT airlifts by simultaneously solving Equations (9), (14) and (15). Chisti et al. (1988) suggested that Equation (14) be replaced by the gas holdup correlation of Hills (1976): U  gr  =  gr  0.24 + 1.35 (U  0 . 93 gr + ULr )  (16)  23.6.5 Mass transfer  Because of its low solubility in fermentation broth, dissolved oxygen can be consumed quickly if it is not continuously replaced. The mass transfer coefficient for oxygen, kLa, is thus a critical scale—up parameter. The rate of mass transfer may be expressed as: dC  dt = k a (C * — CL )^ L L  (17)  where CL is the oxygen concentration in the broth and C* is its saturation concentration in the broth in equilibrium with the gas. kL is the true mass transfer coefficient and aL is the specific gas—liquid interfacial area based on liquid  32 volume; aD, the gas—liquid interfacial area based on total dispersion volume, is often used instead of ai„ and can be easily estimated from Equation (4). ki, has been found to be slightly influenced by the fluid dynamics (Blenke, 1979). The well—known Calderbank and Moo—Young correlation (1961) indicates that ki, is determined mainly by the bubble size. The lumped mass transfer coefficient or the so—called volumetric mass transfer coefficient, ki ad or kLaL, is invariably used for design purposes. A number of correlations for approximating kLa in CT airlifts are listed below (see previous sections for operating conditions and reactor geometry): (1) Bello et al. (1985): for 0.01 _. U gr 5 0.10 m/s kLaD = a1 (1 + AD/AR) -2 li g°.8  (18)  where a1 = 0.75 for water and = 0.79 for 0.15 kmol/m 3 NaC1 solution Or  kLaD = a2 (cgo 1.27  (19)  where a2 = 0.57 for water and = 0.60 for 0.15 kmol/m 3 NaC1 solution or 2 (364 k L a D .h, ror) HAD) ^ ' = a3 U U A Lr^ —  Q  Lr^  (20)  R  where a3 = 1.99 for water and 2.57 for 0.15 kmol/m3 NaC1 solution a4 = 0.87 for water and 0.92 for 0.15 kmol/m 3 NaCl solution  (2) For an air—water system with 0.47x10-2 5 U sgr S 7.7x10-2 m/s, a 13 hole— perforated plate with d o = 0.001 m, DC = 0.11 m, and DCi = 0.058 m, Stejskal and Potucek (1985) found that  33 kLaL = 1.84x10 -3 + 0.35 Ugr  ^  (21)  It was not clear whether U g in Equation (21) was calculated based on the overall reactor diameter or the riser diameter. Koide et aL (1983a,b) reported that kLaD depended on reactor geometry (i.e. kLaD varied with (dc) 1-975 ) for a draft—tube— sparged CT, while kLaD varied with (dc) °.0905 for an annulus—sparged CT. They also reported that neither AB nor Ld influenced gas holdup or kLa for the CT devices studied. Weiland (1984), Wang et al. (1971) and Kriegel et al. (1978) demonstrated a dependence of k La on the ratio (Dci/Dc) for CT airlifts. However, different values of the optimum ratio of (Dci/Dc) for oxygen transfer have been found: 0.8 (Kriegel et al., 1978), 0.74 (Weiland, 1984), and 0.65 (Wang et al., 1971). Merchuk and Siegel (1988) conclude that: (1) kLa increases with increasing U gr ; (2) for a given air sparging rate, increasing UL decreases kLa; (3) the physico—chemical properties of the liquid phase have a less pronounced influence on kLa in airlift reactors than in bubble columns; (4) taller airlifts have higher kLa values; (5) the aeration efficiency (oxygen transferred to the liquid phase divided by power of gas input) remains relatively constant with increasing kLa values and increasing U g, while BC and stirred tank reactors exhibit a sharp decrease in aeration efficiency; (6) the change of flow regime from bubbly to slug flow sharply decreases kLa; (7) CT airlifts have higher kLa than external loop airlifts. Enhanced kLa in CT relative to BC was attributed to different bubble size distributions by Margaritis and Sheppard (1981) and Kawase and Moo—Young  34 (1986). Koide et al. (1984) indicated that the contribution of the CT downcomer to mass transfer is relatively large in an air—water system. In the work of Koide et al. (1984), kLa was not measured directly but calculated using the measured bubble size based on photographic and electrical resistivity methods; ki, values were then estimated using a correlation for single bubbles. The resulting overall kLa (i.e. [(kLa)r (VD)r + (kLa)r(VD)d)/(Vd)T) agreed well with previous data (Koide et al., 1983a,b). However, Bello et al. (1985) showed that mass transfer in the  downcomer was negligible; the overall mass transfer could be adequately estimated by ignoring the contribution of the downcomer. Bello et al. (1985) attributed this to the lower gas—liquid slip velocity in the downcomer. Based on the data of Bello et al. (1985), ULd in their CT airlift was about 0.23 to 0.26 m/s. This suggested that gas bubbles would be held nearly stationary in the downcomer, since the average free terminal velocity for bubbles is about 0.25 m/s. At steady state, the available oxygen in the stagnant bubbles in the downcomer was depleted, so that these bubbles contribute little to the overall mass transfer. Many aspects of the hydrodynamic behavior of airlift bioreactors (e.g. liquid flow rate, gas holdup, and mass transfer coefficients) have been studied in great detail. Although the conflicting information in the open literature often leads to confusion, the previous studies were helpful in formulating a suitable mechanistic model (Chapter 5) to describe the hydrodynamic properties of the proposed airlift bioreactor.  35  3.0 EXPERIMENTAL MATERIALS AND METHODS 3.1 Cells and Cell Maintenance  Three cell lines were used in this study: (1) Vero cells (ATCC CCL81), an anchorage dependent cell line commonly used for vaccine production; (2) BHK cells transfected with an expression vector containing a DNA fragment coding for the amino—terminal lobe of human serum transferrin (hTF/2N) under the control of the metallothionein (MT-1) promoter (Funk et al., 1990); (3) 2E11 hybridoma cell line which produces an anti IL3 murine monoclonal antibody (Ziltner et al., 1988). Frozen stocks of cells were stored in liquid nitrogen in Dulbecco's modified Eagle's medium (DMEM, Gibco) supplemented with 10% dimethyl sulfoxide (DMSO) and 10% fetal calf serum (FCS, Gibco). Cell stocks were maintained in roller bottle cultures with 850 cm 2 surface area at 37 0 C in DMEM or DMEM/F12 (50:50) supplemented with 5% FCS or 5% newborn calf serum (NCS), gentamycin (0.1 mg/L, Gibco), and in some cases, for BHK cultures, methotrexate (0.55 mM, Cyanamid).  3.2 Cell Culture Systems 3.2.1 Ceramic foam  The ceramic foam elements were produced by the Selee Corporation of Hendersonville, North Carolina. The matrix itself consisted of a highly complex pore structure of sintered alumina dodecahedra defined by the ceramic filaments, with a network of interconnecting passages. The true skeletal density is about 4000 kg/m 3 and the void fraction is about 80%. Sintered alumina is non—toxic, can be sterilized at high temperature, and provides chemical resistance, good mechanical stability and high surface area per unit volume. Foam cylinders 36 mm in diameter and 40 mm long were employed in the perfusion system. In addition, tubular  36  Figure 4. Ceramic foam cylinders of 30,50 and 100 PPI. ceramic foam hollow cylinders of outer diameter 66 or 77 mm and inner diameter 57 mm were used as the draft—tube in the airlift systems. For both geometries, three different porosities were used, rated by the manufacturer as 30, 50, 100 pores per inch (PPI; i.e. 1200, 2000 and 4000 pores per m) with average pore diameters of 550, 320, and 140 ,um, respectively. These are shown in Figure 4.  3.2.1.1 Fixed bed perfusion system The experimental system (Figure 5) consisted of a growth chamber containing one ceramic foam cylinder, an external medium reservoir and a seeding chamber. Medium was circulated continuously between the external reservoir and the growth chamber using a peristaltic pump (Model PIOT, Dungeg Inc., Agincourt, Ont.). The circulation rate was typically 130 mL/min and could be readily adjusted. The growth chamber consisted of a short section of Pyrex glass tube of inner diameter 37 mm and had a total volume of approximately 50 mL. Including the void volume, each enclosed ceramic matrix had a total volume of 40 mL. The ceramic cylinder fit tightly into the glass tube without significant  37 clearance at its peripheral surface so that virtually all of the circulated liquid passed through the ceramic matrix. A 2—L stirred tank (LH Fermentation, Slough, UK) functioned as a reservoir in which the circulating medium was reaerated and its pH regulated. Mixing was provided by a 60 mm diameter marine impeller at a speed of 100 rpm. Aeration was supplied by sparging an air—CO 2 (95:5 by volume) gas mixture directly into the medium reservoir. The gas—free medium circulated through the growth chamber so that the cells were isolated from the potential damaging effect of sparging. The operating temperature was maintained at 37°C, while the medium pH was maintained between 7.3 and 7.5 by addition of 0.2 M NaOH solution. Oxygen probes (Ingold) were located in the reservoir vessel upstream from the growth chamber and in line directly after the chamber. This allowed the total oxygen uptake rate to be monitored continuously. Inoculum (at least 1 x10 8 cells) was introduced into the seeding chamber and pumped into the growth chamber. Medium circulation was stopped for about 1 ^Q PUMP  GAS OUTLET /GAS INLET /DO PROBE  '711 PROBE Growth Chamber  GF  o o  0  JACKET  o 0  0  o  oc)dc:).° MEDIUM BOTTLE SEEDING MEDIUM RESERVOIR CHAMBER  Figure 5. Schematic showing perfusion system.  38 to 2 hours to allow cell attachment to take place. The medium was changed whenever the glucose concentration fell below 500 mg/L. DMEM + 5% fetal calf serum (FCS) was generally used. Zn was provided by either direct addition of ZnSO4 solution or by using DMEM/F-12 mixture (50:50) + 5% FCS medium to induce the metallothionein promoter. 3.2.1.2 Airlift system The test system was a modified commercial concentric tube (CT) airlift reactor in which the normal impermeable draft tube was replaced by a porous ceramic draft tube (see Figure 6). The porous draft tube was mounted in a 5 L glass airlift bioreactor of 120 mm inner diameter with a water jacket for temperature control. The reactor was manufactured by LH Fermentation of Slough, UK. The reactor was equipped with a 12 mm—diameter cylindrical sintered glass Pyrex sparger with 170-220 gm pore size. Liquid circulation around and through the porous draft—tube was driven by the sparged gas. The draft—tube consisted of a series of ceramic hollow tubular short cylindrical sections (35 mm in height). Up to 6 sections could be mounted together, stacked end to end, and held in place by a stainless steel wire mesh (grid opening size of 4 mm) located on the inside of the hollow ceramic cylinders. The thickness of the annular ceramic foam cylinders was either 7.5 or 13 mm. In each case, the inner diameter was 51 mm, so that the outer diameters were 66 or 77 mm, respectively. The porous tubular ceramic foams were held in place by glass supporting tubes located at both ends of the draft tube (see Figure 6). The tubular glass support pieces were either 25 mm or 75 mm in length and had a wall thickness of 4 mm. The non—porous glass draft tube had an inner diameter of mm, a wall thickness of 4 mm. and a height of 450 mm.  39  Figure 6. Schematic showing airlift system For BHK cell culture, the draft—tube of the test system was made by stacking two 13 mm thickness, 35 mm tall 30 PPI cylindrical ceramic foam pieces with glass supports at top and bottom (see Figure 6). The operating temperature was maintained at 37°C. Aeration was supplied by sparging an air—CO 2 (95:5 by volume) gas mixture at 100 mL/min into the vessel along the axis, 50 mm below the bottom of the draft tube. A single—hole (6 mm diameter) sparger was used. Inoculum (at least 1 x 10 8 cells) was introduced, with the medium surface just above the top of the porous ceramic draft tube with 0.7 L medium present. The medium volume was then increased gradually to a total of 4 L to accommodate the increasing cell number in the reactor.  40 The effects of several configuration parameters on the airlift hydrodynamic behaviour were examined to optimize the system. These parameters include: (1) ceramic pore opening sizes (30,50 and 100 PPI); (2) ceramic matrix thickness (7.5 or 13 mm); The effect of operational variables on the hydrodynamic characteristics such as gas holdup, overall volumetric mass transfer coefficient (kLa), and liquid circulation rate were also investigated to develop a mathematical model to facilitate scale—up. The operating variables included: (1) gas sparging rate (0.2, 0.6, 1.0, 1.5, 2.0 L/min); (2) liquid properties — effect of serum addition. The differential pressure drop across the draft tube at its base was measured using an electronic pressure transducer (Model PX750-06DI, Omega Engineering Inc., Stamford, Connecticut). The pressure sensor consisted of two tubes inserted from the bottom of the reactor and fixed in positions located just below the lower end of the draft tube with opening of 3 mm diameter. The openings of the sensor tubes were arranged to be perpendicular to the direction of the liquid flow. The pressure transducer had an adjustable span from 0-0.5 to 0-6 inches of water, (i.e. 0-125 to 0-1495 Pa) The transducer had an accuracy of ± 0.5% of calibrated span and a built—in adjustable damping of 0.4 — 4s. A IBM—AT compatible computer (Nimbus, U.K.) was used for data logging. The analog signal from the pressure transducer was converted to a digital signal using a mA input A/D board (Model: PC-61, United Electronic Industries, Newton Center, Massachusetts) then stored by the computer. The listing of the controlling program for data logging using the mA input A/D board appears in Appendix 1. Liquid circulation velocities were calculated by monitoring the time needed for a small colored neutrally buoyant polyurethane particle to travel a known distance along the downcomer. The average liquid circulation velocity was calculated based on 10 such measurements.  41 Mass transfer coefficients were determined by the transient gas in/gas—out method (Sobota et al., 1982). Dissolved oxygen concentration was monitored using Ingold polarographic oxygen probe and Anglican 2000 controllers. The mV output from the Anglican controller was first converted to a digital signal and then registered by the Nimbus computer. Appendix 2 shows the BASIC program listing employed for communication between the controller and the computer.  42 3.2.2 Microcarriers  The porous polystyrene microcarrier beads were manufactured by Microporous Materials Ltd. (Braunston, U.K.). The emulsion process used for the production can be controlled to obtain the desired range of pore sizes. According to the manufacturer, these styrene copolymer beads had a surface area of 50,000 cm 2 / g particles and a porosity greater than 90% by volume. The surface was chemically modified by Microporous Materials Ltd. as listed in Table 3 to facilitate cell attachment and growth. The diameters of the polystyrene particles ranged from 250 to 1000 p.m, with pore sizes from 5 to 80 pm. The material consisted of roughly spherical pore spaces, interconnected by multiple smaller channels 5 to 10 times smaller than the pores. Four pore sizes were tested— 80, 40, 25, and 5 1..tm. Figure 7 shows a scanning electron micrograph of the P40 material. Cytodex-1 (Pharmacia, Uppsala, Sweden) and Cultispher—G (Biolytica, Lund, Sweden) were used for comparison purposes. PHOTO- 12005  Figure 7. Open structure of a macroporous microcarrier P40 particle.  43 3.2.2.1 Pretreatment of microcarrier particles.  Characteristics of the microcarriers used in this work are summarized in Table 3. Commercial microcarriers were prepared according to the instructions provided by the manufacturer. Cytodex-1 microcarriers were added to siliconized glass bottles and were swollen in Ca 2 +, Mg 2 +—phosphate—buffered saline (PBS; 50-100 mL/g Cytodex-1) at room temperature for at least 3 hours. The PBS was then removed and the particles were washed twice with fresh PBS (30-50 mL/g Cytodex) and sterilized by autoclaving for 20 min at 121°C. The supernatant was decanted and the sterile particles were rinsed several times with fresh serum—free medium prior to use. The preparation of Cultispher—G microcarriers was similar to that of Cytodex-1. The thy Cultispher—G microcarriers were swollen and hydrated in Ca 2 +, Mg 2 +—free PBS (50-100 mL/g particles) at room temperature for at least one hour. Most of the microcarriers floated due to entrapped air. Sterilization by autoclaving released the entrapped air. The supernatant was removed and the sterile particles were washed once with fresh sterile PBS and twice with serum— free culture medium (25-50 mL/g) before use. As suggested by the manufacturer, Polyhipe particles (except sulphonated polystyrene) were first wetted using 70% ethanol solution and autoclaved with the 70% ethanol solution at 121°C for 15 min. Then the particles were washed twice with sterile PBS and once with complete medium prior to inoculation.  44 TABLE 3. Properties of Microcarrier Beads Tested Designation  Description  Void  composition  pore  Particle  Volume  size  diameter volume  (%)  (pm)  (.tm)  Settled (mL /g dry wt.)  P80  large pore  90  medium pore  90  polystyrene  12  90% styrene 10% commercial  30-40 500-1000  12  10-30 500-1000  11.4  10-20 500-1000  10.5  80-100 500-1000  12  40-50 500-1000  11.8  30-100 500-1000  14.3  50-60 500-1000  14  divinyl benzene.  polystyrene P25  80-100 500-1000  divinyl benzene.  polystyrene P40  90% styrene 10% commercial  90  90% styrene 10% commercial divinyl benzene.  P5  small pore  90  polystyrene CM80  chloromethyl  (—CH7C1)  styrene  Q80  quaternary  90% styrene 10% commercial divinyl benzene.  90  50/50 styrene and chloromethyl styrene on mole basis.  90  derivitised  CM80 treated with excess ethanol/water (20%) solution of trimethylamine for 8 hours at room  styrene  temperature. DEA80  diethylamino  90  derivitised DEA—LG80  diethylamino  diethylamine for 4 hr. (2.9% N) 90  derivitised CS  collapsed  CM80 refluxed with 10% CM80 refluxed with 100% diethylamine for 4 hr. (2.9% N)  97  sulphonation of polystyrene at  sulphonated  60°C for 6 hours then pH  polystyrene  neutralized with NaOH solution.  S80  sulphonated  (—SO3Na)  polystyrene  90  P80 treated with excess sulphuric  20  250-500  30  30-70 500-1000  11  acids (95-98%) at room temperature for 30 minutes  S40  sulphonated  (—SO3Na)  polystyrene  90  P40 treated with excess sulphuric acid (95-98%) at room temperature for 15 to 240 minutes  30  250-500  12  45 Table 3. continued.. Flex— CM80  flexible  50/50 styrene and chloromethyl  chloromethyl  styrene on mole basis, 10%  polystyrene  divinyl benzene crosslinker  peroxide  Per80  90  60-100 500-1000  11  90  P80 treated with peroxide  50  500-1000  143  treated polystyrene Cultispher—G  Percell  50  100% Gelatin  50  170-270  14-18  Cytodex 1  Pharmacia  0  Dextran matrix with N,N—  0  131  18  diethylaminothyl group substitution  The surface composition and charge of some of the Polyhipe particles are characterized in Table 4. The characterization of the Polyhipe particles was performed by Microporous Materials Ltd. Table 4. Surface characteristics of the Pol hi e particles Zeta potential (mV) —41.34 ± 1.6 —39.68 ±1.28 —28.41 ± 0.12 —17.15 ± 1.15 —45.5 ± 7  C (%)  0 (%)  S (%)  CI (%) 0.14 ND ND ND ND  N (%)  Na  f%)  — P40 95.52 4.34 — — P25 ND* ND ND ND ND P5 ND ND ND ND ND ND ND ND ND Q80 ND ND Flex— ND ND ND ND CM80 PER80 —54.6 ± 5.37 ND ND ND ND ND ND DEA— —2.56 ±9.6 ND ND ND ND ND ND LG80 44.28 ±2.43 94.04 4.01 — DEA-80 0.18 1.8 — *ND denotes not determined Zeta potential was determined from streaming potential measurements in a flat plate system.  3.2.2.2 Microcarrier culture (a) Roller culture  Initially the microcarriers were tested in glass 500 mL roller bottles, siliconized (Dimethyldichlorosilane, Sigma, St. Louis, MO) to prevent attachment of the cells to the bottle surface. Vero and BHK cells were inoculated at 5 x 10 7  46 cells in 20 mL of medium with 500 mg of microcarrier particles. This inoculum was incubated in a static vessel, with periodic agitation, for 4 to 8 h at 37°C to allow cell attachment to take place. The culture volume was then increased to 100 mL with fresh medium containing 10% FCS. For hybridoma cells, the inoculum (5 x 107 cells total) was incubated with the macroporous beads in 50 mL PBS (without Ca++ and Mg++) for 1 h without agitation. The particles were then washed once with 30-50 mL serum containing medium and finally resuspended in 100 mL medium containing 10% FCS. The roller bottles were rotated at 2 rpm. Collapsed sulphonated polystyrene was tested without prior hydration and washing. This material was autoclaved dry (30 min, 121°C) and hydrated by direct addition to 100 mL culture medium with cells (5 x 10 5 cells mL-1). All cultures were perfused semi—continuously by decanting the spent medium and adding fresh medium when the glucose concentration dropped below 1 g/L. Bottles were gassed with an air—0O2 (95:5) gas mixture prior to sealing.  (b) Spinner cultures  Several sizes of Bellco spinner flasks, 100 mL, 250 mL and 1000 mL in volume were used. All flasks were siliconized prior to sterilization to prevent cell attachment to the walls. The cultures were seeded with enough cells suspended in reduced starting volume (i.e. 1/2 to 1/3 of the final volume) to give a final cell concentration of at least 10 5 cells/mL (i.e. 107 cells for a 100 mL spinner flask). Intermittent agitation (2 min agitation every 30 min) was used for Cytodex-1 and the porous particles for the first 3 hours. Subsequently, the stirring speed was maintained just sufficient to suspended the microcarriers. The spinners were either gassed with an air—0O2 (95:5) gas mixture once daily then returned to an incubator (37°C) or kept inside an incubator with a controlled head gas composition and  47 temperature. pH was not controlled and varied from 7.3 initially to 6.8 prior to medium exchange. 3.2.3 Suspension cell culture  Suspension cultures were grown in suspension in 250 or 500 mL spinner flasks (Bellco). Prior to sterilization the flasks were siliconized to prevent cell attachment to the walls. The flat—blade impeller was rotated at 50 RPM to keep the cells in suspension. Some of the BHK cells aggregated and formed granular clumps of about one hundred to several hundred microns in diameter. At least once daily, a 10 mL sample was withdrawn from the stirred flask for analysis and the spinner flask was gassed with the air — CO 2 (95:5) gas mixture. In some cases, the dissolved oxygen concentration of the culture medium was measured using a blood gas analyzer (Model 168, CIBA—Corning).  3.3 Analytical Methods 33.1 Cell numeration 33.1.1 Porous matrix  At the end of each run, cells immobilized inside the ceramic matrix or the porous particles were washed with phosphate—buffered saline (PBS). A solution of 0.1 g/L crystal violet in 0.1 kmol/m 3 citric acid was then added. The mixture was then incubated for 2 to 3 days to release cell nuclei. The stained nuclei were counted using a hemocytometer (Perry and Wang, 1989). In all cases, the standard deviations were less than +10% at a 90% confidence level. During the course of the ceramic matrix perfusion culture and hybridoma microcarrier cultures, samples of the culture medium were examined for cells released from the porous matrix by counting on a hemocytometer. Viable cells in suspension were assessed using the  48 trypan blue (0.4%) exclusion method (Griffith, 1985). In this method, viable cells are impermeable to the dye, whereas dead cells take up the dye. 3.3.1.2 Microcarriers  For Cytodex-1, the culture medium was removed and the carriers were washed with PBS. Cells were released from the carriers by incubation with a 0.25% trypsin-EDTA solution (Gibco, Ontario) at 37°C for 15 minutes.  Dissociated cells were then counted using a hemocytometer.  33.1.3 Suspension cells  Cells suspended in medium were numerated either by means of a microscopic count using a hemocytometer or a particle counter (Elzone 280 PC, Particle Data Inc., Elmhurst, Illinois).  33.2 Assays 33.2.1 Glucose  Glucose was measured using a Beckman Glucose Analyzer II (Brea, CA). j3-D-glucose reacted with oxygen according to the reaction: G H20 Oxidase + 0- D - glucose + Oz Glucose )Gluconic Acid + H202 (22) The dissolved oxygen concentration was monitored by a rhodium/silver polargraphic oxygen electrode. The analyzer response was linear over the range 0.05 to 4.5 g/L.  3.3.2.2 Lactate  L-lactate was analyzed using a YSI L-lactate analyzer (Model 27, Yellow Springs, OH). L-lactate first diffused through a thin polycarbonate membrane with  49 a nominal pore size of 0.01 ,um. The diffusion rate was the rate limiting step as indicated by the manufacturer. Once past the membrane, L—lactate encountered a thin layer of immobilized L—lactate oxidase where the following reaction occurred: L -lactate oxidase L — lactate + 02^>H202 + pyruvate^(23) Hydrogen peroxide diffused toward the cellulose acetate membrane covered platinum anode and gave rise to a probe signal current which was directly proportional to the L—lactate concentration. The analyzer had a useful range of 0.1 to 1.34 g/L, with a precision of ±2%.  33.23 Lactic dehydrogenase (LDH)  Lactic dehydrogenase activity in the culture supernatant (Holscher and Onken, 1988) was determined spectrophotometrically using a commercially available kit (Sigma, No. 340—UV). The basic reaction is Pyruvate + NADH  LDH  >Lactate + NAD^ (24)  Phosphate Buffer (2.85 mL, Sigma, No. 410-3S) and 0.05 mL of culture medium were added directly into a NADH Vial (Sigma, No. 340-2) and incubated for 20 min at 25°C. A volume of 0.1 mL Sodium Pyruvate solution (Sigma, No. 490-1) was then added to the vial. The solution was mixed thoroughly and transferred to a 10 mm light path cuvet. The absorbance [A] of NADH was measured at 340 nm at 15—second intervals for 3 min, with distilled water as reference. The LDH activity (IU/mL) was calculated based on the rate of change of [A] (i.e. Activity (IU/mL) = 9600 * A[A] per min * temperature correction factor). The lower limit of the assay was 30 IU/mL (i.e. a A[A] = 0.003 per min).  50 33.2.4 Transferrin  The transferrin concentration in the BHK culture medium was determined either using an enzyme—linked immunosorbent assay (ELISA) (Roitt et al., 1985) or a particle concentration fluorescence immunoassay (PCFIA) technique (Jervis et al. , 1991). (a) Enzyme-linked immunosorbent assay (ELISA)  The assay plates were coated with the capture antibody by incubating with 50 III. of affinity purified goat anti—transferrin antibody (20 pg/mL) in carbonate— bicarbonate buffer overnight at 4°C. The plates were washed three times with PBS—Tween solution between chemical additions: (1) 200 .tL of blocking agent (1% BSA in PBS—tween) was added to each well and incubated for 2 hours. (2) 50 tI., of appropriately diluted test samples or known standards were then added to each well and incubated for 1 hour. (3) 50 AL of rabbit anti—transferrin (500x dilution) was added to each well and incubated for 1 hour. (4) Alkaline phosphatase conjugated anti—rabbit Ig diluted 1/3000 in PBS was added to each well (50 gL/well). (5) 100 III., of phosphatase substrate (disodium p—nitrophenyl phosphate) in diethanolamine buffer was added to each well. The plates were incubated in the dark for about 30 min and read using a microplate reader (Model: Vmax, Molecular Devices, California).  (b) Particle concentration fluorescence immuno-assay (PCFIA)  Samples, standards or controls (20 1AL each) at appropriate dilution, were added to wells in special FCA 96—well plates. Standards at 0.25, 0.15, 0.1, and 0.05 jig/mL were used for calibration. Polystyrene spheres (20 4L, 0.7 1AM @ 0.25% v/v, Baxter/Pandex Healthcare Corp) coated with goat anti—transferrin capture antibody (Sigma) were added to each well containing a sample. The samples were gently mixed and incubated at room temperature (21°C) for 20 min.  51 Following the first incubation, 20 pL of sheep anti-transferrin-FITC conjugate (ICN, Costa Mesa, California) were added to each sample and incubated for 20 min, at room temperature in the dark, after gentle mixing. The plate was then evacuated using the Pandex FCA. Samples in wells were washed three times with PBS and read using the 485/535 filter pair at 25X gain.  3.3.3 Microscopy 3.3.3.1 Scanning electron microscopy (SEM)  Specimens with adherent cells were prepared for SEM according to the method of Allen (1983). The fragments were fixed in 2.5% glutaraldehyde in 0.1 kmol/m 3 sodium cacodylate buffer, pH 7.3 for 1 hour, washed three times with 0.1 kmol/m 3 sodium cacodylate buffer, treated in osmium tetraoxide for 1 hour and dehydrated by a series of 5-minute ethanol washes (i.e. 30%, 50%,70%, 95%, and 100%). The specimens were dried (Model CPD020, critical point drier, Balzus Union, Liechtenstein), fixed to clean aluminum stubs using double-sided tape or silver paste and sputter coated with gold according to the Nanotech Sputter coater procedure manual before viewing using a scanning electron microscope (Model S4100, Hitachi, Tokyo, Japan or Model 250T, Cambridge Instruments, Cambridge, U.K.).  33.3.2 PEG embedding technique  Porous microcarriers with cells attached were first fixed and dehydrated according to the protocol above. The specimens were then hydrated and kept in 70% ethanol solution before being transferred to 30% polyethylene glycol (PEG; average MW 3350 Da; Sigma) in distilled water at room temperature. The samples were transferred with a pasture pipetter to a 50% PEG solution at 60°C, infiltrated for 30 min and then transferred to 75% PEG solution at 60°C for a further 30  52 minutes. The samples were placed in the 100% PEG for at least 30 min with stirring. Each porous microcarrier was then carefully moved into a single gelatin capsule filled with fresh PEG (60°C). The matrix hardened quickly as it cooled. The capsule blocks were allowed to cool and set overnight before cutting. Blocks were trimmed on a microtome using a glass knife to reveal the center portions of the porous microcarriers. Most of the PEG on the trimmed block was removed with a razor blade. Residual PEG was dissolved in 70% ethanol—water solution. The samples were then dehydrated by a series of 5—minute ethanol washes (i.e. 70%, 95%, and 100%), dried (critical point drying) and gold—coated, following the standard SEM protocols as outlined above, before viewing.  33.3.3 Thin—sectioning microscopy Microcarrier samples were fixed and dehydrated with ethanol according to the standard SEM protocols discussed previously. The 100% ethanol was replaced with 100% propylene oxide before the samples were infiltrated with a graded series of degassed Epon mixture / propylene oxide without the catalyst DMP-30 (i.e. 33.3% Epon /66.6% propylene oxide; 66.6% Epon /33.3% propylene oxide, 100% Epon mixture). The composition of the Epon mixture was as follows: volume (weight) needed  Epon 12 Resin Dodecenylsuccinic Anhydrate (DDSA) Araldite 6005 resin (Air/6005) Dibutyl phthalate (DBP)  25 mL (31 g) 35 mL (55.5 g) 15 mL (17.6 g) 2 mL (1.9 g)  Tri—dimethylaminothylphenol (DMP-30)  0.9 droplets / g mixture  The samples were placed on a rotator and agitated for 24 hours for each infiltration step. DMP-30 catalyst was added to the freshly prepared 100% Epon  53 mixture solution and agitated for an additional 4 hour period. Each microcarrier was then placed into a plastic BEEM capsule filled with the 100% Epon mixture with catalyst (DMP-30). The filled capsules were then placed in an oven at 60°C and allowed to set for 48 hours. The matrix hardened as it cooled. Blocks were trimmed on a microtome using a glass knife. 0.5 ,um thick cut slices were collected from the pool of filtered distilled water located adjacent to the glass knife edge. Each slice was then transferred to a single droplet of water on a glass slide and heat—mounted onto the glass slide. The samples were then stained and viewed under a light microscope.  33.3.4 Confocal micros copy The cells were labeled with fluorescein diacetate (FDA) and propidium iodide (PI). Esterase within living cells cleaves the ester, liberating fluorescence which gives green fluorescence when excited by the laser source (Nikolai et al., 1991). Propidium iodide is taken up only by dead cells and fluoresces red. Stock solutions were prepared by dissolving 5.0 mg/mL FDA in acetone and 20 mg/mL PI in PBS. The FDA/PI (10:90) working solution (0.5 mg FDA and 18 mg PI per mL mixture) was prepared just prior to use. An appropriate volume of the FDA/PI solution was added to the microcarriers and incubated for 5 min at room temperature. The samples then were washed with PBS twice, placed in the concave depression of a glass slide under a coverslip. The stained samples were examined with an epifluorescene Ziess microscope (Axiophot, Carl Ziess inc., Thornwood, New York) linked with a MRC-500 Confocal imaging System (BioRad, Boston, Massachusetts). The computer—linked system was capable of producing and storing 768x512 pixel images. A three—dimensional image could be reconstructed based on series of two—dimensional images collected at various focal plane positions.  54  4.0 RESULTS AND DISCUSSION -Fixed Bed CERAMIC FOAM PERFUSION SYSTEM The test design featured a fixed bed of ceramic cylindrical foam as a support surface, and an external reservoir arranged as shown in Figure 5. The ceramic foam retained cells in the matrix providing physical protection to animal cells. The cells are thus segregated into a realtively small reactor volume, away from the bulk medium. Product separation is inherent in the design.  4.1 Effect of Foam Porosity For a given perfusion rate of culture medium, the level of shear stress experienced by the cells immobilized within the ceramic foam is a function of the pore opening size. Several trials were conducted to test the effect of the ceramic foam porosity on BHK cell growth within the foam. In these early experiments, growth was assessed only from the rate of glucose utilization in the cultures. A detailed investigation of methods for determining cell number in the cultures is reported in Section 4.3. Glucose uptake rate provides an adequate means of monitoring cell growth for simple comparison of different pore size matrices. The experimental results are shown in Figure 8. Good growth of BHK cells was observed on the 30 PPI ceramic matrix. However, growth was significantly less on 50 PPI and 100 PPI cylinders under similar inoculation and operating conditions (described in section 3.2.1.1). Little or no cell growth was observed on the 100 PPI ceramic foam. This might be a consequence of uneven nutrient distribution, flow channeling due to medium by—pass through the clearance space between the outer perimeter of the foam and the inner wall of the growth chamber or excessive liquid shear on the cells immobilized within the small pore ceramic matrix. All subsequent results were obtained using a 30 PPI ceramic cylinder.  •^  •  55 4.5  4.0  3.5-  3.0c^. 0 Es 2.5*C.  g 2.0 o  1.5-  1.0-  0.5-  0.0 0  2  4  6  8  10  Time (days)  Figure 8. Glucose utilization of BHK cells in batch culture on ceramic foam cylinders of various porosity. Each culture was inoculated with lx 10 8 cells and perfused with DMEM medium at 130 mL/min (see Section 3.2.1.1). Cell growth was based on the rate of glucose utilization.  4.2 Cell Growth on the Ceramic Surface The initial attachment and the morphology of BHK cells on the surface of the ceramic matrix was very similar to that seen in T—flasks. Following attachment, the cells extended and flattened themselves on the surface; they continued to grow to form a confluent monolayer (Figure 9A). Finally, they overgrew, forming a multilayer of cells (Figure 9B). When the culture reached steady—state in the perfusion system, the multilayer of the BHK cells was about 0.1 mm thick (approximately 10 cell diameters). Comparison of the cell morphology when the monolayer was just confluent (Figure 9B) with that of the multilayer (Figure 9C) indicates that cell shape and packing changes markedly during the course of the culture.  56 (A)  (B)  57 (C)  Figure 9. SEM photographs of BHK cells grown on porous ceramic. A: attached cells after 1 day; B: a confluent monolayer after 7 days; C: multilayer tissue after 21 days.  4.3 Estimation of Total Cell Number Cells immobilized on the ceramic matrix are not accessible for direct enumeration during the culture. Therefore, indirect methods based on nutrient uptake rate were used to estimate cell number. The total oxygen consumption rate (OCR) of cells in the growth chamber was calculated knowing the medium flow rate and the change in dissolved oxygen concentration following the passage of the medium through the growth chamber. This value was used to estimate the total immobilized cell number assuming that the specific oxygen uptake rate (OCR/cell) did not vary with time or cell concentration during the culture. Satisfactory correlations between the total cell number and the total OCR have been reported previously for a number of cell lines, including BHK (Lydersen et al.. 1985; Lyclersen, 1987). A plot of OCR as a function of time can substitute reasonably  58 well for a growth curve. It was found that OCR and cell number increased in parallel for culture densities less than 10 9 cells/m 2 (Lydersen et  al.,  1985).  However, they reported that the number of cells increased more rapidly than the magnitude of OCR at culture densities in excess of 10 9 cells/m 2 . The cell specific oxygen uptake rate decreases as the cell concentration increases due to decreasing average cell size. At high cell concentration, the actual cell number within the ceramic foam could thus be much higher than that estimated from the OCR. Figure 10 shows the variation of OCR during the course of a culture and the estimated total cell number in the growth chamber based on a reported value of the specific oxygen uptake rate for BHK21 cells (0.20 mmoles/10 9 cells/h; Bognar et al.;  1983). It can be seen that after 15 days, the total oxygen consumption rate  stabilized, equivalent to a final total cell number of 5x10 9 inside the growth chamber. This corresponds to a cell concentration of 1.6x10 11 cells/m 2 of the ceramic surface, or 1.25 x 10 8 cells per mL of total matrix volume. In contrast, the maximum cell concentration achieved in roller bottles was 1 to 2 x10 10 cells per m2.  Glucose uptake rate (GUR) can also be used to monitor culture growth kinetics and to estimate cell number (Bognar  et al.,  1983, Lazar  et  aL,1987).  Figure 11 shows that the GUR of the cell culture reached a steady rate of 2.4 g/L/day after about 15 days. In agreement with the value based on OCR, this corresponds to a total cell number in the matrix of 5 x 10 9 • This estimate was based on our measurement of specific GUR for BHK cells grown in roller bottles of 0.46 g/day/(10 9 cells), the same value as reported by Bognar  et al.  (1983).  59 25  co  1.5  0 0  20-  b  0 0  .  7<  -1.0 cv 2  Estimated cell number  as 15  0  _J  0  E  0.  E u, C 0  10-  'crs  0.  0 0 0  U)  0  - 0.5  C^-  TD  o  -  as  cr) x 5 -  O  o 0^ 5  0 oxygen consumption rate ^0.0 10^15^20 1  Time (days) Figure 10. Oxygen uptake rate and estimated cell number of BHK cells grown on  a porous ceramic support. Cell number was based on the oxygen uptake rate in the perfusion system with DMEM + 5% fetal calf serum and perfusion rate of 128 ml,/min (1x10 8 cells were introduced at time zero as inoculum). Recovery of the stained nuclei from the ceramic matrix after treatment with citric acid accounted for only 70% of the total cell number estimated on the basis of OCR and GUR. The cells suspended in the circulating medium accounted for an additional 5% of the total cell population. The independent estimates of cell number based on the OCR and GUR rates agreed closely (i.e. within 12%). Despite the prolonged treatment with citric acid (up to 72 hours), the tortuous pathways within the ceramic (which facilitate cell immobilization) probably prevent complete recovery of all the immobilized cells. The remaining nuclei were believed to be embedded within the pores of the ceramic form.  60 25  20 TC‹ X  176  0)  2 1.5 "6  U)  E 1.0^  co  a)  0.5  16 -47:5  cn  •^ 0.0 I^.^I^ ^ 1 ^ 30 40 0^10^20 Time  (days)  Figure 11. Glucose utilization of BHK cells and estimated cell number based on the glucose uptake rate in the perfusion system. Cells were grown in DMEM + 5% fetal calf serum (FCS) and a perfusion rate of 128 mL/min.  4.4 Lactate Production Most of the lactate produced in mammalian cell cultures originates from the  metabolism of glucose and glutamine. The specific consumption rate of glutamine, expressed on a molar basis, is much less than that of glucose (Smiley  et al.,  1989).  The molar ratio of glucose to glutamine in DMEM/F-12 is 7 to 1. The molar lactate yield from glucose is twice that from glutamine. Hence, it can be assumed that most of the lactate produced by BHK cells results from the glycolysis of glucose. Figure 12 shows a linear relation between the total lactate produced and the total glucose consumed throughout the culture period in the perfusion system. The conversion ratio of 0.73 g of lactate per g of glucose is similar to that observed for CHO cells (Perry and Wang, 1989). The high conversion of glucose  61 to lactate reflects inefficient use of glucose as an energy source. In the present work, no attempt was made to increase the efficiency of glucose utilization. However, it was observed that the glucose utilization efficiency was increased (i.e. the lactate—to—glucose conversion ratio decreased to 0.43) at high cell concentration in roller bottle cultures. The reason for the increased utilization efficiency in roller cultures was unclear.  60  50 6) a 40 c.) 0  2  a. 30 -  a)  Rf fLf  2, 20 as O  I-  10-  0  I^ 1^ I  0^20^40^60  80  Total Glucose Used (g)  Figure 12. Lactate production based as a function of glucose utilized for BHK cells in the perfusion system (same operating conditions as in Figure 10).  62 4.5 Stability and Viability of Cells on the Matrix  Cell leakage from the ceramic matrices was monitored throughout the course of the culture. At steady state, the concentration of suspended cells in the circulating medium was about 2x10 5 cells/mL of medium, corresponding to less than 5 % of the estimated total cell number. The viability of the released cell population was about 50%. The activity of lactate dehydrogenase (LDH) in the cell—free supernatant was monitored as an indication of cell death within the matrix. LDH, an intracellular enzyme, is released after cell death (Gardner et al., 1990). The relationship between LDH released from freshly disrupted BHK cells and cell concentration is illustrated in Figure 13. LDH activity in the culture medium increased to a maximum LDH activity of 0.06 IU/mL at the steady state cell concentration of 1.25x10 8 cells per mL of matrix. This is equivalent to the amount of LDH released by the sonication of 2 x 10 5 BHK cells per mL of 140 -  0'  120 -  1000 -J  80 Maximum observed [LDH] in the perfusion system  so 40  0  50^100^150  BHK Cell Concentration (cells / mL x 10 -4 )  200  Figure 13. Correlation between measured LDH activity and disrupted BHK cell concentration.  63 medium or 3.6% of the total cell mass present in the system under steady state conditions. However, this value is misleading because LDH released from BHK cells is not stable in cultures (Arathoon and Birch, 1986). The activity of LDH was determined experimentally to decay exponentially with an average half—life of 7 h, i.e.: A (t) = Ao e — kt^  (25)  where A(t) = LDH activity at time t Ao = LDH activity at time zero k = decay constant, in this case k= 0.099 h -1 . t^= time (h) At steady—state in the perfusion matrix system, the medium was changed once every 24 h. The LDH activity measured in the spent medium thus represents the accumulation of LDH released from dead cells over the 24 h period minus the activity lost by inactivation of the enzyme. Assuming that LDH was released continuously into the medium at a constant rate (i.e. constant cell death rate) and decayed at a constant rate, the total accumulated LDH activity measured at the time of medium replacement can be represented by an integral of A(t) over the time between medium changes (24 h in this case). t^t 1Cid ) A t = f A(t) dt = SA eict dt =--A (1— o k 0^0  (26)  The corrected cell death rate (based on A 0) was thus 9.4% of the total cell population per day. At steady state, the cell death rate must be balanced by the cell growth rate. After 15 to 20 days in culture both the OCR and GUR reached steady—state values (Figures 10 & 11) indicating a constant cell mass. At this point a specific growth rate of 0.0039 h -1 would balance the observed cell death rate. Beacuse of the occurrence of cell lysis after cell death, the measured non—viable  64 cell suspended in the medium could only account for 2.5% of the total cell population.  4.6 Effect of Serum during Stationary Growth Phase in the Perfusion Reactor  Serum is the most ill—defined and expensive component of the culture media. Consequently it is desirable to reduce its usage, especially in large—scale cultures. Under steady state conditions, the specific rate of transferrin production of the BHK cells did not change significantly over the first 3 days when the serum level in DMEM/F-12 was reduced from 5% to 2.5% (see Figure 14). Experiments in T—flasks confirmed that serum was critical only for establishing the cell population. The final cell numbers of BHK cultures with 5% serum were comparable to those with 2.5% serum. Once the culture reached  0  ^ ^ 1-^'^t ^ ^ 1 2^3 4 5  Time (days)  Figure 14. Effect of fetal calf serum (FCS) concentration on steady—state transferrin production rate at a perfusion rate of 128 mL/min in DMEM medium (same operating conditions as outlined in Figure 10).  65 confluency, reduction of serum level to 1% did not cause a significant reduction in cell specific productivity; the final transferrin concentration in the medium reached about 12 mg/L prior to medium change, comparable to the transferrin levels with serum containing medium. However, serum provides the attachment factors needed for cell immobilization. In the absence of such attachment factors, cells detached much more easily in the fixed bed perfusion system. The gradual loss in the total productivity in the steady—state perfusion culture after 3 days with the lower serum level (see Figure 14) was likely caused by cell detachment and washout. The suspended cell number in the spent medium increased from 2x10 5 to 3.2x10 5 per mL of medium as the serum level decreased. However, this cell loss cannot fully explain the loss of productivity. Serum is also known to protect cells from the damaging effects of shear stresses (McQueen and Bailey, 1989). This type of damage may have also contributed to the observed loss in productivity of the perfusion system at low serum concentration.  4.7 Effect of Zinc on Transferrin Production Figure 15 shows the cumulative production of transferrin for cells grown in DMEM or in DMEM/F12 with zinc addition. Zinc added to DMEM/F12 (i.e. 10,uM) acts as an inducer of the metallothionein promoter which regulates the expression of the transfected transferrin gene. Under induced conditions at steady— state the culture produced up to 30 mg day 1 (6 mg per 10 9 / cells / day) — about 5 times the productivity of the non—induced cells. The transferrin concentration in the medium of the perfusion system was continuously monitored over the culture period. Figure 15 shows the cumulative amount of transferrin produced by BHK cells grown in the matrix with the perfusion of DMEM/F-12 medium. The steady—state transferrin production rate of  66 6 mg HTF/10 9 cells/day was reached after 15 days. This coincided with the time at which the glucose and oxygen uptake rates stabilized.  E  80  a) 0 -o 60-  2  a.  IL  I-  If;  40-  I-  I^I^I^•^i 0^2^4^6^8  10  ^  12  Time (days)  Figure 15. Induction of transferrin production. 1 x 10 8 cells were introduced at time zero as inoculum and both types of medium contained 5% fetal calf serum. 10 itM zinc was added to the DMEM/F-12 medium at time zero.  4.8 Perfusion Propagator  Bioreactors with volumes greater than 1000 L are now commonly used industrially. The increase in bioreactor volume requires a corresponding increase in the inoculating cell number to achieve the required inoculum cell density. However, traditional techniques for producing inoculum such as roller bottles are insufficient to deal with the high cell number needed. For example, a 1000 L microcarrier culture system requires a minimum inoculum of 5x10 1 ° to 5x10 11 cells for a relatively short lag period and rapid growth (Reuveny and Thoma,  67 1986). Serial propagation is often used to scale—up cell number for the inoculation of microcarrier culture processes. Harvesting cells by treatment of microcarriers with trypsin for each subsequent inoculation is tedious and also increases the risk  of contamination. If the product of interest is not secreted but contained within the cells, a constant supply of cells is required. Since it had been shown previously that the current fixed—bed ceramic perfusion system provided a 50—fold increase in total cell number (i.e. from 1 x 10 8 to 5 x 10 9 BHK cells). It was reasoned that the current system can function well as a source of inocula for scale—up if it could be cycled through a repeated cell growth and cell recovery using trypsin sequence. However, recovery of the immobilized cells within the ceramic matrices by trypsin incubation was somewhat difficult. Figure 16 illustrates the feasibility of using the current perfusion systems for Vero cell production. The perfusion system had an estimated total number of Vero cells of 6.96 x 10 9 during stationary growth phase. The duration of the trypsin incubation period affected the amount of cells which could be harvested from the system and the time needed for the system to recover and reach subsequent confluence. Following incubation with trypsin for 15, 30, and 30 min., 6.0x10 8 , 3.4x10 9 , and 5.1 x 10 9 total viable Vero cells or 8.6%, 49%, and 74% of the total Vero population were recovered from the system at the times indicated by arrows 1, 2, and 3 in Figure 16, respectively. The increase in cell recovery is due to longer trypsin incubation and PBS washes. PBS washes removed residual medium with serum from the growth chamber, hence increased the effectiveness of the trypsin treatment. The second cell recovery was performed without PBS washes while the cell chamber was flushed twice with PBS prior to the third cell recovery.  68 60  13 00 o0  50 -  40 -  30 -  20-  10o Glucose A Lactate  0 0  .^1 200  II^.  400  I  600^800  i^.^r 1000^1200  Time (h) Figure 16. Cycled growth and harvesting of Vero cells in the fixed—bed ceramic perfusion system. 6.0x10 8 , 3.4x10 9 , and 5.1 x 10 9 total viable Vero cells or 8.6%, 49%, and 74% of the total Vero population were recovered from the system at the times indicated by arrows 1, 2, and 3 following incubation with trypsin for 15, 30, and 30 min., respectively.  69  5.0 RESULTS AND DISCUSSION - AIRLIFT SYSTEM The feasibility of using the ceramic matrix for high density culture was established in the previous section using a perfusion culture system. The fixed bed perfusion system is well suited to intermediate scale operation and has the major advantage that cell recovery is relatively easy. However, this system is amenable to scale—up to only a very limited extent. An aspect ratio (i.e. height to diameter ratio) for the porous cylindrical foam of less than 1 is needed to scale—up the fixed bed perfusion reactor since one cannot increase the perfusion rate through the porous matrix indefinitely to meet the increasing oxygen demand without exposing cells to excessive shear stresses. Reactors with an aspect ratio of less than 1 will cause (1) packed density diversity causing channeling and reducing nutrient supply, and (2) large installation area requirement (Murakami et al., 1991). The current fixed bed system cannot be scaled—up beyond 10 L (the volume of the porous matrix) without reduction in the cell concentration within the porous matrix due to high shear stress (see Appendix 4 for the calculation). A major focus of this project has been to develop a simplified and scaleable matrix reactor. An airlift reactor was designed in which the matrix comprises the draft tube of the reactor. Bubbles rising in the draft tube cause a circulatory movement of medium and induce a convective flow of medium through the matrix. This flow provides nutrients to cells immobilized on the inner surface of the matrix and is the most critical parameter affecting the efficiency of the reactor. The cells are protected by the matrix and are not in direct contact with the rising gas bubbles which increases the overall cell viability in the bioreactor when compared to conventional airlift bioreactor. Although CT airlift systems have been studied extensively, none of the studies have investigated the effect of using a porous draft tube. Fluid mixing is  70 enhanced considerably by creating additional junctions between the riser and downcomer due to short—circuiting of some fluid (i.e. splitting the draft tube into several short sections; Blenke, 1979; Chisti 1989). Similarly, the use of a porous draft tube would be expected to provide enhanced mixing characteristics when compared to a conventional airlift reactor. Unlike the external loop airlift packed bed fermenters proposed by Lazar et al. (1987) or the Fibre—bed CT airlift bioreactor used by Chiou et al. (1991), the liquid circulation rate in the porous draft—tube CT airlift system can be regulated without exposing the cells to excessive shear while the mass transfer characteristics in the bulk fluid remain unaffected. Both the draft—tube wall thickness and the matrix porosity could be adjusted to accommodate the required convective flow through the matrix. The thickness of the porous draft tube wall of the proposed system (13 mm) is much less than the packed bed depth (300 mm) used by Chiou et al. (1991) or by Lazar  et al. (1987) and the liquid residence time is, therefore, much shorter. The concentration gradient across the matrix is unlikely to be as serious a problem.  5.1 Pressure Drop Although the porosity of the ceramic matrix must vary depending upon the number of cells attached to the matrix during fermentation runs, simulation experiments without live cells were used to investigate the hydrodynamic characteristics of the system. It would be difficult to conduct pressure drop measurements with live cells due to slow growth of the cells. It requires weeks to establish steady—state in a live cell system. Variation in porosity was used to estimate the effect of cell growth and attachment to the draft tube. In some experiments, the pores of the porous draft tube were blocked by a thin plastic film to simulate a non—porous draft tube.  71 The pressure drops across the various draft tubes measured at the base of the draft tube are shown in Figures 17a and 17b. The non—porous glass draft tube has a wall thickness of 4 mm. Error bars represent two standard deviations, based on more than 100 data points collected by the computer over a 5—minute period, is represented in Figures 17 to 29. As expected, the use of non—porous draft tubes yielded higher pressure drop than porous draft tubes. Generally, the pressure drop across the draft tube increased with decreasing porosity. The pressure drops across all the thin—film blocked matrix draft tubes were very similar. Under identical operating conditions, the pressure drop across the glass draft tube was greater than that of the other non—porous draft tubes because of the differences in draft tube length. The porous section of the draft tube, composed of several pieces of tubular ceramic elements stacked end—to—end, was located between two non—porous glass supporting tubes. (a) o^ glass draft tube  7  —A— nceporas 100 PPI (thick wall) — +— non-porous 30 PPI (thick wall) A^ 100 PPI (thick wall) 50 PPI (tick wall)  6  O  E  ---0- - 30 PPI (tick wall)  5  E 0_ 4 <1  c.  0  0 3 kv kn  2  Q_  0  •^  0.000^0.004^0.008^0.012  ^  Superficial Air Flow Velocity (m/s)  0.016  72  (b) 5 —v—non-pore 50 PPI (thin wall) 30 PPI (thin wall) —x-50 PPI (thin wall) 4- --x-- 100 PPI (thin wall)  0,, I E 30  0  0  2  -  T. cn a) 0  0  0.000^0.004^0.008^0.012  0.016  Superficial Air Flow Velocity (m/s)  Figure 17. Measured pressure drops across the various draft tubes. The airlift contained 5 L PBS solution and sintered glass sparger was used. (a) Long supporting glass pieces were used with thick wall porous draft tubes (360 mm total length). (b) short supporting pieces were used with the thin wall porous draft tube (260 mm total length).  The effect of the overall draft tube length on measured pressure drops is shown in Figures 18 and 19. The overall length of the draft tube included both the porous matrix and the non—porous glass tubes. Differential pressure drops across the porous draft tubes increased with overall draft tube length. For a given riser gas superficial velocity, the amount of liquid perfused through the porous draft tube depends on the vertical location of the porous section in relation to the rest of the draft tube. A draft tube with a short lower glass support has a greater overall  ▪ 73 inward perfusion rate which results in greater riser liquid superficial velocity and hence lower riser gas holdup. The effect of serum addition on the differential pressure drop across the porous draft tube was determined in the absence of antifoam. Generally, the differential pressure drops decreased with serum addition. The magnitude of the differences between the measured differential pressure drops of airlift reactors containing serum and those without serum increased with increasing riser gas superficial velocity. Visual observation suggested that more bubbles were present in the downcomer with increasing serum addition. The stagnant gas bubbles in the downcomer reduced the differential pressure drop across the draft tube. There was little or no difference in the measured differential pressure drops when the serum level was increased from 2 to 5 % (Figures 20 to 23). 1.75  O  E  E  —o— Short top and bottom pieces (260 mm) o Short top and long bottom (310 mm) ^ ^ Long top and short bottom (310rrni) - o Long top and bottom pieces (360 mm) - -  1.50  tZ  ac 1.25 ..o  --  30 PPI, 5L PBS, thick wall  LI 1.00  0 co  2 O tL  2 2o  0.75  0.50  2 0.25 0.00 •^ 0.000^0.004^0.008^0.012^0.016  Riser Gas Superficial Velocity (m/s)  Figure 18. Effect of porous draft tube length on the pressure drop across the thick wall 30 PPI porous draft tube. Pressure drop was measured at the base of the reactor.  • 74 3.0 100 PPI, 6 pieces in 5 LPBS, thick wall  ...•••••  f  E  2 '5  0  ai  -0 = 2.0 -  is  O  0 0.5 • •  (0  a.  0.0 0.000^0.004  Short supporting pieces (260 mm) Long supporting pieces (360 mm) 0.008  0.012^0.016  Riser Gas Superficial Velocity (m/s)  Figure 19. Effect of porous draft tube length on the pressure drop across the thick wall 100 PPI porous draft tube. Pressure drop was measured at the base of the reactor. 6 pieces, 100 PPI, thin wall with short supporting pieces (260 mm total), 5L liquid volume  — A— PBS alone ■ 2% serum in PBS --v-- 5% serum in PBS 0.000  ^  0.004^0.008^0.012  ^  0.016  Riser Gas Superficial Velocity (m/s)  Figure 20. Effect of serum concentration on differential pressure drops across the thin wall 100 PPI porous draft tube, measured at the base of the reactor, at various riser superficial velocities.  ▪ 75 0.8  O  6 Pieces, 30 PPI, thin wall with  -  short supporting pieces (260 mm),  E  5L Liquid volume  E a_ ▪  0.6  -  sa  O  = 0 0 0_  0.4 -  O O  2  < 0.2o_  0  O O O  0_  ^ O A  PBS alone 2% Serum in PBS 5% Serum in PBS  0.0 0.000^0.004^0.008^0.012^0.016  Riser Gas Superficial Velocity (m/s)  Figure 21. Effect of serum concentration on differential pressure drops across the thin wall 30 PPI porous draft tube, measured at the base of the reactor, at various riser superficial velocites. 0  6 pieces, thick wall, 100 PPI, .5L volume ^I I  2.5 - long supporting pieces (360 mm total)^ ___I  E  ,--  E  • 2.0 (Li .0  • 0 ` 0  O  0  --  -  1.5  1.0 -  0. 2 0.5 -  0  • a_  , ,-1  ■ • •  PBS alone 2% Serum in PBS 5% Serum in PBS  0.0 0.000  0.004^0.008 ^0.012  ^  0.016  Riser Gas Superficial Velocity (m/s)  Figure 22. Effect of serum concentration on differential pressure drops across the thick wall 100 PPI porous draft tube, measured at the base of the reactor, at various riser superficial velocities.  76 2.0 6 pieces, 30 PPI, thick wall long supporting pieces (360 mm total)  E  5L PBS  E  °a 1.5 6 _c)  to  1.0 -  0 0 0 Co 0 < 0.5 0 0  0.0 0.000  ■ • • 0.004  0.008  ^  PBS alone 2% Serum in PBS 5% Serum in PBS  0.012  ^  0.016  Riser Gas Superficial Velocity (m/s)  Figure 23. Effect of serum concentration on differential pressure drops across the thick wall 30 PPI porous draft tube, measured at the base of the reactor, at various riser superficial velocities.  5.2 Mathematical Modeling The most critical fluid dynamic parameters in an airlift reactor are the gas holdup and the liquid circulation velocity. It is necessary to establish the relationship between these parameters and gas flow, liquid and gas properties and reactor geometry in order to be able to model this type of reactor. The liquid flow velocity, and gas holdup can be estimated from an overall energy balance over the airlift loop, similar to that given by Eq (10): Power input = rate of, energy dissipation i.e. Ei = ER + ED + EB + ET + EF + Ep^ (27) where Ei = energy input as per Eq. (1) ER = energy dissipation due to wakes behind bubbles in the riser;  77 ED = energy loss in the downcomer due to upflow motion of bubbles as per Eq (10c); EB = energy loss due to fluid turn around at the bottom of reactor; ET = energy loss due to fluid turn around at the top of reactor; EF = energy loss due to friction in the riser and the downcomer; Ep = energy loss due to flow through the porous draft tube.  For CT with a smooth draft tube, EF and ET are negligible compared to other factors (Lee et al., 1986; Chisti et al., 1988 ). While ET can again be negligible, EF in the current airlift system is expected to be significantly greater for the porous draft tube due to the rough draft tube surface. The gas holdup in the downcomer is assumed to be zero. Equation (27) is therefore, reduced to Ei = ER + EB + EF + Ep^  (27a)  EB can be estimated by Eq. (10a) and according to Chisti et al. (1988), EF = 2 Cf p1 ULr (ULT. + U gr) (hD / dR) (ULr AR)^(28) 3 ' Cf P/ ULr n hD (dR/2 ) where dR = riser diameter Cf.= Fanning friction factor, 0.018 for turbulent flow in a rough pipe; ER = plg hD AR Ub E gr^(29) where Ub = terminal bubble rise velocity . The terminal bubble rise velocity was set as an arbitrary, but realistic value of 0.25 m/s for the model calculations as suggested by Lee et al. (1986), Chisti et al. (1988) and Calvo (1989). Also, according to Chiou et al. (1991), we may write Ep = (2 where  7C r  hp Up) (k Lp ,u Up)^  (30)  78 hp = height of the porous portion of the draft tube; Up = liquid superficial velocity through the porous element;  k = Darcy resistance coefficient (i.e. permeability -1); L = draft tube thickness or porous element thickness;  y = liquid viscosity. Up is a function of the longitudinal position, h, and can be estimated by Up (h)= pi g (hD — h) e gr / (,u k L)  (31)  for Reynolds numbers, Re p , less than 30 (Dullien, 1975), where h = vertical distance measured from the base of the airlift hp = dispersion height measured from the base of the airlift The Reynolds number is defined as  Re p =  17) U p 1 3 /I P  (32)  where D = effective average pore diameter P The energy loss due to fluid turn around at the bottom of reactor, EB can be estimated (Chisti et al., 1988) by: EB = 0.5 p1 ULd 3 AD KB^  (33)  where KB = the dimensionless frictional loss coefficient as per Eq (12). AD = downcomer cross—sectional area ULd = downcomer superficial liquid velocity  In this case, the riser gas holdup was not known a priori. The well—known gas holdup correlation of Hills (1976; Eq. 16) was used. Because of the short— circuiting flow through the porous draft tube, E gr , ULr , and Up all varied with height along the reactor. UL r increased as the flow moved upwards, while E gr and  79 Up decreased. For ease of modeling, the airlift was divided vertically into 5 equal  sections. Values of the riser gas holdup, riser superficial liquid velocity and the liquid superficial velocity through the porous draft tube element were calculated from the base up. Simulation runs were also performed for an airlift with 3 or 9 equal divisions. For the range of riser superficial velocities investigated, the resulting simulation values differed less than 1 percent when the airlift was divided into 9 instead of 5 equal sections. However, this difference increased to 6 percent when 3 equal sections, instead of 5 equal sections, were used for the model simulation. The iterative Levenberg—Marquardt method, a quasi—Newton method, was used to solve for the listed constraints simultaneously. Appendix 3 contains the program listing and a simulation example can be found in Appendix 4. The pressure drop across the draft tube can be calculated from the riser gas holdup with the assumption of no gas holdup in the downcomer. Figures 24 and 25 show the calculated and measured pressure drops at various injected rise air superficial velocities. The Darcy resistance coefficient of the porous matrix, k, in Eq. (30) was determined based on ceramic foam cylinders of equivalent porosity using the equation: OP= k•g•L•U p  (34)  For fixed Ups, the pressure drop across the foam cylinder was measured and k was calculated from the slope of AP vs. Up. k was found to be 4.3 x 10 8 , 6.7 x 10 7 and 3.3 x 10 7 m -2 for the 100, 50 and 30 PPI foam cylinders (for water with Reynolds number, Re p , less than 40), respectively. However, the actual Darcy resistance coefficient could vary considerably from the foam cylinders to the hollow cylinders of the same porosity. The porous hollow cylindrical elements were stacked end—to—end to form the draft tube. Some gaps between the hollow ceramic cylindrical elements and between the glass supporting tube and the hollow  80 cylindrical element were inevitable. Those gaps decreased the flow resistance and reduced the k value. The tendency for the proposed model to over—estimate the differential pressure drop across the draft tube can be explained, at least in part, by the differences between the actual Darcy resistance coefficient and the values, based on the cylinders, used for calculation. However, according to Figure 24, the apparent k value for the thick—wall 100 PPI porous draft tube should be about 6.7 x 10 7 instead of the value used, i.e. 4.3 x 10 8 m -2 . The 6.4 fold difference between the two k values seems too large to be justified by the presence of gaps. Large errors associated with the measured pressure drops across the porous draft tube at the base of the reactor also contributed to the difference between the  0.000^0.004^0.008^0.012^0.016  Riser Air Superficial Velocity (m/s) Figure 24. Pressure drop across the thick—wall porous draft tube, measured at the base of the reactor, versus riser superficial air velocity for different pore spacings. The airlift contained 5 L water and the porous draft tube consisted of 6 cylindrical ceramic elements end—to—end. Each point indicates a measured value, while the lines indicate theoretical calculations based on the model.  81 predicted and measured values. An attempt was made to measure the differential pressure drop across the porous draft at various axial positions. The small scale of the airlift used and the swirling motion of liquid in the airlift made such measurement very difficult. In fact, the magnitude of the measurement error was typically 2 to 3 times greater than the collected data. In some instances, the measured differential pressure drop across the porous draft tube was greater in the top section than measured values at the base of the draft tube. The model also predicted values of superficial liquid flow velocity for corresponding riser gas flow values. An attempt was made to compare the predicted superficial liquid velocities in the downcomer with measured values (see  Riser Air Superficial Velocity (m/s)  Figure 25. Pressure drops across the thin—wall porous draft tube at the base of the reactor versus riser superficial air velocity for different pore spacings. The airlift contained 5 L water and the porous draft tube consisted of 6 cylindrical ceramic elements end—to—end. Points indicate measured values, while lines indicate theoretical calculations based on the model.  ^  ▪ 82 Figure 26). Liquid circulation velocities were calculated by monitoring the time needed for a small colored neutrally buoyant polyurethane particle of 3 mm diameter to travel a known distance along the downcomer. Because of the liquid flow through the porous draft tube, the colored particle moved almost randomly down the downcomer instead of in a straight line. Moreover, the particle sometimes travelled too close to the draft tube wall and the frictional force either slowed or stopped the particle movement. Therefore, large errors were associated with the measured downcomer liquid superficial velocities. The model generally under—estimated the values of downcomer superficial velocity.  0.7  5  0.6  0 CD 0.5>  ^30PPI, dick wall ----50 PPI,thick wall ^100 PPI, thick wall 30 PPI, thin wall 50 PPI, thin wall ^100 PPI, thin wall  cr  Ts  0.4 -  • 0.3 CO  E0  c 0.23 O CD D 0.1 cp  0.0 0.00  0.41^0.82^1.23  1.83  Riser Superficial Gas Velocity (m/s)  Figure 26. Average downcomer liquid superficial velocity at various riser gas superficial velocities. Lines indicate the theoretical calculation. Each point indicates measured values for the airlift with a 100 PPI, thick wall porous draft tube of 360 mm. Error estimation was calculated based on 10 measurements.  83 5.3 Mass Transfer Coefficient  Since the porous matrix draft tube occupied only 5% to 11% of the current 5L reactor volume, an air flow rate of 25 cm 3/s, which corresponded to U gr = 0.012 m/s, could provide sufficient mass transfer (kLa of 0.012 s -1 , see Figure 27) to support a BHK cell loading of at least 3.6 x 10 8 cells/mL matrix. Mass transfer coefficients of a similar magnitude were observed by Murakami  et aL  (1991) for  an airlift fiber bed bioreactor. For example, a kLa of 0.02 s -1 was determined at a riser air superficial velocity of 0.0086 m/s for an external airlift with a draft tube diameter of 50 mm. However, to support a BHK culture with a cell concentration of 10 8 cells/mL medium, a kLa value of at least 0.027 s -1 is needed (see Appendix 4 for calculation). The current system could only provide a sufficient kLa to support 4.4 x 10 7 cells / mL medium. 0.020^ 0.018-  I 0.0160.0140.0120.010••■•••  co, 0.008—a— 30 PPI, short supports, thick wall - -o-- 30 PPI, short supports, thin wall A 30 PPI, long supports, thick wall - -7-- 100 PPI, short supports, thick wall -100 PPI, long supports, thick wall --+-- 100 PPI, short supports, thin wall x non-porous glass draft tube  0.0060.0040.002-  o.  0.000  ^  0.004^0.008^0.012  0.016  Riser Gas Superficial Velocity (m/s)  Figure 27. Volumetric mass transfer coefficient at various riser gas superficial velocities. Two standard deviations are represented by the error bar indicated.  84 PBS containing 5% serum was used to determine the effect of serum addition on the mass transfer coefficient. It was found that the addition of serum had a minimal effect on mass transfer as shown in Figure 28. The downcomer gas holdup increases with the addition of serum. However, since stagnant gas bubbles in the downcomer contribute little to the overall mass transfer, the riser gas holdup is likely not affected by the presence of serum. The measured mass transfer coefficients are compared with those predicted by Equation (21) in Figure 29. The empirical equation proposed by Stejskal and Potucek (1985) underpredicted kLa by a considerable margin. Equations (18) through (20) could not be used since the riser gas superficial velocity used in the current system was too low. kLa values, therefore, need to be empirically determined for the current system. 0.015  0.010-  0.005is ;  --n-- glass draft tube in PBS with 5% serum o^ glass draft tube in PBS A - 30 PPI in PBS with 5% serum - -v- 30 PPI in PBS -  -  -  1  0.000 0.000^0.004^0.008^0.012^0.016 Riser Gas  Superficial Velocity (m/s)  Figure 28. Effect of 5% serum addition on mass transfer coefficient. The porous thin— wall 30 PPI draft tube was supported by short glass tubes. Error bars indicate two standard deviations.  85  0.014-  0.012-  0.010-  0.006-  0.004-  0.002  0.000^ 0.000  0.004^0.008^0.012  0.016  Riser Gas Superficial Velocity (m/s)  Figure 29. Comparison of predicted mass transfer coefficients (using Equation 21) with measured values for airlift with 100 PPI thick wall draft tube of 360 mm. The airlift contains 5 L PBS without any serum addition.  5.4 Scale-up Potential Assessed by Proposed Model The proposed airlift reactor could be scaled—up by either increasing the reactor volume or by increasing the volume of the porous matrix within the airlift. The model previously presented considered the scaling parameters of the reactor, such as height, diameter and porous draft tube thickness. In theory, the scale—up potential of the airlift reactor can be assessed using the proposed model. Although the model predicts the correct hydrodynamic trends, it would be unrealistic to design a large scale airlift reactor using the proposed model without further validation. A pilot scale reactor should be constructed to test the model accuracy.  86 However, the proposed model can be used to optimize the draft—tube thickness and maximize cell loading of the current 5—L airlift reactor. The permissive air flow rate must be defined before any scale—up evaluation can proceed. Air flow rate is limited by foam formation due to presence of serum proteins in the medium. Antifoam is usually used to prevent foaming. Antifoam concentration must be kept below the level at which it becomes toxic to the cells. According to Murakami et al. (1991), an air flow rate of 1 vvm (volume of input gas per reactor volume per minute) was the maximum allowable air flow rate for DMEM/F-12 medium with 1.2% fetal calf serum. Since the medium used in the present study contained 5% fetal calf serum, the allowable maximum air flow rate should be lower than 1 vvm. A value of 0.4 vvm is used for all the following scale—up calculations because the reactor has tested and operated at an air flow rate of 0.4 vvm with very little antifoam C addition (less than 800 ppm). The medium perfusion velocity through the porous draft tube can be calculated assuming (1) the height of the draft tube remains the same as before, (2) air flow rate is kept at 0.4 vvm, and (3) Darcy resistance coefficient of the 30 PPI porous draft tube with cells attached is assumed to be constant at 6.7 x 10 7 m -2 . The calculated results for an airlift with different riser cross—sectional areas are shown in Figure 30. The maximum shear stress associated with the perfusion flow can be estimated using (Perry and Wang, 1989): r = g • UP  dF k 4 . (1—em  )  where I"  = shear stress (N/m 2);  d F = diameter of matrix fibers plus cell layer; CM =  void fraction of the porous matrix.  (35)  87 10.0 ^AR = 0.0038 m 2  9.0  8 0 U) St  a)  A R = 0.0019 r11 2  ^ A R = 0.0007 m 2  11) C\1  -  8.0  - -- A R = 0.0003 m 2  7.0  6.0  5.0  4.0  3.0 10^15^20^25^30^35^40  Porous Draft Tube Wall Thickness (mm) Figure 30. Calculated average perfusion velocity through the porous matrix versus selected draft tube wall thickness for the tested 5 L airlift with various riser area, AR. On this basis, the maximum allowable perfusion fluid velocity should not exceed 0.04 m/s to keep the maxium shear stress below 2 N/m 2 . The assumptions used to evaulate the maximum shear stress are (1) e m = 0.7, (2) k = 6.7 x 10 7 m -2 and (3) dF = 1 mm. Most of the predicted perfusion velocities shown in Figure 30 are around 0.04 m/s. Because of the reactor configuration, the draft tube thickness usually can not exceed 50 mm, it is necessary to reduce the input gas flow rate to decrease the shear stress on cells. The perfusion rates through the porous matrix predicted by the model can supply sufficient oxygen to support cell concentrations of greater than 1x10 9 cells/mL matrix (or 10 12 cells total), if oxygen concentration in the bulk fluid can be maintained at air saturation. The porous  88 matrix can occupy as much as 37% of the current reactor volume. For larger airlift reactors, the draft tube thickness could be increased to the extent that the medium perfusion rate through the porous matrix could not sustain high levels of cell concentration within the matrix without exposing the cells to excessive shear forces. The maximum theoretical allowable draft tube thickness is about 690 mm to support 10 8 cells/mL matrix at a shear stress of 2 N/m 2 or 345 mm to support a cell concentration of 2 x 10 8 cells/mL at the same shear stress level of 2 N/m 2 . The porous matrix volume to total reactor volume ratio would thus decrease as the reactor volume increases. However, the ultimate factor which controls the cell loading, as in the case of the current 5 L reactor is not the thickness of the porous draft tube but the mass transfer characteristics. The current airlift system could only support a total BHK cell population of 2.5 x 10 11 cells at an air flow of 0.4 vvm. To improve and maintain uniform mass transfer throughout the bioreactor, an airlift bioreactor should consist (1) a well—designed gas—liquid separator (i.e. expanded cross—sectional area at the top of the reactor) to reduce gas—holdup in the downcomer, (2) a high aspect ratio (from 6:1 to 12:1) to increase volumetric mass transfer rate of oxygen and (3) a porous draft tube of uniform thickness but with a controlled pore size distribution (decreasing in size from the top to the bottom) to compensate for the increased differential pressure drop across the draft tube at the bottom and, therefore, maintain an uniform medium perfuse flux throughout the entire draft tube length.  5.5 Long Term Culture  The porous draft tube airlift system was for animal cell culture. The draft tube of the tested system consisted of two 30 PPI thick—wall tubular ceramic  89 pieces mounted between two supporting glass tubes. The total length of the draft tube was 220 mm and the porous ceramic provided a total volume of 182 mL. The system was inoculated with 4 x 10 8 BHK cells suspended in 1000 mL DMEM + 10% NCS. It was difficult to achieve an even inoculation of cells throughout the porous matrix. Intermittent gas sparging (sparging for 1 min out of every 20 min) to provide enough liquid circulation to distribute cells throughout the matrix was used for the first 4 hours while cell attachment occurred. Continuous gas sparging was then initiated and the medium level was raised to a position 50 mm above the top of the draft—tube, i.e. there was 1.7 L medium in total. The medium was changed whenever the glucose concentration dropped below 0.5 g/L. The total medium volume was eventually increased to 4 L as the cell population increased. Depending on the length of the draft tube used, the inoculum volume, required to cover the entire porous matrix, could be as large as the total reactor volume. Large numbers of cells were, therefore, needed to initiate the bioreactor. As a result of the small inoculum volume, the draft tube length tested is only one— third of the normal length. Many of the inoculating BHK cells did not attach to the ceramic matrix and were washed out of the reactor after the first medium change. This may explain the long period (i.e.700 h) needed for the system to reach steady state operation (see Figure 31). With a similar ratio of inoculating cells to porous matrix volume, the fixed—bed ceramic perfusion system took about 360 h to reach steady state. Figure 31 shows the cumulative amount of glucose consumed and the total lactate produced by the BHT( cells. The glucose—to—lactate conversion ratio in this case was 0.79, indicating that glucose was used less efficiently than in the perfusion system (see Section 4.4). The use of the single hole sparger did not provide sufficient mass transfer of oxygen and resulted in higher lactate yield. The number of immobilized cells within the matrix was also somewhat lower than for  90 the fixed bed perfusion system (i.e. 9.7 x 10 7 vs. 1.3 x 10 8 cells / mL matrix). Nevertheless, the airlift system was successfully operated for 62 days, despite the less—than—optimum oxygen supply.  rn  350  0  o w 300 0 0  0a  -  o  250  °  o  °  o o C8  0 0 0  OA OA  a) 0 100 cp  E  0 0 A A  50  0  o Glucose A Lactate 11  .^t^.  1^•^I  0^200 400 600 800 1000 1200 1400 1600  Time (h) Figure 31. Cumulative glucose used and lactate produced by BHK cells in the porous draft—tube airlift bioreactor. Inoculum of 4 x 10 8 cells was introduced to the DMEM + 10% NCS medium.  91  6.0 RESULTS AND DISCUSSION — POROUS MICROCARRIERS Although it is possible to scale up the previously described airlift reactor beyond 100 L in volume, the large cell number required as inoculum makes the process difficult to start up. Moreover, representative cell samples from the airlift reactor can not be obtained without sacrificing reactor. On the other hand, it is relatively easy to initiate and scale—up microcarrier cultures. Cell culture using conventional microcarriers in a conventional stirrer bioreactor is common. Cell growth occurs on the particle surface only. The use of porous particles provides additional surface area for cell growth and protects cells from the shear forces in the bulk fluid. Suspended bead immobilization systems can be used in a number of different reactor configurations including suspended beds or stirred tank bioreactors. However, most of the commercially available macroporous beads are collagen based (collagen, gelatin, or collagen—glycosaminoglycan). These materials are obtained from undesired complex biological materials such as skin, tendons, or ligaments which are difficult to standardize, regulate and are prone to contamination if not sterilized properly. Ceramic particles could, in principle, be used as microcarriers. However, the abrasive nature of the small ceramic particles and their high specific gravity precluded their use as suitable microcarriers. Siran porous glass beads are also abrasive in nature and have high specific gravity. The cell loading on Siran porous glass particles was reduced drastically when these particles were used in stirred reactors instead of fixed bed bioreactors (Kratje et  al., 1991). Polystyrene carriers have a number of potential advantages over the porous microcarriers currently available. Polystyrene microcarriers are made from an inexpensive, chemically well defined material, which is amenable to a variety of surface modifications. The material is sterilizable and is already accepted as a cell  92 culture support in the form of plates and flasks. Modified polystyrene porous particles were therefore tested in this study as a new type of microcarrier.  6.1 Cell Attachment Rate 6.1.1 Effect of surface chemistry group modifications Cell growth depends on the microcarrier surface characteristics such as wettability, and chemical group expression. Previous data showed that cells grown on dextran carriers with a moderate degree of substitution with DEAE anion— exchange groups (1.5 meq/g) surpass those grown on beads with higher substitution level (Levine et al., 1979). Polystyrene dishes have been used for cell culture for more than two decades; they are also commonly used to manufacture microcarriers. Cells do not attach readily to untreated polystyrene (e.g. bacteriological grade petri plates). Modifications of the properties of the polystyrene surfaces makes them more suitable for cell culture. These modification processes include treatment with sulphuric acid (Maroudas, 1977; Thomas et al., 1986), treatment with chromic acid (Kiemperer and Knox, 1977; Curtis et al., 1983), and glow discharge plasma treatment (Ramsey et al., 1984; Lee et al., 1991). The exact mechanism responsible for enhancing cell attachment, cell spread and growth on the polystyrene surfaces is still unknown. Some researchers have suggested the importance of carboxyl groups (Ramsey et al., 1984), while others attribute the improved performances to the presence of hydroxyl group (Curtis et al., 1983). The importance of surface carbonyl groups has also been suggested (Eriel et al., 1991). It might be expected that the surface properties of a support would be critical for cell adherence. In the initial investigation, the rate of cell attachment or entrapment was, therefore, compared for the various Polyhipe surface modifications in this study. A carrier loading of 9 g /L medium was used to  93 provide excess carrier surface area. 10 mL aliquots from a single batch were added to 100 mL siliconized glass roller bottles. A 0.1 mL sample of freely suspended cells was taken from each roller periodically and enumerated using a hemocytometer and a cell counter. The cell viability was greater than 95% for all runs. The glass bottles were agitated once every 30 min. Cell attachment/entrapment was calculated from the decrease in suspension cell number. Error bars indicate two standard deviations, calculated using data from duplicate runs, on the figures which follow. Cells suspended in medium without added particles in siliconized glass rollers were used as controls. Control roller bottles provided a measure of cell attachment to the glass roller surface and clump formation. Figures 32 and 33 show that most of the cell attachment took place in the first 120 to 150 min and that the suspended cell number remained roughly constant thereafter. For both BHK and Vero cells Polyhipe S80 had the fastest attachment rate and attained the highest attached/entrapped cell number for the surface modifications tested. In Figure 33, due to that a large number of the Vero cells in the control cultures attached to the wall of -the glass bottle, the total attached Vero cells includes cells attached to the microcarrier as well as those attached to the bottle.  •  94 4x107 3x107 3x107 co 2x107 C.) Y 2x107 1:3 1 X1 0 7 a)  _c as  1x10 '  0 5x10"  -- 0- Cortrol --0-- P80 CM80 —V—S80 —0--- Flex-CM80 100^200^300  ^4()0  Time (min)  Figure 32. Effect of surface modifications on BHK cell attachment rate. Each roller has a starting suspension cell concentration of 7.5x10 6 cells /mL and contains 10 mL DMEM with 10% newborn calf serum. 1 .6x107 -  1.4x107  1.2x107  -  2 toxid a) -o  .c 8.0x10as < 6.0x1CP  0 -  -^- - Control  4.0x1CP -  --0-- P80 A--- CM80 -^Flex-CM80  2.0x1CP -  —0--  S80  0.0  0^100^200 '  300  400  Time (min)  Figure 33. Effect of surface modifications on Vero cell attachment rate. Each roller has a starting suspension cell concentration of 1.9x10 6 cells /mL and contains 10 mL DMEM with 10% newborn calf serum.  95 The influence of surface modification on the rate of attachment of Vero cells was less dramatic. However, S80 beads still had the highest cell attachment/entrapment rate among all Polyhipe particles tested (Figure 33). Similar experiments were performed on polystyrene microcarriers with other types of surface chemistry modifications. They included PER80, DEA80, and DEALG80 microcarriers. Vero cells attached equally well to the positively charged DEA80 and the negatively charged S80 microcarriers (see Figure 34). However, BHK cells attached less well to the DEA80 beads than to the S80 microcarriers as shown in Figure 35. Enhanced cell adhesion to polystyrene surfaces treated with sulphuric acid has also been observed by several other investigators (Thomas et al., 1986; Maroudas, 1976; Curtis et al., 1983; Lydon and Foulger, 1988). It should  be noted that the treatment by concentrated sulphuric acid at ambient temperature, as in this case, produces mainly hydroxylation and limited sulphonationl. The resulting S particles were hydrophilic in nature and wetted easily. All other types of Polyhipe could not be wetted without the use of 70% ethanol.  1  Gregory, D. Microporous Materials Ltd., Braunston, Deventry, Northants, UK, personal  communication (1992).  96 2.00x107 - -0-•- Control - -0-- D EA80 DEA-LG80 1.50x107  -  —V— S80 0 - PER80 -  -  -  w0 0  tooxid0  fiS  4  0  as o 5.00x106  0.000 100  Time (min)  Figure 34. Effect of surface modification on Vero cell attachment to polystyrene microcarriers. Each roller bottle contains 10 mL DMEM with 10% newborn calf serum and has a starting suspension cell concentration of 2.0x10 6 cells /mL. 3.0x107 2.5x107  -  co 2.0x107  CO v a) 1.5x107 cs vs  is 1 .0x107 o F-  5.0x106  -  - ^ Control - --A. PER80 - p- DEA80  -  - O DEA-LG80 0.0 mo  0  100  20C  300  Time (min)  Figure 35. Attachment of BHK cells to treated polystyrene microcarriers. Each roller contains 10 mL DMEM medium supplemented with 10% newborn calf serum and has a starting suspension cell concentration of 1.6x10 6 cells /mL.  ^  97 6.1.2 Cell attachment/entrapment rate for various microcarriers Of all the modified polystyrene surfaces tested, the S80 microcarriers were found to be most suitable for animal cell culture. The performance of the S80 microcarriers in terms of cell attachment rate was, therefore, compared to that of the commercial microcarriers, Cytodex-1, Cultispher—G and collagen—coated P80 supplied by Solo Hill. Because of the smaller particle diameters and relatively low particle densities, Cytodex-1 and Cultispher—G microcarriers had more "exposed" outer surface area than an equal mass of Polyhipe (with the particle diameters ranged from 500 to 1000 pm.) However, cell attachment/entrapment rates were essentially identical for all tested microcarriers (Figures 36 and 37). 6x1& ^ - Control-0-- Cytodex-I ^A CG —V— S80 ----0- - Collegen-Coated ^  5x1CP -  t-i; 4x1& -  0  a) 3x1CP -  _c C.) 2x1CP  1x1CP  ^0  -  ip^I^  0  ^  50  ^  I ^ ^ 100^150 250 200  Time (min) Figure 36. Vero cell attachment to various microcarriers. Equal masses of each type of microcarriers (9 WL) were used. Each 10 mL cell inoculum had a concentration of 7 x 10 5 cells/mL DMEM.  98 2.5x107 --0---Collegen-Goated-O-S80 A Cytodex-1- v- CG -^-Control  100  400  200  Time (min) Figure 37. BHK cell attachment to various microcarriers. Each roller contains 10 mL DMEM with a initial suspension cell concentration of 2 x 10 6 cells /mL and 9 g/L of carrier. 6.1.3 Sulphuric acid treatment  The degree of surface hydroxylation varies with the duration of treatment with sulphuric acid. The effect of this treatment on cell attachment/entrapment rate was investigated using S40 beads for treatment time of 15, 60, and 240 min and for S80 beads, treated with sulphuric acid for 60 min, Only small differences in terms of surface chemistry were found (Table 5, data supplied by the manufacturer). Table 5. Surface characteristics of S microcarriers S80 S40-15 S40-60 S40-240  Treatment duration 60 min 15 min 60 min 240 min  Zeta potential (mV) -43.77 ± 5.24 -57.85 ± 2.56 -50.81 ± 3.32 -39.04 ± 3.84  C (%)  0 (%)  S (%)  Cl (%)  64.44 78.85 79.98 70.71  25 15.21 14.10 21.13  5.14 2.52 2.53 3.75  0.07 0.07  N (%) Na (%) 3.66 2.27 1.34 1.60  1.76 1.15 1.98 2.74  99 As is evident from Figures 38 and 39, cell attachment/entrapment rates were similar for all the S beads, despite differences in pore sizes (i.e. 80 um vs. 40 um) /  and treatment time.  3.0x10  6  U) 0  2.5x10 6-  c  0  2.0x10 6 U)  0  1.5x106-  ar E' a)1 . CX1 0 6 0  4  —0— S80  C.)  --0-- S40-15  0 5.0x10 5 -  A S40-60 - -p- - S40-240  Y  -. .-0-- -  CO 0.0  Control  0^100  ^300 I Time (min)  Figure 38. Attachment of BHK cells to sulphuric acid treated Polyhipe microcarriers as a function of time following treatment length of 15, 60 and 240 min. BHK cells were suspended in the 10 mL DMEM / 10% NCS medium.  6.1.4 Influence of inoculation procedure The addition of a concentrated cell inoculum to a dry microcarrier preparation might enhance the initial entrapment and cell penetration into the carrier particles. However, the entrapped air within the pores of the dry S Polyhipe prevented the uptake of medium. Most of the particles were not wetted immediately upon the addition of the medium and floated on the medium surface.  -  100 Non—wetted S Polyhipe, like other non—wetted polystyrene surfaces, did not support cell attachment/entrapment (Figure 40). The difference in porosity of the particles had little effect on cell attachment rate. The decrease in suspension Vero cell concentration in the medium was due to cells attached to the bottom of the glass roller. Autoclaving of the S particles suspended in PBS released most of the entrapped air. The particles formed a suspension upon the addition of DMEM/ 10% NCS medium. The S beads prepared according to this protocol exhibited enhanced cell adhesion properties, as one can see from Figures 38 and 39. 4.0x1&  E 3.5x1CP 3.0x1CP 0  a) 2.5x105 o_ .c 2.0x1CP c 0  c  1 .5x10 -  0  O 1.0x1of a)  0 2 5.0x105 a) 0.0 0  —0— S80 —0-- S40-15 A S40-60 S40-240 - - 0- - Control ^  100^200  ^  300  Time (min) Figure 39. Attachment of Vero cells to sulphuric acid treated Polyhipe microcarriers as a function of time following treatment length of 15, 60 and 240 min. Vero cells were suspended in the 10 mL DMEM / 10% NCS medium.  101 3.0x10  E U)  2.5x10  Tf) 0  2.0x1rY  c 1.5x10 0 0  2  a)  1.0x10  0  (7) c 5.0x105 0.  U)  0.0  —0— S80-Wet --0-- S80-Dry A S40-Wet S40-Dry -^- Control 0^50^100^150  ^  200  ^  250  Time (min) Figure 40. Cell attachment to dry particles. Vero cell suspensions were added directly to the sterile dry S40 and S80 microcarriers.  6.1.5 Effect of particle diameter The S40 (15 min concentrated sulphuric acid treatment) were sieved and divided into three different particle diameter ranges: less than 0.3 mm, between 0.3-0.5 mm and between 0.5-1.0 mm. The purpose of the experiment was to determine the importance of outer "exposed" surface area. High microcarrier loading (9 g/L) of all three diameter ranges S40 microcarriers provided excess total surface area even with complete cell attachment. However, the area of the particle surface exposed to the cell suspension was significantly higher for the particles with the smallest diameter. If the cells penetrate quickly into the inner pores of the S40 particles at the start, then the cell attachment/entrapment rate should not be affected by the particle diameter. The number of Vero cells attached  102 to the smaller diameter particles was 50% higher than the corresponding number attached to the larger particles during the first 60 min (i.e. 1.5 x 10 7 vs. 1 x 10 7 total attached cells, as shown in Figure 41) suggesting that initial attachment was predominantly to the outer exposed particle surface.  E  co 75 0  c O cTs .  E  a) 0 C O  0 a)  0 0 a) > c0 cac_ a) a_ =ca  CI)  I^ I  0^100^200  ^  300  Time (min)  Figure 41. Effect of particle diameter on the attachment rate of Vero cells to Polyhipe particles treated with sulphuric acid. Each roller conatins 10 mL DMEM medium with 10% newborn calf serum and 9 g S40 carrier /L. 6.1.6 Influence of the inoculum cell concentration  The inoculum concentration might be expected to influence the cell attachment rate. However, for the range of Vero inoculum concentrations used, there was little or no difference among the normalized attachment rates (Figure 42). The high carrier loading of 9 g/L provided sufficient surface area for initial  103 attachment up to 2.8 x 10 6 Vero cells. The saturation cell number was about 3.5 x 10 8 BHK and 6 x 10 8 Vero cells per gram of S40 microcarriers, respectively. 90  1^1^1^•^I^If  Vero cells with 9 g/L S40 beads zr 80)  —1:1-5.65x16 cells/mL  (.2  --0--2.82x16 cells/mL A  1.41x1CP cells/mL  - -v- -0.71x16 cells/mL - -0 -control -  • 50a) _c 0 as 40as  r 0^20^40^60^80^100 120 140 160  Time (min)  Figure 42. Inoculum concentration effect on the Vero cell attachment rate. The control was inoculated with 3 x 10 6 cells/mL. Each roller conatins 10 mL DMEM medium with 10% newborn calf serum and 9 g S40 carrier /L.  6.1.7 Effect of medium composition on cell attachment rate  Cell attachment rate can be strongly influenced by the medium composition (Hu et al., 1985). The presence of serum was shown to inhibit attachment and entrapment of 2E11 hybridoma cells (See Section 6.2.7). For Vero cells there appeared to be no significant effect of medium composition on attachment rate (Figure 43) for the range of conditions investigated.  104  J E  5.0x1d  5  -  c 4.0x10iT55 a) 0 3.0x10  -  0  a)  E  2.0x1&  -  —o—DMEM + 10% NCS — 0—MEWF-12 + 10% NCS a) A^ DMEM/F-12 1.0x10- -v- -PBS ----'-DMEM + 10% NCS (Control) 0.0  I 0^50^100^150^200^250  300  Time (min) Figure 43. Effect of medium composition on Vero cell attachment rates to S40 microcarriers. The initial pH of the various media was 7.2. Cell viability was greater than 90% for all cells except those in PBS. Viability of the Vero cells in PBS decreased from 95% to 60% after 4 hours. The control contained no microcarrier and cells were suspended in DMEM with 10% NCS.  6.2 Long Term Microcarrier Cultures Previous work showed that polystyrene particles treated with sulphuric acid had the highest cell attachment rates. However, this does not guarantee that the long term cell culture performance of S Polyhipe will be superior to carriers with different surface modifications. Long term cell culture experiments were necessary to determine the effect of particle surface chemistry modification on cell growth characteristics.  105 6.2.1 Biomass evaluation  The sponge—like open pore structure of the styrene based microcarriers is illustrated in Figure 7. Direct evaluation of biomass within these microcarriers is difficult. Cell growth could not be monitored by light microscopy due to the opaque nature of the Polyhipe particles. Glucose uptake rate (GUR) was used as an indicator of cell growth for the fixed bed ceramic perfusion system. For cultures in exponential growth, under constant operating conditions, the rate of glucose consumption is proportional to the cell concentration. However, the use of GUR could be misleading depending on changes, for instance, in glucose depletion or growth inhibition at decreased pH. Experiments were conducted in T— flasks to determine the effect of glucose concentration on cell glucose uptake rate. At lower glucose concentration the GUR of BHK cells was reduced and the relationship between GUR and cell number changed as illustrated by Figure 44. For porous bead cell loading calculations, the GUR was, therefore, determined in the glucose concentration range between 4.5 and 2 g/L after each medium exchange. Based on those experiments, BHK cells use about 0.46 g/10 9 cells/day, while Vero cells utilize 0.27 g glucose /10 9 cells/day. Before adopting GUR as the main criterion upon which to estimate cell numbers, several methods of enumeration were compared. Direct cell number enumeration techniques were investigated. Treatment of Polyhipe microcarriers with trypsin released about 10% of the estimated total cell number, based on GUR. Alternatively, the particles were resuspended in a solution of crystal violet (0.1 g/L) with 0.1 M citric acid to digest the cell mass and liberate nuclei for counting. Even after prolonged treatment (72 h) and repeated flushing of the crystal violet / citric acid solution through a bed of particles, only about 60 % of the estimated cell number based on GUR could be accounted for. Presumably the pore structure of the macroporous beads entraps cells and prevents complete recovery. Nuclei  106 counts recovered from cells grown on Cytodex microcarriers were within 10% of the cell number estimated from GUR.  I^•^•  a  1.8 1.6  1. 4 112 1.2 as^• CC 1.0  cu  r). a) In 0  0.8 0.6 0.4 —0— 1 x 108 cells/ mL --o-- 5 x 107 cells / mL  0.2 0.0 00  1^•^I  I^I  I^•^I^•^I  0.5^1.0^1.5^2.0^2.5^3.0^3.5^4 0  Medium Glucose Concentration (g/L) Figure 44 Effect of medium glucose concentration on BHK cell glucose uptake rate. Cells were allowed to grow to confluency in parellel T—flasks and the medium was replaced with fresh DMEM medium with 5% fetal calf serum. Trypsin solution was utilized for direct cell numeration.  Cellular ATP has been shown to be a reliable measure of viable cell mass over the course of suspension hybridoma 2E11 batch cultures (Sonderhoff et al., 1992). In the present work, known amounts of the porous microcarriers were extracted with trichloracetic acid (2.5%) and the ATP released was assayed. The total cell number immobilized in the microcarrier was then calculated based on the average value of ATP obtained from BHK suspension cells. The cell number estimation based on the ATP assay was close to the estimate from the glucose uptake rate. Table 6 summarizes the results for steady state cell number estimation  107 on CM80 microcarriers (0.5 g/L) based on three different methods. The pattern of glucose utilization for the CM80 BHK culture is shown in Figure 45; the medium was changed whenever the glucose concentration in the medium dropped below 1 gil, . The culture reached steady state after 250 h.  Table 6. Comparison of estimated cell number (cells / mL beads) by alternative methods  Methods  Nuclei Count  ATP Concentration 4 fmol/cell  Estimated cell number per mL carriers  1.2x108  2.1x108  Glucose Uptake Rate ts^. 0.46 x 10 -9 g/cell/day 2.5 x 10 8  4001  C  0 3I.. a' 0 0 C 0 0 0 20 0 0 = (79  3  0  •^1^• 0^50^100^150^200^250  Time (h)  Figure 45. Glucose concentration in BHK CM80 roller cultures of BHK cells. DMEM medium with 5% newborn calf serum (NCS) was changed whenever the glucose concentration dropped below 1 g/L. 5 x 10 7 cells were introduced as inoculum.  108 6.2.2 Cell growth on Polyhipe in roller bottles 6.2.2.1 Vero cell growth, effect of carrier surface modification  The performance of the porous polystyrene particles with the following surface chemical group modifications was tested using Vero cells. The modifications included CM, DEA, S, 0, and CS. The pattern of glucose utilization for Vero cells growing on CM Polyhipe is shown in Figure 46. Figure 47 shows scanning electron micrographs of Vero cells growing on DEA80 Polyhipe (5 g/L). In Figure 47 the laminar structure of the DEA80 particle is evident. Cells attached to the laminae are not easily distinguished in this micrograph. Similar culture conditions were used for all the Polyhipe particles. Figure 48 gives the cumulative glucose utilization profile for these cultures. It shows that the glucose consumption is significantly less for the cells grown on the Q Polyhipe and CS Polyhipe. The 5  a  4 -o  C 0  ""6 3 0  U  C O 0  0  =^I^=^I 0^100^200^300^400  500  Time (h)  Figure 46. Glucose concentrations in semi-continuous perfusion of Vero cells on CM80. The roller culture was inoculated with 1 x 10 8 cells and contained 0.5 g CM80 beads/L DMEM with 10% Fetal calf serum.  109 profiles for the other cultures appear to be virtually identical. The glucose utilization rate (from the slope of the glucose concentration between 2 and 4.5 g/L vs. time plot, e.g. Figure 46) of vero cells on particles with different modifications was used to estimate cell number as a function of culture time for the Polyhipe cultures. In most cases the cell number increased rapidly for about 200 h and then stabilized at a value of approximately 1.25 x 10 8 cells/mL of matrix. Different behaviour was found for the vero cells on Q80 and CS80 polystyrene particles (see Figure 49). Figure 49 showed that polystyrene microcarriers modified by addition of chloromethyl, diethylamine group, and sulphonation can support vero cell growth equally well. Surprisingly, vero cells growth on the modified polystyrene particles (P40 and P80) was comparable to that on the charged DEA80, S80 and CM80 particles. It should be noted that the non—wetted Polyhipe polystyrene particles do not support any cell growth.  Figure 47. Vero cells on DEA80 particles.  •  110 3.0 —  2.5 —  0.5 —  0.0  If^  •  0^100^200^300  400  500  Time (h)  Figure 48. Cumulative glucose utilization by Vero on particles with different modifications. Operating conditions are outlined in Figure 46. 1.50 0  1.25  6 6 ^  ^  •  6 ^ 6  .e? o  1.00 —  0^X  ^  O  0  xx  A  x  "5 0.75-  J  E  (I)  ^  0.50—  CD  X 6  2 > 0.25 —  CbO ^ 6 p  00  4+  Ff  x  A  o S80 ^ CM80 o DEA80 A P40 v P80 + 080 x CS80  x  +  +  0.00 0^100^200^300  ^  400  500  Time (h) Figure 49. Calculated Vero cell densities on various Polyhipe particles based on glucose utilization rates. Operating conditions are outlined in Figure 46.  ^  111 6.2.2.2 BHK cell growth, effect of carrier surface modification  The growth characteristics of BHK cells on Polyhipe were similar to those of Vero cells. In roller bottles, the cells grew predominantly on and within the particles. The performance of Polyhipe compares well to that of Cytodex 1 in these cultures as shown in Figure 50. DEA80 Polyhipe was used here as it represents the surface modification most comparable to DEAE dextran (Cytodex 1). The DEA80 beads had an exchange capacity of 1.9 meq/g, similar to the reported optimum exchange capacity range (1.5 — 2.5) for DEAE Sephadex microcarriers (Levine et al., 1979). The optimal exchange capacity depends on both the cell type and the microcarrier material (Himes and Hu, 1987); hence, the optimal exchange capacity of the DEA80 beads may not be 1.9 meq/g, particularly because the charge groups are on the cell growth surface, in contrast to the DEA80 30.00 -  25.00 -  ^  a  13  O O  ^20.00  -^  O  8 8  10.00 O  O  8  5.00 -  ^ Cytodex-1 I o DEA80 0.00  ^  0  ^  100  T  200^300  400  ^  500  Time (h)  Figure 50. Cumulative glucose used by BHK cells grown on 5 g/L of microcarriers in DMEM/10% fetal calf serum with 10 ,uM zinc. 2 x 10 8 cells were introduced to the roller cultures as inoculum.  112 particles where the exchange capacity is uniformly distributed throughout the particle. Both cultures utilized the same weight of carrier (i.e. 5 g L -1 ). However, due to the difference in density, the volume of the Polyhipe is approximately 67 % that of the Cytodex—I. The final cell concentrations in these cultures were approximately 6.1 x 10 7 cells mL-1 of Cytodex-1 and 8.5 x 10 7 cells mL -1 of DEA80. It should be noted that the microcarrier volumes were over—estimated since the settled volumes of the microcarriers, suspended in PBS, were used for the cell concentration calculation. The actual microcarier cell concentrations could be 30% higher. The particle diameter of DEA80 was greater than that of Cytodex1 (i.e. 500 to 1000 gm vs. 131 gm). Therefore, the outer 'exposed' surface area to volume ratio of the DEA80 beads was significantly lower than that of the Cytodex-1. BHK cell growth within the pores of the DEA80 particles presumably compensates for the lower outer surface area.  °  BHK cells on 2 g/L carriers^4i1 20.00  -  is. c; o  a 6 15.00-  )  o a) ci) D a) en o 10.00c) =  5  o  Tu  o ^ O ^  1-2 5.00-  0.00  I^I^1  0^100^200^300  Cytodex-1 Cultispher-G CM80 P80  400  500  Time (h)  Figure 51. Cumulative glucose uptake of BHK cells on various types of microcarriers with a concentration of 2 g/L. Operating conditions as per Figure 50.  113 Growth of BHK cells on CM80, Cytodex I and Cultispher—G particles (2 g/L) in roller bottles was also investigated. Figure 51 shows that the growth characteristics of BHK on the three types of microcarriers were similar based on the glucose utilization rates. Final cell concentrations were estimated to be 9.4 x 10 7 cell mL -1 of Cytodex, 1.4 x 10 8 cells mL-1 of CM80 and 1.0 x 10 8 cells mL-1 of Cultispher—G. Cultispher—G particles have a lower void volume within the particles compared to Polyhipe particles (i.e. 50% vs. 90%). This likely accounts for the lower cell concentration obtained with the Cultispher—G cultures. BHK cells grew equally well on Cytodex-1 and CM80 particles at microcarrier concentrations of 2 or 5 g/L. However, the increase in microcarrier concentration did not lead to a proportional increase in glucose uptake rate. Instead, a 2.5 fold increase in carrier concentration gave only a 1.5 to 1.6 fold increase in the glucose uptake rate (see the slopes of Figures 51 and 52). A decrease in the cell number per unit volume of microcarrier with increased carrier loading was also noted by Smiley et al. (1989) for CHO cells grown on Cytodex-1 microcarrier. Croughan et al. (1988) found that increasing the microcarrier concentration by a factor of two resulted in a 25% reduction in average growth rate of FS-4 cells. They attributed this to an increased frequency of collisions between the microcarriers. They also found that the growth kinetics of FS-4 cells were strongly influenced by the cell concentration. Nutrient limitations caused by insufficient medium exchange, uncontrolled medium pH, and high oxygen consumption rate are all factors that may also prevent the cell concentration from increasing in proportion to the microcarrier concentration.  114 I^I^.^1 BHK cells on 5 g/L carriers  30.00  t  0  g  25.00 o 0  8^_  o  -a, 20.00 go a) U) = U) O 0 =  15.00 O O  (5 10.00 cc o I-  o  8  o0 o®^0 O 8  g  0 g  o°8 O^ur  5.00  0.00 0  o O^iI1 cr u ^  100  ^  o v O 200^300  Cytodex-1 Cultispher-G CM80  ^  400  ^  500  Time (h) Figure 52. Microcarrier (5g/L) cumulative glucose consumption of BHK roller cultures in 100 mL DMEM medium. Each roller culture was inoculated with 2x10 8 cells. 6.2.23 Transferrin production  The cumulative transferrin produced versus time for various microcarriers is shown in Figure 53. The transferrin production rate of BHK cells grown on Cytodex was greater than for those grown on CM80 or Cultispher-G microcarriers. At steady state (after 100 h), the transferrin production rates were 115, 101 and 90 gg/10 9 cells/h for BHK cells grown on 2 g/L of Cytodex, CM80 and Cultispher-G microcarriers, respectively. Mignot et al. (1990) reported lower cell specific productivity of CHO cells (i.e. ,ug/cells/h) when grown on CultispherG cultures compared to Cytodex-3. Protein secretion can be limited to the exposed cell surface area available for protein transport. Observed under the microscope, cells grown on the Cytodex-1 microcarriers generally formed a monolayer of cells or a multilayer less than 5 cells thick. The Cytodex-1  115 microcarrier also has a much smaller particle diameter. The total exposed cell surface area to cell number ratio is higher for cells on Cytodex-1 than for the other two types of microcarriers. The low level of agitation used in this study may also introduce a diffusion limitation which is unfavorable to the cells grown within the pores of the CM80 and Cultispher—G cultures. Either of these factors may explain the observed reduction in the transferrin production rate.  Time (h) Figure 53. Cumulative transferrin produced in BHK roller cultures containing 2 g microcarrier/L DMEM medium, supplemented with 10 fiM zinc. 6.2.3 Particle pore size effect  The particle pore size had little or no effect on cell attachment rate as one can see from Figures 38 and 39. Presumably cells attached initially to the particles' outer surfaces. To investigate the effect of the pore size on long term culture, a series of 100 mL glass rollers with 2 g/L particles suspended in 30 mL medium  ▪  116 was used. Each roller was inoculated with 10 7 BHK cells. BHK cell growth was similar on polystyrene particles with different pore size (Figure 54). One of the P5 cultures failed to grow and had to be re—inoculated at 200 h. All cultures eventually reached similar steady state glucose uptake rates corresponding to cell concentrations of 2.2 — 2.4 x 10 8 cells/ mL particles. 23.33  0 ^  0 0g 0.  20.00  O 0 g O 0 •  16.67  a• a tt 0  0•  0  13.33 0  is  •  Oft^• •  10.00  6.67 rj  i1  3.33  0.00  •  0  I  0 i •  ^ ta a2 ^ a a°1gI  Is  i**• i e • ••• • 100  •  •  •  •  ^ P5 o • P25 • P40 ♦^A • • o P80 •  •  200^300^400  •  500^600  Time (h)  Figure 54. Total glucose used by BHK cells on polystyrene particles of different pore sizes. Each roller was inoculated with 10 7 BHK cells and contained 2 g microcarrier /L medium. Failure to observe differences in cell growth with particles of different pore sizes might be caused by mass transfer limitations which restrict growth within the particle. A second series of experiments using duplicate 250 mL spinner flasks with low particle loading (0.5 g/L) was performed. It was reasoned that the high agitation introduced by the use of the spinner, low particle loading and frequent medium exchange (i.e. medium was changed whenever glucose concentration dropped below 2 g/L) would reduce any diffusion limitation. Before cell culturing,  117  all particles were sieved to ensure that each type of microcarrier tested had a particle diameter between 250 and 500 ,um. P25 particles were not wetted and most of the particles were washed out after the first medium change. Cell growth was similar for all particles except for P25 cultures (Figure 55). The glucose uptake rates stabilized at 0.018 g/h per L medium (except for the P25 BHK cultures) equivalent to a cell concentration of 1.56 x 10 8 cells/mL carriers. 17.50  •^1 Effect of pore size on the immobilized BHK cells growth on polystyrene beads  15.00^  o P5 P25 P40 x x P80 A v  12.50  o  10.00-  ^ X^3K 0  7.50 -  X  1  O  + ^  Cu  2.50-  0.00  O O  O  o 5.00 -  x  e  -  0 + 6 ^^ O 15 ^  cr,  ▪ = CO  ^  0  O O x_ X g g X  V vV A A A .  1^•^1^•^1^'^1^•^1^•  0^50 100 150 200 250 300  350 400 450  Time (h)  Figure 55. Cumulative glucose used for BHK polystyrene microcarrier spinner culture. Each spinner contains 0.5 g particle/L DMEM / 10% NCS medium. In these experiments, the BHK cells formed cell clumps with diameters up to 0.5 mm. These clumps were difficult to separate from the similar sized microcarriers. Thus, the measured BHK glucose uptake rates reflected the total cell population within the spinners instead of those on the porous beads. To overcome this problem the experiments were repeated using Vero cells. Vero cells  118 grow preferentially on surfaces and do not usually form cell clumps. The glucose uptake rate of Vero cultures should therefore indicate the attached cell population only. However, no differences in total glucose uptake rates or lactate production rates were evident for any of cultures (see the slopes of Figures 56 and 57).  17.5  15.0-  12.5-  b)  .  FD:1  -  co  10.0-  (i)  =  7.5 -  a  Ts 1— es 5.0 -  135 v P25 o + P40 x * P80 A  2.5fl  0.0  I  I  0^100^200^300^400  500  Culture Time (h)  Figure 56. Cumulative glucose consumed by Vero cells on P microcarriers. Each spinner was inoculated with 7.8 x 10 7 Vero cells and contained 0.5 g microcarrier per L DMEM medium. These results for both BHK and Vero cells indicated that pore size did not influence cell growth. Since the P5 pore size of 5 yin (nominal) is smaller than the size of either cell type, it appears that cell growth was mainly on the outer surface of the particles.  119 12.5 to  4  10.0 -  0  0  0 0 0  8 2.5-  8X  0  0  0  ^^P5 A v P25 o + P40 x x P80  0.0 ^ ^ ^ ^ 200^300 400 0^100 500  Time (h) Figure 57. Cumulative lactate produced by Vero cells on P microcarriers. Operating conditions are the same as outlined in the legend of Figure 56. Confocal microscopy was used in an attempt to determine the degree of cell penetration into the particles. However, the laser beams did not penetrate the polystyrene structures resulting in dark shadows in the microscope images (Figure 58). Thin sectioning was performed on the Vero polystyrene microcarrier cultures. Figure 59 shows the Vero cell growth on the various P particles. The pore size of the P5 particles was too small for cell entry; most of the Vero cells grew on the outer surface of the particles. Little difference in pore sizes between the P40 and P25 particles was found. Cells appeared to grow equally well on both types of particle. P80 particles had many large void spaces throughout the particles, indicated by the empty spaces in the middle of the thin sectioned microscope image. The true effect of pore size on cell loading might be masked by the  120 inconsistencies found with the particle structures and size distributions. According to the manufacturer, it was difficult to manufacture particles of 80 aum pore size with the same structural consistency as particles with 40 ,um pore size. The larger void space reduced the total available space for cell attachment and entrapment. Therefore, particles with 40 ,um pore size were used for subsequent long term culture studies.  Figure 58. Confocal images of BHK cells growing on a P80 Polyhipe particle. The particle has been stained with fluroscein diacetate (green) to indicate viable cells and propidium iodide (red) to indicate dead cells. Overlayed green and red images are shown in white ( = 100 ,um).  121 (A)  (B)  122 (C)  (D) r •  11.  .**^-  Figure 59. Thin sectioned microscopy images of Vero cells on polystyrene particles after 500 h culturing (same operating conditions as per Figure 56). Particle pore sizes are 5, 25, 40 and 80 um in images (A), (B), (C) and (D), respectively. Scale bar shown indicates 100 ,um distance.  123 6.2.4 Minimum cell inoculum requirement  A minimum inoculum concentration is needed to initiate cell attachment and subsequent cell growth for animal cell microcarrier cultures (Hu et al., 1985; Forestell et al., 1992). This cell number depends on the medium composition, the medium pH level, the microcarrier type and the cell type. An inoculum of at least 1-2 cells per bead is generally recommended to minimize the number of unpopulated microcarriers (Forestell et al., 1992). The effect of initial inoculum concentration on Vero cell grown on S40 and Cytodex-1 was examined using either 100 or 250 mL spinner cultures containing 2 g beads per L DMEM medium. Duplicate spinners were inoculated with 1.36 x 10 5 , 2.72x 105 , 4.08x 10 5 and 5.44x 10 5 cells per mL medium — corresponding to 1, 2, 3, and 4 Vero cells per Cytodex-1 microcarrier, respectively. The volume of the larger S40 particles was up to 20 times greater than that of Cytodex-1. Therefore, the approximate S40 microcarrier number per g dry weight could be 20 times less for Cytodex-1. Due to the irregular shapes of the particles, the exact number of S40 carrier particles per g dry weight was difficult to estimate. Figures 60 to 63 illustrate that cells reached confluency faster on Cytodex-1 than on S40 beads, especially at low inoculation concentration, even though the S40 microcarriers were inoculated with higher cell—to—bead ratios. With the lowest inoculation concentration (i.e. level #1 of Figure 60 to 63), the glucose uptake rate of the Cytodex-1 cultures reached steady state after 200 h. However, it took 350 h for the corresponding S40 cultures to reach steady state. Despite the lower steady state glucose uptake rate for S40 Vero cultures (i.e. 0.075 vs. 0.1 g/L/h), the final estimated cell concentration was higher (2.7 x 10 8 and 2.5 x 10 8 cells / mL carrier for S40 and Cytodex, respectively). The less dense Cytodex-1 carriers had more carrier volume on the basis of equal mass. The high Vero cell loading on the S40 particle is confirmed  ^ ^  124 by the scanning electron micrographs (Figure 64). The cut—open particles had Vero cells growing throughout the entire particles. Vero Cylodex-1 cultures with different  4o-  inoculating cell concentrations  i a a  ^At  0  ^X  ^3  .^Tic x  30-  x  -6  x  *A  D to  cn  v  x8  s  0  0  A 6 X g 3K  V  a  8  r 8  § 20E1 0  1 AX? M (5 8  10-  Level #1 + Level #2 x Level #3 ^ X Level #4 ,^ o  ^  I w 08  o •  I a i 1  0 411--1,----1--1--11-15a 0^100^200^300^400^500^600  Time (h)  Figure 60. Cumulative glucose used by Vero cells on Cytodex-1 microcarriers. 45 Vero S40 cultures with different 40- inoculating cell concentrations o 35-  o  ..-.. 30-  0 + x  V  Kt  x  i 1  X  0  0 20z ^X  +  V +^^ X ii A X i I +^^ 0 0 ^ x ii + + 0  t +  ^  0 0  0  ^^  2+A a + t,^a o  x ik, AE A^^ ^^o xiI ° 0 0 0 0 0  5-  * 0  2  iii ^^ 0A+  10-  0  X  x  Level #1 Level #2 Level #3 Level #4  V , 0 25D  ^x  x  o ° o ° 1 88 00 A 8 1 ODD  100  ^  200^300^400  ^  500  ^  600  Time (h)  Figure 61. Cumulative glucose used by Vero cells on S40 microcarriers.  125 40 Vero Cytodex-1 cultures with different inoculating cell concentrations  30-  0  3  x ^  0 0  -a 2 20X  fU  8  _J  8 10-  ill^  0  ^ 0 Level #1 o^+ Level #2 A^x Level #3  0  0 0  .*O1  v^I( Level #4  0  I 0^100^200  400  500  600  Time (h)  Figure 62. Cumulative lactate produced by Vero cells on Cytodex-1 microcarriers.  35-  Vero S40 cultures with different inoculating cell concentrations x  30-  'LT -  6 25-  ^ o  o + x Yi  Level #1 Level #2  x  Level #3 Level #4  x  0^)K  x  +t ^  ^ 0  0  ^ 0 ^ 0 0 O  5 10 -n  0 ),( v  0I  i x ^a 118 0 g 0^ 0 • • wlt I 6^xg  5-  0  100  ^ 4s1  ^  200^300^400  500^600  Time (h)  Figure 63. Cumulative lactate produced by Vero cells on S40 microcarriers.  126 (A)  Figure 64. Scanning electron micrographs of vero cells on S40 carrier after 600 h culture time. (A) and (B) are two views of the same particles.  127 6.2.5 Cell growth on S40 Polyhipe in spinners  BHK Cell growth of the S40 Polyhipe was compared with that of Cytodex-1 and Cultispher—G microcarriers. Carrier concentration was 2.0 g/L. The culture medium was changed whenever the glucose concentration dropped below 2 g/L. The diffusion limitations and the severity of the concentration gradients within the particles should be reduced by frequent replacement of medium. 250 mL spinner cultures were initiated with a high inoculum concentration (i.e. 10 8 BHK cells each). As indicated in Figure 65, little difference was noted in the cumulative glucose used or the lactate produced in the three microcarrier cultures over the 600 h culture time. 60  80 70  50-  8A 8A _  5A OA 0A ^ A OA  -6) 40-o co  RA^_  S 300  10-  0  0A  o RR  60 O  50 r-  R - 40 a)  02^RR^- 30 Coa OA^R a CD 0-  R PRRIR .01 R^ ^ 20 (r4,^ 840 R OO o  R^ a  gl A n n nn  Cytodex  10  A Cultispher-G  0  0^100^200^300^400^500^600  Time (h) Figure 65. Cumulative glucose used and lactate produced by BHK cells on microcarriers (2 g/L) in 200 mL medium. Each 250 mL spinner culture was inoculated with 10 8 cells. The glucose uptake rates of all three cultures reached steady state values of approximately 0.1 WL/h after 200 h as shown in Figure 66. The estimated total cell  128 number in each of the 250 mL spinners was 1.04 x 10 9 cells (or 5.21 x 10 6 BHK cells/mL medium). Because S40 particles are more dense than Cytodex-1 and CG particles, the carrier volume based on unit mass is reduced. Since the process can be scaled up in both process intensity (i.e. carrier loading) and volume, it is more appropriate to calculate the cell loading based on the carrier volume instead of the total culture fluid volume. The BHK cell loading per unit carrier volume was 33% higher for the S40 carriers based on the greater GUR (see the slopes of Figure 67). It should be noted that the S40 microcarrier had a greater inner particle voidage to accommodate cell growth compared to Cultispher—G (90% vs. 50%). The estimated BHK cell density on the S40 was 2.0 x 10 8 cells/ mL carrier after a 600 h culture period, in contrast to 1.3 x 10 8 cells / mL carrier for both the Cytodex-1 and Cultispher—G.  0.12  o  0.11  0  0  66  OAA0600  0.10 ^ ......^ 0  ...7.1  A  0.09^ ^  o  o  as 0.08^ CC a^^ ...= 0.07 2 al^ A I sI^ 0  2^ ORCIOR `-' Og^ s 0  6,06 ° o 8 fl A 4^EtA8  H  ".=  D 0.06^^ A O ri) O  0  c.) 0.05^A =  C7  0.04 0.03 0.00  ..r  1 0  o S40 ^ Cytodex A Cultispher-G I^'^I^r^.^I^,^i  1 00^200^300^400^500^600  Figure 66. Glucose utilization rates of various BHK microcarrier spinner cultures. Operating conditions as outlined in Figure 65.  129 0.50 0.45 -  0 0  0.40 • 6- 0.35 c.)  P  .0  P^A  -J  E 0.30 -  A'0  P" o' cn^. ^PP^ A 0" A' o ' Fp 0.25 EI)3^s o U) ^ P' go  Op^ts, ,si  f,,D 0.20 0 )  3 0.15 -  ( -  -  To • 0.10 -  ^  6  —0— 340 —0-- Cytodex-I A Cultispher-G  ,ffn 11'1  0.05 0.00 s 0  o'  o' o'^6  1 Oo  °^  —  200  ^  300^400  ^  —  500  ^  600  ^  700  Time (h) Figure 67. Unit carrier cumulative glucose used by BHK cells on microcarriers. Operating conditions as outlined in Figure 65.  6.2.6 Cell penetration depth  Figure 68 shows the SEM of S40 microcarrier particles after being cultured for 600 h. The beads shown are near spherical and covered with BHK cells. The sectioned particle shows that BHK cells grew throughout the center of the particle. Mass transfer limitation was evident as indicated by the cell concentration gradient; with the highest cell density at the outer perimeter of the particle. The theoretical oxygen penetration depth could be used to estimate the maximum distance for cells to infiltrate into the particle. It should be noted that cells only need to penetrate 30% of the diameter to utilize 70% of the sphere volume. The following assumptions were made to characterize oxygen diffusion and consumption:  130 (1) The system can be considered to be at steady state; (2) The molecular diffusivity of oxygen is constant throughout the particle (i.e. homogeneous); (3) Diffusion in the particle follows Fick's law; (4) There is neglible external film diffusion resistance; An oxygen mass balance around a spherical support particle then gives : i  (I-- r 2dCs )=Q X = Q D eff r 2 dr^dr^02  (36)  Where D e ff = effective diffusivity; r = radial coordinate; Q = total oxygen consumtion rate; X = biomass; 00 2 = specific oxygen uptake rate. The boundary conditions are: (1) at r = ri (outer perimeter) C s =Cb * (2) at r = rd (radial position at which oxygen concentration is zero) Cs =0 dC s _ 0 dr  —  Equation (36) can be solved with these boundary conditions to give: r )] C^ C = Q [0.2 _ rd 2 ) _ 2 rd 2 in ( ___ s 6D eff^rd  (37)  The penetration depth of oxygen (PD = ri — rd) for the case of no external diffusion limitation can be estimated to be * 6 1 D eff p _ eff b ] 4 D^Q [  (38)  131 The theoretical cell penetration depth of 262 ,um into a particle agreed well with the actual value of 150-200 ,um (see Figure 68) with additional assumptions: (1) Dissolved oxygen at 50% air saturation (Cb* = 0.2 mmol 02/L at 37°C in water) (2) Void fraction of the particles reduced to 20-40% after cell growth infiltration (i.e. the effective difusivity was assumed to be about 50% of the diffusivity in the bulk medium or a reduction of effective diffusivity from 2.65x10 -5 cm 2/s to about 1.3 x 10 -5 cm 2/s). (3) Biomass of 2 x 10 8 cells/mL immobilized within the particle (i.e. Q = 40 mmoles/L/h). Combinations of other dissolved oxygen levels and various appropriate void fractions could also give rise to a PD of 262 ,um. The measured dissolved oxygen concentrations varied from 100% air saturation after the medium had just been replaced to 40% air saturation after the fresh medium had been in place for 24 h. The actual void fraction of the particle is also difficult to assess. The void fraction near the outer perimeter of the particle is low, while the interior of the particle is almost completely free of cells.  132  Figure 68. (A) Scanning electron micrograph of S4() particles covered with BEAK cells (B) A cut—opened S40 particle revealing the central portion of the microcarrier.  133 6.2.7 Hybridoma cell growth in Polyhipe particles  Collagen—based macroporous beads are often used to grow hybridoma cells (Almgren, 1991). Hybridoma cells only attach loosely to surfaces. Thus physical entrapment within the three dimensional structure of the beads is thought to be the main mechanism of immobilization (Almgren, 1991). The polystyrene macroporous beads were tested as carriers for the growth of 2E11 hybridoma cells. In the first series of experiments the particles were seeded using the same protocol as for the Vero and BHK roller cultures. Under these seeding conditions the cells did not grow selectively within the pores of the particle. Most of the cells in the culture failed to be trapped within the carrier. Himes and Hu (1985) measured increased rates of CHO cell attachment to Cytodex (Pharmacia) microcarriers when DMEM/serum was replaced by Ca++ and Mg++ free PBS. A second series of experiments was therefore performed in which the cells were incubated with the macroporous beads in PBS (without Ca++ and Mg++) for 1 h without agitation. The particles were then washed once with serum containing medium and finally resuspended in medium containing 10% fetal calf serum. This technique promoted attachment and growth of 2E11 cells within chloromethyl beads (but was not effective for the unmodified polystyrene or those treated with sulphuric acid). Proteins in the culture medium might bind preferentially to the particle surfaces and reduce the mammalian cell attachment. Thus, in the absence of serum, there may be an opportunity for direct binding of the hybridomas to the surfaces of the beads. Figures 69 and 70 show scanning electron micrographs of the surface of the CM80 beads containing hybridoma cells. These confirm that cells were entrapped in the pores, even near the surface of the particles. Figure 71 shows the changes of cell concentration freely suspended in the culture fluid and the glucose concentration as a function of time. It can be seen that the cells grow rapidly  134 outside the particles and failed to be trapped within the carrier after each medium change. The glucose uptake rate increased (see the slope of the Figure 72) despite suspension cell removal during each medium change, indicating an increase in the number of cells within the particles. Release of hybridomas from these surface regions are suspected to be responsible for the continued presence of suspension cells, particularly during the startup periods of these cultures. The loss of cells to the surrounding medium prolonged the period required for the hybridomas to establish a high steady state immobilized cell density (400 h compared with 150 h for BHK cells under similar inoculation and growth conditions). The maximum cell concentration in the macroporous bead culture was about 3 x 10 7 cells/mL of carrier, about 20% of the concentration achieved using the Vero cells. Under steady state conditions, the antibody concentration of the 2E11 cells grown on CM80 macroporous beads reached 40 mg/L of monoclonal antibody every 2 days (when the culture medium was changed). Antibody production rate increased from 0.63 to 0.89 mg/L/h while the glucose uptake rate remained relatively stable (see Figure 72). This indicated that the CM80 Polyhipe particles retained most of the starting hybridoma cell population, since the antibody production rate is related to cell growth rate. Antibody productivity increased as the cell growth rate decreased.  135  Figure 69. Hybridoma cells growing within the pores of chloromethyl Polyhipe particle  Figure 70. Enlarged view of hybridoma cells growing in a pore of a chloromethyl Polyhipe particle.  ▪ 136 100  480  1  0)  O  c 3o  80  •t;  E  C a) U C O  40 TD  2-  C 0  O  0  C  20 a_ 0  0  50^100^150^200  ^  300  Time (h)  Figure 71. Growth of hybridoma cells with chloromethyl Polyhipe (5 g/L) in a roller bottle. 2 x 10 7 cells suspended in PBS without serum were used as inoculum; PBS solution was later replaced by 100 mL DMEM with 10% fetal calf serum. 1.0 •  Antibody  o  Glucose  • 0.8  •  •  rn 0.6  a)  CD CC  O 0  •  0 0.4 =  2  Ca O  0_  /— to-  0 02  ■I1  Steady-state GUR 0 0  i ^200  400  , 600  0.0 800  Time (h)  Figure 72. Cumulative glucose utilization and antibody production rate of hybridoma cells entrapped in 5 g/L of CM80 macroporous beads.  • 137 7.0 RESULTS AND DISCUSSION —EFFECT OF CULTURE SYSTEMS ON CELL GROWTH AND CELL PRODUCTIVITY 7.1 Comparison of Cell Specific Transferrin Productivity of Different Culture Systems  BHK cells were also grown in standard suspension cultures to compare the rate of transferrin production with cells grown in the fixed bed ceramic perfusion system. Cultures of 250 mL in spinner flasks were seeded with 1 x 10 8 cells in 5% FCS, DMEM/F-12 medium. Figure 73 illustrates the growth kinetics of BHK cells cultivated in suspension. The cell concentration was about 5.5 x 10 6 cells/mL during the stationary phase in the spinner flask. The transferrin production rate of the suspended BHK cells during the exponential phase of growth in the spinner flask (5.5 mg HTF/1x10 9 cells/day maximum or an yield of 12 mg HTF/ g– glucose used) was comparable to that of cells immobilized in the ceramic matrices 25  lactate  3.0 -0..,^  2.^ 0 0-0-0-0 0 - \^^ 6^\^0 (V  ,A — A-•_•• 20— , ..._....._, ^-^0 _1 .__I / - -- A o^13^ co glucose \^0 •^transferrin 0-^ till^ • u3"b /^ / ,^ c . ' i o 2.0 -^ 7<. ^ 15= 0 A c^ c O^ 0 , )s a) Ca 1.5a) ♦^cell concentration co , .•• • 0^ - 100 o as^ o 0^ \^ .-J^ -  o  8  1 .0 '-'^  A'  C_) .1 ;\.. 0^ . a) a) / P PP^ co^(1) v) c o^o 5 Es 0-- ^A = I\ET /0 0_ 0 r.)--^. .__...t ii CD 4^ ^ • • 4 0.0-^ ^ I _ t•—/ ,  A. --^  I^'^I  ^  0^40^80^120^160^200  0  Process Time (h)  Figure 73. Transferrin production by BHK cells in spinner suspension culture with DMEM/F-12 medium supplemented with 10 ,uM zinc and 5% fetal calf serum.  138 (i.e. 6 mg HTF/10 9 cells/day maximum or an yield of 13 mg HTF/ g—glucose). The transferrin productivity of BHK cells on S40 microcarrier was also compared with cells on Cultispher—G and Cytodex-1 microcarriers. Figure 74 shows that the transferrin productivity per g of carrier was similar for the three microcarriers tested. However, because of their higher density, the transferrin productivity per unit carrier volume for the S40 beads was 33% greater than for Cytodex-1 and CG particles; this is shown in Figure 75. Transferring the cultures from DMEM to DMEM/F-12 did not induce increased transferrin production rate (see slopes of Figure 74). In fact, the transferrin production rate decreased from 3.46 mg/10 9 cells/day to 2.30 mg/10 9 cells/day (or 7.5 to 5 mg HTF/ g—glucose), suggesting that the low glucose concentration of DMEM/F-12 might have resulted in diffusion limitations. It is likely that the induction due to the small amount of zinc present in F-12 (about 3 500  400 -  (73  -0a)  O od  300 DMEM F-12  C.)  Pin 0 OW av-•  2  ,o ,e1 ,o n 0 ^ 4y— o, 00 DMEM w 5OL M zinc o":  200  ro;fr  100 - DMEM —0— S40 —0— Cytodex —A— Cultispher-G 0  0^100^200^300^400  ^  500  ^  600  Time (h)  Figure 74. Total transferrin produced by BHK on microcarriers (2 g/L) in 200 mL medium. Each 250 mL spinner culture was inoculated with 10 8 cells.  139 ,uM) could not overcome the mass transfer limitation. The addition of zinc (50 ,uM) to DMEM/10% NCS medium enhanced cellular transferrin productivity slightly (Figure 74). The transferrin productivity increased to 4.38 mg/10 9 cells/day (i.e. 9.5 mg HTF/ g—glucose). 4  1  0  /  3-  a)  U  e  2 .r. a). ca 2 -  .= 0 •1r)  E  ^^  ns cr)  crEr  E  I-  o  p' 46 .4  JA/  45  1-  —0— S40 —0— Cytodex-I A Cultispher-G  —  —  I^I^ I 0d 0^100^200^300^400 _ 500^600  Time (h) Figure 75. Transferrin production per unit carrier volume by BHK cells on various microcarriers. Operating conditions as outlined in Figure 74. The S40 microcarrier spinner cultures gave a maximum specific transferrin productivity of 4.4 mg/10 9 cells/day (i.e. yield of 9.6 mg HTF / g—glucose). This is lower than the 6.0 mg/10 9 cells/day (i.e. yield of 13.0 mg HTF / g—glucose) obtained with the ceramic perfusion system or with the suspension spinner culture. The difference in the observed cell productivity may be due to variations within the two batches of cells. The effect of the culture system on cell productivity cannot be fully determined unless all the inocula comes from a single batch. In Figure 76, BHK cells harvested from a single roller bottle were divided equally to  140 inoculate roller, spinner and S40 (0.5 g/L) microcarrier cultures. Each culture system contained 200 mL DMEM medium. DMEM/F-12 medium was used after the cultures reached steady state (at 200 h). Prior to the formation of large cell clumps in the spinner suspension culture, a large proportion of the cells in the suspension culture were lost when the medium was changed; this retarded the spinner suspension growth. The estimated final cell concentrations were 5.2 x 10 6 , 2.55 x 10 6 , 2.51 x 10 6 cells/mL for the roller, the microcarrier and the spinner cultures, respectively.  25-  ^  S40  O  Roller  O  A Suspension O  O  O  8 5-  ^  ^^ A  A  ^  O A  0 ^ • 0  p A A A  A  A  A  I^I^ 1^•^I 50^100^150^200^250  300  ^  350  Time (h) Figure 76. Cumulative glucose used by BHK cells.Each culture system contained 200 mL DMEM medium. DMEM/F-12 medium was used after the cultures reached steady state (at 200 h). Each culture was inoculated with 1 x 10 8 cells. The high cell loading in the roller system also resulted in a high production of transferrin (shown in Figure 77). However, the increase in the product could not be fully explained by the increase in cell number. The maximum cell specific  141 transferrin productivity rates were 10.3, 4.5 and 4.7 mg/10 9 cells/day (or maximum yields of 21.7 9.8, 10.2 mg HTF/ g—glucose) for the roller, the microcarrier and the suspension culture, respectively. The reason for the high cell specific transferrin productivity of the roller system is unknown. However, the BHK cells in the roller culture utilized the glucose more efficiently and produced less L—lactate (see Figure 78). The glucose—to—lactate conversion ratio decreased from 0.83 to 0.55 for the BHK roller culture, and more transferrin was produced per unit mass glucose used (Figure 79). It should be noted that the reduction of glucose—to—lactate conversion ratio occured only during multilayer cell growth. The conversion ratio remained constant around 0.8 when cell number in the roller was kept low. The lower glucose—to—lactate conversion ratio in the roller culture either suggests better oxygenation despite the potential mass transfer limitation to cells in the inner layer or reduction of available glucose to cells in the inner layer. The measured dissolved oxygen concentration of the roller culture was consistently higher than that of the spinner culture (i.e. 60% vs. 30% air saturation after the medium was replaced for 24 h). However, the _optimum dissolved oxygen concentration for cellular production was not determined. Spier and Griffiths (1984) found that significant increases in cell yield and multiplication rate for various mammalian cells including BHK cells, when the oxygen partial pressure was 9-10% compare to the normal 21%. A stirred system with controlled dissolved oxygen concentration is needed to eliminate the mass transfer limitation encountered in the current spinner flask systems and to determine the optimum oxygen concentration.  142  —0— S40 (0.5 g/L) —  A-  ^  Suspension Roller culture  Time (h) Figure 77. Cumulative transferrin produced by BHK cells from various culture systems. ^1.00^ V ^ V ^ V V V—V—V—V—V—V—V  __A ^  o  -  ^ ^^ ^ ^ ^ ^ 0  -  0 ^ ^  co  z  N 0.750 0) 0)  0 A 0  2  —A—A —A—A—A  0.50To cr, —0— S40 A Roller —p— Suspension —  0.25 0  —  •^I^•^I^• 50^150^200^250 100  300  ^  350  Time (h)  Figure 78. Glucose—to—lactate conversion ratios for various BHK cultures.  • 143 24 22 13  20  A S40 O Roller Suspension ^  0 ,' o O 9,  o 12  A  0  0^,// 0 ,-05 0  8 ■23 17,  C  6  A  A  4  E  2  0 ^ ^ ^ ^ 200 400 0 100 300 Time  (h)  Figure 79. Transferrin produced per g glucose utilized for various BHK culture systems.  7.1 Comparison of Cell Loading and Large Scale CelLCulture Suitablity The  steady state cell concentration in all three porous carrier systems tested  reached 1 x 10 8 cells/ mL matrix. The three—dimensional configuration of the porous matrix mimics tissue structure more closely than the conventional two— dimensional configuration and favours multilayer cell growth. The fixed bed ceramic perfusion system offers a significant increase in cell density compared to the commercial Opticell system (i.e. 1.25x10 8 vs. 2x10 7 cells/cm 3 ) for BHK cells despite its lower surface area to volume ratio when compare to Opticell (i.e. 12 versus 32 cm 2 surface area per cm 3 volume). The cell loading within the porous matrix of current fixed bed perfusion system is superior to those of other fixed bed reactor proposed by other researchers. i.e. 3.2 x 10 6 CHO cells/mL matrix for the fiber bed reactor used by Perry and Wang (1989), 1 x 10 8 hybridoma cells/mL  144 inner pores space for the ceramic perfusion system employed by Applegate and Stephanopoulos (1990), 1 x 10 7 hybridoma HB32 cell/mL void volume of the glass bead packed bed reactor used by Ramirez and Mutharasan (1989), and 1.4 x 10 7 Vero cells/mL matrix for the fixed bed glass "Porospher" tested by Looby and Griffiths (1988). The advantages of the current fixed bed system are: (1) the structure consists of multiple interconnected channels instead of the straight parallel channels found in the Opticell systems; this greatly reduces the possibility of channel blockage due to over—grown cells. (2) The non—toxic ceramic element can be easily cleaned, sterilized and reused. We have used the same ceramic element for more than ten different culture experiments. (3) the start—up procedure is relatively easy and the inoculating volume requirement is low. However, the fixed bed ceramic system can only be scaled up to a limited extend (i.e. maximum porous matrix volume of 10 L) without suffering from reduction in cell concentration. The airlift system with a porous draft tube also permits greater cell loading than those reported by other researchers. For instances,. the maximum cell loading in the external airlift packed bed reactor used by Murdin et al. (1989) reached only 2.5 x 10 6 hybridoma cells/mL polyester foam. The fiber—bed airlift bioreactor used by Chiou et al. (1991) achieved a maximum cell concentration of 6.8 x 10 7 CHO cells/ mL in a packed glass fiber bed while the maximum BHK cell concentration in the current airlift system was 9.7 x 10 7 cells/mL ceramic foam. The airlift system eliminates the needed for an external pump to circulate the medium through the porous matrix and simplifies reactor construction and scaleup. However, this system is somewhat difficult to initiate. The difficulties associated with initiating the airlift system (i.e. high inoculum volume requirement) and the inability of obtaining representative samples from the culture reactor may hinder the system from being used as a suitable large scale cell  145 culture system. The airlift system could be used for the cultivation of suspension cells since large scale airlift reactors are commonly used for hybridoma cell growth. The porous draft tube could entrap suspension cells and potentially increase the cell concentration in the reactor. Inoculation procedure for the cultivation of suspension cells is also much simpler when compared to adherent cells; the inoculum volume does not need to cover the entire draft tube. With the input gas flow, the convective flow from the bulk fluid through the porous draft tube exists regardless of the liquid height. As cell density in the bulk fluid increases, medium can then be added to the reactor to increase fluid volume. The maximum cell concentrations obtained from the porous microcarrier system were similar to those reported for various commercial porous microcarrier culture systems. For example, cell concentrations exceeding 10 8 cells/mL matrix, typically 2-3 x 10 8 cells/mL, have been reported for culture systems utilizing Verax, and Cultispher—G beads. Since these values were supplied by the manufacturers, they usually represented the best results obtained from optimized processes. In our comparative studies, cell loading per volume of carrier was 20% higher in S40 Polyhipe microcarrier than in Cultispher—G particles (see Section 6.2.5). Cell density attained in Polyhipe also exceeds that reported for other noncommercial porous particles, such as the polyurethane foam inveatigated by Matsushita et al. (1990) or the reticulated polyvinyl formal resin foam used by Yamaji and Fukuda (1991) and Yamaji et al. (1989). Although the fixed bed perfusion system can only be scaled up to a very limited extend, its scale is well suited for use as a cell propagator to supply the required inoculating cells for other type of reactors. Both the airlift system and the porous microcarriers system can be scaled up by both increasing the volume and the carrier loading of the system. The airlift system was somewhat difficult to initate when compared to the microcarrier system. However, once the cell  146 population was established within the airlift, the reactor is simple to operate and easy to maintain for extended periods. Cell recovery from the microcarrier system is difficult, complicating the process of cell propagation. The inaccessibility of cells within the matrix is not a disadvantage as long as the product of interest is secreted from the cells to the bulk medium. With its ease of scaling—up, simple inoculation procedure, and the ability to obtain representative samples from the reactor, the porous microcarrrier system is the best general choice for large—scale high cell density animal cell culture.  147  8.0 CONCLUSION & RECOMMENDATIONS The feasibility of using open—porous matrices as support materials for large scale animal cell cultivation has been demonstrated. The following major conclusions can be drawn from the experimental data and modelling results presented in the previous chapters:  Fixed bed perfusion system • The current fixed bed ceramic perfusion system offers a superior cell loading compared to the Opticell system (i.e. 1.25x10 8 vs. 2x10 7 cells/cm 3) for BHK cells. The advantages of the current ceramic foam are: (1) reduction of channel blockage due to over—grown cells, and (2) re—usable non—toxic ceramic element. • The ceramic foam perfusion system offers the benefits of an immobilized culture system and ease of product removal with the additional benefit of continuous culture monitoring by oxygen consumption rate monitoring. • The ceramic perfusion system can be used as a cell propagator. Vero cells could be grown and harvested from the fixed bed ceramic foam repeatedly providing a stable source of cells.  Airlift reactor • The airlift system supported long term cell culture at high cell density comparable to that produced in the externally circulated fixed bed ceramic perfusion system. BHK cells were grown successfully in the airlift system for more than 8 weeks. • The differential pressure drop across the porous draft tube was greatly influenced by the porosity of the tubular ceramic elements. The differential pressure drop increased with decreasing porosity and decreasing draft tube length.  148 • The presence of serum in the reactor fluid increased the downcomer gas holdup, thereby reducing the differential pressure drop across the draft tube. However, the mass transfer coefficient for oxygen was not affected. • The model proposed for the airlift reactor adequately described the overall hydrodynamic trends. It allows the prediction of the riser gas holdup, superficial liquid velocity and liquid perfusion rate through the porous draft tube. • Based on the mass transfer characteristics, the current airlift system could support a cell density of at least 3.6 x 10 8 cells / mL matrix or 5 x 10 7 cells/mL of total reactor volume.  Porous microcarrier system • Attachment rates of BHK and Vero cells were more rapid on sulphonated polystyrene microcarriers (S80, S40) than on Polyhipe with other surface modifications and were similar to the attachment rates for Cytodex and Cultispher—G microcarriers. • Vero and BHK cells grew as well on the unmodified polystyrene as on the chloromethyl, diethyl amino or sulphonated polystyrene surfaces. At steady—state in batch—fed perfusion cultures, Vero and BHK cell densities of greater than 1 x 10 8 cells/mL carrier on glucose uptake rates and ATP recovery were estimated. • Cell recovery from the porous Polyhipe was difficult. Only about 10 % of the estimated total cell population could be harvested following trypsinization. Thus the recovery of viable cells from Polyhipe for seeding large—scale bioreactors appears to be a problem. About 60 % of the cells could be accounted for on the basis of nuclei released from the Polyhipe particles by digestion with citric acid. • Hybridoma cells were not effectively entrapped by Polyhipe particles when the particles were seeded in the presence of serum. Chloromethyl Polyhipe did bind and retain a significant fraction of hybridoma cells when the initial contacting was  149 done without serum. Maximum cell concentrations were about 3 x 10 7 cells/ mL carrier, i.e. less than a third of the cell densities reached by adherent cells. • Maximum Vero and BHK cell number per unit volume of carrier was about 20% higher on the sulphonated polystyrene surfaces than for Cytodex-I and Cultispher-G microcarriers. Performance based on carrier weight was equivalent. • Scanning electron microscopy confirmed the attachment and entrapment of Vero and BHK cells within both types of porous matrices. Scanning electron micrographs indicated possible mass transfer limitations within the porous S40 microcarriers.  The following recommendations are made for future studies to expand the scope of this present study. • The effect of using serum-free culture medium on cell loading within the porous matrix and cellular productivity in the proposed bioreactors should be investigated. • The effect of serum concentration on cell recovery in the perfusion propagator should be examined. Higher cell recovery rate might be obtained by reducing the serum concentration in the medium prior to cell harvesting. • A large scale airlift reactor should also be used for the hydrodynamic studies. This would allow the axial pressure drop profile across the porous draft tube to be determined, which could then be used to validate the proposed model. The overall gas holdup of the airlift could also be determined by measuring the difference between the height of the liquid and the height of dispersion. The scale of the current airlift is too small and does not permit such measurements to be taken without errors of magnitude two to three times greater than the collected data. • The increase of the differential pressure drop across a porous draft with time due to cell attachment and growth should be monitored and the porosity change because of cell growth within the porous matrix should be investigated.  150 • The proposed airlift system was difficult to initiate for the cultivation of anchorage—dependent cells. However, large—scale airlift reactors (i.e. over 1000 L) are commonly used for suspension cell culture. The proposed airlift system should be tested for the cultivation of suspension cells, such as hybridoma cells. By adjusting the porosity of the porous draft tube, entrapment of suspension cells may take place and possibly increase the cell concentration within the reactor. • Dissolved oxygen was not controlled in the spinner microcarrier cultures and diffusion limitations within the microcarrier were evident. The oxygen limitation is most likely caused by pore blockage near the outer surface of the microcarrier due to over—grown cells and the low mass transfer coefficient of the spinner flasks used. A better oxygenated stirred tank system should be used to increase the dissolved oxygen concentration in the bulk medium and hence, reduce the mass transfer limitations within the porous beads. • Increased bead to bead collision and high shear stress resulting from increased particle loading and high agitation can cause reduction in unit reactor cell loading, as in the case of Cytodex-1 cultures. The performance of the porous microcarriers at high concentration (i.e. greater than 5 g /L) and in stirred bioreactors with high agitation should be evaluated and compared with the performance of non—porous microcarriers. • The nature of the observed enhanced cellular transferrin productivity in roller cultures may be due to the dissolved oxygen concentration being optimum. The effect of dissolved oxygen concentration on cell productivity should to be further investigated.  151  NOMENCLATURE al) = gas—liquid interfacial area per unit volume of dispersion (m -1 ) aL = gas—liquid interfacial area per unit volume of liquid (m -1 ) AB = free area for liquid flow between riser and downcomer at the bottom of the CT airlift reactor (m 2) AD = cross—sectional area of downcomer (m 2) AR = cross—sectional area of riser (m 2) Cf = Fanning friction factor Cs = dissolved oxygen concentration Cb * = dissolved oxygen concentration in the bulk medium Cbi = dissolved oxygen concentration at the outer perimeter of the shpere dB = average bubble diameter (m) DC = Column diameter (m) Da =Column inner diameter (m) Deff = effective diffusivity (m 2s -1 ) d = diameter of matrix fibers plus cell layer (m) F dR = riser diameter (m) D = some effective average pore diameter (m) Eg = energy loss due to fluid turn around at the bottom of the reactor (W) ED = energy loss in downcomer due to upflow of bubbles (W) EF = energy loss due to friction in the riser and the downcomer (W) Ei = the total power input (W) Ep = energy loss due to flow through the porous draft tube (W) ER = energy dissipation due to wakes behind bubbles in the riser (W) ET = energy loss due to fluid turn around at the top of the reactor (W) h = vertical distance above the sparger (m)  152 hp = dispersion height (m) hL = unaerated liquid height (m) hp = length of the porous section of the draft tube (m) k = Darcy resistance coefficient (m -2) keff = effective diffusion coefficient (m 2/s) kL = mass transfer coefficient (m/s) KB = dimensionless frictional loss coefficient as per Eq. (12) L = porous element thickness or porous draft tube thickness (m) LD = the clearance between the base and the draft—tube (m) PD = the penetration depth of oxygen (m) Ph = pressure at the top (Pa) Q = flow rate of the gas (m 3 / s) Q02 = specific oxygen uptake rate (mole/s)  r = radius of particle (m) ri = radial position at outer perimeter of the sphere rd = radial position at which oxygen concentration is zero R = gas constant Re p = Reynolds number T = temperature (°C) Ub = bubble rising velocity (m/s) Ujj = superficial liquid circulation velocity in the downcomer (m/s) ULr = superficial liquid circulation velocity in the riser (m/s) U0 = velocity difference between the gas velocity inside the orifice of the sparger and the gas velocity just above the sparger (m/s). Up = liquid superficial velocity through the porous element (m/s) VD = dispersion volume (m 3 ) X = biomass (g)  153 Eg = gas hold—up  gr = gas hold—up in the riser E d = gas hold—up in the downcomer g E  EM = void fraction of the porous matrix  p1 = liquid density (kg/m 3)  9 = sparger efficiency (typically around 0.06) 1.t = liquid viscosity (Pa s) T = shear stress (N/m 2)  AP = differential pressure drop (Pa)  154  REFERENCES Adema, E., Shnek, D., Cahn, F., and Sinsky, A.J., " Use of porous microcarriers in agitated cultures", Biopharmacology, 3, 20-23 (1990). Allen, T.D., "The application of SEM to cells in culture: selected methodologies", Scanning Electron Microscopy, 4, 1963-1972 (1983).  Almgren, J., C. Nilsson, A.—C. Petersson and Nilsson, K., "Cultispher — macroporous gelatin microcarrier — new applications", In Production of Biologicals from Animal Cells in Culture. R.E. Spier, J.B. Griffiths & B. Meignier, Eds. 434-438. Butterworths—Heinemann, Oxford, UK, (1991). Applegate, A., and Stephanopoulos, G.N., "Development of a single—pass ceramic matrix bioreactor for large—scale mammalian cell culture", AIChE Fall meeting, Chicago, (1990). Arathoon, W.R., and Birch, J.R., "Large—scale cell culture in biotechnology", Science, 232, 1390-1395 (1986).  Bello, R.A., "A characterization study of airlift contactors for applications to fermentations", PhD Thesis, University of Waterloo, Ontario, (1981). Bello, R.A., Robinson, C.W. and Moo—Young, M., "Gas holdup and overall volumetric oxygen coefficient in airlift contactors", Biotechnol. Bioeng., 28, 369-381 (1985). Bello, R.A., Robinson, C.W., and Moo—Young, M., "Liquid circulation and mixing characterisitcs of airlift contactors", Can. J. Chem. Eng., 62, 573-577 (1984). Bentley, P.K., Gates, R.M.C., Lowe, K.C., de Pomerai, D.I., and Walker J.A.L., "In vitro cellular response to a non—ionic surfactant, pluronic F-68", Biotech. Lett., 11(2), 111-114 (1989).  155 Bleim, B., and Katinger, H.W.D., "Scale-up engineering in animal cell technology: Part II", Trends in Biotechnology, 6 (9), 224-230 (1988). Blenke, H., "Loop reactors", Adv. Biochem. Eng., 13, 121-214 (1979). Bognar, E.A., Pugh, G.G., and Lydersen, B.K., "Large scale propagation of BHK 21 cells using the Opticell culture system", J. Tissue Culture Methods, 8, (4), 147154 (1983). Boraston, R., Thomson, P.W., Garland, S., Birch, J.B., "Growth and oxygen requirements of antibody producing mouse hybridoma cells in suspension culture", Develop. Biol. Stand. 55, 103-111 (1984). Bugarski, B., King, G.A., Jovanovic, G., Daugulis, A.J., and Goosen, M.F.A., "Performance of an external loop airlift bioreactor for the production of monoclonal antibodies by immobilized hybridoma cells", AppL Microbiol. Biotechnol., 30, 264-269 (1989).  Cahn, F., "Biomaterials aspects of porous microcarriers for animal cell culture", Trends in Biotechnology, 8, 131-136 (1990).  Calderbank, P.H., and Moo-Young, M, "Mass -transfer in fermentation equipment", Trans. Inst. Chem. Eng., 39, 337 (1961) Calvo, E.C., "A fluid dynamic model for airlift loop reactors", Chem. Eng. Sci., 44, (2), 321-323 (1989) Chakravarty, M., Begum, S., Singh, H.D., Baruah, J.N., and Iyengar, M.S., "Gas Hold-up distribution in a gas-lift column", Biotechnol. Bioeng. Symp., 4, 363378 (1973) Chiou, T.W., Murakami, S., Wang, D.I.C., and Wu, W.T., "A fiber-bed bioreactor for anchorange-dependent animal cell cultures: part I. bioreactor design and operations", Biotechnol. Bioeng., 37, 755-761 (1991). Chisti, M.Y., and Moo-Young, M., "Airlift reactors: characteristics, applications and design considerations", Chem. Eng. Commun., 60, 195-242 (1987).  156 Chisti, M.Y., Halard, B., and Moo—Young, M., "Liquid circulation in airlift reactor", Chem .Eng. Sci., 43 (3), 451-457 (1988). Chisti, M.Y., and Moo—Young, M., "Perdiction of liquid circulation velocity in airlift reactors with biological media", J. Chem. Technol. Biotechnol, 42, 211219 (1988). Chisti, M.Y., Airlift Bioreactors, p.147-148, Elsevier Science Publisher, New York, NY, (1989). Cho, M.H., and Wang, S.S., "Enhancement of oxygen transfer in hybridoma cell culture by using a perflurocarbon as an oxygen carrier", Biotechnol. Lett., 10 (12), 855-860 (1988). Croughan, M.S., Hamel, J.F.P., and Wang, D.I.C., "Effects of microcarrier concentration in animal cell culture", Biotechnol. Bioeng., 32, 975-982 (1988). Curtis, A.S.G., Forrester, J.V., McInnes, C., and Lawrie,F., "Adhersion of cells to polystyrene surfaces", J.Cell Biol., 97, 1500-1506 (1983). de Bruyne, N.A., "The design of bench scale reactors", in Spier ,R.E., and Griffiths, J.B., (eds), Animal Cell Biotechnology, -Vol. 3, p.159, Academic Press, Toronto, (1988). de St. Groth, S.F., "Automated production of monoclonal antibodies in a cytostat", J. Immunol. Methods, 57, 121 (1983).  Dean, R.C., Jr., Karkare, S.B., Phillips, P.G., Ray, N.G., and Runstadler, P.W., Jr., Continuous cell culture with fluidized sponge beads in Large scale animal cell culture technology, Lydersen, B.K., ed., 145-167, Hanser Publishers, N.Y. ,  (1987). Deckwer,W.D., Burckhart, R., and Zoll,G., "Mixing and mass transfer in tall bubble columns", Chem. Eng. Sci., 29, 2177-2188 (1974). Dullien, F.A.L., "Single flow through porous media and pore structure", Chem. Eng., 10, 1-34 (1975).  157 Dussap, G., and Gros, J.B., "Energy consumption and interfacial mass transfer in an airlift fermentor", Chem. Eng. J., 25,151-162 (1982). Emery, A.N., Lavery, M., Williams, B., Handa, A., "Large—scale hybridoma culture" in Webb C. and Mavituna F. (eds): Plant and animal cell culture:  process possibilities, Chichester: Ellis Horwood, 137-146 (1987). Ertel, S.I., Chilknoti, A., Horbett, T.A., and Ratner, B.D., "Endothelial cell growth on oxygen—containing films deposited by radio—frequency plasmas: the role of surface carbonyl groups", J. Biomater. Sci. Polymer Edn., 3, (2), 163-183 (1991). Familletti, P.C., and Fredericks, I.E., "Techniques for mammalian cell immobilization", Bio/Technol., 6 (1), 41-44 (1988). Feder, J., "The determination of large—scale animal cell perfusiono culture systems", in Upstream processes: equipment and techniques, 125-152, Alan Liss, Inc., (1988). Foran, D.J., Calm, F., and Eikenberry, E.F., "Assessment of cell proliferation on porous microcarriers by means of image analysis", Analytical and Quantitative  Cytology and Histology, 13, (3), 215 222 (1991). -  Forestell, S.P., Kalogerakis, K., and Behie, L.A., "Development of the optimal inoculation conditions for microcarrier cultures", Biotechnol. Bioeng., 39,305— 313 (1992). Frangos, J.A., Eskin, S.G., J.A., Mclntire, L.V., and Ives, C.L., "Flow effects of prostacylin production by cultured human endothelial cells", Science, 227,1477 (1985). Frangos, J.A., Mclntire, L.V., and Eskin, S.G., "Shear stress induced stimulation of mammalian cell metabolism", Biotechnol. Bioeng., 32,1053-1060 (1988). Freeman, W., and Davidson, J.F., "Holdup and liquid circulation in bubble columns", Trans. Inst. Chem. Eng., 47, T251—T262 (1969).  158 Funk, W. D., MacGillivray, R.T.A., Mason, A.B., Brown, S.A., and Woodworth, R.C., Expression of the amino-terminal half molecule of the human serum transferrin in cultured cells and characterization of the recombinant protein. Biochemistry, 29, 1654-1660 (1990).  Gardner, A.R., Gainer, U.L., and Kirwan, D.J., "Effect of stirring and sparging on cultured hybridoma cells", Biotechnol. Bioeng., 35, 940-947 (1990). Griffiths, J.B., "Advances in animal cell immobilization technology" in Animal Cell Biotechnology, Vol. 4, Spier, R.E., and Griffiths, J.B., eds., 150-166, Academic Press, (1990). Guy, C., Carreau, P.J., and Paris, J., "Comments on the paper: "liquid circulation and mixing characteristics of arilift contactors", Can. J. Chem. Eng., 64, 521523 (1986). Handa, A., Emery, A.N., and Spier, R.E., "On the evaluation of gas-liquid interfacial effects on hybridoma viability in bubble column bioreactors", Develop. Biol. Stand. 66, 241-254 (1987).  Handa-Corrigan, A., "Large scale in vitro hybridoma culture: current status", Bioltechnology, 6, (7), 784-786 (1988).  Handa-Corrigan, A., Emery, A.N., and Spier, R.E., "Effect of gas-liquid interfaces on the growth of suspened mammalian cells: mechanisms of cell damage by bubbles", Enzyme Microb. TechnoL , 11 (4), 230-235 (1989). Hatch, R.T., Experimental and theoretical studies of oxygen transfer in an air-lift fermenter, PhD Thesis, Massachusetts Institute of Technology, Cambridge, Massachusetts (1973). Hills, J.H., "The operation of a bubble column at high throughputs I - Gas holdup measurements", Chem. Eng. J., 12, 89-99 (1976).  159 Himes, V.B. and W.-S. Hu., "Attachment and growth of mammalian cells on microcarriers with different ion-exchange capacities", Biotechnol. Bioeng., 29, 1155-1163 (1987). Hirtenstein, M.D., Clack J, and Richards, R., in Rajan, J. (eds), Tissue culture in medical research, p 97, Pargamon, Oxford, (1980). Hu, W.S., Merier, J., and Wang, D.I.C., "A mechanistic analysis of the inoculum requirement for the cultivation of mammalian cells on microcarriers", Biotechnol Bioeng., 27,585-595 (1985). HOIscher, M., and Onken, U., "Influence of bovine serum albumin on the growth of hybridoma cells in airlift loop reactors using serum free medium", Biotechnol. Lett, 10 (10), 689-694 (1988). Hiilscher, M., Scheibler,U., and Onken, U., "Selective recycle of viable animal cells by coupling of airlift reactor and cell settler", Biotechnol. Bioeng., 39, 442-446 (1992). Jervis, E., Lee, D.W., and Kilburn, D.G., "Monitoring cell culture productivity by a rapid fluorescence immunoassay technique", In R.E. Spier, J.B. Griffiths & B. Meignier, Eds., Production of Biologicals from Animal Cells in Culture, Proceedings of the 10th ESACT Meeting, 597-599. Butterworths-Heinemann, Oxford UK., (1991). Jones, A.G., "Liquid circulation in a draft-tube bubble column", Chem. Eng. Sci., 40, (3), 449-462 (1985). Katinger, H.W.D., Scherier, W., and Kromer, E., "Bubble coumn reactor for mass propagation of animal cells in suspension culture", German Chem. Eng, 2,3138 (1979). Katinger, H.W.D., and Scheirer, W., "Status and development of animal cell technology using suspension culture techniques", Acta Biotechnol, 2:3-41 (1982).  160 Kawase, Y., and Moo-Young, M., "Mixing and mass transfer in concentric-tube airlift fermentoers: Newtonian nad non-Newtonian Media", J. Chem. Technol. Biotechnol., 36, 527-538 (1986).  Keller, J., Dunn, U., and Heinzle, E., "Improved performance of the fluidized bed reactor for the cultivation of animal cells", Proceedings of the 11th ESACT Meeting (1991). Kilburn, D.G., and Webb, F.C.,"The cultivation of animal cells at controlled dissolved oxygen partial pressure", Biotechnol. Bioeng., 10, 801-814 (1968). Klemperer, H.G. and Knox, P., "Attachment and growth of BHK and liver cells on polystyrene: effect of surface groups introduced by treatment with chromic acid", Lab. Pract., 26, 179-180 (1977). Koide, K., Horibe, K., Kitaguchi, H., and Suzuki, N., "Contributions of annulus and draft tube to gas-liquid mass transfer in bubble columns with draft tube", J. Chem. Eng. Japan, 17, 5, 547-549 (1984).  Koide, K., Kurematsu, K., Iwamoto, S., Iwata, Y., and Horibe, K., "Gas holdup and volumetric liquid-phase mass transfer coefficient in bubble column with draft tube and with gas dispersion into tube", J. Chem. Eng. Japan, 16, (5), 413-419 (1983a). Koide, K., Sato, H., and Iwamoto, S., "Gas holdup and volumetric liquid-phase mass transfer coefficient in bubble column with draft tube and with gas dispersion into annulus", J. Chem. Eng. Japan, 16, (5), 407-413 (1983b). Kratje, R.B., Jager, V., and Wagner, R., "Comparison of the production efficiency of mammalian cells grown in a fludized bed and in a stirred tank bioreactor", Proceedings of the 11th ESACT Meeting, (1992). Kratje, R.B., and Wagner, R., "Evaluation of production of recombinant human interleukin-2 in fluidized bed bioreactor", Biotechnol. Bioeng., 39, 233-242 (1992).  161 Kriegel, E., Pilepp, E., Scheffler, U., "Entwicklung eines bioreaktors nach dem umlaufprinzip", Tech. Mitt. Krupp. Forsch—Ber., 36, 41-43 (1978) Kunas, K.T., and Papoutsakis, E.T., "Damage mechanisms of suspended animal cells in agitated bioreactos with and without bubble entrainment", BiotechnoL Bioeng., 36, 476-483 (1990).  Kwong, C.W., Hsieh, J.H., Syu, M.J., and Chou, S.S., "Anti—hepatitis B surface antigen monoclonal antibody IgM production in suspension and immobilized bioreaktors", Biotechnol. Lett., 11, (6), 377-382 (1989). Lambert, K.L., Boraston, R., Thompson, P.W., and Birch, J.R., "Production of monoclonal antibodies using large scale cell culture", Dev. Ind. Microbiol., 27, 101-106 (1987). Lazar, A., Reuveny, S., Muzrahi, A., Avtalion, M., Whiteside, J.P., and Spier, R.E., "Production of biologicals by animal cells immobilized in polyurethane foam matrix", European Society for Animal Cell Technology (ESACT) General Meeting, 8th, Tiberias, Israel, (1987). Lee, C.H., Glasgow, L.A., Erickson, L.E., and Patel, S.A., "Liquid circulation in airlift fermentors", in Ho, C.S, and Oldshue, J.Y., (eds.), Biotechnology Processes Scale—up and Mixing, 50-59, AICHE, New York, (1986).  Lee, J.H., Park, J.W., and Lee, H.B., "Cell adhersion and growth on polymer surfaces with hydroxyl groups prepared by water vapour plasma treatment", Biomaterials, 12, 443-448 (1991).  Lefrancois, L., Mariller, C.G., and Mejane, J.V., Brevet d'Invention, France, 1, 102, 200, (1955). Lehmann, J., Vorlop, J., and Buntemeyer, H., "Bubble—free ractors and their developments for continuous culture with cell recycle", in Spier, R.E., and Griffiths, J.B., (eds.), Animal Cell Biotechnology, Vol. 3, 222-237, Academic Press, Toronto, (1988).  162 Levine., D.W., Wnag., D.I.C., and Thilly, W.G., "Optimization of growth surface parameters in microcarrier cell culture", Biotechnol. Bioeng., 21, 821-845 (1979). Lim, F. & A.M. Sun., " Microencapsulation islets as bioartificial endocrine pancreas", Science, 210, 908-910 (1980). Looby, D., and Griffiths, B., "Fixed bed porous glass sphere (poro spher) bioreactors for animal cells", Cytotechnology, 1, 339-346 (1988). Looby, D., and Griffiths, B., "Immobilization of animal cells in porous carrier culture", TIBTECH, 8, 204-209 (1990). Ludwig , A., Kretzmer, G., and Schugerl, K., "Determination of a critical shear stress level applied to adherent mammalilan cells", Enzyme Microbiol. Technol., 14, (3), 209-213 (1992).  Lydersen, B.K., Pugh,C.G., Paris, M.S., Sharma, B.P. and Noll, L.A., "Ceramic matrix for large scale animal cell culture", Bio/technology, Jan, 63-67 (1985). Lydersen, B.K., in Lydersen, B.K., ed., Perfusion cell culture system based on ceramic matrices in large scale animal cell culture technology, 170-192, Hanser Publishers, N.Y., (1987). Lydon, M.J., and Foulger, C.A., "Cell—substratum interactions: serum spreading factor", Biomaterials, 9, 525-527 (1988). Margaritis, A., and Sheppard, J.D., "Mixing Time and oxygen transfer characteristics of a double draft tube airlift fermentor", Biotechnol. Bioeng., 23, 2117-2135 (1981). Maroudas, N.G., "Sulphonated polystyrene as an optimal substratum for the adhesion and spreading of mesenchynal cells in monovalent and divalent saline solutions", J. Cell. Physiol., 90, 511-520 (1977). Matsushita., T., Ketayama, M., Kamihata, K., and Funatsu, K., "Anchorage— dependent mammalian cell culture using polyurethane foam as a new  163 substrratum for cell attachment", Appl. Microbiol. Biotechnol., 33, 287-290 (1990). McQueen, A., and Bailey, J.E., "Influence of serum level, cell lines, flow type and viscosity of flow induced lysis of suspension mammalian cells", Biotechnol. Lett., 11, (8), 531-536 (1989).  Merchuk, J.C. and Siegel, M.H., "airlift reactos in chemical and biological technology", J. Chem. Technol. Biotechnol., 41, 105-120 (1988). Merchuk, J.C., "gas Hold—up and liquid velocity in a two—dimensional air lift reactor", Chem. Eng. Sci., 41, 11-16 (1986a) Merchuk, J.C., "Hydrodynamics and hold—up in airlift reactors", Encyclopedia of Fluid Mechanics, Vol 3, Chermisinoff, N.P., ed., Chapter 49, 1485-1511 (1986b). Mignot, G., Faure, T., Ganne, V., Arbeille, B., Pavirani, A., and Romet—Lemonne, J.L., "Production of recombinant Von Willebrand factor by CHO cells cultured in macroporous microcarriers", Cytotechnology, 4, 163-171 (1990). Miltenburger, H.G., and David, P., "Mass production of insect cells in suspension", Develop. Biol. Stand., 46, 183-186 (1980). Murakami, S., Chiou, T.W., and Wang , D.I.C., "A fiber—bed bioreator for anchorange—dependent animal cell cultures: part II. scaleup potential", Biotechnol. Bioeng., 37, 762-769 (1991).  Murdin, A.D., Thorpe, J.S., Groves, D.J., and Spier, R.E., "Growth and metabolism of hybridomas immobilized in packed beds: comparison with static and suspension cultures", Enzyme Microbiol. Technol., 11 (6), 341-346 (1989). Nikolai, T.J., and Hu, W.S., "Cultivation of mammalian cells on macroporous microcarriers", Enzyme Microbiol. Technol., 14, (3), 203-208 (1992).  164 Nikolai, T.J., Madhusudan, V.P., Geotghebeur, S., and Hu, W.S., "Improved microscopic observation of mammalian cells on microcarriers by fluorescent staining", Cytotechnology, 5, 141-146 (1991). Nilsson, K., F. Buzsaky & K. Mosbach, "Growth of anchorage-dependent cells on macroporous microcarriers", Bio/Technol. 4, 989-990 (1986). Nollert, M.U., Diamond, S.L., and McIntire, L.V., "Hydrodynamic shear stress and mass transport modulation of endothelial cell metabolism", Biotechnol. Bioeng., 38, 588-602, (1991). Onken, U. and Weiland, P., "Airlift fermeters: construction, behaviour, and use", Adv. Biotechnol. Processes 1, 67-95 (1983). Perry, S.D., and Wang, D.I.C., Fiber bed reactor design for animal cell culture, Biotechnol. Bioeng. 34, 1-9 (1989). Piret, J.M & C.L. Cooney, "Immobilized mammalian cell cultivation in hollow fiber bioreactors", Adv. Biotechnol., 8, 763-783 (1990). Radlett, P.J., Telling, R.C., Whiteside, J.P., and Maskell, M.A., "The supply of oxygen to submerged cultures of BHK 21 cells", Biotechnol. Bioeng., 14, 437445 (1972). Ramsey, W.S., Hertl, W., Nowlan, E.D., and Binkowski, N.J., "Surface treatments and cell attachment", In Vitro, 20, (10), 802-808 (1984). Reiter, M., Hohenwarter, 0., Gaida, T., Zach, N., Schmatz, C., Bluml, G., Weigan, F., Nilsson, K., and Katinger, H., " The use of macroporous geltain carriers for the cultivation of mammalian cells in fluidized bed reactors", Cytotechnology, 3:271-277 (1990). Reuveny, S., and Thoma, R.W., "Apparatus and methodology for microcarrier cell culture", Adv. Appl. Microbiol. 31:139-179 (1986).  165 Robinson, C.W., "On power input and liquid phase mixing times in pneumatically—agitated contactors:response to critique of Guy et al.", Can. J. Chem. Eng., 64, 523-528 (1986).  Roitt., I., Brostoff, J., and Male, D., Immunology, 25.5-25.6, Gower Medical Publishing, N.Y., (1985). Schiigerl K., Lucke, J., and Oels, U., "Bubble column bioreactors", Adv. Biochem. Eng., 7, 1-84 (1977).  Shah, Y.T., Kelkar, B.G., and Godbole, S.P., " Design parameters estimations for bubble column reactors", AIChE J., 28, (3), 353-379 (1982). Siegel, M.H., Hallaile, M., and Merchuk, J.C., "Airlift reactos: design, operation, and application", Adv. Biotechnol. Processes 7, 79-124 (1988) Siegel, M.H., Merchuk, J.C. and Schiigerl, K., "airlift reacto analysis: interrelationships between riser, downcomer, and gas—liquid separator behaviour, including gas recirculation effects", AIChE J., 32, 10, 1585-1596 (1986). Sinskey, A.J., Fleischaker, R.J., Tyo, M.A., Giard, - D.J., and Wang, D.I.C., "Production of cell derived products, virus amd interferon", Ann. New York Acad. Sci., 369, 47-59 (1981).  Smiley, AL., Hu, W—S, Wang, D.I.C., Production of human immune interferon by recombinant mammalian cells cultivated on microcarriers, Biotechnol. Bioeng. 33, 1182-1190 (1989). Sobotka, M., Prokop, A., Dunn, I.J., and Einsele, A., "Reviews of methods for the measurement of oxygen transfer in microbial systems", Tsao, G.T., ed. in Annual Reports on Fermentation Processes, Acadamic Press, London, (1982).  Sonderhoff, S.A., Kilburn, D.G., and Piret, J.M., "Analysis of mammalian cell biomass based on cellular ATP", Biotechnol. Bioeng., 39, 859-864 (1992).  166 Spier, R.E., and Griffths, B., "An examination of the data and concepts germane to the oxygenation of cultured animal cells", Develop. Biol. Stand., 55, 81-92 (1984). Spier, R.E., and Whiteside, J.P., "The production of foot-and-mouth disease virus from BHK21 C13 cells grown on the surface of glass spheres", Biotechnol. Bioeng., 18, 649-657 (1976). Stathopoulos, N.A., and Hellums, J.D., "Shear stress effects on human embryonic kidney cells in vitro", Biotechnol. Bioeng., 27, 1021-1026, (1985). Stejskal, J. and Potucek, F., "Oxygen transfer in liquids", Biotechnol. and Bioeng., 27, 503-508 (1985). Telling, R.C., and Redlett, P.J., "Large scale cultivation of mammalian cells", Adv. Appl. Microbiol. 13, 91-95 (1971). Tharakan, J.P., and Chau, P.C., "A radial flow hollow fiber reactor for the large scale cuilture of mammalian cells", Biotechnol. Bioeng., 28, 1064-1071 (1986). Thomas, K.D., Tighe, B.J., and Lydon, M.J., "Cellular interactions with synthetic polymer surfaces II. surface modification of polystyrene for cell culture", in Advances in Biomaterials, Christel, P., Meunier, A., and Lee, A.J.C., eds, Elsevier Science Publishers, Amsterdam, 379-383 (1986). Thomas, N.H., and Janes, D.A., "Fluid dynamic considerations in airlift bioreactors", in Ho., C., and Oldshue, J.Y., eds, Biotechnology Processes Scale-up and Mixing, AICHE, N.Y., N.Y., (1987). Tolbert, W.R., Feder, J., and Kimes, R.C., "Large-scale rotating filter perfusion system for high density growth of mammalian suspension cultures", In Vitro, 17, 885-890 (1981). Tramper, J.D., and Vlak, J.M., "Bioreactor design for growth of shear-sensitive insect cells", in Webb C. and Mavituna F., eds: Plant and animal cell culture: processes possibilities, Chichester: Ellis Horwood, 125-136 (1987).  167 Tramper, J.D., Joustra, D., and Vlak, J.M., "Bioreactor design for growth of shear-sensitive Mammalian and insect cells", Adv. Biotechnol. Processes, 7, 199-228 (1988a). Tramper, J.D., Smit, D., Straatman, J., and Viak, J.M., "Bubble column design for growth of fragile insect cells", Bioprocess Eng. , 3:37-41 (1988b). Van Wazel, A.L., "Cultivation of anchorage-dependent cells and their applications", J. Chem. Technol. Biotechnol., 32, 318-323 (1982). Van Wazel, A.L., and Van der Velden de Groot, C.A.M., "Large scale cultivation of animal cells in microcarrier culture", Process Biochem., 13, 6-8 (1978). Vournakis, J.N., and Runstadler, P.W., "Microenvironment: the key to improve cell products", Bio !technology , 7, 143-145 (1989). Weiland, P., "Wissenchaftliche Forschungsarbeit", Chem. Ing. Tech., 56, 64-65 (1984). Whiteside, J.P., and Spier, R.E., "The scale-up from 0.1 to 1000 liter of a unit process system based on 3 mm diameter glass spheres for the production of four strains of FMDV from BHK monolayer cells!, Biotechnol. Bioeng., 23, 551-565 (1981). Wood, L.A. and Thompson, P.W., "Application of the air lift fermenter", Internaction Conference on Bioreactor Fluid Dynaimcs, (BHRA Fluid Engineering Center, Cranfield, Bedford, England), 157-172 (1986). Yamaji, H., Fukuda, H., Nojima, Y., and Webb, C., "Immobilization of anchorage-independent animal cells using reticulated polyvinyl formal resin biomass support particles", Appl. Microbiol. Biotechnol., 30, 609-613 (1989). Yamaji, H., and Fukuda, H., "Long-term cultivation of anchorage-independent animal cells immobilized within reticulated biomass support particles in a circulating bed fermentor", Appl. Microbiol. Biotechnol, 34, 730-734 (1990).  168 Ziltner, H.J., I. Clark—Lewis, B. Fazekas De St. Groth, P.C. Orban, L.E. Hood, S.B.H. Kent & J.W. Schrader., J. Immunol. 140,1182-1187 (1988).  169  APPENDICES Appendix 1. Model PC-61 A/D board data logging program  program mA_Card_data; uses crt, dos; const mA_Card = $330; number_of channels = 8; Gain = 0.68; Resistance = 270 {ohms}; maxfreq = 30; maconvers = 185.60508; maoffset = 3; type out_file = text; var i, j, k : integer; sint^: longint; portnum : integer; milliamps, sfreq^: real; ch^: char; outfile : out_file; filename : string; procedure initialise; begin clrscr; gotoxy(20, 1); write('mA Card (PC-61) test program (V1.3)'); port[mA_Card + 3] := $83; port[mA_Card + 2] := $FF; port[mA_Card + 0] := 0 end {initialise}; procedure cursor_off; begin port[$3B4] := 10; port[$3D4] := 10; port[$3B5] := 13; port[$3D5] := 13  170 end {cursor_off}; procedure cursor_on; begin port[$3B4] := 10; port[S3D4] := 10; port[$3B5] := 11; port[$3D5] := 6 end {cursor_on}; procedure write_headings; begin gotoxy(1, 4); write('Setup procedure: '); gotoxy(6,5); write('Switches 1, 2, 5 and 6 must be on, the others off.'); gotoxy(6, 6); write('With known input, calibrate board with coarse and fine trimpots.'); gotoxy(6, 7); write('Note voltage, between pins 39-36 of U6 (7109), as calibration figure.'); gotoxy(20, 10); writeln('Channel^Current(mA)'); gotoxy(1, 22); write('Press any key to exit'); end {write_headings}; function Read_Voltage : integer; var temp : integer; neg : boolean; begin port[mA_Card + 2] := $FF; repeat until (port[mA_Card + 2] and 1) = 1; port[mA_Card + 2] := $3F; repeat until (port[mA_Card + 2] and 1) = 0; port[mA_Card + 2] := $1F; temp := port[mA_Card + 1] and $4F; neg := (port[mA_Card + 1] and $80) = $80; port[mA_Card + 2] := $2F; temp := 256 * temp + port[mA_Card + 1]; port[mA_Card + 2] := $3F; if neg then temp := -temp; Read_Voltage := temp (* Read_Voltage := random(4096); *) end { Read_Voltage }; procedure openfile; begin write(output, 'Name of file (DIR:\FILENAME) '); readln(input, filename); assign(outfile, filename);  171 rewrite(outfile); end {openfile}; procedure closefile; begin close(outfile); end; procedure setup; begin write('Sample frequency (ie: 1, 5, 10, 15, 30 Hz) '); readln(sfreq); write(Time average sample interval (number of samples) '); readln(sint); write('Port number (0-7) '); readln(portnum); port[mA_Card + 0] := portnum; end; procedure survey; var i, j, sample, voltage, org_h, org_m , org_s, org_s 100, takevery : integer; runsum, timer : real; mampavg : real; firstsample : boolean; hour, min, sec, sec100 : word; begin takevery := trunc(maxfreq/sfreq); firstsample := TRUE; sample := 1; writeln('*******SURVEY BEGINS*******); WRITELN('PRESS ANY KEY TO END SURVEY'); gotoxy(22,9);write('TIME');gotoxy(35,9);write('AVERAGE'); repeat runsum := 0; for i := 1 to sint do begin for j :=1 to (takevery-1) do begin voltage := Read_Voltage; end; voltage := Read_Voltage; runsum := runsum + voltage;  172 (* gotoxy(22,9+i); write(i); gotoxy(25,9+i);write(voltage:4); *) end; mampavg := (trunc(runsum/sint)—MAOFFSET)/MACONVERS; gettime(hour, min, sec, sec100); if firstsample then begin firstsample := FALSE; org_h :=hour; org_m := min; org_s:=sec; org_s100:=sec100; end; timer :=(hour—org_h)*3600 + (min—org_m)*60 + sec — org_s + (sec100— org_s100)/100; writeln(outfile, timer:6:2, ", mampavg:8:6); sample := sample + 1; gotoxy(22,10); write(timer:6:2);gotoxy(35,10);write(mampavg:4:2); until keypressed; end; begin clrscr; openfile; setup; initialise; survey; (* cursor off; write_ headings; repeat for j := 1 to number of channels do begin port[mA Card + 0] := j — 1; gotoxy(22, 11+j); write(j:2); milliamps := (Read_Voltage/Gain)/Resistance; gotoxy(37, 11+j); if abs(milliamps) > (4096/Gain/Resistance) then write('Overflow') else begin write(milliamps:7:3, ") end { if }; end { for until keypressed; cursor on; gotoxy(1, 22); write(Test terminated normally'); read(input, ch) *) closefile; end. }  173 Appendix 2. Anglican controller data logging program listing  CLS PRINT "open" OPEN "com1:1200,n,8,2,rs,asc,ds0" FOR RANDOM AS #1 LEN = 256 PRINT "done" LINE INPUT "Enter file name to store DO data: "; filename$ OPEN filename$ FOR OUTPUT AS #2 TIMER ON send$ = "!01R1;" PRINT "press any key to begin , press any key to end" begin$ = INPUT$(1) timeini = TIMER DO PRINT #1, send$  ret$ = INPUTS(LOC(1), #1) a = ABS(LEN(ret$) — 4) PRINT (TIMER — timeini), ","; LEFT$(ret$, a) PRINT #2, (TIMER — timeini), ","; LEFT$(ret$, a) FOR i = 1 TO 6000 NEXT i LOOP UNTIL INKEY$ <> "" PRINT "stop" CLOSE #1 CLOSE #2 END  ^  174 Appendix 3. MathcadTm airlift model program listing Energy input: E^ 7= Q-P y in 1 .  hp  +  PT  K B := 11.40. (  AD  0.79  Ab  Energy Dissipations: given  1.)dissipations in the riser due to bubble wakes 6 1'2+ 6 3+ 6 4+ 6 5 E R =A R.p-gh a V IR ^ 5  2.) Energy dissipation in the downcomer due to upflow bubble motions with respect to liquid: assumption: e in the downcomer is zero, so energy dissipated in the dowmcomer is zero 3.) Energy losses due to flow through porous draft tube 9) I 6 511 Ui=p-g.hp- (hp  c  II 2=p•g- h D -  10 L 7  10 L k p. 1)Ie3 11 2Lk  3=p-gihr) hD -  , 3)1  U 4 =frgilip - (n.  2 - 1 -1 10 L k p. 6  6 111 thp -^ 1 ^2 '10^L k p,  where: U=flow velocity through porous draft tube ii=viscosity L=draft tube thickness  175 2^ h P V LR2=V LR1U 5.— 5 ^ hp V LR3 =V LR2 + 3 -U r^5  h  2 -^ P V LR4 =V LR3 +^-U 5  E 3>0  2^h 5  V LR5' V LR4 + ^--U  E 4?°  2^ h p LR6=VI,R4 + --u 5 V ^  E5.0  Q  AR V LR1 + V 1,R2 (^ 0.24 + 1.35. Q+ — A R^2  "3  Q  AR 2 .1  LR3 + V LR2 0.24 + 1.35( Q + V A R ^2  093  Q  3 /2  AR 0.24 + 1.35(^ Q V la4 + A R + ^2  V LR3 093  Hills (1976) valid fortrgr greater than 3 miser  176  Q AR 0.24 + 1.35  Q V iR5 +V LR4 /393 A R^2  Q AR  H 5- ^  Q V LR6 + V LR5) 0.24 + 1.35^+ ( A R^2  E  0.93  (1.1. 1 +U 2 +U 3 +13 4 +UT ^ • (2.7E-r) 5^  4.) Energy losses due flow through the top and the bottom sections of the airlift  I  3 (B 3 K l ) - —2- 0.5-A. R + 0.018-h D.w.r E^ n= p. (Via  AD  Total energy balance yields:  E 1 =E R + E BT Ep  E BT ->0  177 Appendix 4 Sample Calculation  (a) Estimated kLa requirement Literature reported oxygen consumption rate (OCR) for BHK cells is about 0.20 mmoles/10 9 cells/h. For 10 8 cells/mL medium in the proposed reactors, OCR is 20 mmoles/L/h. OCR = kLa (C*—C) where C* = 0.2 mmoles/L at 37°C C = 0 to detemine the minimum kLa needed The required kLa is, therefore, estimated to be 0.027 s -1 . (b) kLa determination A mass transfer coefficient can be conveniently defined by a simple mass balance for a given reactant or product species in a bioreactor. For example, considering the transfer of oxygen from air bubbles passing through a fermenter. Oxygen transfer rate =  dC dt  (39)  where C is the dissolved oxygen concentration in the bulk liquid at any time t, C * is the oxygen concentration in the liquid at the gas—liquid interface at infinite time (equivalent to the saturation concentration), a is the interfacial area and kL is the liquid—phase mass transfer coefficient. Depending on the type of flow pattern inside the reactor, Eq (39) can be incorporated in an overall oxygen balance in the liquid phase, and thus oxygen supply rates can be readily evaluated. kLa of a well—mixed batch process can be obtained by integrating equation (39):  178  f  dC  (39a)  = kLa fdt^ c (C -C)^0  O  r  ln(  C  -  C -C  ) = (kLa) t  ^  (39b)  If a fractional approach to equilibrium (E) is defined as the ratio of the mass transfer at any instant (i.e. C—00) to the maximum possible transfer (i.e. C * —00), or E— C - C  ° ^(40) C * -C o  then eq (40) may be written in term of E as —ln(1—E) = (kLa) t^  (40a)  This equation is more useful because the values of E=o (for zero dissolved oxygen) and E=1 (saturation condition) are easier to set and monitor on chart recorders and meters. The mass transfer coefficient, kLa, is the slope of a plot of ln(1/(1—E)) vs. t. (c) Scale—up of the fixed bed perfusion system In order to maintain a cell concentration of 1.3 x 10 8 cells/mL matrix, it is necessary to supply sufficient oxygen to the immobilized cells. A greater perfusion rate is needed to supply oxygen as the total cell population increases with the increased fixed bed length. However, increasing in the perfusion rate also increased shear stress. According to Bleim and Katinger (1988), it would be ideal  179 to keep the average shear stress within the fixed bed (calculated using Equation 32) around 0.5 N/m 2 . Assuming that (1) the Darcy resistance coefficient of the 30 PPI porous draft tube with cell attached = 6.7 x 10 7 m -2 , (2) void fraction of the matrix with cells = 0.7, and dF = 1 mm, the maximum allowable perfusion velocity is estimated to be 1 cm/s. The perfusion rate of 1 cm/s allowed the fixed bed system to be scaled—up 108 times in volume over the current system (i.e. to a maximum of 5.43 L matrix volume). The maximum size of the ceramic matrix is about 10L even if the average shear stress was allowed to increase from 0.5 to 1.0 N/m 2 . (d) Airlift simulation example Several parameters regarding to the physical configuration of the bioreactor, such as draft tube wall thickness, permeability of the draft tube and the reactor operating conditions, such as air flow rate, dispersion height are needed for the model simulation. The values of the input parameters are listed below: Q = 3.33 x 10 -5 m 3/s;^AR = 0.0022 m 2;^L = 0.013 m hp = 0.55 m;^AD = 0.00665 m 2 ; k = 6.7 x 10 7 nr2 hp = 0.21 m;^Ab = 0.004 m 2;^r= 0.025 m The resulting outputs are: egr varied from 0.8% to 1.7% Up varied from 35 to 105 mm/s Ufr varied from 0.23 to 1.12 m/s Rep = 19 calculated based on average U p of 60 mm/s. It should be noted that there was about 1.9 L,/s of medium perfused through the porous matrix while an average value of 1.8 Lis of medium flow down the downcomer. The volume ratio of medium perfused through the matrix to medium flow down the downcomer increases as the porosity of the draft tube matrices decreases.  180  Appendix 5 Raw data Raw data of Figure 8 Glucose Concentration (g/L) Time (day) 100 PPI 50 PPI 30 PPI 3.9 4.02 4.02 0 3.6 3.88 3.74 1 3.3 3.75 3.3 2 3.72 2.95 2.5 3 2.6 3.68 4 1.1 0.6 4.5 3.65 225 5 1.9 6 1.55 7 13 8 Raw data of Figure 10 Time (day)  oxygen uptake rate (it-moles/min) 0.96 1.74 2.2 4.07 6.97 658 9.17 9.18 10.53  0 2.92 3.92 4.17 4.91 5.93 6.41 7.14 7.92 822  Data for Fames 11 and 12 Total glucose used (g) Time (days) 0 0.96 1.92 2.92 3.92 4.17 4.91 5.93 6.41 7.14 7.92 822 8.98 9.24 9.93  0 0372 0.888 1.44 1.998 2.136 2.912 4.04 4.428 5.258 6.033 6.198 7.2 7.4 8.06  estimated cell number (x 10 -8 ) 0.02 0.1 0.175 0.22 0.41 0.7 0.7 0.828 0.822 -  Estimated^cell number (x 10 -8 ) 0.210598 0.251359 0.29606 0.30163 030163 0.434957 0.585473 0.520168 0.52862 0578962 0.419451 0.507723 0.567299 0.468957 0.446745  Total lactate produced (g) 0 0.258 0.618 1.17 1.992 2.112 2.74 3.67 3.97 4.72 5.34 5.53 6.32 6.67 7.15  181 Data for Figures 11 and 12 continued 10.09 8.17 10.93 9.47 11.26 9.92 11.43 10.09 11.96 11.05 12.18 11.48 12.43 11.9 12.89 12.67 13.47 13.9 13.89 14.57 13.93 14.62 14.39 15.9 14.91 16.7 15.11 17.3 15.39 18.06 15.91 18.9 16.17 1956 1635 20.03 16.98 20.93 17.14 21.38 17.27 21.8 17.91 23 18.03 23.19 18.9 24.98 19.1 25.29 19.21 25.6 19.91 27.06 20.27 27.95 20.91 29.18 21.91 31.2 22.04 31.4 22.26 32.1 22.89 3327 23.11 33.85 23.22 34.15 23.91 35.43 24.09 35.79 24.22 36.4 24.46 37.15 24.9 38.4 25.27 39.8 25.9 41.5 26.06 41.8 26.19 42.4 26.96 44.6 27.94 47.62 28.9 50.82 29.21 51.83 29.91 53.2  0.60737 0.791103 0.642293 0.763946 1.023332 0.987647 0.911391 1.031141 1.009761 0.773163 1.095815 1.174207 1.233277 1.552793 1.176543 1.128761 1399342 1.097739 1.152467 1.64219 1387435 0.939766 0.989348 0.980293 1.187005 1.332582 1.238571 1.194049 1.071163 0.966973 1.282685 1369283 122106 1.457511 1.245201 1.047576 1.81856 2.124266 1.621168 1.641549 2.228261 2.091962 1.242777 1.76369 2.030576 1.613799 1.743196 1.791141 1.417174  7.195 8.07 839 8.59 9.44 9.68 10.05 10.64 11.66 12.12 12.17 13.15 13.73 14.06 14.67 15.28 15.76 16.16 16.72 17.09 17.3 18.19 18.34 19.67 19.92 20.17 21.23 21.93 22.79 24.22 2438 24.91 25.81 26.2 26.46 27.5 27.85 28.3 29.11 29.95 31.1 325 32.81 332 34.96 37.5 40.04 40.09 42  182 Data for Figures 11 and 12 continued 30 53.5 30.2 53.9 30.9 55.3 31.2 55.96 31.9 57.5 32.2 58 32.9 59.6 33.2 60.32 33.9 61.69 345 63.06 35 63.86 35.45 65.36 663 35.9 36.47 67.97 36.91 68.5 37.21 69.62 70.9 37.9 38.22 72.2 38.9 73.3 39.19 7433 75.62 39.88 Data for Figure 13 BHK Cell Concentration (x 10 -4 ) 4 20 40 90 170 Data for Figure 14 Time (day) 5% serum 0 0.84 1.17 134 1.87 2.09 2.34 2.8 338 3.84 4.3 4.82  Total transferin produced (mg) 0 7.4 15.6 24 35.5 37.81 3838 53.82 65.12 87.22 92.02 107.32  1.43763 1.449277 1.086957 1.141304 1.195652 1.050723 1.074016 1.273293 1.184005 1.152304 1.055255 1.340582 1.473429 1363783 1.123473 1.341815 1.518587 1.608038 1543516 1.404723 1.473174  42.31 42.58 43.08 43.64 44.76 45.45 46.11 46.8 47.8 49.2 49.74 50.87 51.56 52.95 53.37 54.26 55.06 55.95 56.67 57.48 58.02  LDH (IU/L) 522 60 74.2 97.4^139.2  Time (day) 2.5% serum 0 0.2 0.48 1 1.26 1.44 1.99 2.23 2.36 3 3.12 3.99 4.19 4.3 5  Total transferin produced (mg) 0 1.5 12 20.1 22.2 24 34.9 40.8 43.5 61.9 63 73.2 73.7 76.7 83  183 Data for^ Figure 15 DMEM Time (day) Total Transferrin produced (mg) 0 0 1.06 0.729 2.94 1.729 2.729 3.99 3.729 5.23 4.719 6.46 11.2 5.948 6.792 13.6 7.715 15.3 8.715 18.2 9.729 23.9 10.72 27.9 11.69 33.8  Data for Figure 16 Time (11) Total glucose used (g) 0 0 18 037 26 0.38 42 0.51 66 0.9 90 1.45 114 2.24 142 2.9 144 2.9 162 3.65 186 4.9 210 5.9 212 5.9 234 7.27 258 8.64 261 8.64 282 10.55 306 11.91 308 11.91 330 14.16 354 15.43 355 15.43  DMEM/F-12 Time (day) 0 0.96 1.92 2.92 3.92 4.17 4.91 5.93 6.41 7.14 7.92 8.22 8.98 9.24 9.93 10.09 10.93 11.26 11.43  Total lactate produced (g) 0 0.152^_ 034 0.399 0.808 1.156 1.914 2.764 2.764 3.507 4.248 5.367 5.367 5.517 8.43 8.43 10.02 11.039 11.039 12.966 13.84 13.84  Total Transferrin produced (mg) 0 0.631 2.772 4.44 8.184 15.96 18.216 23.284 29.224 33.632 45.523 53.923 59.863 66.963 74.663 79.263 86.663 94.863 103.26  184 Data for Figure 16 continued 378 17.67 403 18.96 404 18.96 21.14 426 429 21.27 22.5 455 455 22.5 480 24.91 500 26.03 501 26.03 28.7 523 546 30.11 549 30.11 552 30.29 571 32.69 595 33.95 596 33.95 619 34.72 642 35.63 666 36.3 667 36.3 691 36.81 715 3732 715 37.32 739 37.72 763 38.6 787 39.25 815 40.07 835 4055 40.55 835 859 41 882 41.87 906 42.54 930 43.26 931 43.26 44.47 955 45.48 980 1002 46.21 1009 46.45 46.45 1009 1026 47.46 1051 48.85 1075 49.88 1076 49.88 1099 51.19 52.17 1123 .  15.666 16.446 16.446 18.078 18.028 18.708 18.708 21.117 21.223 21.223 23.188 24.088 24.088 24.262 26.321 27.355 27355 28.193 28.864 29.469 29.469 29.888 30341 30341 30.826 31.513 32.129 33.1 33.348 33.348 33.784 34.545 35.074 35.588 35.588 36.388 37.331 37.903 37.845 37.845 38.522 39.6 4032 40.32 41.256 41.792  185 Data for Figure 17 Non-porous galss draft tube AP (mm F170) Upy (m/s) 0.00 0.00E+00 0.82 1.63E-03 2.18 4.90E-03 3.63 8.16E-03 5.54 1.22E-02 6.60 1.63E-02 Thick wall Uty. (m/s) 0.00E+00 1.63E-03 4.90E-03 8.16E-03 1.22E-02 1.63E-02 Thin wall UEr (m/s) 0.00E+00 1.63E-03 4.90E-03 8.16E-03 1.22E-02 1.63E-02  Ugr (m/s) 0.00E+00 1.63E-03 4.90E-03 8.16E-03 1.22E-02 1.63E-02  ±AP 0.00 0.03 0.09 0.13 0.15 0.16  Non-porous thick wall 100 PPI AP (mm 1110) 0.00 0.57 1.50 2.26 3.13 3.75  ±AP 0.00 0.05 0.09 0.15 0.12 0.12  50 PPI AP (mm I-1,0) 0.00 0.37 0.78 1.11 1.43 1.80  ±AP 0.00 0.04 0.07 0.04 0.07 0.07  30 PPI AP (mm H70) 0.00 0.29 0.67 1.08 1.54 1.75  ±AP 0.00 0.03 0.04 0.05 0.07 0.11  50 PPI AP (mm 1-1,0) 0.00 0.10 0.32 0.44 0.57 0.74  ±AP 0.00 0.00 0.03 0.00 0.00 0.08  30 PPI AP (mm F1 70) 0.00 0.09 0.21 0.38 0.49 059  ±AP 0.00 0.01 0.03 0.04 0.04 0.09  ±AP 0.00 0.04 0.07 0.08 0.12 0.21  Non-porous thick wall 100 PPI AP (mm f170) 0.00 0.45 126 2.26 3.25 3.92  100 PPI AP (mm I-190) 0.00 0.52 1.11 156 2.19 2.48  ±AP 0.00 0.06 0.06 0.10 0.15 0.17  100 PPI AP (mm I-170) 0.00 0.57 1.62 2.17 3.33 3.86  ±AP 0.00 0.04 0.07 0.24 0.29 0.17  non-porous 50 PPI thin wall AP (mm H2O) 0.00 0.62 1.47 2.54 352 4.17  ±AP 0.00 0.06 0.17 0.08 0.24 0.05  ,  186 Data for Fieure 18 draft tube Long^top configuration short^bottom Ilex (m/s) AP ±AP (mm H70) 0.00 0.00 0.00E+00 0.10 0.01 1.63E-03 0.35 0.03 4.90E-03 0.58 8.16E-03 0.03 1.22E-02 0.77 0.05 1.63E-02 0.88 0.08  Long^top long^bottom AP ±AP (mm H70) 0.00 0.00 0.20 0.02 056 0.02 0.80 0.04 1.25 0.04 1.48 0.04  Data for Finure 19 draft tube long top and^long bottom configuration Uar (m/s) ±AP AP (mm 1170) 0.00 0.00E+00 0.00 1.63E-03 0.52 0.06 4.90E-03 1.11 0.06 1.56 0.10 8.16E-03 1.22E-02 2.19 0.15 2.48 0.17 1.63E-02 Data for Finure 20 PBS Medium type Ugr (m/s) AP (mm H20) 0.00E+00 0 1.63E-03 0.81447 1.84029 4.90E-03 8.16E-03 2.61897 1.22E-02 3.62619 1.63E-02 4.26652  Data for Finure 21 Medium PBS type Ugr (m/s) AP (mm H20) 0.00E+00 0.000 1.63E-03 0.147 4.90E-03 0.299 8.16E-03 0.408 1.22E-02 0.580 1.63E-02 0.687  ±AP 0 0.04906 0.04676 0.0287 0.09903 0.18381  ±AP 0.000 0.013 0.022 0.030 0.039 0.041  PBS + 2% FCS AP (mm 1190) 0 0.77453 1.57774 2.47252 3.03661 3.71127  PBS + 2% FCS AP (mm 1-170) 0.000 0.107 0.226 0.315 0.392 0.511  Short^top long^bottom AP ±AP (mm H70) 0.00 0.00 0.15 0.03 0.45 0.03 0.69 0.00 1.05 0.04 1.26 0.00  Short^top short^bottom AP ±AP (mm 1190) 0.00 0.00 0.09 0.00 0.27 0.05 0.49 0.06 0.62 0.04 0.77 0.09  Short top and^short bottom AP (mm H90) 0.00 0.43 0.88 1.13 1.66 1.73  ±AP 0 0.0291 0.03361 0.04021 0.07778 0.08947  ±AP 0.000 0.007 0.010 0.012 0.018 0.032  ±AP 0.00 0.03 0.07 0.07 0.09 0.15  PBS + 5% FCS AP (mm 11,0) 0 0.81874 1.75501 2.35886 3.31138 3.79387  PBS + 5% FCS AP (mm H70) 0.000 0.090 0.228 0.314 0.381 0.470  ±AP 0 0.04435 0.04535 0.04805 0.06089 0.12408  ±AP 0.000 0.008 0.008 0.011 0.018 0.014  187 Data for Figure 22 PBS Medium type Ugr (m/s) AP (mm H70) 0.000 0.00E+00 0.523 1.63E-03 1.106 4.90E-03 1.558 8.16E-03 2.186 1.22E-02 1.63E-02 2.484  ±AP 0.000 0.056 0.060 0.105 0.145 0.171  Data for Figure 23 PBS Medium type Ugr (m/s) AP (mm H70) 0.000 0.00E+00 0.294 1.63E-03 4.90E-03 0.666 1.081 8.16E-03 1.537 1.22E-02 1.63E-02 1.751  ±AP 0.000 0.026 0.045 0.048 0.073 0.109  PBS + 2% FCS AP (mm H70) 0.000 0.547 0.926 1.156 1.642 1.924  PBS + 2% FCS AP (mm H2O) 0.000 0.309 0.542 0.774 1.071 1.278  ±AP 0.000 0.036 0.043 0.064 0.077 0.082  ±AP 0.000 0.017 0.025 0.033 0.048 0.071  PBS + 5% FCS AP (mm H70) 0.000 0563 0.949 1.317 1.701 2.053  PBS + 5% FCS AP (mm H2O) 0.000 0.299 0.582 0.769 1.074 1305  ±AP 0.000 0.050 0.059 0.072 0.069 0.106  ±AP 0.000 0.019 0.054 0.038 0.065 0.075  Data for Figures 24 and 25  Ugr U^(m/s)  0.00E+00 1.63E-03 4.90E-03 8.16E-03 1.22E-02 1.63E-02  100 PPI (thick) AP (mm H2O) 0.00 0.88 1.94 2.74 3.58 4.32  SOPPI (thick) AP (mm H70) 0.00 0.60 1.20 1.63 2.06 2.43  Model prediction 100 PPI 30 PPI (thick) (thin) AP (mm AP H2O) (mm 11,0) 0.00 0.00 0.48 0.46 0.93 0.91 125 122 1.58 1.52 1.77 1.86  SOPPI (thin) AP (mm I-170) 0.00 056 1.14 156 2.00 2.38  30 PPI (thin) AP (mm I-170) 0.00 0.75 1.70 2.48 3.21 3.90  Data for Figure 26  Ugr (m/s) 0.00E+00 1.63E-03 4.90E-03 8.16E-03 122E-02 1.63E-02  100 PPI (thick) ULd (m/s) 0.00 0.06 0.10 0.13 0.16 0.18  5OPPI (thick) Uhl (m/s) 0.00 0.11 0.20 0.25 0.31 0.36.  Model prediction 30 PPI 100 (thick) PPI (thin) ULd Uj (m/s) (m/s) 0.00 0.00 0.15 0.05 0.27 0.08 0.35 0.10 0.43 0.11 0.50 0.13  SOPPI (thin) Uu (m/s) 0.00 0.05 0.13 0.17 0.21 0.24  30 PPI (thin) U1 (m/s) 0.00 0.10 0.17 0.23 0.28 0.32  measured^data 100 PPI (thick) U1 ±ULd (m/s) (m/s) 0.00 0.00 0.08 0.02 0.03 0.14 0.18 0.07 0.12 0.25 0.28 0.11  ^  188 Data for Figure 27 reactor 3OPPI^thick wall config. with short^supports Upy (m/s) kT a (1/s) std. dev. 0.60E+00 0.00E+00 0.00E+00 1.63E-03 3.44E-03 1.15E-04 4.90E-03 7.60E-03 5.04E-04 8.16E-03 1.29E-02 9.50E-04 122E-02 1.46E-02 9.06E-04 1.63E-02 _ 137E-02 1.23E-03  3OPPI^thin wall with short^supports kr a (1/s) std. dev. 0.00E+00 0.00E+00 2.81E-03 1.68E-04 7.65E-03 2.04E-04 9.85E-03 2.84E-04 1.20E-02 432E-04 1.37E-02 3.97E-04  30PPI^thick wall with long^supports kT a (1/s) std. dev. 0.00E+00 0.00E+00 2.98E-03 4.13E-05 7.72E-03 4.47E-04 9.92E-03 7.80E-04 1.70E-02 3.13E-04 1.70E-02 1.23E-03  10OPPI^thick wall with short^supports 14 a (1/s) std. dev. 0.00E+00 0.00E4-00 2.92E-03 2.33E-04 7.32E-03 1.82E-04 8.65E-03 2.64E-04 1.18E-02 4.12E-04 1.21E-02 7.12E-04  100 PPI^thin wall with short^supports kT a (1/s) std. dev. 0.00E+00 0.00E+00 3.21E-03 1.96E-04 8.97E-03 6.80E-04 1.16E-02 2.42E-04 1.48E-02 5.32E-04 1.63E-02 3.23E-04  100 PPI^thick wall with long^supports kT a (1/s) std. dev. 0.00E+00 0.00E+00 3.97E-03 6.85E-04 6.83E-03 2.60E-04 1.22E-02 7.40E-04 1.21E-02 4.68E-04 1.36E-02 1.06E-03  glass draft tube with serum kr a (1/s) 0.00E+00 3.67E-03 8.67E-03 1.09E-02 136E-02 1.42E-02  thin 3OPPI wiht pbs  reactor config. U97 (m/s) 0.60E+00 1.63E-03 4.90E-03 8.16E-03 1.22E-02 1.63E-02  Data for Figure 28 reactor glassdraft config. tube with pbs Upy (m/s) 14 a (1/s) 0.60E+00 0.00E+00 1.63E-03 3.15E-03 4.90E-03 6.48E-03 8.16E-03 1.00E-02 1.22E-02 1.14E-02 1.63E-02 1.40E-02  std. dev. 0.00E+00 1.29E-04 2.29E-04 3.84E-04 3.21E-04 4.51E-04  std. dev. 0.00E+00 1.41E-04 1.60E-04 9.51E-05 2.12E-04 2.28E-04  Data for Figure 29 100PPI^with thick wall and reactor configuration long supports kT a (1/s) std. dev. Upy (m/s) 0.00E+00 0.00E+00 0.60E+00 1.63E-03 3.97E-03 6.85E-04 4.90E-03 6.83E-03 2.60E-04 1.22E-02 7.40E-04 8.16E-03 1.21E-02 1.22E-02 4.68E-04 1.63E-02 1.36E-02 1.06E-03  kT a (1/s) 0.00E+00 2.81E-03 7.65E-03 9.85E-03 1.20E-02 137E-02  Model predictions 14 a (1/s) 1.84E-03 2.41E-03 3.56E-03 4.71E-03 6.14E-03 7.58E-03  std. dev. 0.00E+00 1.68E-04 2.04E-04 2.84E-04 4.32E-04 3.97E-04  glass draft tube ki. a (1/s) 0.00E+00 3.15E-03 6.48E-03 1.00E-02 1.14E-02 1.40E-02  std. dev. 0.00E+00 1.29E-04 2.29E-04 3.84E-04 3.21E-04 4.51E-04  thin 30 PPI with serum kT a (Vs) 0.00E+00 3.10E-03 7.80E-03 1.02E-02 1.26E-02 138E-02  std. dev. 0.00E+00 5.05E-05 1.84E-04 4.02E-04 5.58E-04 7.05E-04  189 Data For Figure 30 Wall^Thickness (mm) 10 15 20 30 40  AR = 0.038 m 2  AR = 0.0019 m 2  AR = 0.0007 m 2  AR = 0.0003 m 2  0.0517 0.0408 0.0347  0.07867 0.06061 0.0493 0.0398  0.0856 0.069 0.059 0.04712 0.04025  0.095 0.078 0.0659 0.053 0.045  Data for Figure 31 Time (h) Total glucose used (g) 0 0.00 27 0.20 47 0.44 49 0.44 71 0.99 77 1.10 95 1.51 96 1.29 126 1.29 119 1.33 120 133 126 137 144 2.22 171 2.80 190 3.18 214 3.74 215 3.74 225 3.93 238 4.34 246 4.34 263 4.63 264 4.63 270 5.06 287 6.28 294 6.83 312 7.53 337 8.48 359 9.13 360 9.13 365 9.38 383 10.65 391 11.64 407 13.04 414 13.56 430 14.71 438 15.20 455 16.00 462 16.54 481 17.87 482 17.87  Total lactate produced (g) 0.00 0.27 0.54 1.06 1.32 1.40 1.71 1.71 2.13 2.23 2.23 2.30 2.86 3.45 3.91 4.73 4.73 5.03 5.29 5.29 5.51 5.51 5.84 732 7.32 10.48 11.48 11.38 1138 1135 12.70 13.23 14.33 14.72 15.64 16.39 16.67 17.05 18.01 18.01  190 Data for Figure 31 continued 504 19.23 511 19.65 527 20.99 527 20.99 535 21.70 24.02 551 553 24.15 24.86 558 24.86 558 575 27.41 581 28.35 581 28.35 599 31.28 604 31.73 604 31.73 623 35.26 626 35.80 626 35.80 647 39.43 648 39.43 44.05 674 676 44.05 695 47.94 701 48.80 48.80 701 719 52.62 725 53.97 727 53.97 743 59.05 60.49 750 60.49 750 767 66.65 772 67.81 772 67.81 73.81 791 794 74.13 74.13 794 815 82.01 816 82.01 839 90.09 839 90.09 863 98.01 868 98.81 868 98.81 887 105.93 892 106.85 892 106.85 911 113.97 114.89 916  19.30 19.55 20.36 20.36 21.18 22.80 22.54 22.93 22.93 24.90 25.72 2532 26.94 27.72 27.72 30.08 29.87 29.87 32.09 32.09 35.60 35.60 38.97 38.93 38.93 42.09 43.59 43.59 47.91 48.91 48.91 54.13 55.02 55.02 60.61 58.90 58.90 67.45 67.45 74.33 74.33 81.25 81.47 81.47 87.21 87.96 87.96 95.37 94.57  191 Data for Figure 31 continued  916 936 940 941 941 959 965 965 987 988 1013 1013 1033 1033 1056 1058 1085 1079 1082 1085 1104 1128 1128 1152 1152 1175 1175 1199 1200 1223 1224 1248 1248 1271 1274 1295 1300 1300 1319 1320 1347 1347 1367 1367 1391 1398 1398 1415 1415  114.89 122.01 122.69 122.89 122.89 131.73 133.01 133.01 143.65 143.65 154.13 154.13 163.65 163.65 174.69 174.69 175.93 182.21 182.21 182.77 190.01 195.65 195.65 203.05 203.05 21133 211.33 219.61 219.61 228.17 228.17 236.85 236.85 245.61 245.61 254.65 255.17 255.17 263.33 26333 272.69 272.69 281.57 281.57 291.21 291.93 291.93 296.77 296.77  94.57 100.96 101.34 101.77 101.77 109.08 110.28 110.28 118.71 118.71 127.58 127.58 135.22 135.22 139.66 139.66 140.99 146.16 146.16 146.95 152.78 156.98 156.98 163.14 163.14 169.71 169.71 176.72 176.72 183.65 183.65 190.77 190.77 19733 19733 204.93 20538 20538 212.77 212.77 220.42 220.42 226.93 226.93 235.01 235.18 235.18 238.63 238.63  192 Data for Figure 31 continued 302.21 1439 302.21 1439 307.57 1463 307.57 1463 Data for Figure 32 Time Control std. dev. (min) 0 -0 211000 405172 60 150 220500 311834 240 --330 332750 232991 420 1.08E6 119501  242.89 242.89 246.79 246.79  P80  std. dev.  CM80  S80  std. dev.  -0 -0 -0 365000 410829 1.202E6 345421 2.517E6 132228 1.188E6 -774500 1.466E6 3.272E6 811405 651500 544472 3.169E6 348839 14500 -1.074E6 85913 1.241E6 126218 3.307E6 856306 825000 137178 2.084E6 568867 3.108E6 426031  Data for Figure 33 P80 Time Control std. dev. (min) 0 0 0 0 60 3.208E6 3.277E6 8.098E6 150 8.758E6 1.011E6 1.187E7 240 1.037E7 1.771E6 1.196E7 330 1.187E7 1.575E6 _1.383E7 Data for Figure 34 Time Control std. dev. (min) 0 0 0 60 265000 18296 150 275000 367412 240 419000 431688 330 388000 439113  std. dev.  std. dev.  CM80  std. dev.  0 3.495E6 3.46E6 2.597E6 1.452E6  0 4.55E6 8.076E6 1.076E7 1.199E7  0 2.199E6 602808 1.777E6 841457  FlexCM80 0 4.518E6 9.19E6 1.048E7 1.142E7  std. dev.  S80  DEA80 std. dev.  DEALG80 0 0 0 760000 206475 447000 139E6 424264 1.28E6 1.59E6 375473 1.18E6 1.55E6 381395 1.16E6  .  0 0 362392 659000 713735 1.51E6 936916 1.64E6 773663 1.52E6  std. dev.  S80  std. dev.  0 152027 1.513E6 572756 1.694E6  0 6.38E6 1.223E7 1.354E7 1.53E7  0 3.125E6 2.383E6 2.721E6 2.238E6  std. dev. PER80 std. dev. 0 25043 57540 3977 29256  Data for Figure 35 Time Control std. dev. (min) 0 0 0 60 2.349E6 2.955E6 150 1.173E7 7.213E6 240 1.496E7 282135 330 1.212E7 485075  0 1.65E7 2.305E7 2.473E7 2.647E7  Data for Figure 36 Time Control std. dev. (min) 0 0 0 60 156000 4680 150 135000 17400 240 161000 74600  Cytodex std. dev. Culti- std. dev. S80 std. dev. spher -1 0 0 0 0 0 0 340000 151000 80500 87000 228000 12400 373000 33000 389000 44500 424000 25900 432000 50100 387000 93300 460000 30100  .  S80  std. dev. PER80 std. dev. DEA80 std. dev. 0 1.361E6 4.838E6 3.608E6 2.868E6  Flex- std. dev. CM80 0 -1.098E6 595030 1.739E6 53386 2.551E6 40305 2.612E6 432749 2.94E6 37830  0 0 0 0 7.191E6 620368 6.993E6 395979 1.189E7 7.56E6 1.601E7 2.445E6 1.434E7 5.4E6 1.72E7 3.665E6 1.75E7 4.503E6 1.979E7 252437  0 0 98100 313778 954000 327832 1.07E6 233610 987000 184643  DEALG80 0 1.142E7 1.569E7 1.532E7 1.878E7  std. dev. 0 3.935E6 272236 3.435E6 3.656E6  Collegen std. dev. Coated 0 0 190000 89600 441000 4240 534000 3010  193 Data for Figure 37 Time Collegen std. dev. S80 std. dev. (min) Coated 0 0 0 0 0 60 654000 297161 541000 42544 150 1.4E6 130932 1.38E6 59072 240 1.53E6 144780 1.6E6 40010 330 1.67E6 68872 1.79E6 51453  Cytodex std. dev. -1 0 0 693000 131757 1.97E6 171355 1.89E6 225838 1.99E6 161927  Data for Figure 38 Time S80 std. dev. (min) 0 3E6 0 60 1.77E6 106000 150 848000 53000 240 521000 55200 330 315000 22600  3E6 1.55E6 795000 415000 400000  Data for Figure 39 Time S80 std. dev. (min) 0 3.58E6 0 117000 60 2E6 150 1.52E6 184000 240 910000 14100 330 598000 190000 Data for Figure 40 Time S80- std. dev. (min) WET 0 2.744E6 0 30 1.644E6 95812 120 691777 5342 240 376333 31112  .  Culti- std. dev. spher 0 0 1.07E6 13199 1.71E6 171055 1.69E6 125022 1.75E6 73173  Control std. dev. 0 339000 186000 468000 116000  0 289029 510148 710170 414600  S40-15 std. dev. S40-60 std. dev. S40-240 std. dev. Control std. dev. 3E6 1.57E6 946000 488000 466000  0 99000 67200 81300 58700  0 49500 2120 99700 10600  3E6 1.81E6 846000 444000 355000  0 170000 48100 15600 9900  0 3E6 2.84E6 332000 2.78E6 177000 2.59E6 311000 1.96E6 283000  S40-15 std. dev. S40-60 std. dev. S40-240 std. dev. Control std. dev. 0 3.58E6 0 3.58E6 0 3.58E6 3.61E6 38900 2.1E6 220000 2.74E6 166000 231E6 258000 3.51E6 203000 132E6 14100 1.64E6 38900 1.81E6 148000 3.02E6 431000 660000 184000 1000000 138000 1.2E6 108000 2.91E6 108000 580000 89600 653000 95500 605000 87800 2.75E6 0 S80DRY 2.744E6 2.687E6 1.98E6 1318E6  std. dev.  S40- std. dev. WET 0 2.744E6 0 112076 1.925E6 37830 51618 904833 37476 178190 522666 16027  Data for Figure 41 Time Control std. dev. <300 pm std. dev. 300 - 500 (min) pm 0 2.04E6 0 2.04E6 0 2.04E6 60 1.71E6 299000 544000 26800 1.04E6 98800 532000 150 1.56E6 81200 89600 240 1.45E6 94300 76000 11300 295000 1410 115000 19000 330 1.72E6 38700  S40DRY 2.744E6 2.846E6 2324E6 1.825E6  std. dev. Control std. dev. 0 258565 148963 180233  2.744E6 0 2.483E6 183140 1.876E6 73539 1.241E6 94045  std. dev. > 500 pin std. dev. 0 158000 65100 49500 14100  2.04E6 1.17E6 624000 340000 186000  0 71400 39600 31100 5660  .  194 Data for Figure 42 Time 5.6x10 6 std. dev. 2.8x10 6 std. dev. 1.4x10 6 std. dev. 0.7x 0 6 std. dev. Control std. dev. (min) 0 0 0 0 0 0 0 0 0 0 0 1.7 30.5 1.8 23.5 11.8 60 21.2 25.9 12.1 -1.29 0.03 58.9 5.4 64.6 3.8 72.8.-90 53.2 2.1 2.0 10.0 3.5 1.9 79.7 1.1 84.2 0.8 81.7 1.3 28.2 150 58.9 2.8 Data for Figure 43 Time DMEM std. dev. DMEM/ F-12 (min) serum serum 5.8E6 0 5.77E6 0 35 4.32E6 76400 4.69E6 95 3.78E6 187000 4.48E6 185 2.11E6 101000 2.45E6 275 1.35E6 112000 _ 2.39E6  std. dev. DMEM std. dev.  0 890000 1.08E6 233000 418000  6.09E6 3.97E6 3.55E6 3.11E6 2.78E6  PBS  42400 6.12E6 381000 4.03E6 607000 4.09E6 37500 2.91E6 294000 3.03E6  std. dev. Control (DMEM serum) 0 6.13E6 76700 6.15E6 85200 6.08E6 320000 5.62E6 770000 4.83E6  std. dev.  35400 84500 47000 427000 28600  195 Data for Figure 44 Medium Glucose Conc. WO 1 1.5 2.2 3.4 4  Unit GUR (g/h/L) 0.8 1.3 1.7 1.8  Data for Figures 45 and 46 Time (h) Glucose BILK cells Concentration WO 0 3.3 27 1.67 51.5 0.88 51.5 4.14 72 2.83 96 1.5 120 0.83 120 4.1 144 2.62 168 1.3 168 4.17 192.75 2.1 217 0.95 217 4.17 240 2.1 264 0.88 264 4.08 288 2 312 0.86  Unit GUR (g/h/L) 0.5 0.9 1 0.95 1  Time (h) Vero cells 0 52 72 96 96 120 144 144 168 197 197 218 240 240 264 288 288 312 336 336 363 388 388 408 432 432 456  Glucose Concentration WO  4 258 1.51 0.7 3.11 2.03 1.08 4.23 2.28 0.82 4.34 2.09 0.76 4.23 1.7 0.46 4.26 1.7 0.45 4.24 1.63 0.47 4.12 2 0.68 4.12 1.79  196 Data for Figure 48 Time CM80 DEA80 (h) 0 0 0 52 0.142 0.086 72 0.249 0.158 96 0.33 0.264 120 0.438 0.348 144 0.533 0.463 168 0.728 0.561 197 0.874 0.798 218 1.099 0.926 240 1.232 1.156 264 1.485 1.297 288 1.609 1.591 312 1.865 1.688 336 1.99 1.907 363 2.251 2.048 388 2367 2.262 408 2579 2.38 432 2.711 2.583 456 2.944 2.717  P40 0 0.086 0.158 0.264 0.348 0.463 0.561 0.798 0.926 1.156 1.297 1.591 1.688 1.907 2.048 2.262 2.38 2.583 2.717  Time (1) 0 21 43 67 91 115 139 166 190.5 211 235 259 283 307 322 356  P80  S80  0 0.029 0.071 0.156 0.284 0.362 0.508 0.649 0.853 0.97 1.181 1.306 1.507 1.632 1.857 1.983  0 0.029 0.071 0.156 0.284 0.362 0.508 0.649 0.853 0.97 1.181 1.306 1.507 1.632 1.857 1.983  Data for Figure 49 (cell concentration x 10 -8 Time CM80 DEA80 P40 Time P80 (h) (h) 0 0.60 039 0.39 0.0 0.24 52 0.60 039 0.39 21.0 0.24 72 0.64 0.59 059 43.0 0.40 96 0.58 0.58 67.0 0.65 120 0.62 91.0 0.63 144 0.65 0.65 139.0 0.83 168 0.97 190.5 1.04 197 1.05 1.05 235.0 1.03 218 1.24 283.0 1.00 240 1.20 1.20 322.0 1.38 264 1.16 288 1.20 1.20 312 1.17 336 1.06 1.06 363 1.06 388 1.07 1.07 408 1.17 432 1.04 1.04  S80 0.24 0.24 0.40 0.65 0.63 0.83 1.04 1.03 1.00 1.38  Time (h) 0 27 51.5 72 96 120 144 168 193 217 240 264 288 312 336.5 362 386 408 432  Time (h) 0.0 27.0 51.5 72.0 96.0 120.0 168.0 193.0 217.0 240.0 264.0 288.0 312.0 336.5 386.0 408.0  Q80 0 0.05 0.106 0.165 0.222 0.274 0.29 0.35 0.393 0.427 0.46 0.497 0.531 0.57 0.614 0.642 0.718 0.788 0.858  080 0.31 0.31 038 039 0.33 0.21 031 0.23 0.21 0.22 0.22 0.22 0.25 0.21 0.47 0.45  Time (h) 0 24 48.5 74 98 120 144 168 192 216 242.5 266.5 290 311.5 335.5 359 383 410  CS80 0 0.032 0.098 0.176 0.261 0.347 0.46 0.51 0.602 0.745 0.836 0.996 1.087 1.224 1.325 1.54 1.646 1.894  Time CS80 (h) 0.0 030 24.0 0.30 48.5 0.42 74.0 0.49 120.0 0.64 168.0 0.44 216.0 0.69 266.5 0.82 290.2 0.84 3115 0.56 335.5 0.87 383.0 1.00  197 Data for Figure 50 Time (h) Cytodex 0 0 0.43 24 0.97 45.5 1.8 70 2.37 94 2.37 94 3.76 118 4.92 142 4.92 142 169 6.78 7.78 193 193 7.78 214 9.92 238 11.12 11.12 238 262 13.3 262 13.3 285.5 15.45 285.5 15.45 309.5 17.75 17.75 309.5 337 20.26 337 20.26 22.61 3595 3595 22.61 382 24.96 382 24.96 406 26.21 406 26.21 28.53 430 430 2853  DEA80 0 0.5 1.12 2.02 2.68 2.68 4.01 5.42 5.42 7.49 8.52 8.52 10.25 11.56 1156 13.31 13.31 15.24 15.24 17.25 17.25 19.6 19.6 21.65 21.65 23.88 23.88 26 26 283 28.3  Data for Figure 51 Time (h) Cytodex CG 0 0.00 0.00 0.29 6.25 0.23 24 1.26 1.28 30 1.53 1.58 485 2.52 2.57 72 3.12 3.19 72 3.12 3.19 97.5 4.53 4.73 123 5.48 535 123 5.35 5.48 144 6.50 6.81 150 6.87 7.14 168 7.55 7.79 168 7.55 7.79 174 7.97 8.16  Time (h) 0 6 24 30 48 54 72 78 96 97 102 1215 146.5 146.75 170  CM80 0.00 0.13 0.61 0.99 1.99 2.32 3.38 3.78 532 5.32 5.74 7.37 8.07 8.07 10.13  Time (h) P80 0 0.00 24.5 0.56 50.5 158 74.75 2.85 97 3.52 120.5 5.26 1455 6.22 169 8.38 192.25 9.71 218 11.56 243.5 12.29 265 13.84 288.5 14.71 312.5 16.52 336 17.22  198 Data for Figure 51 continued 192 9.28 9.62 198 9.58 9.91 9.98 10.40 216 216 9.98 10.40 11.63 11.44 240 11.77 11.62 245.5 267 12.52 12.54 267 12.52 12.54 291 13.83 13.66 313 14.61 14.53 14.61 14.53 313 319 14.97 14.89 336.75 16.19 15.94 16.26 343 16.52 361 17.26 16.95 361 16.95 17.26 367 17.61 17.22 385 18.80 18.34 18.65 391 19.11 409 19.70 19.27 Data for Figure 52 Time (h) Cytodex 0 0 24 0.43 45.5 0.97 70 1.8 94 237 94 2.37 118 3.76 142 4.92 142 4.92 169 6.78 193 7.78 193 7.78 214 9.92 238 11.12 238 11.12 262 13.3 262 133 285.5 15.45 285.5 15.45 309.5 17.75 309.5 17.75 337 20.26 337 20.26 359.5 22.61 359.5 22.61 382 24.96  DEA80 0 0.5 1.12 2.02 2.68 2.68 4.01 5.42 5.42 7.49 8.52 8.52 10.25 11.56 11.56 1331 1331 15.24 15.24 17.25 17.25 19.6 19.6 21.65 21.65 23.88  192 216 240 264 290.5 315 336 360 366.25 384 390 408.5  11.32 12.21 13.80 14.66 16.22 16.98 18.11 19.06 19.36 20.55 20.85 21.47  Time (h) CM80 0 0 27 1.63 51.5 2.42 51.5 2.42 72 3.73 96 5.06 120 5.73 120 5.73 144 7.21 168 8.53 168 8.53 192.75 10.6 217 11.75 217 11.75 240 13.82 264 15.04 264 15.04 288 17.12 312 18.26 312 18.26 336.5 20.63 362 21.86 362 21.86 386 24.37 25.46 408 408^, 25.46  360.5 387.25 412  19.16 20.01 21.53  199 Data for Figure 52 continued 23.88 382 24.96 26.21 26 406 26 26.21 406 28.3 28.53 430 28.3 28.53 430 Data for Figure 53 CM80 Time (h) 0 0 16.929 48 21.629 54 22.829 72 22.829 72 24.039 78 30.699 96 97 30.699 32.434 102 38.914 121.5 41514 146.5 146.75 41.514 47.694 170 47.694 170 60.904 192 64.914 216 64.914 216 74.303 240 79.094 264 79.094 264 87.046 290.5 89.774 315 89.774 315 96.739 336 360 99.703 360 99.703 101.021 366.25 106.656 384 108.128 390 109.153 408.5 408.5 109.153  Time (h) 0 72 123 168 216 267 313 361 409  432  27.71  Cytodex 0 9.9324 19.5638 30.1068 433213 63.2213 75.7213 87.2213 106.656  Time (h) 0 6 24 24 30 72 123 168 216 267 313 361 409  CG 0 2.2 4.7 4.7 6.229 9.9324 19.5638 33.9233 47.1378 55.7378 65.7378 73.2378 813378  200 Data for Figure 54 Time P5 P5 (h) 0 0.000 0.000 23.5 0.785 0.725 47 1.595 1.445 68.5 2.230 2.060 92.25 2.680 2.410 116.5 3.023 2.695 117 3.023 2.695 141.5 3.858 3.380 165.5 4.793 4.113 191 5.873 5.050 213 6375 5.645 214 6.375 5.645 236.5 7.030 6.282 260.75 7.825 7.087 285 8.425 7.557 309 8.788 7.987 309.5 8.788 7.987 334 9342 8.467 359.5 9.998 9.107 381 10.748 9.982 405 11.573 10.702 429 12.793 11.977 453.5 13.843 13.067 477.5 14.853 14.092 503.5 16.908 16.317 528.75 18.233 17.697 548.5 20.103 19347 573 21.523 20.772 Data for Figure 55 Time PS P5 (h) 0 0.000 0.000 22 0.290 0320 475 0.455 0.730 71.5 0.745 0.990 95.5 1.250 1.420 118 1515 1.790 118 1.515 1.790 142 1.930 2.280 142 1.930 2.280 166.5 2355 2.910 166.5 2355 2.910 190 2.735 3510 190 2.735 3.510 216 3.415 4.240 216 3.415 4.240 240.5 4.245 5.240 240.5 4.245 5.240 264 5.295 6.100  P25 0.000 0.375 0.725 1.165 1.575 2.050 2.050 2.650 3.445 4.460 5.105 5.105 5.760 6.680 7.250 7.710 7.710 8.415 9.515 10.410 11.965 13.400 15.212 16350 18.215 19.535 21.155 22535  P25  Time (h) 0.000 0 0.285 23.5 0.645 47 1.075 68.5 1.490 92.25 1.910 116.5 1.910 117 2.460 141.5 3.273 165.5 4.315 191 4.915 213 4.915 214 5.645 236.5 6.475 260.75 285 309 309.5 334 359.5 381 405 411 429 453.5 477.5 503.5 528.5 548.5 573  P40 0.000 0.560 1.200 1.735 2.115 2.480 2.480 3.100 3.900 4.950 5.530 5.530 6.220 6.970 7.575 7.965 7.965 8.550 9.205 10.030 10.748 10.975 11.845 12.925 13.800 15.330 16.650 17.955 19395  P40  P80  P80  0.000 0.000 0.000 0.565 0.140 0.355 1.225 0.235 0.780 1.815 0.480 1.335 2.235 0.570 1.745 2.520 0.700 2.125 2.520 0.700 2.125 3.080 0.950 2.640 3.893 0.950 3.255 4.710 0.920 4.215 5.335 0.950 5.010 5335 0.950 5.010 5.900 1.050 5.595 6.630 1.520 6.395 7.105 2.025 6.940 7.500 2.450 7.395 7.500 2.450 7.395 7.960 3.050 7.850 8.660 3.810 8.745 9.400 4.510 9.705 10.095 5.157 10590 10370 5.345 10.847 11.290 6305 11.792 12340 7.280 12.972 13.205 8.120 14.107 14.720 9.735 15.837 15.980 10.995 17.257 17.485 12.415 18.802 18.875 13.860 20.317  P25  P25  P40  P40  P80  P80  0.000 0355 0.835 1.085 1.535 1.785 1.785 2.135 2.135 2.425 2.425 2.555 2.555 2.755 2.755 2.945 2.945 3.065  0.000 0360 0.830 1.160 1.470 1.720 1.720 2.000 2.000 2.160 2.160 2.470 2.470 2.650 2.650 3.020 3.020 3.260  0.000 0.320 0.630 0.930 1.170 1.270 1.270 1.710 1.710 2.070 2.070 2310 2.310 2.510 2.510 3.180 3.180 3.750  0.000 0.280 0.710 1.060 1.470 1.860 1.860 2.390 2390 3.100 3.100 3.890 3.890 4.510 4.510 5.760 5.760 6.770  0.000 0.345 0.805 1.255 1.685 2.020 2.020 2.885 2.885 3.645 3.645 4325 4.325 5.205 5.205 6.215 6.215 7.415  0.000 0350 0.760 1.220 1.650 2.070 2.070 2.480 2.480 3.130 3.130 3.820 3.820 4.400 4.400 5.310 5.310 6.170  201 Data for Figure 55 continued Time P5 P25 PS (h) 5.295 6.100 3.065 264 6.095 6.690 3.225 287 287 6.095 6.690 3.225 312 6.915 7.730 3.325 6.915 7.730 3.325 312 336 8.005 8.640 3.475 336 8.005 8.640 3.475 360 8.835 9520 3.565 8.835 9.520 3.565 360 387.5 9.865 10.890 3.705 387.5 9.865 10.890 3.705 412.5 10.075 11.890 3.855 412.5 10.075 11.890 3.855 431.5 11.185 12.690 3.945 Data for Figure 56 Time P5 P5 (h) 0 0.000 0.000 23 0.390 0.390 485 0.620 0.740 71 1.110 1.060 95 1.650 1.450 121 2340 2.010 2340 2.010 121 148 3.050 2.420 167 3.710 3.040 191 4.290 3.720 191 4.290 3.720 215 5.030 4.570 239 5.740 5.160 239 5.740 5.160 263 6390 5.440 290 7.180 5.770 290 7.180 5.770 314 7.960 6.080 335 8580 6.400 335 8.580 6.400 359 9.310 6.670 383.5 9.970 6.980 3835 9.970 6.980 407 10.810 7.440 431 11.420 7.830 11.420 7.830 431 455.5 12.290 8.280 481.5 13.110 8.870 4815 13.110 8.870 503 13.790 9.250  P25 0.000 0.285 0585 1.095 1.525 2205 2.205 2.965 3.555 3.985 3.985 4.785 5.575 5.575 6.275 7.115 7.115 7.935 8.555 8.555 9.215 9.965 9.965 10.805 11.505 11.505 12335 13.265 13265 13.915  P25  P40  P40  P80  P80  3.260 3.520 3.520 3.710 3.710 3.780 3.780 4.010 4.010 4.290 4.290 4.710 4.710 4.950  3.750 3.980 3.980 4.710 4.710 5.410 5.410 6.120 6.120 7.100 7.100 8.400 8.400 9.060  6.770 7.580. 7.580 8.840 8.840 10.250 10.250 11.200 11.200 12.760 12.760 13.600 13.600 14.410  7.415 8.235 8.235 9.725 9.725 11.255 11.255 12.115 12.115 13.695 13.695 14.745 14.745 15.625  6.170 6.810 6.810 7.810 7.810 8.910 8.910 9.890 9.890 11.310 11310 12.240 12.240 13.020  P25  P40  P40  P80  P80  0.000 0.440 0.750 1390 2.130 2.910 2.910 4.150 4.850 5.310 5.310 6.420 7.090 7.090 7.760 8590 8.590 9.510 10.080 10.080 10.940 11590 11.590 12520 13.170 13.170 14.150 15.020 15.020 15.860  0.000 0510 0.820 1.460 2.090 2.860 2.860 3.940 4.660 5.160 5.160 6.080 6.800 6.800 7.640 8.430 8.430 9.030 9.730 9.730 10.400 11.080 11.080 11.870 12.470 12.470 13.380 14.220 14.220 14.880  0.000 0.460 0.860 1.410 2.090 2.770 2.770 3.800 4.480 5.020 5.020 5.930 6.660 6.660 7.400 8.170 8.170 8.850 9570 9.570 10.230 10.880 10.880 11.650 12.290 12290 13.180 14.070 14.070 14.860  0.000 0.485 0.915 1565 2.095 2.685 2.685 3.655 4295 4.855 4.855 5.775 6.605 6.605 7.375 8315 8315 9.075 9.805 9.805 10.715 11565 11.565 12.645 13.345 13345 14.355 15.435 15.435 16.405  0.000 0.600 0.990 1.560 2.180 2.750 2.750 3.780 4500 5.000 5.000 5.990 6.750 6.750 7550 8.440 8.440 9.310 10.100 10.100 11.000 11.880 11.880 12.890 13.630 13.630 14.770 15.770 15.770 16.760  202 Data for Figure 57 P80 P25 P40 P40 P80 Time P5 P5 P25 01 ) 0 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 23 0.216 0.238 0.250 0.248 0.249 0.293 0.264 0.304 48.5 0.455 0.501 0.526 0.585 0.601 0.664 0.706 0.784 1.129 1.129 1.147 71 0.826 0.781 0.865 1.058 1.011 1.649 1.483 1.530 95 1.243 1.066 1.254 1.608 1.551 1.743 1.639 1.340 1.698 2.108 2.025 2.085 1.801 121 1.743 1.639 1.340 1.698 2.108 2.025 2.085 1.801 121 2.269 1.740 2.378 3.137 2.882 2.939 2.553 2.507 148 167 2.740 2.152 2.782 3.625 3.453 3.423 3.053 3.075 3.035 2.547 3.261 3.922 3.755 3.865 3.458 3.456 191 191 3.035 2.547 3.261 3.922 3.755 3.865 3.458 3.456 215 3.544 3.058 3.933 4.731 4.509 4.561 4.140 4.251 239 4.064 3.496 4.605 5.236 5.139 5.028 4.744 4.704 4.064 3.496 4.605 5.236 5.139 5.028 4.744 4.704 239 263 4.550 3.672 5.146 5.885 5.691 5.594 5.318 5.288 290 5.019 3.916 5.721 6.368 6.167 6.045 5.919 5.858 290 5.019 3.916 5.721 6.368 6.167 6.045 5.919 5.858 5.598 4.165 6.357 7.009 6.735 6.616 6.564 6.566 314 335 6.019 4372 6.807 7.425 7.081 7.013 6.999 7.014 335 6.019 4.372 6.807 7.425 7.081 7.013 6.999 7.014 359 6.596 4.601 7.370 8.117 7.696 7.595 7.759 7.795 3835 7.150 4.888 7.938 8.605 8.137 8.043 8344 8.406 3835 7.150 4.888 7.938 8.605 8.137 8.043 8.344 8.406 407 7.728 5.222 8.568 9.332 8.687 8.613 9.173 9.178 431 8.238 5.545 9.108 9.852 9.111 9.083 9.701 9.703 431 8.238 5.545 9.108 9.852 9.111 9.083 9.701 9.703 455.5 8.906 5.884 9.738 10.617 9.723 9.712 10.463 10577 4815 9.417 6.269 10.279 11.104 10.224 10.205 11.132 11.145 4815 9.417 6.269 10.279 11.104 10.224 10.205 11.132 11.145 503 9.955 6.604 10.818 11.791 10.757 10.806 11.958 11.964 Data for Figure 60 Time Level Level #1 #1 0 0 0 21 0.13 0.09 46.75 0.16 0.25 69.5 0.44 0.6 92 0.44 0.83 92 0.76 0.83 116.75 124 1.41 141.5 2.01 2.07 141.5 2.01 2.07 165.5 2.59 2.88 165.5 2.59 2.88 189 3.62 3.75 189 3.62 3.75 214 4.77 5.1 214 4.77 5.1 238.5 6.03 5.78  Level #2  Level #2  0  0  0.3 0.57 1.03 1.62 1.62 2.41 3.45 3.45 4.82 4.82 6.21 6.21 7.67 7.67 9.34  0.22 0.42 0.99 1.52 1.52 234 3.31 3.31 4.64 4.64 6.11 6.11 7.62 7.62 9.27  Level #3 0 0.37 0.78 1.5 2.24 2.24 331 4.49 4.49 5.86 5.86 7.47 7.47 9.17 9.17 10.94  Level #3 0 0.16 0.46 1.09 1.55 1.55 2.7 3.65 3.65 4.68 4.68 6.14 6.14 7.91 7.91 9.36  Level #4  Level #4  0  0  0.26 0.64 1.29 1.96 1.96 3.18 4.07 4.07 5.47 5.47 6.91 6.91 8.29 8.29 8.7  0.17 0.5 1.32 2.1 2.1 3.68 4.69 4.69 6.1 6.1 7.92 7.92 9.94 9.94 11.98  203 Data for Figure 60 continued Time Level Level Level #1 #1 (h) #2 5.78 9.34 6.03 238.5 7.68 7.36 11.4 261 7.68 7.36 11.4 261 9.4 9.01 13.9 285 9.4 9.01 13.9 285 11.36 11.24 15.79 309 309 11.36 11.24 15.79 17.2 333.5 12.89 12.77 333.5 12.89 12.77 17.2 356.5 14.56 14.54 19.14 356.5 14.56 14.54 19.14 381 16.33 16.4 20.76 381 1633 16.4 20.76 404.5 17.79 18.23 22.44 404.5 17.79 18.23 22.44 428 19.77 20.21 24.35 428 19.77 20.21 2435 452 21.65 22.29 26.01 452 21.65 22.29 26.01 476 23.52 24.29 27.86 476 23.52 24.29 27.86 25.51 25.97 29.96 500 500 25.51 25.97 29.96 522.5 27.41 27.47 31.8 522.5 27.41 27.47 31.8 550.5 29.57 2937 34.12 550.5 29.57 29.37 34.12 574.5 31.86 31.2 36.42 574.5 31.86 31.2 36.42 597 33.53 32.81 38.23 Data for Figure 61 Time Level Level (h) #1 #1 0 0 0 21 0.26 0.11 46.75 0.2 0.2 69.5 0.4 0.32 92 0.41 0.24 116.75 0.71 033 141.5 1.02 0.68 141.5 1.02 0.68 165.5 1.22 0.71 165.5 122 0.71 189 1.64 1.06 189 1.64 1.06 214 1.8 1.17 214 1.8 1.17 238.5 2.46 1.48 238.5 2.46 1.48  Level #2 0 0.25 031 057 0.82 133 1.95 1.95 2.45 2.45 3.56 3.56 4.67 4.67 5.88 5.88  Level #2 9.27 10.88 10.88 12.63 12.63 1456 14.56 15.99 15.99 17.98 17.98 19.89 19.89 21.8 21.8 23.88 23.88 26.39  Level #3 10.94 12.96 12.96 14.91 14.91 16.77 16.77 1839 18.39 20.57 20.57 22.87 22.87 24.88 24.88 26.98 26.98 28.85 28.85 31.79 31.79 33.46 33.46 35.37 35.37 37.29 37.29 39.09 39.09 40.58  Level Level Level #4 #3 #4 9.36 8.7 11.98 10.9 10.41 14.05 14.05 10.9 10.41 12.58 12.31 16.24 12.58 12.31 16.24 14.4 13.95 18.62 14.4 13.95 18.62 16.88 15.24 20.6 16.88 15.24 20.6 18.25 17.15 22.66 17.15 22.66 18.25 19.64 19.01 24.48 19.64 19.01 24.48 21.2 20.55 26.34 21.2 20.55 26.34 22.71 22.56 28.16 22.71 22.56 28.16 24.36 25.89 30.05 24.36 25.89 30.05 26.66 27.86 31.97 26.66 27.86 31.97 29.03 30.36 33.52 29.03 30.36 33.52 30.49 32 34.9 30.49 32 34.9 31.93 34.07 36.95 31.93 34.07 36.95 33.61 36.26 38.71 33.61 36.26_ 38.71 35 38.24 40.18  Level #2  Level #3  Level #3  Level #4  Level #4  0  0  0  0  0  0.13 0.21 0.5 0.62 1.07 1.61 1.61 2 2 2.62 2.62 3.46 3.46 4.73 4.73  0.26 0.46 0.8 1.04 1.37 1.89 1.89 2.06 2.06 2.74 2.74 3.31 3.31 4.34 4.34  0.23 0.46 0.86 1.09 1.61 2.4 2.4 3.15 3.15 4.32 4.32 5.63 5.63 7.51 7.51  0.25 0.42 0.68 0.89 1.33 1.96 1.96 2.21 2.21 3.1 3.1 3.81 3.81 5.08 5.08  0.3 0.5 0.81 1.02 2.06 2.13 2.13 2.62 2.62 3.64 3.64 4.39 4.39 5.37 5.37  204 Data for Figure 61 continued Time Level Level Level #1 #2 (It) #1 1.83 7.07 261 2.94 1.83 7.07 261 2.94 2.44 8.58 285 3.83 8.58 285 3.83 2.44 10.72 309 4.67 2.97 2.97 10.72 309 4.67 5.69 3.58 12.41 333.5 5.69 3.58 12.41 333.5 14.17 356.5 6.86 4.75 356.5 6.86 4.75 14.17 538 16.02 381 8.06 381 8.06 538 16.02 404.5 9.26 6.27 17.7 404.5 9.26 6.27 17.7 19.6 428 10.95 7.69 428 10.95 7.69 19.6 452 12.33 9.25 21.61 452 12.33 9.25 21.61 476 14.46 10.9 23.54 14.46 10.9 23.54 476 500 16.52 12.53 25.82 500 16.52 12.53 25.82 5225 18.36 14.04 27.97 522.5 1836 14.04 27.97 550.5 20.59 15.73 3031 550.5 20.59 15.73 3031 574.5 22.95 17.78 32.61 5743 22.95 17.78 32.61 597 24.92 19.74 34.24 Data for Figure 62 Time Level Level #1 #1  Level #2  Level #2 5.81 5.81 7.46 7.46 8.52 8.52 9.93 9.93 12.03 12.03 13.95 12.17 13.94 13.94 16.05 16.05 18.28 18.28 21.77 21.77 23.25 23.25 24.85 24.85 26.54 26.54 28.26 28.26 29.8  Level #3 4.86 4.86 5.94 5.94 7.13 7.13 8.68 8.68 10.61 10.61 12.92 12.92 153 153 17.84 17.84 20.22 20.22 22.8 22.8 25.06 25.06 27.14 27.14 29.43 29.43 31.46 31.46 33.28  Level #3 9.15 9.15 11.3 11.3 14.09 14.09 16.24 1624 18.36 18.36 21.88 21.88 24.13 24.13 26.36 2636 28.29 28.29 30.33 3033 33.27 33.27 35.49 35.49 3734 37.54 39.85 39.85 41.41  Level #4 6.41 6.41 8.16 8.16 10.42 10.42 12.04 12.04 13.83 13.83 15.44 15.44 16.92 16.92 19.17 19.17 21.13 21.13 23.45 23.45 25.53 25.53 27.54 27.54 29.73 29.73 32.15 32.15 34.13  Level #4 6.52 6.52 7.65 7.65 9.71 9.71 11.55 11.55 13.32 13.32 15.46 15.46 17.24 17.24 19.1 19.1 21.25 21.25 24.4 24.4 26.34 2634 28.25 28.25 30.11 30.11 31.87 31.87 33.17  Level #2  Level #3  Level #3  Level #4  Level #4  0  0  0  0  0  0  0  0  0  21 46.75 69.5 92 92 116.75 141.5 141.5 165.5 165.5 189 189 214 214 238.5  0.034 0.134 0.33 0.753 0.753 1.06 1.69 1.69 2.31 2.31 3.01 3.01 4.01 4.01 4.686  0.104 0.347 0.717 1.457 1.457 2.071 2.907 2.907 4.067 4.067 5.127 5.127 6.325 6.325 7.348  0.198 0.617 1.153 2.015 2.015 2.623 3.706 3.706 4.89 4.89 6.22 6.22 7.619 7.619 8.86  0.149 0.485 0.967 1.832 1.832 2.609 3.448 3.448 4.674 4.674 5.884 5.884 7.17 7.17 8.497  0.052 0.16 0364 0.779 0.779 1.08 1.715 1.715 2.454 2.454 3.06 3.06 4.063 4.063 4.433  0.12 0.339 0.729 1.416 1.416 2.021 2.834 2.834 3.977 3.977 5.156 5.156 6.243 6.243 7.469  0.131 0.381 0.779 1.44 1.44 2302 3.038 3.038 3.888 3.888 5.09 5.09 65 6.5 7.651  0.142 0.46 1.032 1.977 1.977 3.186 3.982 3.982 5.174 5.174 5.283 6.647 6.735 6.735 8.323  205  Data for Figure 62 continued Time Level Level Level #2 (II) #1 #1 8.86 238.5 4.686 7.348 6.067 9.074 10.614 261 261 6.067 9.074 10.614 285 7.565 11.41 12.456 285 7.565 11.41 12.456 309 9.964 12.898 14.03 309 9.964 12.898 14.03 333.5 11.303 14.195 15.557 333.5 11.303 14.195 15.557 356.5 12.772 15.982 17.756 356.5 12.772 15.982 17.756 381 14.481 17.094 19.245 381 14.481 17.094 19.245 404.5 15.829 18.848 21.396 404.5 15.829 18.848 21.396 428 17.504 20.658 23.705 428 17.504 20.658 23.705 452 19.19 21.879 25.098 452 19.19 21.879 25.098 20.764 23.241 27.344 476 20.764 23.241 27344 476 500 22.594 25.11 28.738 500 22.594 25.11 28.738 5225 24.713 27273 30.946 522.5 24.713 27.273 30.946 550.5 26.457 .29.035 32.526 5505 26.457 29.035 32526 574.5 28.466 31.073 33.744 574.5 28.466 31.073 33.744 597 29.864 32585 35.157 Data for Figure 63 Time Level Level #1 #1 0 0 0 21 0.059 0.024 46.75 0.091 0.104 69S 0.198 0.158 92 0.366 0.259 116.75 0.498 0.304 1413 0.77 0.513 141.5 0.77 0.513 165.5 0.99 0.671 165.5 0.99 0.671 189 1.33 0.868 0.868 189 1.33 214 1.68 0.945 214 1.678 0.945 238.5 1.994 1.129  Level #2 0 0.064 0.191 0.404 0.724 1.005 1.55 1.55 2.13 2.13 2.9 2.9 4.2 4.196 4.666  Level #2 8.497 10.079 10.079 11.837 11.837 13.145 13.145 14.448 14.448 16359 16.359 17.772 17.772 19.257 19.257 21.367 21367 23.715 23.715 25.329 25.329 27.438 27.438 29.038 29.038 30.717 30.717 32.666 32.666 34222  Level #3 4.433 5.81 5.81 7.305 7.305 9.045 9.045 10.527 10.527 12293 12.293 13.899 13.899 15.818 15.818 17.639 17.639 19.278 19.278 20.877 20.877 22.482 22.482 24.016 24.016 25.451 25.451 27.118 27.118 28398  Level #3 7.469 8.987 8.987 10.597 10.597 12.092 12.092 13.423 13.423 15376 15376 17.546 17.546 19.62 19.62 21.458 21.458 23.594  Level #4 7.651 9.053 9.053 10.598 10.598 11.933 11.933 14.065 14.065 15346 15346 16.744 16.744 18.292 18.292 19.575 19.575 20.802 20.802 22.731 22.731 24543 24543 25.912 25.912 27.205 27.205 28.66 28.66 29.844  Level #4 8.323 10.245 10.245 12.22 12.22 14.185 14.185 16.088 16.088 18.058 18.058 19.35 1935 21.325 21.325 22.838 22.838 24.296 24.296 25.973 25.973 27.488 27.488 28.925 28.925 30313 30313 31.977 31.977 33217  Level #2 0 0.057 0.199 037 0.659 0.849 1.306 1.306 1.546 1.546 2.097 2.097 2.683 2.683 3.557  Level #3 0 0.124 0.352 0.619 0.964 1.17 1.671 1.671 2.032 2.032 2.432 2.432 2.975 2.975 3.669  Level #3 0 0.116 0.332 0397 1.067 1.235 1.977 1.977 2.759 2.759 3.567 3.567 4.578 4.578 5.788  Level #4 0 0.072 0.227 0388 0.802 1.008 1.621 1.621 1.997 1.997 2.574 2.574 3.164 3.164 4.075  Level #4 0 0.094 0.253 0519 0.881 1.076 1.619 1.619 2.161 2.161 2.761 2.761 3.36 3.36 3.934  206 Data for Figure 63 continued... Time Level Level Level #1 #2 #1 (11) 238.5 1.994 1.129 4.666 261 2.544 1.461 5.712 2.544 1.461 5.712 261 3.15 1.876 7.115 285 1.876 7.115 3.15 285 3.803 2.375 8.754 309 3.803 2.375 8.754 309 333.5 4.669 2.691 10346 333.5 4.669 2.691 10.346 356.5 5.451 4.667 11.844 356.5 5.451 4.667 11.844 6.407 6.054 13.33 381 381 6.407 6.054 13.33 6.951 14.889 404.5 7.58 6.951 14.889 404.5 7.58 8.796 8.016 16.337 428 8.796 8.016 16.337 428 452 9.984 9.275 18.002 452 9.984 9.275 18.002 476 11.596 10.439 19.578 476 11.596 10.439 19.578 500 13.233 12.087 21.581 500 13.233 12.087 21.581 522.5 14.778 13.648 23.617 5225 14.778 13.648 23.617 550.5 16.379 15.039 25.534 550.5 16.379 15.039 25.534 574.5 18.471 16.65 27.631 574.5 18.471 16.65 27.631 597 20.151 18.343 29.256  Level #2 3.557 4.381 4.381 5.823 5.823 6.768 6.768 8.068 8.068 10 10 11.511 11.511 13.152 13.152 14.767 14.767 16.484 16.484 18.818 18.818 20.197 20.197 21.922 21.922 23.199 23.199 24531 24531 26.049  Level #3 3.669 4.336 4.336 5.385 5.385 6.476 6.476 7.868 7.868 9.563 9.563 11.369 11369 13.461 13.461 15.581 15.581 17.701 17.701 19.816 19.816 21.937 21.937 24.149 24.149 26.037 26.037 27.98 27.98 29.602  Level #3 5.788 7.122 7.122 9.016 9.016 11.216 11.216 13.057 13.057 15.15 15.15 17.497 17.497 19.508 19.508 21.403 21.403 23.045 23.045 24.539 24.539 26.728 26.728 28.993 28.993 30.402 30.402 32.128 32.128 33.789  Level #4 4.075 5.173 5.173 6.723 6.723 8.488 8.488 9.916 9.916 11.717 11.717 13.055 13.055 14.43 14.43 16.215 16.215 17.796 17.796 19546 19.546 21.596 21.596 23.728 23.728 25.715 25.715 27.984 27.98429.669  Level #4 3.934 4.917 4.917 5.575 5.575 7.181 7.181 8.818 8.818 10.422 10.422 12.121 12.121 13.633 13.633 15.124 15.124 16.798 16.798 18.818 18.818 20.561 20.561 22.633 22.633 23.939 23.939 25309 25.309 26.809  Data for Figures 65-67 & 74-75  ^  ^ ^ Cytodex CG I^S40 ^ ^ Time glucose lactate glucose^unit^TF unit glucose lactate glucose unit^TF unit glucose lactate glucose unit ^TF^unit ^ ^ (h)^rate^vol. glu^vol. TF rate^vol. glu^vol. TF rate vol. glu^vol. TF  1  0 49 93 117 142 165.5 189 214 240 261.5 285 309 334 358 382 407.5 429 453 477 501 525 549.75 575 598  0.00 1.83 3.80 5.25 7.00 9.15 11.50 14.35 16.42 18.49 20.72 22.99 25.10 27.35 2956 31.74 33.76 36.04 38.47 41.10 43.87 46.43 49.02 51.36  0.00 1.23 2.57 3.65 5.10 7.05 9.23 11.49 13.18 14.75 16.51 18.32 20.18 22.01 23.75 25.39 26.89 28.43 30.31 32.24 34.13 36.49 39.08 41.45  621  53.85  44.08 2.10E-02 44.88 413.14 344.28 53.31  8.21E-03 8.21E-03 1.05E-02 1.30E-02 1.62E-02 1.92E-02 2.14E-02 1.94E-02 1.76E-02 1.91E-02 1.90E-02 1.79E-02 1.78E-02 1.86E-02 1.78E-02 1.79E-02 1.89E-02 1.96E-02 2.11E-02 2.25E-02 2.19E-02 2.06E-02 2.04E-02 2.10E-02  0.00 1.53 3.17 4.38 5.83 7.63 9.58 11.96 13.68 15.41 17.27 19.16 20.92 22.79 24.63 26.45 28.13 30.03 32.06 34.25 36.56 38.69 40.85 42.80  0.00 17.41 37.81 51.40 66.10 82.20 99.51 119.31 136.21 151.91 165.61 179.71 19751 207.31 220.81 239.61 246.84 267.04 290.24 310.74 334.54 354.04 371.74 391.24  0.00 14.51 31.51 42.83 55.08 68.50 82.93 99.43 113.51 126.59 138.01 149.76 164.59 172.76 184.01 199.68 205.70 222.53 241.87 258.95 278.78 295.03 309.78 326.03  0.00 1.69 3.83 5.57 7.42 9.70 12.03 14.85 17.00 18.80 21.07 23.44 25.67 27.98 30.06 32.20 34.17 36.11 38.45 40.84 43.22 45.64 48.25 50.80  0.00 1.22 2.58 3.79 5.34 7.42 9.68 11.96 13.86 15.21 17.05 18.89 20.75 22.58 24.32 26.01 27.53 28.95 30.74 32.59 34.32 36.42 39.03 41.71  8.21E-03 8.21E-03 1.05E-02 1.30E-02 1.62E-02 1.92E-02 2.14E-02 1.94E-02 1.76E-02 1.91E-02 1.90E-02 1.79E-02 1.78E-02 1.86E-02 1.78E-02 1.79E-02 1.89E-02 1.96E-02 2.11E-02 2.25E-02 2.19E-02 2.06E-02 2.04E-02 2.10E-02  0.00 1.53 3.17 4.38 5.83 7.63 9.58 11.96 13.68 15.41 17.27 19.16 20.92 22.79 24.63 26.45 28.13 30.03 32.06 34.25 36.56 38.69 40.85 42.80  0.00 17.30 4730 67.00 86.10 105.60 125.00 142.40 160.60 172.80 187.90 201.90 216.60 228.10 242.30 257.40 269.60 286.90 310.70 335.60 356.10 377.30 397.00 417.70  0.00 14.51 31.51 42.83 55.08 68.50 82.93 99.43 113.51 126.59 138.01 149.76 16459 172.76 184.01 199.68 205.70 222.53 241.87 258.95 278.78 295.03 309.78 326.03  0.00 1.70 3.56 4.92 636 8.20 10.29 12.73 14.80 16.74 18.93 21.28 23.46 25.69 27.74 29.86 31.72 33.70 35.94 38.36 40.94 43.35 45.89 48.35  44.27 2.10E-02 44.88 439.90 34428 51.00  0.00 1.23 2.50 3.55 4.92 6.52 8.41 10.40 12.13 13.60 15.36 17.20 19.06 20.81 22.48 24.16 25.60 27.04 28.80 30.71 32.52 34.50 37.05 39.58  8.21E-03 8.21E-03 1.05E-02 1.30E-02 1.62E-02 1.92E-02 2.14E-02 1.94E-02 1.76E-02 1.91E-02 1.90E-02 1.79E-02 1.78E-02 1.86E-02 1.78E-02 1.79E-02 1.89E-02 1.96E-02 2.11E-02 2.25E-02 2.19E-02 2.06E-02 2.04E-02 2.10E-02  0.00 1.53 3.17 4.38 5.83 7.63 9.58 11.96 13.68 15.41 17.27 19.16 20.92 22.79 24.63 26.45 28.13 30.03 32.06 34.25 36.56 38.69 40.85 42.80  0.00 17.80 42.80 58.30 73.60 88.80 105.60 125.30 140.70 156.00 170.40 183.10 195.10 208.23 220.43 232.83 247.53 266.43 288.83 311.83 331.53 350.43 366.93 387.93  0.00 14.51 3151 42.83 55.08 68.50 82.93 99.43 113.51 126.59 138.01 149.76 16459 172.76 184.01 199.68 205.70 222.53 241.87 258.95 278.78 295.03 309.78 326.03  42.22 2.10E-02 44.88 406.93 344.28  207  208 Data for Figure 71 Time (h) 123.5 144 168 241 336 384 480 576.5 672 744 840 Data for Figure 71 Time (h) 0 24 48.5 74 98 98.5 120 144 168 192 192.5 216 2423 266.5  Cell Concentration (x 10 -7 ) 0.51 1.95 2.31 2.34 2.41 3.21 3.02 2.83 1.91 2.37 2.68  Glucose Concentration WO 3.71 3.58 2.84 1.54 0.82 4.11 3.73 3.17 1.91 1.37 4.21 3.22 1.68 0.69  Data for Figure 72 Time (h) Total^glucose Antibody used (g/L) production rate Ong,a1 10 0 0 -24 0.12 -48 031 -72 -0.76 99 2.01 -112 2.72 -112.1 -2.72 -144 2.84 168.5 3.09 -192.5 3.68 --216.5 4.64 -240.5 5.6 -266 5.84 266.1 5.84 -291 6.35 --  Cell leakage (x 10 -4 ) 4.3 12 35 86 77 77 6 22 62 10.1 10.1 17.4 70 50  209 Data for Figure 72 continued 336^8.82 384^9.07 389^9.07 408^11.09 408.1^11.09 435^13.66^0.63197 435.1^13.66 458.5^16.24^0.68376 458.6^16.24 480^18.3 504.5^19.4^0.54466 504.6^19.4 528.5^19.92 552.5^21.08 576.5^22.42^0.38943 576.6^22.42 603.5^23.83 627.5^25.28^0.64833 627.6^25.28 648.5^27 672.5^28.49^0.89087 672.6^28.49 696.5^30.53 720.5^31.73^0.73069 720.6^31.73 744^33.56 771.5^34.73^0.82515 771.6^34.73 794^36.8 Data for Figure 73 Time (h)^Glucose^Lactate Transferrin^Cell concentration /......_ concentratioiEgaCszj i^concentratioin centratioin m L^x10-5/ mLj 0 2.98 0.21 0.00^11.5 17 2.76 0.37 0.74 13.3 24 2.54 0.43 1.03 13.6 41 2.49 0.54 1.19 15.6 48 2.41 0.81 1.21 21.1 65 1.09 2.06 2.55 24.1 72 1.9 1.12 5.30 33.1 90 1.43 1.59 8.80 40.6 102 1.13 1.81 10.90 50 116 0.76 2.21 13.60 65 124 0.53 2.31 1530 71 137 0.3 2.59 18.20 103 144 0.22 2.63 19.2 102 161 0.19 2.81 19.5 101 168 0.16 2.82 20.3 96 185 0.15 2.82 20.2 49.6 193 0.14 2.82 20 72.8 209 0.14 2.83 20 24 216 0.13 2.83 20 25  210 Data for Figure 76 Time (10 S40 .0 0 0.775 24 0.775 24 1.985 48 2.865 72 2.865 72 95 4.045 5.235 117.5 5.235 117.5 142 7.195 7.195 142 8.325 166.5 8.325 166.5 190.5 9.885 190.5 9.885 216 10.655 10.655 216 240 12.335 12.335 240 12.905 262 264 12.905 14.615 285.5 14.615 285.5 15.245 310 15.245 310  Data for Figure 77 S40 Time (h) 0 0 9.9 24 22.55 72 41.45 117.5 142 166.5 61 190.5 77.3 216 240 89.6 262 285.5 103.05 310 334 117.8 357.5  Suspension 0 0.78125 0.78125 1.83125 2.74125 2.74125 4.24125 5.43125 5.43125 7.76125 7.76125 10.25125 10.25125 12.40125 12.40125 14.75125 14.75125 16.86125 16.86125 18.94125 18.94125 21.51125 21.51125 23.90125 23.90125  roller 0 0.53125 0.53125 1.25125 1.67125 1.67125 2.40125 3.01125 3.01125 3.01125 3.01125 3.63125 3.63125 4.29125 4.29125 5.04125 5.04125 6.19125 6.19125 6.99125 6.99125 8.13125 8.13125 8.88125 8.88125  Suspension 0 5.4 16.5  roller 0 6.8 24.2 47.8 68.05 90.1 116.1 151.05 191.45 230.65 276.15 325.4 378.4 436.5  37.95 51.6 71.6 95.25 119.95  211 Data for Figure 78 S40 Time 0 0.848 24 0.848 72 0.848 117.5 0.848 142 0.848 166.5 0.848 190.5 0.848 216 0.848 240 0.848 262 0.848 285.5 0.848 310 0.848 Data for Figure 79 Time (It) S40 0 10.29 24 6.57 72 5.91 117.5 142 166.5 7.89 190.5 216 1025 240 262 10.93 285.5 310 11.21 334 3575 10.43  Suspension 0.836 0.821 0.767 0.691 0.629 0.577 0.555 0.555 0.555 0.555 0.555 0.555  roller 0.940 0.940 0.940 0.940 0.940 0.940 0.940 0.940 0.940 0.940 0.940 0.940  Suspension 10.13 9.86 10.55  roller 8.71 8.83 8.82 8.74 8.78 10.32 13.57 16.86 18.99 18.20 19.09 21.17 22.13 2251  10.45 10.03 11.36 11.57 10.82  4B/  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0058547/manifest

Comment

Related Items