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


831-ubc_1993_spring_phd_lee_daniel.pdf [ 11.72MB ]
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

Full Text

HIGH DENSITY ANIMAL CELL CULTURESYSTEMS USING POROUS SUPPORTSbyDaniel W. LeeB.A.Sc., The University of British Columbia, 1987M.A.Sc., The University of Waterloo, 1989A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THEREQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHYinTHE FACULTY OF GRADUATE STUDIESDEPARTMENT OF CHEMICAL ENGINEERINGWe accept this thesis as conforming to the required standardTHE UNIVERSITY OF BRITISH COLUMBIAOctober, 1992© Daniel W. Lee, 1992In presenting the thesis in partial fulfillment of the requirements for an advanced degreeat the university of British Columbia. I agree that the Library shall make freelyavailable for reference and study. I further agree that permission for extensive copyingof this thesis for scholarly purposes may be granted by the head of my department or byhis or her representatives. It is understood that copying or publication of this thesis forfinancial gain shall not be allowed without my written permissionDepartment of Chemical EngineeringThe University of British Columbia,Vancouver, CanadaDec. 28th, 1992DE-6 (2/88)iiABSTRACTMonolithic porous ceramic and porous polystyrene microcarriers were examinedas supports for large scale mammalian cell culture. Vero cells and transformed babyhamster kidney (BHK) cells which produced human transferrin were grown in threedifferent reactor configurations: a fixed bed ceramic perfusion system, an airlift systemwith draft tube made of porous ceramic and a stirred tank configuration which usedporous polystyrene microcarriers. The porous matrices provide an increased surfacearea for cell attachment and growth and protect the entrapped or immobilized cells fromshear stress in the bulk fluid. Steady state cell concentrations in all three systems werefound to be in excess of 10 8 cells per mL porous matrix. All three systems intensifiedthe culture process by increased cell mass per unit volume — a minimum five—foldincrease in reactor volumetric cell density over simple suspension cultures. The airliftand microcarrier stirred tank system offer the potential to scale—up by increasing reactorvolume. The fixed bed ceramic perfusion system can be used as a cell propagator toproduce the required inoculum for other large scale bioreactors.Unlike the Opticore of the Opticell systems, the ceramic foam element can bereused. Its multiple interconnected channel structure greatly reduces the possibility ofchannel blockage due to over—grown cells. The biologically inert porous polystyrenemicrocarriers tested have distinct advantages, in terms of product purification, over thecollagen based porous microcarriers such as Cultispher—G and Informatrixmicrocarriers. Cell attachment rates onto the porous polystyrene microcarriers treatedwith sulphuric acid were comparable to those of Cytodex-1 and Cultispher—G particleswhile the cell growth and productivity per unit carrier volume were 20% superior to thetwo tested commercial microcarriers.The use of airlift eliminates the need for a separate oxygenator or spin filter forgas exchange. The input gas created a differential pressure drop across the porous drafttube and forced the medium to perfuse through the porous draft tube. A simple111mathematical model was formulated to describe the hydrodynamic behaviour of theporous draft tube airlift system. The model allows gas holdup, liquid superficialvelocity, and liquid perfusion rate through the porous draft tube to be predicted for agiven gas input. For the cases examined, the predictions are in satisfactory agreementwith the overall trend of experimental measurements.ivTABLE OF CONTENTSABSTRACT^ iiTABLE OF CONTENTS^ ivLIST OF TABLES ixLIST OF FIGURES^ xACKNOWLEDGMENTS xvi1.0 INTRODUCTION^ 11.1 Current Trend to Intensify Cultures^  21.2 Objectives^ 32.0 LITERATURE REVIEW 52.1 Bioreactor Design Criteria ^  52.1.1 Shear effect 52.1.2 Oxygen supply^ Bubble—free oxygenation^ Oxygen transfer using external oxygenators^ Direct sparging^ 82.2 High Density Immobilized Cell Culture Systems^ 92.2.1 Microcarrier culture systems^ Conventional systems Systems utilizing porous microcarriers^ 12(a) Verax^ 12(b) Cultispher—G microcarrier^ 13(c) Siran porous glass beads 14(d) Informatrix porous microcarriers^ 142.2.2 Microcapsulation^  152.2.3 Fixed bed bioreactor  15V2.2.4 Hollow fiber systems^ 172.3 Airlift Bioreactors^  182.3.1 Classification of airlift reactors^ 192.3.2 Applications of airlift in animal cell culture^ Cell retention in airlift reactors 202.3.3 Flow patterns^ 212.3.4 Power input 232.3.5 Gas velocity^ 232.3.6 Hydrodynamics Effect of sparger^ Dispersion characteristics^ Liquid circulating velocity, UL Gas holdup, 6g^ Mass transfer 313.0 EXPERIMENTAL MATERIALS AND METHODS^ 353.1 Cells and Cell Maintenance^ 353.2 Cell Culture Systems 353.2.1 Ceramic foam^ 353.2.1.1 Fixed bed perfusion system^ 363.2.1.2 Airlift system^ 383.2.2 Microcarriers^ 423.2.2.1 Pretreatment of microcarrier particles.^ 433.2.2.2 Microcarrier culture^ 45(a) Roller culture  45(b) Spinner cultures^ 463.2.3 Suspension cell culture 473.3 Analytical Methods^ 47vi3.3.1 Cell numeration^ 473.3.1.1 Porous matrix 473.3.1.2 Microcarriers^ 483.3.1.3 Suspension cells 483.3.2 Assays^ 483.3.2.1 Glucose^ 483.3.2.2 Lactate 483.3.2.3 Lactic dehydrogenase (LDH)^ 493.3.2.4 Transferrin^ 50(a) Enzyme—linked immunosorbent assay (ELISA) ^ 50(b) Particle concentration fluorescence immuno-assay (PCFIA)^ 503.3.3 Microscopy^ 513.3.3.1 Scanning electron microscopy (SEM) ^ 513.3.3.2 PEG embedding technique^ 513.3.3.3 Thin—sectioning microscopy 523.3.3.4 Confocal microscopy^ 534.0 RESULTS AND DISCUSSION —Fixed Bed Ceramic Foam PerfusionSystem^ 544.1 Effect of Foam Porosity^ 544.2 Cell Growth on the Ceramic Surface^ 554.3 Estimation of Total Cell Number 574.4 Lactate Production^ 604.5 Stability and Viability of Cells on the Matrix^ 624.6 Effect of Serum during Stationary Growth Phase in the PerfusionReactor^ 644.7 Effect of Zinc on Transferrin Production^ 65vii4.8 Perfusion Propagator^ 665.0 RESULTS AND DISCUSSION — Airlift System^ 695.1 Pressure Drop^ 705.2 Mathematical Modeling^ 765.3 Mass Transfer Coefficient 835.4 Scale—up Potential Assessed by Proposed Model^ 855.5 Long Term Culture^ 886.0 RESULTS AND DISCUSSION — Porous Microcarriers^ 916.1 Cell Attachment Rate^ 926.1.1 Effect of surface chemistry group modifications ^ 926.1.2 Cell attachment/entrapment rate for various microcarriers^ 976.1.3 Sulphuric acid treatment^ 986.1.4 Influence of inoculation procedure^ 996.1.5 Effect of particle diameter  1016.1.6 Influence of the inoculum cell concentration^ 1026.1.7 Effect of medium composition on cell attachment rate^ 1036.2 Long Term Microcarrier Cultures^  1046.2.1 Biomass evaluation 1056.2.2 Cell growth on Polyhipe in roller bottles^ 1086.2.2.1 Vero cell growth, effect of carrier surfacemodification^  1086.2.2.2 BHK cell growth, effect of carrier surfacemodification^  1116.2.2.3 Transferrin production^  1146.2.3 Particle pore size effect  1156.2.4 Minimum cell inoculum requirement^ 1236.2.5 Cell growth on S40 Polyhipe in spinners  127viii6.2.6 Cell penetration depth^ 1296.2.7 Hybridoma cell growth in Polyhipe particles^ 1337.0 RESULTS AND DISCUSSION —Effect of Culture Systems on Cell Growthand Cell Productivity^  1377.1 Comparison of Cell Specific Transferrin Productivity of DifferentCulture Systems^ 1377.1 Comparison of Cell Loading and Large Scale Cell Culture Suitablity^ 1438.0 CONCLUSION & RECOMMENDATIONS ^ 147NOMENCLATURE^ 151REFERENCES 154APPENDICES^ 169Appendix 1. Model PC-61 A/D board data logging program^ 169Appendix 2. Anglican controller data logging program listing  173Appendix 3. MathcadTM airlift model program listing^ 174Appendix 4 Sample Calculation ^  177Appendix 5 Raw data^ 180LIST OF TABLESTable 1. Typical desirable conditions for mammalian cell growth^ 2Table 2. Commercially available microcarriers 11Table 3. Properties of microcarrier beads tested^ 44Table 4. Surface characteristics of the Polyhipe particles^ 45Table 5. Surface characteristics of S microcarriers 98Table 6. Comparison of cell number (cells / mL beads) by alternative methods^ 107ixLIST OF FIGURESFigure 1. Schematic illustration of Opticell bioreactor^  16Figure 2. Typical airlift bioreactor configuration  18Figure 3. Airlift bioreactor configurations — (a) external loop (b) internal loopairlift^  19Figure 4. Ceramic foam cylinders of 30,50 and 100 PPI^ 36Figure 5. Schematic showing perfusion system 37Figure 6. Schematic showing airlift system^ 39Figure 7. Open structure of a macroporous microcarrier P40 particle^ 42Figure 8. Glucose utilization of BHK cells in batch culture on ceramic foamcylinders of various porosity^ 55Figure 9. SEM photographs of BHK cells grown on ceramic surface.^ 57Figure 10. Oxygen uptake rate and estimated cell number based on the oxygenuptake rate^ 59Figure 11. Glucose utilization of BHK cells in the perfusion system with 5%FCS and a perfusion rate of 128 mL,/min ^ 60Figure 12. Lactate production based on glucose utilized for BHK cells in theperfusion system.^ 61Figure 13. Correlation between measured LDH activity and disrupted BHK cellconcentration^ 62Figure 14. Effect of fetal calf serum (FCS) concentration on steady—statetransferrin production rate at a perfusion rate of 128 mL/min inDMEM medium^ 64Figure 15. Induction of transferrin production.^  66xxiFigure 16. Vero cell growth on ceramic perfusion system with repeatedharvesting. ^ 68Figure 17. Measured pressure drops across the various draft tubes^ 72Figure 18. Effect of porous draft tube length on the pressure drop across thethick wall 30 PPI porous draft tube  ^ 73Figure 19. Effect of porous draft tube length on the pressure drop across thethick wall 100 PPI porous draft tube^ 74Figure 20. Effect of serum concentration on differential pressure drops acrossthe thin wall 100 PPI porous draft tube, measured at the base of thereactor, at various riser superficial velocities^ 74Figure 21. Effect of serum concentration on differential pressure drops acrossthe thin wall 30 PPI porous draft tube, measured at the base of thereactor, at various riser superficial velocities^ 75Figure 22. Effect of serum concentration on differential pressure drops acrossthe thick wall 100 PPI porous draft tube, measured at the base of thereactor, at various riser superficial velocities^ 75Figure 23. Effect of serum concentration on differential pressure drops acrossthe thick wall 30 PPI porous draft tube, measured at the base of thereactor, at various riser superficial velocities^ 76Figure 24. Pressure drop across the thick—wall porous draft tube, measured atthe base of the reactor, versus riser superficial air velocity fordifferent pore spacings ^ 80Figure 25. Pressure drops across the thin—wall porous draft tube at the base ofthe reactor versus riser superficial air velocity for different porespacings  ^ 81Figure 26. Average downcomer liquid superficial velocity at various riser gassuperficial velocities^  82xiiFigure 27. Volumetric mass transfer coefficient at various riser gas superficialvelocities^  83Figure 28. Effect of 5% serum addition on mass transfer coefficient^ 84Figure 29. Comparison of predicted mass transfer coefficients with measuredvalues for airlift with 100 PPI thick—wall draft tube of 360 mm^ 85Figure 30. Calculated average perfusion velocity through the porous matrixversus draft tube wall thickness for the tested 5 L airlift with variousriser area, AR^ 87Figure 31. Cumulative glucose used and lactate produced by BHK cells in theporous draft—tube airlift bioreactor^ 90Figure 32. Effect of surface modifications on BHK cell attachment rate^ 94Figure 33. Effect of surface modifications on Vero cell attachment rate^ 94Figure 34. Effect of surface modification on Vero cell attachment to polystyrenemicrocarriers^ 96Figure 35. Attachment of BHK cells to treated polystyrene microcarriers^ 96Figure 36. Vero cell attachment to various microcarriers. Equal masses of eachtype of microcarriers (9 g/L) were used^ 97Figure 37. BHK cell attachment to microcarriers 98Figure 38. Attachment of BHK cells to sulphuric acid treated Polyhipemicrocarriers as a function of time following treatment length of 15,60 and 240 min^ 99Figure 39. Attachment of Vero cells to sulphuric acid treated Polyhipemicrocarriers as a function of time following treatment length of 15,60 and 240 min.^ 100Figure 40. Cell attachment to dry particles^  101Figure 41. Effect of particle diameter on the attachment rate of Vero cells toPolyhipe particles treated with sulphuric acid ^  102Figure 42. Inoculum concentration effect on the Vero cell attachment rate^ 103Figure 43. Effect of medium composition on Vero cell attachment/entrapmentrates^ 104Figure 44. Effect of medium glucose concentration on BHK cell glucose uptakerate.^  106Figure 45. Glucose concentration of BHK CM80 roller culture in fed—batchoperation^  107Figure 46. Glucose concentrations in semi—continuous perfusion of Vero cellson CM80^ 108Figure 47. Vero cells on DEA80 particles ^  109Figure 48. Cumulative glucose utilization by Vero on particles with differentmodifications ^  110Figure 49. Calculated Vero cell densities on various Polyhipe particles based onglucose utilization rates^  110Figure 50. Cumulative glucose used by BHK cells grown on 5 g/L ofmicrocarriers^  111Figure 51. Cumulative glucose uptake of BHK cells on various types ofmicrocarriers with a concentration of 2 g/L^ 112Figure 52. Microcarrier (5g/L) cumulative glucose consumption of BHK rollercultures in DMEM medium^ 114Figure 53. Cumulative transferrin produced in BHK roller cultures containing 2g microcarrier/L DMEM medium, supplemented with 10 ,uM zinc^ 115Figure 54. Total glucose used by BHK cells on polystyrene particles of different^chamber sizes    116Figure 55. Cumulative glucose used for BHK polystyrene microcarrier spinnerculture^  117Figure 56. Cumulative glucose consumed by Vero cells on P microcarriers^ 118xivFigure 57. Cumulative lactate produced by Vero cells on P microcarriers^ 119Figure 58. Confocal images of a P80 particle^ 120Figure 59. Thin sectioned microscopy images of Vero cells on polystyreneparticles^  122Figure 60. Cumulative glucose used by Vero cells on Cytodex-1 microcarriers ^ 124Figure 61. Cumulative glucose used by Vero cells on S40 microcarriers ^ 124Figure 62. Cumulative lactate produced by Vero cells on Cytodex-1microcarriers^  125Figure 63. Cumulative lactate produced by Vero cells on S40 microcarriers^ 125Figure 64. Scanning electron micrographs of Vero cells on S40 carrier after 600h culture time^ 126Figure 65. Cumulative glucose used and lactate produced by BHK cells onmicrocarriers (2 g/L) in 200 mL medium^ 127Figure 66. Glucose utilization rates of various BHK microcarrier spinnercultures ^  128Figure 67. Unit carrier cumulative glucose used by BHK cells on microcarriers ^ 129Figure 68. Scanning electron micrograph of S40 particles^  132Figure 69. Hybridoma cells growing within the pores of chloromethyl Polyhipeparticle^  135Figure 70. Enlarged view of hybridoma cells growing in a pore of achloromethyl Polyhipe particle^  135Figure 71. Growth of hybridoma cells with chloromethyl Polyhipe (5 g/L) in aroller bottle^  136Figure 72. Cumulative glucose utilization and antibody production rate ofhybridoma cells entrapped in 5 g/L of CM80 macroporous beads^ 136Figure 73. Transferrin production by BHK cells in spinner suspension culture ^ 137XVFigure 74. Total transferrin produced by BHK on microcarriers (2 g/L) in 200mL medium^ 138Figure 75. Transferrin production per unit carrier volume by BHK cells onvarious microcarriers^  139Figure 76. Cumulative glucose used by BHK cells^  140Figure 77. Cumulative transferrin produced by BHK cells from various culturesystems^ 142Figure 78. Glucose to lactate conversion ratios for various BHK cultures^ 142Figure 79. Transferrin produced per g glucose utilized for various BHK culturesystems^ 146ACKNOWLEDGMENTSFirst, I would like to thank my supervisors, Dr. Doug Kilburn and Dr. John Gracefor their useful suggestions, support and encouragement throughout this work. I wouldwish to thank my committee members, Dr. Norm Epstein and Dr. Ross MacGillivray foruseful discussions.I would also like to express my appreciation to all my friends in the twodepartments (Chemical Engineering and Microbiology) and in the BiotechnologyLaboratory, especially to Dr. James Piret and Eric Jervis, for their help and usefulsuggestions.The exceptional technical assistance provided by the technicians and staff in theDepartment of Chemical Engineering, Department of Microbiology, and BiotechnologyLaboratory, especially by Doug Haddow, Randy Dean and Gary Lesnicki, is alsogratefully acknowledged.Financial support from the B.C. Science Council under a Science and TechnologyDevelopment Fund core grant and the British Columbia Foundation for Non—animalResearch is gratefully appreciated.xvi1.0 INTRODUCTIONMany medically important pharmaceutical proteins, such as tissueplasminogen activator (tPA), cannot be produced in microbial systems byrecombinant DNA technology and can only be expressed using animal cells ashosts. Such proteins made by bacteria might not be folded in the properconfiguration; addition of sugar, phosphate or alkyl groups to the basic amino acidbackbone might not be done properly.Although cell culture has a number of similarities to microbial fermentationsused for brewing and antibiotics manufacture, there are important differences.Animal cells are much more sensitive to conditions in their environment; animalcells are characterized by their fragility due to their relatively large size (typicaldiameter about 15 ,um) and their lack of a protective cell wall. The mammalian cellbilayer membrane provides little protection against external disturbances such asturbulence in the external fluid. In addition to physical fragility, several otherlimitations of animal cells restrict their productivity in conventional large scalefermenters:(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 107cells/cm3) 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 ofbioreactors for their growth more critical. Some of these limitations can beovercome by appropriate bioreactor design.A well designed bioreactor should expose the cells to low levels of shear andprovide sufficient mass transfer and mixing so that there is an adequate chemicalcell culture environment throughout. To successfully grow animal cells inbioreactors requires that the key chemical and physical aspects of their nativeenvironment be reproduced faithfully in the bioreactors. Table 1 lists the typicalenvironmental requirements for mammalian cell growth.Table 1. Typical Environmental Conditions for Mammalian Cell GrowthTemperature^32 to 40, usually 37°C^pH^ 7.0 to7.5Fluid shear stress^< 2 N/m2Nutrients—dissolved oxygen^30 to 80% air saturation—glucose^0.1 to 4.0 g/LMetabolic products—ammonium ion^< 4 mM—lactate^not critical if pH controlledbetween 7.2-7.5Satisfying these key requirements becomes increasingly difficult as the cellconcentration increases. Process intensification to improve reactor productivityinvolves increasing the cell concentration which requires higher rates of masstransfer. In conventional bioreactor systems, the improvement in mass transfercharacteristics is achieved by increasing the agitation and gas sparging rate. Bothchanges increase the shear stress on the cells in the reactor.1.1 Current Trend to Intensify CulturesLess than a decade ago, the primary means of producing large amounts ofbiologicals using animal cells was the roller bottle. In a few cases, simple stirredvessels were used. With the increasing number of valuable therapeutic proteinsbeing produced by genetically manipulated animal cells, various bioreactors havebeen developed. Cell immobilization has proven to be one of the most effectiveways of increasing cell concentration and prolonging the protein productionperiod. It allows continuous or semi—continuous perfusion of medium withoutwashing the cells out of the reactor. During the stationary growth phase, the needfor serum and other complex medium components is reduced due to increasedconcentration of cell—derived products (growth factors, etc.), while theconcentration of product is generally increased at high cell concentration(Lydersen, 1987; Croughan et al., 1988). Some current immobilized cell culturesystems provide good mass transfer characteristics and provide physical protectionto the fragile animal cells (e.g. by encapsulation). However, most of these currentsystems suffer from overly complex operating procedures, and expensive cellimmobilization materials; most are not readily scaled up. Bioreactor efficiencymust be increased further in order to produce the therapeutic proteinseconomically.1.2 ObjectivesThe overall objective of this work was to investigate new methods for large—scale culture of animal cells at high cell concentration. Such cultures require somemethod of immobilizing the cells within the culture. The use of porous media,either as monolithic blocks or as particles, was investigated. These porousmaterials offer significant advantages over many conventional non—poroussupporting substrates. The specific objectives of the study were:(1) To design a new, high cell density bioreactor for the growth of anchoragedependent cells using fixed ceramic foam matrix.(2) To investigate the use of a new porous chemically modified polystyrenematerial as a microcarrier for cell culture.The bioreactors were designed to provide high surface to volume ratios for cellgrowth, an adequate oxygen supply without direct contact between the cells andgas—liquid interfaces and minimum shear.Three different bioreactor designs (packed bed, airlift and stirred tankconfigurations) using two different inorganic porous substrates (ceramic andpolystyrene) for cell culture were examined in this study. A further objective wasto formulate a mathematical model to facilitate scale—up of the airlift bioreactor.2.0 LITERATURE REVIEW2.1 Bioreactor Design CriteriaA well designed bioreactor for animal cell culture must satisfy manyrequirements. Two of the key design criteria for large—scale bioreactors are: (1)minimal shear induced stresses, and (2) sufficient supply of oxygen to maintainhigh cellular productivity and yield.2.1.1 Shear effectIt has been realized that shear can influence cell culture processes in variousways — 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). Shearstress can be associated with either liquid motion or gas—liquid interfaces. In theabsence of gas sparging, critical stirring speeds from 60 to 400 rpm have beenreported beyond which cell death occurred (Hirtenstein et al. , 1980; de St. Groth,1983; Telling and Radlett, 1971). The wide range of stirring speeds reflectsdifferences in reactor geometry and cell lines and emphasizes the fact thatagitation rate itself, without reference to vessel volume, impeller size and othergeometric factors, does not quantify the hydrodynamic stress. Stathopoulos andHellums (1985) found that liquid—induced shear stress above 2.6 N/m 2 caused amarked reduction in the viability of human embryonic kidney cells. At a lowershear stress (0.65 N/m 2) urokinase production was actually stimulated comparedto no shear stress controls; similar findings were reported by Frangos et al. (1985,1988). Shear stress greater than 1 to 5 N/m2 was reported to be detrimental toanimal cells in suspensions. A critical shear stress of 0.75 to 1.0 N/m 2 wasreported for adherent BHK cells (Ludwig et al., 1992).Handa et al. (1987) studied the gas—liquid interfacial effect on the viability ofhybridoma cells in bubble columns. They concluded that the survival ofhybridomas depended on (1) cell type, (2) bubble size (smaller bubbles being moreharmful), 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—to-diameter ratio). The findings of Emery et al. (1987) and Katinger and Scheirer(1982) support the hypothesis that cell death in sparged systems occurs mainly inthe region of bubble disengagement from the free liquid surface. Overall cellviability was found to increase with increasing reactor volume due to reduction inthe cell exposure time to disengaging bubbles. More recently, Handa—Corrigan etal. (1989) proposed two possible cell damage mechanisms associated with bubbledisengagement from the liquid surface in sparged systems: damage due to rapidoscillations caused by bursting bubbles, and damage due to shearing in drainingliquid film (or lamellae) in foams.Tramper et al. (1988) believe that, for cells in -suspension, shear stressesassociated with liquid motion and rising of air bubbles are less than thoseassociated with injection of air bubbles into the medium and their bursting at thesurface. Kunas and Papoutsakis (1990) showed that in the absence of a vortex andbubble entrainment, hybridoma CRL-8018 cell damage due to stresses in the bulkturbulent liquid occurred only at very high agitation rates (above 700 rpm in a 2 Lreactor with a 70 mm diameter impeller). They further showed that theentrainment and motion of very fine bubbles in the absence of a vortex did notcause growth retardation, even at an agitation rate of 600 rpm. This indicates thatcell death in bioreactors, either with or without sparging, is primarily the result ofair entrainment, bubble break—up and surface bursting. The elimination of directcontact between cells and unstable gas—liquid interfaces should decrease celldamage in a bioreactor.2.1.2 Oxygen supplyDissolved oxygen is one of the most rapidly metabolized nutrients ofmammalian cells. Initial concentrations of other nutrients such as glucose andamino acids are usually greater than ten—fold higher than that of oxygen. Tosustain optimum cell growth, the dissolved oxygen concentration must be keptabove some critical value. Reported values of critical dissolved oxygenconcentration vary from 8 to 70% of saturation (Kilburn and Webb, 1968; VanWezel and van der Velden de Groot, 1978; Radlett et al., 1972; Sinskey et al.,1981; Boraston et aL, 1984). The critical dissolved oxygen concentration dependson 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 ,umole02 per 106 cells per hour (Spier and Griffiths, 1984; de Bruyne, 1988). A vastvariety of systems have been developed to provide- oxygen in tissue culturebioreactors. Bubble—free oxygenationGiven the damaging effects of gas bubbles, bubble—free aeration systemsshould offer significant advantages. Surface aerators which increase the turbulenceof air—liquid interface can enhance oxygen mass transfer. However, scale—upwould be problematic for surface—aerated systems of high aspect ratio. Surfaceaeration alone could not provide sufficient mass transfer for systems with high cellloading. An alternative method for aeration involves the use of gas permeablemicroporous membranes (Miltenburger and David, 1980). The immersedmembrane increases the total surface area for oxygen transfer. The membrane isoften mounted onto a rotating shaft to enhance the mass transfer rate further. Thefeasibility of scale—up using multiple membranes in parallel has beendemonstrated for reactors up to 150 L in volume (Lehmann et al., 1988). However,repeated autoclavings may cause pinholes in the membrane, hence increasingdowntime and the maintenance costs. More exotic methods, such as the use ofperfluorocarbon as oxygen carrier (Cho and Wang, 1989), have also beeninvestigated. Although the costly perfluorocarbon particles can be recycled, theadditional downstream recovery cost of the perfluorocarbon means that thetechnique is not feasible on an industrial scale (Yamaji et al., 1989). Oxygen transfer using external oxygenatorsAeration systems in which cell free medium is continuously recirculatedthrough an oxygenator are often used for cell cultivation (for examples in hollowfiber systems, OpticellTM, and Verax systems). High shear conditions for supplyingoxygen can be used in the oxygenator (e.g. high fluid velocity, high speed stirringand direct sparging) without danger of cell damage. Such systems are relativelycomplex and require the use of a mechanical pump to circulate medium from theoxygenator to the bioreactor. This increases the risk of contamination and systembreakdown.2.1.23 Direct spargingDirect sparging is the simplest method for providing oxygen to cell cultures.Foaming is inevitable due to the presence of proteins or serum in the culturemedium. However, foaming can be decreased if the cells are grown in serum—freemedium or in improved bioreactors with reduced serum requirements (e.g. Veraxsystem). The addition of an antifoam agent can also control foaming (Handa-Corrigan et al., 1989; Bently et al., 1989). However, the presence of an antifoamagent complicates downstream processing. At the low levels of agitation usuallyemployed in animal cell culture vessels, the entrapment and break—up of airbubbles is minimal. The oxygen transfer rate is limited by the short residence timeand small interfacial area of the bubbles. A higher air flow rate or finer bubble sizeis required to achieve a higher oxygen transfer rate. Either of these measures mayincrease cell damage. Scale—up by increasing the reactor aspect ratio can increasethe rate of oxygen transfer and the residence time of the bubbles withoutdetrimental effects on cell viability (Boraston et al., 1984). Aspect ratios as highas 6:1 and 12:1 are sometimes used. Bubble columns and airlift systems have beenused for animal cell cultivation with considerable success. However, most reportedapplications of direct sparging in these cell culture systems have been restricted tocells 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 SystemsThe use of high cell density immobilized cultuie systems for large—scalegrowth of animal cells is becoming increasingly important. Immobilization of thecells provides an intrinsic separation of cells from medium which facilitatesdownstream processing. This together with reduced nutrient requirements, leads tosignificant economies of operation. Ideal immobilized cell culture systems shouldprovide a high surface area for cell attachment, good mass transfer characteristicsbetween the cells and the surrounding medium and a low liquid shear stressenvironment for cell growth. Current techniques for immobilizing cells usuallyutilize 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.,101987) or Siran glass beads (Kratje et al., 1992). (B) Encapsulation in stirred tanksor 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 (Perryand Wang, 1989), stainless steel matrices (Familletti and Fredericks, 1988), orpolyurethane foam (Matsushita et al., 1990). (D) Hollow—fiber systems (Tharakanand Chau, 1986).2.2.1 Microcarrier culture systems2.2.1.1 Conventional systemsMicrocarrier particles are widely used for growing anchorage—dependentcells. The provision of sufficient culture surface area for cell attachment andgrowth 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 inindustry since the introduction of Cytodex-1. Many other types of microcarriershave 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 spargedmedium by a rotating filter which is often fixed onto the agitator shaft stirrer toform a cage. The centrifugal force resulting from the rotational motion enhancesmass transfer and also delays fouling of the filter membrane. The spin filtersystem is particularly effective for microcarrier systems (Cho and Wang, 1988). In11addition to isolating cells from gas sparging, it facilitates medium removal incontinuous perfusion culture systems. For microcarrier cultures, direct spargingposes a more challenging problem. Microcarriers tend to rise to the liquid—gasinterface carried by the air bubbles to remain there due to the small densitydifference between the microcarrier and the growth medium. This problem can beavoided 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 theincreased microcarrier density.Table 2. Commercially available microcarriersManufacturer CommercialDesignationChemical compositionNon porous Pharmacia Cytodex 1 DEAE—DextranPharmacia Cytodex 3 collagen coatedDEAE—DextranNunc Biosilon PolystyreneLux CytosphereBioplasPolystyrenePolystyrenecollagen coatedpolystyreneGlassSolo HillSolo Hill Collagen _Solo Hill BioglasIBF Micarcel G PolyacrylamideBioRad Bio—Carriers PolyacrylamidePolystyreneBioRad PSVentrex Ventregel GelatinPorous Verax Verax CollagenPercell Biolytica Cultispher—G GelatinPercell Biolytica Cultispher—H GelatinKirin Cellsnow celluloseSchott Siran GlassBiomat Informatrix Collagen-gl ycosamino—glycan122.2.1.2 Systems utilizing porous microcarriersPorous microcarrier systems (see Table 2) offer significant advantages oversolid bead microcarriers. These systems are distinct from the conventional surfacemicrocarrier culture system in that the cells are immobilized at high densitiesinside the matrix pores where they are protected from the fluid shear. Problems inlong term cultures due to shear damage are reduced. Although primarily conceivedfor adherent cell applications, some porous carriers can also be used with non-adherent cells such as hybridomas (Almgren et al., 1991). Most of the systemsinvolve the use of proprietary macroporous matrices. In an attempt to mimic thecell culture environment in mammals, most of these macroporous beads arecollagen based (collagen, gelatin, or collagen-glycosaminoglycan). Macroporousbeads can be inoculated directly from the bulk medium in the same fashion asconventional microcarriers. Suspended bead immobilization systems can be usedin a number of different reactor configurations including fluidized beds or stirredtank bioreactors. Porous microcarrier systems can be scaled-up easily in bothprocess intensity (cell density) and volume. -(a) VeraxThe Verax bioreactor system (Lebanon, New Hampshire) uses a proprietaryweighted collagen sponge matrix (specific gravity 1.2 to 2.5). The porousmicrobeads, fluidized in the bioreactor, form a thick slurry (55% solids byvolume), and the reactor is operated under continuous culture conditions forextended periods of time. Oxygen is supplied by a hollow fiber external gasexchanger which is connected to the fluidized bed reactor. Oxygen-rich mediumfrom the gas exchanger is pumped into the base of the reactor chamber to fluidizethe small microbeads. Oxygen-depleted medium from the top of the reactorchamber is then returned to the gas exchanger. Under conditions of perfusion13(medium replacement), a high cell density is developed within the fluidizedbioreactor and microbeads (typically 4 x 10 7 cell /mL of reactor volume and 2 to 3x 108 cells/mL inside the sponge bead matrix). Over 85 different cell lines, bothsuspension and anchorage—dependent cells, have been tested and culturedsuccessfully using this system (Griffiths, 1990). The 90% void volume sponge—matrix microbeads can have diameters ranging from 200-600 gm and pore sizesfrom 30-100 gm (Dean et al., 1987). This system has been scaled up to 2000liters. The microbeads are coated with collagen both on the interior and exteriorcarrier surfaces (Vournakis and Runstadler, 1989). Conventional heat sterilizationcannot be used because it causes denaturation of the collagen. The cost of thesterile microbeads somewhat offsets the advantages of the system, but it isprimarily its complexity that has limited its acceptance by industry.(b) Cultispher-G microcarrierCultispher—G (CG) microcarriers (Biolytica, Lund, Sweden) are made ofcross—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 of1.04 g/mL Cell densities up to 3 x 108 cells/ mL carrier have been reported(Nikolai and Hu, 1992; Mignot et al., 1990). Since gelatin is already heatdenatured, CG particles can be sterilized by autoclaving. The CG microcarrier wasoriginally designed for use in stirred—tank bioreactors. Recently, a new type of CGmicrocarriers with higher density (achieved by inclusion of a proprietary titaniumcompound) and large particle diameter (430-600 ptm) has been synthesized foruse in fluidized bioreactors (Reiter et al., 1990).14(c) Siran porous glass beadsSiran porous glass spheres (Schott Glaswerke, Mainz, Germany) have wideranges of particle sizes (30-5000 gm), internal pore sizes (10-400 gm), drydensities (0.7-1.2 g/mL), and internal void fraction (up to 70%). In cell culturethese glass spheres have been used in both packed bed (Looby and Griffiths, 1988)and two—phase fluidized bed bioreactors with separate oxygenator (Kratje andWagner, 1992; Keller et aL, 1991). Cell densities up to 2.6 x 108 cells/mL carrierhave been reported (Kratje and Wagner, 1992). The effect of bioreactorconfiguration on cell loading has been investigated (Kratje et al., 1991). Cellloading on the glass microbeads was reduced four—fold when the glass beads wereutilized in a stirred reactor instead of a fluidized bed reactor. In a fixed bedconfiguration the biomass loading of Chinese Hamster Ovary (CHO) was threefold higher than in a fluidized bed reactor (Griffiths, 1990). Present evidenceindicates that the porous glass spheres are unsuitable for cell cultivation in stirredtank bioreactors (Kratje et al., 1991; Griffiths, 1990).(d) Informatrix porous microcarriersInformatrix microcarriers supplied by Biomat Corporation (Belmount,Massachusetts) are made of a collagen—glycosaminoglycan copolymer with meandiameters 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—glycosaminoglycancopolymers are more resistant to collagenase degradation and have bettermechanical properties than native collagen. The growth and differentiation of cellscan be influenced by the composition of collagen—glycosaminoglycan (Cahn,1990).152.2.2 MicrocapsulationCells may be protected from the adverse effects of gas sparging byentrapment inside beads or microcapsules (Lim and Sun, 1980). Recently,Bugarski et al. (1989) and Kwong et al. (1989) were able to immobilizehybridoma cells in alginate microcapsules. A two—fold increase in maximum cellconcentration was observed for cells immobilized in the alginate microcapsules(i.e. 3.5 x 106 cells/mL—alginate) compared to suspension cells in a conventionalairlift system (Kwong et al. 1989). The secreted product can also be entrapped inthe microcapsules facilitating downstream processing (Bugarski et al., 1989).However, microencapsulation techniques are both complicated and cannotcurrently be scaled up for large scale industrial processes (Yamaji et al., 1989).2.2.3 Fixed bed bioreactorOne of the best examples of a fixed bed bioreactor is the OpticellTM system(Charles River Biotechnical Services Inc., Wilmington, Massachusetts). Thisconsists of a ceramic matrix, pump, oxygen and pH probes, gas permeator(oxygenator), medium reservoir, and feedback controller arranged as shown inFigure 1. The cylindrical ceramic matrix cartridge contains multiple channelsrunning the length of the cylinder. Each channel has a square cross—section withsides of approximately 1 mm and wall thickness of about 0.15 mm. The ceramicmatrix provides a surface area of 25-40 cm 2 per cm3 volume. The mostcommonly used ceramic element has a nominal surface area of 4.25 m 2 . Twotypes of ceramic have been used. One provides a relatively smooth surface; theother is highly irregular, its rough surface covered with many small cavities. Thesize of the cavities can be manipulated to occupy up to 40% of the total surfacearea, with a mean diameter up to 50 Jim. The cavities allow cell entrapment to takeplace and provide added surface area for cell attachment.CeramicCultureChamberPump16Vero cell densities up to 5.7 x 105 cells/cm2 ceramic surface ( 1.4 x 107cells/mL of ceramic cartridge) have been reported (Lydersen et al., 1985). Seedingthe Opticell reactor is problematic due to unevenly distributed cell inoculumpopulation. A complex procedure is used to ensure uniform seeding of the matrix.Sufficient medium flow through the ceramic is needed to prevent growth— limitinggradients from developing along the channel of the ceramic matrix. Oxygenelectrodes are provided to allow the consumption rate of dissolved oxygen to bemonitored continuously based on the inlet and outlet dissolved oxygenconcentrations and the medium flow rate. The oxygen supply is maintained byvarying the medium flow rate. However, at high medium flows cells are exposedto excessive shear, causing detachment and blockage. The complex pumping loopbetween the external gas permeator and the ceramic cartridge is fragile and proneto contamination. Channel blockage due to over—grown cells is another commonproblem associated with the OpticellTM system. The high cost of the single useFigure 1. Schematic illustration of Opticell bioreactor.17ceramic matrix and difficulties in scale—up have also hindered its acceptance.2.2.4 Hollow fiber systemsThe use of hollow fiber bioreactors is a well established technique for cellculture (Feder, 1988; Tharakan and Chau, 1986). In these reactors, the mediumflows through the fibers, and nutrients and wastes diffuse radially through the fiberwall to/from the cells in the shell space. The fibers are packed parallel to eachother inside the shell space and are connected to a manifold at each end. Theoverall configuration is similar to that of a single—pass shell and tube heatexchanger. Cells grow in the extracapillary space (ECS). Cell densities in excessof 108 cells/mL can be established in the ECS (Tharakan and Chau, 1986). It isalso possible to concentrate a high molecular weight product in the ECS.Concentration gradients pose major problems because of the heterogeneousculture growth conditions (Piret and Cooney, 1990). Scale—up of the culturevolume is limited by the number and the length of fibers that can beaccommodated within the shell.182.3 Airlift BioreactorsA 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 theriser section is usually sparged with gas. The different gas—holdups in the gas-sparged riser and the unsparged downcomer cause different fluid bulk densities inthree flow regions which in turn induce fluid circulation in the reactor. Thebehavior of an airlift reactor is influenced by the interaction of these sections.Since the first development of the airlift fermenter by Lefrancois and hiscollaborators (1955), the device has been used for most types of fermentationranging from single cell protein (SCP) production by microbial cells to morerecent monoclonal antibody production by hybridoma cells. The simplicity of thedesign and construction, the well defined flow pattern, and the relatively lowpower input make the airlift reactor an attractive alternative to conventional stirredtank fermenters. Pneumatic agitation in airlift fermenters eliminates shaft sealsand bearings associated with stirred tank reactors, hence reducing the possibilityof contamination and mechanical failure.Gas separator--_____Riser0 0 oo 00 0000oo000 00DowncomerDraft TubeBaseFigure 2. Typical airlift bioreactor configuration192.3.1 Classification of airlift reactorsVarious criteria have been proposed to classify airlift reactors (Onken andWeiland, 1983; Blenke, 1979). The two basic classes are: (1) external loop whenthe riser and downcomer are two separate conduits, usually round vertical pipesconnected 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 bafflewhich divides the column into a riser and a downcomer (see Figure 3b). Internalloop reactors can be further subdivided into two sub—classes, draft—tube (orconcentric tube) and split—column. Simplicity of construction is the mostappealing characteristic of the internal loop airlift reactor. Special attention is paidhere to the draft—tube type internal loop airlift reactor with Newtonian fluid sincethese are the features of the system examined in this study.2.3.2 Applications of airlift in animal cell cultureThe use of airlift bioreactors to grow animal cells in suspension was firstreported by Katinger et al. (1979). Airlift reactors can transfer sufficient oxygen(a)^ (b) ^ C00Riser 000 00 00 0O0OO 0o0000000O00F \\DowncomerFigure 3. Airlift bioreactor configurations: (a) external loop (b) internal loop airlift20for conventional batch cultures of animal cells (maximum cell concentration < 107cells/mL) at a relatively low shear rate (Wood and Thompson, 1986). Most data onairlift cell culture in the literature have been provided by the research group ofCelltech Ltd (Berkshire, U.K.), primarily for suspension cell culture. Celltech hassuccessfully grown more than 35 different cell lines, primarily hybridoma cells, inairlift 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 antibodyproduction rate. The effect of dissolved oxygen concentration (DO2) over therange of 8 to 100% was examined by Boraston et al. (1984). They found littleeffect on growth rate, glucose utilization rate, or maximum cell density of mousehybridoma NB1. However, they did acknowledge that the optimum DO2 variesand must be determined for individual cell lines. Hiilscher and Onken (1988)found that the concentration of bovine serum albumin (BSA) in the serum—freemedium significantly influences the growth of mouse hybridoma XR6—G10—B3.Hybridoma cell death rate was found to decrease with increasing BSAconcentration. Cell retention in airlift reactorsHiilscher et al. (1992) suggested the use of an external cell settler and cellrecycle to increase cell concentration in airlift reactors. Cells suspended in theeffluent 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). Theincorporation of a packed—bed for cell immobilization into airlift bioreactorconfigurations has also been proposed to increase cell loading (Murdin et al.,1989; Femilletti, and Fredericks, 1988; Lazar et. al., 1987; Chiou et al., 1991). Theuse of an airlift, where the liquid circulation is driven by the sparged bubbles in21the riser, eliminates the need for auxiliary mechanical pumping. The downcomerof the airlift is packed with a suitable support material for cell attachment andgrowth. Relatively high cell densities ( 6.8 x 10 7 CHO cells/mL of packed—bedvolume; 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 lowliquid 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) orcontinuous/perfusion mode.However, the near plug flow of medium through the packed bed createslongitudinal gradients of nutrient and waste products making scale—up of suchsystems difficult. The application of airlift reactors in commercial biotechnologyprocesses remains limited, due in part to the lack of basic design parameters in theliterature. Unfortunately, the available information in the literature is oftenconflicting and shows large variations. Precise comparisons of different studies areoften difficult since the operating conditions and reactor geometry have not beenclearly reported.23.3 Flow patternsThe riser section of a concentric tube (CT) airlift can sometimes beconsidered as a conventional bubble column (BC). Shah et. al (1982) identifiedthree flow regimes for BC:22(1) Homogeneous bubble flow, usually for superficial gas velocity, U gr, less than0.05 m/s(2) Slug flow — bubbles occupy the entire column cross—section; the frequencyand the length of spherical cap slug increases with increasing gas flow rate; slugflow often occurs in reactors with diameters less than 0.15 m.(3) Churn— turbulent flow — transitional region where the gas does not form acontinuous phaseAirlift and bubble column (BC) reactors for animal cell cultures should beoperated in the homogeneous bubble flow regime because of the requirement forlow 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 liquidmode. As observed by Onken and Weiland (1983) and Merchuk (1986b), non—hindered liquid circulation in CT's tends to delay the transitions between the flowregimes discussed above. A priori determination of the flow regime is difficult,since it is affected by many parameters, such as the physical configuration of thereactor and the properties and velocity of the liquid phase. Shah et. al. (1982)suggested an approximate guideline of flow regime dependency on gas velocityand 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 ina BC. As the gas flow rate is increased, causing a parallel increase in the liquidvelocity, bubbles are entrained in the flow stream to the downcomer. Oscillationbetween swirling flow, wavy flow and straight flow is observed at liquid velocitiesnear the terminal rising velocity of the bubbles.2323.4 Power inputThe 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 hydrostaticpressure region at the bottom to the top of the liquid—gas interface, and (2) kineticenergy transferred to the fluid by the jet of gas entering the reactor, i.e.•E Q= RTIn Ph -Fp1gh +q) U201^Ph^2Q^ (1)where Ph = pressure at the topQ = flow rate of the gasT = temperatureh = height of liquid column above the sparger(p = sparger efficiency (typically around 0.06) to account for thefact that the gas velocity just above the sparger is lower thanthe velocity in the orifice.U0 = velocity difference between the gas velocity inside theorifice of the sparger and the gas velocity just above thesparger.pi = liquid densityThe second term of Eq. (1) is usually negligible for airlift fermenters (Robinson,1986). However, Guy et al. (1986) pointed out that discrepancies (especially forpressurized vessels) have arisen in some cases as a result of neglecting the secondterm in Eq. (1).23.5 Gas velocityTwo factors dealing with the gas velocity may explain variations in the openliterature. The first is the term, "superficial velocity" U g, which may be based on24either the entire cross—sectional area of the vessel or the riser cross—sectionalarea. The superficial velocity based on the entire cross—sectional area should beused only when comparing the power input to a BC and a CT airlift. In all othersituations, the superficial velocity should be based on the riser cross—sectionalarea. The second factor is the effect of the axial variation of the dispersion heighton the superficial gas velocity. Chisti and Moo—Young (1987) showed that thehydrostatic pressure can affect the superficial gas velocity significantly. Theyproposed that only the "true" superficial velocity should be used, i.e.u _ Qm R T^ . in [Ph +Pi g 1g AR hL pl g^Phwhere AR = cross—sectional area of riserhi, = unaerated liquid height(2)2.3.6 Hydrodynamics23.6.1 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, singlenozzles, and perforated plates or pipes, are most commonly used in fermentationprocesses. Porous plates are more expensive and have higher operating costs dueto greater pressure drops. Spargers coupled with a liquid jet nozzle, driven by anexternal pump, are rarely used in fermentation processes due to their need forassociated pumping machinery and the higher shear force imposed on cells.The location of the sparger in the airlift device also has a strong influence onthe overall hydrodynamic behavior. Chisti and Moo—Young (1987) found bettergas distribution when the sparger was placed just inside the riser section (i.e.25above the base connecting zone). However, very little is known about the effect ofthe position of the sparger on the overall behavior of airlift devices. Dispersion characteristicsThe gas bubble size in a typical turbulent bioreactor is generally controlledby the equilibrium between the dynamic forces which tend to break up the bubblesand the surface tension forces which preserve their size and shape. The averagesize of the bubbles is usually independent of its size at formation. Dussap andGros (1982) proposed the following expression for predicting gas-liquidinterfacial area, aD , in aqueous sodium sulphite in a CT (volume = 0.015 m3, DC= 0.11 m, Dci =0.0756 m, h=1.8 m).aD = 3.66(-1-)E• 0 77VDfor 200 W/m3 Ei/VD  1500 W/m 3(3)where VD = dispersion volumeIf one assumes spherical bubbles, the average bubble diameter, dB can becalculated knowing that6£d = gB aDwhen £g = gas that dB =1 . 64 • Eg .VD(4)j-0.77^(5)Chisti et al. (1987) proposed the following equation for estimating the bubble risevelocity (Ub) in a BC (air-water dispersion):26Ub = 0.284 + 2.7 Ug (m/s)^ (6)More accurate prediction of Ub in a BC can be made using the followingequations (Chisti et al. , 1987):For Ug < 0.05 m/s in an air-water dispersion,Ub = 0.284 + 1.1 Eg (m/s)^ (7a)for Ug > 0.05 m/s,Ub = 0.284 + 11.2 Egg (m/s)^ (7b)Hills (1976) correlated the bubble rise velocity in the riser for UL r > 0.3 m/s andan air-water dispersion as follows:Ub = 0.24 + 1.35 (Ug + 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 ofinterest for airlift columns is independent of bubble size and approximately equalto 0.23 m/s. This is true for a CT containing bubbles of volume equivalentdiameters from 0.002 m to 0.01 m in an air-water system which does not containsignificant amounts of impurities (i.e. surface tension not less than 0.055 N/m).2.3.63 Liquid circulating velocity, ULAs illustrated by Equation (8), the upflow of liquid in airlift devicesincreases the velocity of gas bubbles in the riser section, thereby lowering theriser gas hold-up, Eg, and hydrodynamic pressure difference between the riser anddowncomer. Airlifts have a much more uniformly distributed gas phase across thecolumn cross-section, with a maximum Eg near the column wall, compared to BC27(Onken and Weiland, 1983). UL affects all parameters which characterize thereactor behavior, such as Eg, mixing time, and mass transfer. Unlike Ug, ULcannot be varied independently. Several investigators (e.g. Onken and Weiland,1983; Merchuk, 1986a; Bello et al., 1984) showed that UL varies with (Ug) 13 ,where j3 was approximately 0.33 to 0.4, depending on the reactor geometry andflow regime (Onken and Weiland, 1983). p was found to decrease as the flowregime 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 bepredicted byULr = 0.66 (AD/AR)0.78 ugr0.33 (9)Equation (9) indicates that ULr increases with increasing (AD/AR) ratio.However, both Weiland (1984) and Blenke (1979) found that UL r (for two phaseflow) reached a maximum for (AD/AR) = 0.59. In fact, Equation (9) agreesreasonably well with literature values for (AD/AR) ratios up to 0.89. Significantdeviation was observed for higher values of (AD/AR) (e.g. AD/AR = 1.12 in thework 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 betweenthe 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 ofChakravarty et al. (1974). The devices used by Blenke (1979) and Hatch (1973)had smaller Hu values of 70 mm and 33 mm, respectively.Two general theoretical approaches have been used to predict UL r . One isbased on a momentum balance, while the second is based on an overallmacroscopic energy balance. Bello (1981) found that the predicted ULr, when28based on a momentum balance, was insensitive to the flow regime andcorresponded more closely to measured values than values obtained from anenergy balance. However, the model proposed by Bello (1981) could not be usedwithout 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 rmeasured by Hatch (1973) and Jones (1985) in CT for both the bubbly and slugflow regimes. More recently, Chisti et al., (1988) and Calvo (1989) havedemonstrated that predicted values of Uj based on an overall energy balance cansatisfactorily describe most of the available liquid circulation data. They utilizedEi = 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) ofthe reactor; andEF = 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 arenegligible compared to other factors in Equation (10). Equation (10) then reducestoEi = ER + ED + EBwhere, according to Chisti et al., (1988), one can writeEB = 0.5 pi (VLd)3 KB AD (1 — Egd);^ (10a)ER = Ei —pi g hD Ufr AR Egr;^(10b)ED = pi g hp ULd AD Egd (10c)29For CT, Chisti et a/. (1988) proposed2.g hD (£ gr —£ gd )1).5. )ULr =  K . (A / A )2^(i_ EgdB R DKB is the dimensionless frictional loss coefficient and can be estimated from thecorrelation (Chisti et al. 1988),KB = 11.402 (AD/AB) °389^(12)where AB = free area for liquid flow between riser and downcomer at the bottomof the CT airlift reactor. Note that AB is affected by the clearance between thebase 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. Thismay explain the dependence of UL 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 largerthan those used by Chisti et a/. (1988) and were found to depend on (AR/AD). Thelarger KB values used by Calvo (1989) seemed to compensate for the errorintroduced by neglecting ED which, according to Lee et a/. (1986), accounted for50 — 60% of the total energy loss in the CT. Gas holdup, egIn order to predict ULr, the overall gas holdup, eg, is needed. eg isinfluenced not only by UL but also by the gas residence time, oxygen transfer, and30liquid mixing. Merchuk (1986b) found that the gas hold-up was lower for an airliftthan for a BC. In addition, a maximum hold-up, which characterizes the behaviorof the BC near the critical superficial velocity, does not appear in thecorresponding 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 eg (i.e.Egr + egd) in the CT and the corresponding BC when Ug, based on the entirecross-section, is used for comparison. This occurs even though the net effect ofUL and Ub in the riser section tended to reduce the gas holdup. The downcomergas holdup is increased by the downward liquid flow. Increasing the liquidvelocity increases the_number of the bubbles dragged down the downcomer andthe distance traveled by the large bubbles inside the downcomer. The gas-liquidseparation ability near the top of reactors can influence the overall gas holdupsubstantially. External loop airlifts generally have a much lower e gd (i.e. 300 to700%) than CT airlifts for the same unit volume energy input (Bello et al. 1985).Many empirical correlations for estimating 6g in CT airlifts have been proposed(e.g. 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:Egr = 3.4 x 10-3 (1 + AD/AR) -1 (Ei/VD)(2/3) (13)Egr = 0.16 (1 + AD/AR) (Ugr / ULr)a (14)with a = 0.56 (water); 0.58 (salt solution) andEgd = 0.89 Egr for air-water only (15)The suggested correlations were for an air-water or a salt solution ( 0.15 kmolm 3 NaCl) with bubbly flow (i.e. Ug = 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 ionicstrength may also affect gas holdup. The downcomer area or diameter is often31incorporated in expressions for predicting eg. It should be noted that most gasholdup expressions have been based on U g for fresh gas input only. Therecirculated gas is generally not considered. Merchuk (1986a, 1986b), Freedmanand Davidson (1969) and Schiigerl et al. (1977) have all reported that the use ofmultiple—orifice spargers gives higher eg (up to 30% higher) than single orificespargers for all UL and Ug tested. However, the differences reported by Merchuk(1986) were generally insignificant (<5%). Conflicting results were also reportedby Siegel et al. (1986) who found no appreciable difference in the gas holdup orthe gas recirculation rate caused by changing the sparger configuration and spargerorifice size.One can predict ULr, Egr, Egd in CT airlifts by simultaneously solvingEquations (9), (14) and (15). Chisti et al. (1988) suggested that Equation (14) bereplaced by the gas holdup correlation of Hills (1976):gr = 0.24 + 1.35 (Ugr + ULr )0 . 9323.6.5 Mass transferBecause of its low solubility in fermentation broth, dissolved oxygen can beconsumed quickly if it is not continuously replaced. The mass transfer coefficientfor oxygen, kLa, is thus a critical scale—up parameter. The rate of mass transfermay be expressed as:dCdt = kLaL (C * — CL )^ (17)where CL is the oxygen concentration in the broth and C* is its saturationUgr (16)concentration in the broth in equilibrium with the gas. kL is the true mass transfercoefficient and aL is the specific gas—liquid interfacial area based on liquid32volume; aD, the gas—liquid interfacial area based on total dispersion volume, isoften used instead of ai„ and can be easily estimated from Equation (4). ki, hasbeen found to be slightly influenced by the fluid dynamics (Blenke, 1979). Thewell—known Calderbank and Moo—Young correlation (1961) indicates that ki, isdetermined mainly by the bubble size.The lumped mass transfer coefficient or the so—called volumetric masstransfer coefficient, ki ad or kLaL, is invariably used for design purposes. Anumber of correlations for approximating kLa in CT airlifts are listed below (seeprevious sections for operating conditions and reactor geometry):(1) Bello et al. (1985):for 0.01 _. Ugr 5 0.10 m/skLaD = a1 (1 + AD/AR)-2 lig°.8where a1 = 0.75 for water and = 0.79 for 0.15 kmol/m 3 NaC1 solutionOrkLaD = a2 (cgo1.27where a2 = 0.57 for water and = 0.60 for 0.15 kmol/m 3 NaC1 solutionor(364kLaD .h, ror) HAD) —2'^ = a3 QULr^ULr^ARwhere a3 = 1.99 for water and 2.57 for 0.15 kmol/m3 NaC1 solutiona4 = 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 Usgr S 7.7x10-2 m/s, a 13 hole—perforated plate with do = 0.001 m, DC = 0.11 m, and DCi = 0.058 m, Stejskal andPotucek (1985) found that(18)(19)(20)33kLaL = 1.84x10-3 + 0.35 Ugr^ (21)It was not clear whether U g in Equation (21) was calculated based on the overallreactor diameter or the riser diameter. Koide et aL (1983a,b) reported that kLaDdepended 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. Theyalso reported that neither AB nor Ld influenced gas holdup or kLa for the CTdevices studied. Weiland (1984), Wang et al. (1971) and Kriegel et al. (1978)demonstrated a dependence of kLa on the ratio (Dci/Dc) for CT airlifts. However,different values of the optimum ratio of (Dci/Dc) for oxygen transfer have beenfound: 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 lesspronounced 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 bypower of gas input) remains relatively constant with increasing kLavalues and increasing Ug, while BC and stirred tank reactors exhibit asharp decrease in aeration efficiency;(6) the change of flow regime from bubbly to slug flow sharply decreaseskLa;(7) CT airlifts have higher kLa than external loop airlifts.Enhanced kLa in CT relative to BC was attributed to different bubble sizedistributions by Margaritis and Sheppard (1981) and Kawase and Moo—Young34(1986). Koide et al. (1984) indicated that the contribution of the CT downcomer tomass 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 bubblesize based on photographic and electrical resistivity methods; ki, values were thenestimated 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 etal., 1983a,b). However, Bello et al. (1985) showed that mass transfer in thedowncomer was negligible; the overall mass transfer could be adequatelyestimated by ignoring the contribution of the downcomer. Bello et al. (1985)attributed this to the lower gas—liquid slip velocity in the downcomer. Based onthe 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 steadystate, 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. liquidflow rate, gas holdup, and mass transfer coefficients) have been studied in greatdetail. Although the conflicting information in the open literature often leads toconfusion, the previous studies were helpful in formulating a suitable mechanisticmodel (Chapter 5) to describe the hydrodynamic properties of the proposed airliftbioreactor.353.0 EXPERIMENTAL MATERIALS AND METHODS3.1 Cells and Cell MaintenanceThree cell lines were used in this study: (1) Vero cells (ATCC CCL81), ananchorage dependent cell line commonly used for vaccine production; (2) BHKcells transfected with an expression vector containing a DNA fragment coding forthe amino—terminal lobe of human serum transferrin (hTF/2N) under the control ofthe metallothionein (MT-1) promoter (Funk et al., 1990); (3) 2E11 hybridoma cellline which produces an anti IL3 murine monoclonal antibody (Ziltner et al., 1988).Frozen stocks of cells were stored in liquid nitrogen in Dulbecco's modifiedEagle's medium (DMEM, Gibco) supplemented with 10% dimethyl sulfoxide(DMSO) and 10% fetal calf serum (FCS, Gibco). Cell stocks were maintained inroller bottle cultures with 850 cm 2 surface area at 370C in DMEM or DMEM/F-12 (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 Systems3.2.1 Ceramic foamThe ceramic foam elements were produced by the Selee Corporation ofHendersonville, North Carolina. The matrix itself consisted of a highly complexpore structure of sintered alumina dodecahedra defined by the ceramic filaments,with a network of interconnecting passages. The true skeletal density is about 4000kg/m3 and the void fraction is about 80%. Sintered alumina is non—toxic, can besterilized at high temperature, and provides chemical resistance, good mechanicalstability and high surface area per unit volume. Foam cylinders 36 mm in diameterand 40 mm long were employed in the perfusion system. In addition, tubular36Figure 4. Ceramic foam cylinders of 30,50 and 100 PPI.ceramic foam hollow cylinders of outer diameter 66 or 77 mm and inner diameter57 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 poresper inch (PPI; i.e. 1200, 2000 and 4000 pores per m) with average pore diametersof 550, 320, and 140 ,um, respectively. These are shown in Figure Fixed bed perfusion systemThe experimental system (Figure 5) consisted of a growth chambercontaining one ceramic foam cylinder, an external medium reservoir and a seedingchamber. Medium was circulated continuously between the external reservoir andthe growth chamber using a peristaltic pump (Model PIOT, Dungeg Inc.,Agincourt, Ont.). The circulation rate was typically 130 mL/min and could bereadily adjusted. The growth chamber consisted of a short section of Pyrex glasstube 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 40mL. The ceramic cylinder fit tightly into the glass tube without significant37clearance at its peripheral surface so that virtually all of the circulated liquidpassed through the ceramic matrix. A 2—L stirred tank (LH Fermentation, Slough,UK) functioned as a reservoir in which the circulating medium was reaerated andits pH regulated. Mixing was provided by a 60 mm diameter marine impeller at aspeed of 100 rpm. Aeration was supplied by sparging an air—CO 2 (95:5 byvolume) gas mixture directly into the medium reservoir. The gas—free mediumcirculated through the growth chamber so that the cells were isolated from thepotential damaging effect of sparging. The operating temperature was maintainedat 37°C, while the medium pH was maintained between 7.3 and 7.5 by addition of0.2 M NaOH solution. Oxygen probes (Ingold) were located in the reservoir vesselupstream from the growth chamber and in line directly after the chamber. Thisallowed the total oxygen uptake rate to be monitored continuously.Inoculum (at least 1 x108 cells) was introduced into the seeding chamberand pumped into the growth chamber. Medium circulation was stopped for about 1GAS OUTLET/GAS INLET/DO PROBE'711 PROBE^Q PUMPGrowth Chamber GF 0o oo00 ooc)dc:).°JACKETMEDIUM BOTTLESEEDING MEDIUM RESERVOIRCHAMBERFigure 5. Schematic showing perfusion system.38to 2 hours to allow cell attachment to take place. The medium was changedwhenever the glucose concentration fell below 500 mg/L. DMEM + 5% fetal calfserum (FCS) was generally used. Zn was provided by either direct addition ofZnSO4 solution or by using DMEM/F-12 mixture (50:50) + 5% FCS medium toinduce the metallothionein promoter. Airlift systemThe test system was a modified commercial concentric tube (CT) airliftreactor in which the normal impermeable draft tube was replaced by a porousceramic draft tube (see Figure 6). The porous draft tube was mounted in a 5 Lglass airlift bioreactor of 120 mm inner diameter with a water jacket fortemperature control. The reactor was manufactured by LH Fermentation ofSlough, UK. The reactor was equipped with a 12 mm—diameter cylindricalsintered glass Pyrex sparger with 170-220 gm pore size. Liquid circulation aroundand through the porous draft—tube was driven by the sparged gas. The draft—tubeconsisted of a series of ceramic hollow tubular short cylindrical sections (35 mmin height). Up to 6 sections could be mounted together, stacked end to end, andheld in place by a stainless steel wire mesh (grid opening size of 4 mm) located onthe inside of the hollow ceramic cylinders. The thickness of the annular ceramicfoam cylinders was either 7.5 or 13 mm. In each case, the inner diameter was 51mm, so that the outer diameters were 66 or 77 mm, respectively. The poroustubular ceramic foams were held in place by glass supporting tubes located at bothends of the draft tube (see Figure 6). The tubular glass support pieces were either25 mm or 75 mm in length and had a wall thickness of 4 mm. The non—porousglass draft tube had an inner diameter of mm, a wall thickness of 4 mm. and aheight of 450 mm.39Figure 6. Schematic showing airlift systemFor BHK cell culture, the draft—tube of the test system was made bystacking two 13 mm thickness, 35 mm tall 30 PPI cylindrical ceramic foam pieceswith glass supports at top and bottom (see Figure 6). The operating temperaturewas maintained at 37°C. Aeration was supplied by sparging an air—CO 2 (95:5 byvolume) gas mixture at 100 mL/min into the vessel along the axis, 50 mm belowthe 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 justabove the top of the porous ceramic draft tube with 0.7 L medium present. Themedium volume was then increased gradually to a total of 4 L to accommodate theincreasing cell number in the reactor.40The effects of several configuration parameters on the airlift hydrodynamicbehaviour 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 asgas holdup, overall volumetric mass transfer coefficient (kLa), and liquidcirculation rate were also investigated to develop a mathematical model tofacilitate 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 measuredusing an electronic pressure transducer (Model PX750-06DI, Omega EngineeringInc., Stamford, Connecticut). The pressure sensor consisted of two tubes insertedfrom the bottom of the reactor and fixed in positions located just below the lowerend of the draft tube with opening of 3 mm diameter. The openings of the sensortubes were arranged to be perpendicular to the direction of the liquid flow. Thepressure 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 spanand 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 pressuretransducer was converted to a digital signal using a mA input A/D board (Model:PC-61, United Electronic Industries, Newton Center, Massachusetts) then storedby the computer. The listing of the controlling program for data logging using themA input A/D board appears in Appendix 1. Liquid circulation velocities werecalculated by monitoring the time needed for a small colored neutrally buoyantpolyurethane particle to travel a known distance along the downcomer. Theaverage liquid circulation velocity was calculated based on 10 such measurements.41Mass transfer coefficients were determined by the transient gas in/gas—out method(Sobota et al., 1982). Dissolved oxygen concentration was monitored using Ingoldpolarographic oxygen probe and Anglican 2000 controllers. The mV output fromthe Anglican controller was first converted to a digital signal and then registeredby the Nimbus computer. Appendix 2 shows the BASIC program listing employedfor communication between the controller and the computer.PHOTO- 12005423.2.2 MicrocarriersThe porous polystyrene microcarrier beads were manufactured byMicroporous Materials Ltd. (Braunston, U.K.). The emulsion process used for theproduction can be controlled to obtain the desired range of pore sizes. Accordingto the manufacturer, these styrene copolymer beads had a surface area of 50,000cm2 / g particles and a porosity greater than 90% by volume. The surface waschemically modified by Microporous Materials Ltd. as listed in Table 3 tofacilitate cell attachment and growth. The diameters of the polystyrene particlesranged from 250 to 1000 p.m, with pore sizes from 5 to 80 pm. The materialconsisted of roughly spherical pore spaces, interconnected by multiple smallerchannels 5 to 10 times smaller than the pores. Four pore sizes were tested— 80, 40,25, and 5 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.Figure 7. Open structure of a macroporous microcarrier P40 particle.433.2.2.1 Pretreatment of microcarrier particles.Characteristics of the microcarriers used in this work are summarized inTable 3. Commercial microcarriers were prepared according to the instructionsprovided by the manufacturer. Cytodex-1 microcarriers were added to siliconizedglass bottles and were swollen in Ca2+, Mg2+—phosphate—buffered saline (PBS;50-100 mL/g Cytodex-1) at room temperature for at least 3 hours. The PBS wasthen removed and the particles were washed twice with fresh PBS (30-50 mL/gCytodex) and sterilized by autoclaving for 20 min at 121°C. The supernatant wasdecanted and the sterile particles were rinsed several times with fresh serum—freemedium prior to use. The preparation of Cultispher—G microcarriers was similar tothat of Cytodex-1. The thy Cultispher—G microcarriers were swollen and hydratedin Ca2+, Mg2+—free PBS (50-100 mL/g particles) at room temperature for at leastone hour. Most of the microcarriers floated due to entrapped air. Sterilization byautoclaving released the entrapped air. The supernatant was removed and thesterile 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 15min. Then the particles were washed twice with sterile PBS and once withcomplete medium prior to inoculation.44TABLE 3. Properties of Microcarrier Beads TestedDesignation Description VoidVolume(%)composition poresize(pm)Particlediameter(.tm)Settledvolume(mL /gdry wt.)P80 large porepolystyrene90 90% styrene 10% commercialdivinyl benzene.80-100 500-1000 12P40 medium porepolystyrene90 90% styrene 10% commercialdivinyl benzene.30-40 500-1000 12P25 polystyrene 90 90% styrene 10% commercialdivinyl benzene.10-30 500-1000 11.4P5 small porepolystyrene90 90% styrene 10% commercialdivinyl benzene.10-20 500-1000 10.5CM80(—CH7C1)chloromethylstyrene90 50/50 styrene and chloromethylstyrene on mole basis.80-100 500-1000 12Q80 quaternaryderivitisedstyrene90 CM80 treated with excessethanol/water (20%) solution oftrimethylamine for 8 hours at roomtemperature.40-50 500-1000 11.8DEA80 diethylaminoderivitised90 CM80 refluxed with 10%diethylamine for 4 hr. (2.9% N)30-100 500-1000 14.3DEA—LG80 diethylaminoderivitised90 CM80 refluxed with 100%diethylamine for 4 hr. (2.9% N)50-60 500-1000 14CS collapsedsulphonatedpolystyrene97 sulphonation of polystyrene at60°C for 6 hours then pHneutralized with NaOH solution.20 250-500 30S80(—SO3Na)sulphonatedpolystyrene90 P80 treated with excess sulphuricacids (95-98%) at roomtemperature for 30 minutes30-70 500-1000 11S40(—SO3Na)sulphonatedpolystyrene90 P40 treated with excess sulphuricacid (95-98%) at roomtemperature for 15 to 240 minutes30 250-500 1245Table 3. continued..Flex— CM80 flexiblechloromethylpolystyrene90 50/50 styrene and chloromethylstyrene on mole basis, 10%divinyl benzene crosslinker60-100 500-1000 11Per80 peroxidetreatedpolystyrene90 P80 treated with peroxide 50 500-1000 143Cultispher—G Percell 50 100% Gelatin 50 170-270 14-18Cytodex 1 Pharmacia 0 Dextran matrix with N,N—diethylaminothyl groupsubstitution0 131 18The surface composition and charge of some of the Polyhipe particles arecharacterized in Table 4. The characterization of the Polyhipe particles wasperformed by Microporous Materials Ltd.Table 4. Surface characteristics of the Pol hi e particlesZeta potential(mV)C (%) 0 (%) S (%) CI(%)N (%) Naf%)—P40 —41.34 ± 1.6 95.52 4.34 — 0.14 —P25 —39.68 ±1.28 ND* ND ND ND ND NDP5 —28.41 ± 0.12 ND ND ND ND ND NDQ80 —17.15 ± 1.15 ND ND ND ND ND NDFlex—CM80—45.5 ± 7 ND ND ND ND ND NDPER80 —54.6 ± 5.37 ND ND ND ND ND NDDEA—LG80—2.56 ±9.6 ND ND ND ND ND NDDEA-80 44.28 ±2.43 94.04 4.01 — 0.18 1.8 —*ND denotes not determined Zeta potential was determined from streaming potentialmeasurements in a flat plate system. Microcarrier culture(a) Roller cultureInitially the microcarriers were tested in glass 500 mL roller bottles,siliconized (Dimethyldichlorosilane, Sigma, St. Louis, MO) to prevent attachmentof the cells to the bottle surface. Vero and BHK cells were inoculated at 5 x 10 746cells in 20 mL of medium with 500 mg of microcarrier particles. This inoculumwas incubated in a static vessel, with periodic agitation, for 4 to 8 h at 37°C toallow cell attachment to take place. The culture volume was then increased to 100mL with fresh medium containing 10% FCS. For hybridoma cells, the inoculum (5x 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 thenwashed once with 30-50 mL serum containing medium and finally resuspended in100 mL medium containing 10% FCS. The roller bottles were rotated at 2 rpm.Collapsed sulphonated polystyrene was tested without prior hydration andwashing. This material was autoclaved dry (30 min, 121°C) and hydrated bydirect addition to 100 mL culture medium with cells (5 x 10 5 cells mL-1). Allcultures were perfused semi—continuously by decanting the spent medium andadding 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 culturesSeveral sizes of Bellco spinner flasks, 100 mL, 250 mL and 1000 mL involume were used. All flasks were siliconized prior to sterilization to prevent cellattachment to the walls. The cultures were seeded with enough cells suspended inreduced starting volume (i.e. 1/2 to 1/3 of the final volume) to give a final cellconcentration 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 andthe porous particles for the first 3 hours. Subsequently, the stirring speed wasmaintained just sufficient to suspended the microcarriers. The spinners were eithergassed 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 and47temperature. pH was not controlled and varied from 7.3 initially to 6.8 prior tomedium exchange.3.2.3 Suspension cell cultureSuspension cultures were grown in suspension in 250 or 500 mL spinnerflasks (Bellco). Prior to sterilization the flasks were siliconized to prevent cellattachment to the walls. The flat—blade impeller was rotated at 50 RPM to keepthe cells in suspension. Some of the BHK cells aggregated and formed granularclumps of about one hundred to several hundred microns in diameter. At leastonce daily, a 10 mL sample was withdrawn from the stirred flask for analysis andthe 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 ablood gas analyzer (Model 168, CIBA—Corning).3.3 Analytical Methods33.1 Cell numeration33.1.1 Porous matrixAt the end of each run, cells immobilized inside the ceramic matrix or theporous particles were washed with phosphate—buffered saline (PBS). A solution of0.1 g/L crystal violet in 0.1 kmol/m3 citric acid was then added. The mixture wasthen incubated for 2 to 3 days to release cell nuclei. The stained nuclei werecounted using a hemocytometer (Perry and Wang, 1989). In all cases, the standarddeviations were less than +10% at a 90% confidence level. During the course ofthe ceramic matrix perfusion culture and hybridoma microcarrier cultures, samplesof the culture medium were examined for cells released from the porous matrix bycounting on a hemocytometer. Viable cells in suspension were assessed using the48trypan blue (0.4%) exclusion method (Griffith, 1985). In this method, viable cellsare impermeable to the dye, whereas dead cells take up the dye. MicrocarriersFor Cytodex-1, the culture medium was removed and the carriers werewashed with PBS. Cells were released from the carriers by incubation with a0.25% trypsin-EDTA solution (Gibco, Ontario) at 37°C for 15 minutes.Dissociated cells were then counted using a hemocytometer.33.1.3 Suspension cellsCells suspended in medium were numerated either by means of amicroscopic count using a hemocytometer or a particle counter (Elzone 280 PC,Particle Data Inc., Elmhurst, Illinois).33.2 Assays33.2.1 GlucoseGlucose was measured using a Beckman Glucose Analyzer II (Brea, CA).j3-D-glucose reacted with oxygen according to the reaction:G0- D - glucose + Oz  Glucose Oxidase + H20 )Gluconic Acid + H202 (22)The dissolved oxygen concentration was monitored by a rhodium/silverpolargraphic oxygen electrode. The analyzer response was linear over the range0.05 to 4.5 g/L. LactateL-lactate was analyzed using a YSI L-lactate analyzer (Model 27, YellowSprings, OH). L-lactate first diffused through a thin polycarbonate membrane with49a nominal pore size of 0.01 ,um. The diffusion rate was the rate limiting step asindicated by the manufacturer. Once past the membrane, L—lactate encountered athin layer of immobilized L—lactate oxidase where the following reactionoccurred:L -lactate oxidaseL — lactate + 02^>H202 + pyruvate^(23)Hydrogen peroxide diffused toward the cellulose acetate membrane coveredplatinum anode and gave rise to a probe signal current which was directlyproportional to the L—lactate concentration. The analyzer had a useful range of 0.1to 1.34 g/L, with a precision of ±2%.33.23 Lactic dehydrogenase (LDH)Lactic dehydrogenase activity in the culture supernatant (Holscher andOnken, 1988) was determined spectrophotometrically using a commerciallyavailable kit (Sigma, No. 340—UV). The basic reaction isPyruvate + NADH LDH >Lactate + NAD^ (24)Phosphate Buffer (2.85 mL, Sigma, No. 410-3S) and 0.05 mL of culture mediumwere added directly into a NADH Vial (Sigma, No. 340-2) and incubated for 20min 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 a10 mm light path cuvet.The absorbance [A] of NADH was measured at 340 nm at 15—secondintervals 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 was30 IU/mL (i.e. a A[A] = 0.003 per min).5033.2.4 TransferrinThe transferrin concentration in the BHK culture medium was determinedeither using an enzyme—linked immunosorbent assay (ELISA) (Roitt et al., 1985)or a particle concentration fluorescence immunoassay (PCFIA) technique (Jerviset al. , 1991).(a) Enzyme-linked immunosorbent assay (ELISA)The assay plates were coated with the capture antibody by incubating with50 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 withPBS—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) 50tI., of appropriately diluted test samples or known standards were then added toeach well and incubated for 1 hour. (3) 50 AL of rabbit anti—transferrin (500xdilution) was added to each well and incubated for 1 hour. (4) Alkalinephosphatase conjugated anti—rabbit Ig diluted 1/3000 in PBS was added to eachwell (50 gL/well). (5) 100 III., of phosphatase substrate (disodium p—nitrophenylphosphate) in diethanolamine buffer was added to each well. The plates wereincubated 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, wereadded to wells in special FCA 96—well plates. Standards at 0.25, 0.15, 0.1, and0.05 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—transferrincapture antibody (Sigma) were added to each well containing a sample. Thesamples were gently mixed and incubated at room temperature (21°C) for 20 min.51Following the first incubation, 20 pL of sheep anti-transferrin-FITC conjugate(ICN, Costa Mesa, California) were added to each sample and incubated for 20min, at room temperature in the dark, after gentle mixing. The plate was thenevacuated using the Pandex FCA. Samples in wells were washed three times withPBS and read using the 485/535 filter pair at 25X gain.3.3.3 Microscopy3.3.3.1 Scanning electron microscopy (SEM)Specimens with adherent cells were prepared for SEM according to themethod of Allen (1983). The fragments were fixed in 2.5% glutaraldehyde in 0.1kmol/m3 sodium cacodylate buffer, pH 7.3 for 1 hour, washed three times with 0.1kmol/m3 sodium cacodylate buffer, treated in osmium tetraoxide for 1 hour anddehydrated by a series of 5-minute ethanol washes (i.e. 30%, 50%,70%, 95%, and100%). The specimens were dried (Model CPD020, critical point drier, BalzusUnion, Liechtenstein), fixed to clean aluminum stubs using double-sided tape orsilver paste and sputter coated with gold according to the Nanotech Sputter coaterprocedure manual before viewing using a scanning electron microscope (ModelS4100, Hitachi, Tokyo, Japan or Model 250T, Cambridge Instruments, Cambridge,U.K.).33.3.2 PEG embedding techniquePorous microcarriers with cells attached were first fixed and dehydratedaccording to the protocol above. The specimens were then hydrated and kept in70% ethanol solution before being transferred to 30% polyethylene glycol (PEG;average MW 3350 Da; Sigma) in distilled water at room temperature. The sampleswere transferred with a pasture pipetter to a 50% PEG solution at 60°C, infiltratedfor 30 min and then transferred to 75% PEG solution at 60°C for a further 30volume (weight) neededEpon 12 ResinDodecenylsuccinic Anhydrate (DDSA)Araldite 6005 resin (Air/6005)Dibutyl phthalate (DBP)Tri—dimethylaminothylphenol (DMP-30)25 mL (31 g)35 mL (55.5 g)15 mL (17.6 g)2 mL (1.9 g)0.9 droplets / g mixture52minutes. The samples were placed in the 100% PEG for at least 30 min withstirring. Each porous microcarrier was then carefully moved into a single gelatincapsule 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. Blockswere trimmed on a microtome using a glass knife to reveal the center portions ofthe porous microcarriers. Most of the PEG on the trimmed block was removedwith 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 thestandard SEM protocols as outlined above, before viewing.33.3.3 Thin—sectioning microscopyMicrocarrier samples were fixed and dehydrated with ethanol according tothe standard SEM protocols discussed previously. The 100% ethanol was replacedwith 100% propylene oxide before the samples were infiltrated with a gradedseries 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:The samples were placed on a rotator and agitated for 24 hours for eachinfiltration step. DMP-30 catalyst was added to the freshly prepared 100% Epon53mixture solution and agitated for an additional 4 hour period. Each microcarrierwas then placed into a plastic BEEM capsule filled with the 100% Epon mixturewith catalyst (DMP-30). The filled capsules were then placed in an oven at 60°Cand allowed to set for 48 hours. The matrix hardened as it cooled. Blocks weretrimmed on a microtome using a glass knife. 0.5 ,um thick cut slices were collectedfrom 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 andheat—mounted onto the glass slide. The samples were then stained and viewedunder a light microscope.33.3.4 Confocal micros copyThe cells were labeled with fluorescein diacetate (FDA) and propidiumiodide (PI). Esterase within living cells cleaves the ester, liberating fluorescencewhich 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. Stocksolutions were prepared by dissolving 5.0 mg/mL FDA in acetone and 20 mg/mLPI in PBS. The FDA/PI (10:90) working solution (0.5 mg FDA and 18 mg PI permL mixture) was prepared just prior to use. An appropriate volume of the FDA/PIsolution was added to the microcarriers and incubated for 5 min at roomtemperature. The samples then were washed with PBS twice, placed in theconcave depression of a glass slide under a coverslip. The stained samples wereexamined 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 ofproducing and storing 768x512 pixel images. A three—dimensional image could bereconstructed based on series of two—dimensional images collected at variousfocal plane positions.544.0 RESULTS AND DISCUSSION -Fixed Bed CERAMIC FOAMPERFUSION SYSTEMThe test design featured a fixed bed of ceramic cylindrical foam as a supportsurface, and an external reservoir arranged as shown in Figure 5. The ceramicfoam retained cells in the matrix providing physical protection to animal cells. Thecells are thus segregated into a realtively small reactor volume, away from thebulk medium. Product separation is inherent in the design.4.1 Effect of Foam PorosityFor a given perfusion rate of culture medium, the level of shear stressexperienced by the cells immobilized within the ceramic foam is a function of thepore opening size. Several trials were conducted to test the effect of the ceramicfoam 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. Adetailed investigation of methods for determining cell number in the cultures isreported in Section 4.3. Glucose uptake rate provides an adequate means ofmonitoring cell growth for simple comparison of different pore size matrices.The experimental results are shown in Figure 8. Good growth of BHK cellswas observed on the 30 PPI ceramic matrix. However, growth was significantlyless on 50 PPI and 100 PPI cylinders under similar inoculation and operatingconditions (described in section Little or no cell growth was observed onthe 100 PPI ceramic foam. This might be a consequence of uneven nutrientdistribution, flow channeling due to medium by—pass through the clearance spacebetween the outer perimeter of the foam and the inner wall of the growth chamberor excessive liquid shear on the cells immobilized within the small pore ceramicmatrix. All subsequent results were obtained using a 30 PPI ceramic cylinder.558 104•^6Time (days)0^.0Es 2.5-*C.g 2.0o• 1.5-1.0-0.5-0.0Figure 8. Glucose utilization of BHK cells in batch culture on ceramic foamcylinders of various porosity. Each culture was inoculated with lx 10 8 cells andperfused with DMEM medium at 130 mL/min (see Section Cell growthwas based on the rate of glucose utilization.4.2 Cell Growth on the Ceramic SurfaceThe initial attachment and the morphology of BHK cells on the surface ofthe ceramic matrix was very similar to that seen in T—flasks. Followingattachment, the cells extended and flattened themselves on the surface; theycontinued to grow to form a confluent monolayer (Figure 9A). Finally, theyovergrew, forming a multilayer of cells (Figure 9B). When the culture reachedsteady—state in the perfusion system, the multilayer of the BHK cells was about0.1 mm thick (approximately 10 cell diameters). Comparison of the cellmorphology when the monolayer was just confluent (Figure 9B) with that of themultilayer (Figure 9C) indicates that cell shape and packing changes markedlyduring the course of the culture.(A)56(B)(C)57Figure 9. SEM photographs of BHK cells grown on porous ceramic. A: attachedcells after 1 day; B: a confluent monolayer after 7 days; C: multilayer tissue after21 days.4.3 Estimation of Total Cell NumberCells immobilized on the ceramic matrix are not accessible for directenumeration during the culture. Therefore, indirect methods based on nutrientuptake rate were used to estimate cell number. The total oxygen consumption rate(OCR) of cells in the growth chamber was calculated knowing the medium flowrate and the change in dissolved oxygen concentration following the passage of themedium through the growth chamber. This value was used to estimate the totalimmobilized cell number assuming that the specific oxygen uptake rate (OCR/cell)did not vary with time or cell concentration during the culture. Satisfactorycorrelations between the total cell number and the total OCR have been reportedpreviously 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 reasonably58well for a growth curve. It was found that OCR and cell number increased inparallel for culture densities less than 109 cells/m2 (Lydersen et al., 1985).However, they reported that the number of cells increased more rapidly than themagnitude of OCR at culture densities in excess of 10 9 cells/m2. The cell specificoxygen uptake rate decreases as the cell concentration increases due to decreasingaverage cell size. At high cell concentration, the actual cell number within theceramic 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 theestimated total cell number in the growth chamber based on a reported value of thespecific oxygen uptake rate for BHK21 cells (0.20 mmoles/109 cells/h; Bognar etal.; 1983). It can be seen that after 15 days, the total oxygen consumption ratestabilized, equivalent to a final total cell number of 5x10 9 inside the growthchamber. This corresponds to a cell concentration of 1.6x10 11 cells/m2 of theceramic surface, or 1.25 x 108 cells per mL of total matrix volume. In contrast, themaximum cell concentration achieved in roller bottles was 1 to 2 x10 10 cells perm2 .Glucose uptake rate (GUR) can also be used to monitor culture growthkinetics 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 2.4g/L/day after about 15 days. In agreement with the value based on OCR, thiscorresponds to a total cell number in the matrix of 5 x 10 9• This estimate wasbased on our measurement of specific GUR for BHK cells grown in roller bottlesof 0.46 g/day/(10 9 cells), the same value as reported by Bognar et al. (1983).0000591.5co 20-as 1500.Eu, 10-C0C^-cr)x 5 -Oob.7<-1.0 cv20E_J'crs0.U)TD- 0.5-oasEstimated cell number000 00 oxygen consumption rate0^1^0.010 15^20Time (days)Figure 10. Oxygen uptake rate and estimated cell number of BHK cells grown ona porous ceramic support. Cell number was based on the oxygen uptake rate in theperfusion system with DMEM + 5% fetal calf serum and perfusion rate of 128ml,/min (1x108 cells were introduced at time zero as inoculum).Recovery of the stained nuclei from the ceramic matrix after treatment withcitric acid accounted for only 70% of the total cell number estimated on the basisof OCR and GUR. The cells suspended in the circulating medium accounted for anadditional 5% of the total cell population. The independent estimates of cellnumber 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 tortuouspathways within the ceramic (which facilitate cell immobilization) probablyprevent complete recovery of all the immobilized cells. The remaining nuclei werebelieved to be embedded within the pores of the ceramic form.525602520 TC‹X17621.5 "6E1.0^coa)160.5 -47:5cn0.0I^.^I^•^10^10 20 30^40Time (days)Figure 11. Glucose utilization of BHK cells and estimated cell number based onthe 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 ProductionMost of the lactate produced in mammalian cell cultures originates from themetabolism 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 molarlactate yield from glucose is twice that from glutamine. Hence, it can be assumedthat most of the lactate produced by BHK cells results from the glycolysis ofglucose. Figure 12 shows a linear relation between the total lactate produced andthe 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 thatobserved for CHO cells (Perry and Wang, 1989). The high conversion of glucose0)U)6050 -6)a 40 -c.)02a. 30 -a)RffLf2, 20 -asOI-10-I^ 1^ I0^20 40 60Total Glucose Used (g)08061to lactate reflects inefficient use of glucose as an energy source. In the presentwork, 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 cellconcentration in roller bottle cultures. The reason for the increased utilizationefficiency in roller cultures was unclear.Figure 12. Lactate production based as a function of glucose utilized for BHKcells in the perfusion system (same operating conditions as in Figure 10).624.5 Stability and Viability of Cells on the MatrixCell leakage from the ceramic matrices was monitored throughout the courseof the culture. At steady state, the concentration of suspended cells in thecirculating medium was about 2x10 5 cells/mL of medium, corresponding to lessthan 5 % of the estimated total cell number. The viability of the released cellpopulation was about 50%. The activity of lactate dehydrogenase (LDH) in thecell—free supernatant was monitored as an indication of cell death within thematrix. LDH, an intracellular enzyme, is released after cell death (Gardner et al.,1990). The relationship between LDH released from freshly disrupted BHK cellsand cell concentration is illustrated in Figure 13. LDH activity in the culturemedium increased to a maximum LDH activity of 0.06 IU/mL at the steady statecell concentration of 1.25x108 cells per mL of matrix. This is equivalent to theamount of LDH released by the sonication of 2 x 10 5 BHK cells per mL of140 - 0' 120 -Maximum observed [LDH]in the perfusion systemso -50^100^150BHK Cell Concentration (cells / mL x 10 -4 )Figure 13. Correlation between measured LDH activity and disrupted BHK cellconcentration.400 200100-0-J 8063medium or 3.6% of the total cell mass present in the system under steady stateconditions. However, this value is misleading because LDH released from BHKcells is not stable in cultures (Arathoon and Birch, 1986). The activity of LDH wasdetermined 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 tAo = LDH activity at time zerok = 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 onceevery 24 h. The LDH activity measured in the spent medium thus represents theaccumulation of LDH released from dead cells over the 24 h period minus theactivity lost by inactivation of the enzyme. Assuming that LDH was releasedcontinuously into the medium at a constant rate (i.e. constant cell death rate) anddecayed at a constant rate, the total accumulated LDH activity measured at thetime of medium replacement can be represented by an integral of A(t) over thetime between medium changes (24 h in this case).t^tAt = f A(t) dt = SAo eict dt =-- 1 -A (1— Cid )k0^0(26)The corrected cell death rate (based on A 0) was thus 9.4% of the total cellpopulation per day. At steady state, the cell death rate must be balanced by the cellgrowth rate. After 15 to 20 days in culture both the OCR and GUR reachedsteady—state values (Figures 10 & 11) indicating a constant cell mass. At this pointa 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—viable64cell suspended in the medium could only account for 2.5% of the total cellpopulation.4.6 Effect of Serum during Stationary Growth Phase in the Perfusion ReactorSerum is the most ill—defined and expensive component of the culturemedia. Consequently it is desirable to reduce its usage, especially in large—scalecultures. Under steady state conditions, the specific rate of transferrin productionof the BHK cells did not change significantly over the first 3 days when the serumlevel in DMEM/F-12 was reduced from 5% to 2.5% (see Figure 14).Experiments in T—flasks confirmed that serum was critical only forestablishing the cell population. The final cell numbers of BHK cultures with 5%serum were comparable to those with 2.5% serum. Once the culture reached1-^'^t0^1^2 3^4^5Time (days)Figure 14. Effect of fetal calf serum (FCS) concentration on steady—statetransferrin production rate at a perfusion rate of 128 mL/min in DMEM medium(same operating conditions as outlined in Figure 10).65confluency, reduction of serum level to 1% did not cause a significant reduction incell specific productivity; the final transferrin concentration in the mediumreached about 12 mg/L prior to medium change, comparable to the transferrinlevels with serum containing medium. However, serum provides the attachmentfactors needed for cell immobilization. In the absence of such attachment factors,cells detached much more easily in the fixed bed perfusion system. The gradualloss in the total productivity in the steady—state perfusion culture after 3 days withthe lower serum level (see Figure 14) was likely caused by cell detachment andwashout. The suspended cell number in the spent medium increased from 2x10 5 to3.2x105 per mL of medium as the serum level decreased. However, this cell losscannot fully explain the loss of productivity. Serum is also known to protect cellsfrom the damaging effects of shear stresses (McQueen and Bailey, 1989). Thistype of damage may have also contributed to the observed loss in productivity ofthe perfusion system at low serum concentration.4.7 Effect of Zinc on Transferrin ProductionFigure 15 shows the cumulative production of transferrin for cells grown inDMEM 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 theexpression 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 5times the productivity of the non—induced cells.The transferrin concentration in the medium of the perfusion system wascontinuously monitored over the culture period. Figure 15 shows the cumulativeamount of transferrin produced by BHK cells grown in the matrix with theperfusion of DMEM/F-12 medium. The steady—state transferrin production rate ofE 80a)02-o 60-a.ILI-40-If;I-I^I^I^•^i0^2 4 6 8Time (days)10^12666 mg HTF/109 cells/day was reached after 15 days. This coincided with the timeat which the glucose and oxygen uptake rates stabilized.Figure 15. Induction of transferrin production. 1 x 108 cells were introduced attime zero as inoculum and both types of medium contained 5% fetal calf serum. 10itM zinc was added to the DMEM/F-12 medium at time zero.4.8 Perfusion PropagatorBioreactors with volumes greater than 1000 L are now commonly usedindustrially. The increase in bioreactor volume requires a corresponding increasein the inoculating cell number to achieve the required inoculum cell density.However, traditional techniques for producing inoculum such as roller bottles areinsufficient to deal with the high cell number needed. For example, a 1000 Lmicrocarrier culture system requires a minimum inoculum of 5x10 1 ° to 5x10 11cells for a relatively short lag period and rapid growth (Reuveny and Thoma,671986). Serial propagation is often used to scale—up cell number for the inoculationof microcarrier culture processes. Harvesting cells by treatment of microcarrierswith trypsin for each subsequent inoculation is tedious and also increases the riskof contamination. If the product of interest is not secreted but contained within thecells, a constant supply of cells is required.Since it had been shown previously that the current fixed—bed ceramicperfusion system provided a 50—fold increase in total cell number (i.e. from 1 x108 to 5 x 109 BHK cells). It was reasoned that the current system can functionwell as a source of inocula for scale—up if it could be cycled through a repeatedcell growth and cell recovery using trypsin sequence. However, recovery of theimmobilized cells within the ceramic matrices by trypsin incubation wassomewhat difficult. Figure 16 illustrates the feasibility of using the currentperfusion systems for Vero cell production. The perfusion system had an estimatedtotal number of Vero cells of 6.96 x 10 9 during stationary growth phase. Theduration of the trypsin incubation period affected the amount of cells which couldbe harvested from the system and the time needed for the system to recover andreach subsequent confluence. Following incubation with trypsin for 15, 30, and 30min., 6.0x108 , 3.4x109, and 5.1 x 109 total viable Vero cells or 8.6%, 49%, and74% of the total Vero population were recovered from the system at the timesindicated by arrows 1, 2, and 3 in Figure 16, respectively. The increase in cellrecovery is due to longer trypsin incubation and PBS washes. PBS washesremoved residual medium with serum from the growth chamber, hence increasedthe effectiveness of the trypsin treatment. The second cell recovery was performedwithout PBS washes while the cell chamber was flushed twice with PBS prior tothe third cell recovery.400.^120060681 350 -40 -30 -20-10-o GlucoseA Lactate00o000Figure 16. Cycled growthperfusion system. 6.0x10 849%, and 74% of the totaltimes indicated by arrowsand 30 min., respectively.II^. I600^800Time (h)and harvesting of Vero cells in the fixed—bed ceramic, 3.4x109 , and 5.1 x 109 total viable Vero cells or 8.6%,Vero population were recovered from the system at the1, 2, and 3 following incubation with trypsin for 15, 30,i^.^r1000^1200695.0 RESULTS AND DISCUSSION - AIRLIFT SYSTEMThe feasibility of using the ceramic matrix for high density culture wasestablished in the previous section using a perfusion culture system. The fixed bedperfusion system is well suited to intermediate scale operation and has the majoradvantage that cell recovery is relatively easy. However, this system is amenableto scale—up to only a very limited extent. An aspect ratio (i.e. height to diameterratio) for the porous cylindrical foam of less than 1 is needed to scale—up the fixedbed perfusion reactor since one cannot increase the perfusion rate through theporous matrix indefinitely to meet the increasing oxygen demand without exposingcells to excessive shear stresses. Reactors with an aspect ratio of less than 1 willcause (1) packed density diversity causing channeling and reducing nutrientsupply, and (2) large installation area requirement (Murakami et al., 1991). Thecurrent fixed bed system cannot be scaled—up beyond 10 L (the volume of theporous matrix) without reduction in the cell concentration within the porous matrixdue to high shear stress (see Appendix 4 for the calculation).A major focus of this project has been to develop a simplified and scaleablematrix reactor. An airlift reactor was designed in which the matrix comprises thedraft tube of the reactor. Bubbles rising in the draft tube cause a circulatorymovement of medium and induce a convective flow of medium through thematrix. This flow provides nutrients to cells immobilized on the inner surface ofthe 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 risinggas bubbles which increases the overall cell viability in the bioreactor whencompared to conventional airlift bioreactor.Although CT airlift systems have been studied extensively, none of thestudies have investigated the effect of using a porous draft tube. Fluid mixing is70enhanced considerably by creating additional junctions between the riser anddowncomer due to short—circuiting of some fluid (i.e. splitting the draft tube intoseveral short sections; Blenke, 1979; Chisti 1989). Similarly, the use of a porousdraft tube would be expected to provide enhanced mixing characteristics whencompared to a conventional airlift reactor. Unlike the external loop airlift packedbed fermenters proposed by Lazar et al. (1987) or the Fibre—bed CT airliftbioreactor used by Chiou et al. (1991), the liquid circulation rate in the porousdraft—tube CT airlift system can be regulated without exposing the cells toexcessive shear while the mass transfer characteristics in the bulk fluid remainunaffected. Both the draft—tube wall thickness and the matrix porosity could beadjusted to accommodate the required convective flow through the matrix. Thethickness of the porous draft tube wall of the proposed system (13 mm) is muchless than the packed bed depth (300 mm) used by Chiou et al. (1991) or by Lazaret al. (1987) and the liquid residence time is, therefore, much shorter. Theconcentration gradient across the matrix is unlikely to be as serious a problem.5.1 Pressure DropAlthough the porosity of the ceramic matrix must vary depending upon thenumber of cells attached to the matrix during fermentation runs, simulationexperiments without live cells were used to investigate the hydrodynamiccharacteristics of the system. It would be difficult to conduct pressure dropmeasurements with live cells due to slow growth of the cells. It requires weeks toestablish steady—state in a live cell system. Variation in porosity was used toestimate the effect of cell growth and attachment to the draft tube. In someexperiments, the pores of the porous draft tube were blocked by a thin plastic filmto simulate a non—porous draft tube.71The pressure drops across the various draft tubes measured at the base of thedraft tube are shown in Figures 17a and 17b. The non—porous glass draft tube hasa wall thickness of 4 mm. Error bars represent two standard deviations, based onmore than 100 data points collected by the computer over a 5—minute period, isrepresented in Figures 17 to 29. As expected, the use of non—porous draft tubesyielded higher pressure drop than porous draft tubes. Generally, the pressure dropacross the draft tube increased with decreasing porosity. The pressure drops acrossall the thin—film blocked matrix draft tubes were very similar. Under identicaloperating conditions, the pressure drop across the glass draft tube was greater thanthat of the other non—porous draft tubes because of the differences in draft tubelength. The porous section of the draft tube, composed of several pieces of tubularceramic elements stacked end—to—end, was located between two non—porous glasssupporting tubes.(a)76O5EE0_ 4<1c.00 3kvkn2Q_0 o^ glass draft tube—A— nceporas 100 PPI (thick wall)—+— non-porous 30 PPI (thick wall)A^ 100 PPI (thick wall)- 50 PPI (tick wall)---0- - 30 PPI (tick wall)• ^0.000 0.004^0.008^0.012^0.016Superficial Air Flow Velocity (m/s)5—v—non-pore 50 PPI (thin wall)30 PPI (thin wall)—x-50 PPI (thin wall)4- --x-- 100 PPI (thin wall)0.01600.000^0.004^0.008^0.012(b)720,, -IE 3-0002 -T.a)0cnSuperficial Air Flow Velocity (m/s)Figure 17. Measured pressure drops across the various draft tubes. The airliftcontained 5 L PBS solution and sintered glass sparger was used. (a) Longsupporting glass pieces were used with thick wall porous draft tubes (360 mmtotal length). (b) short supporting pieces were used with the thin wall porous drafttube (260 mm total length).The effect of the overall draft tube length on measured pressure drops isshown in Figures 18 and 19. The overall length of the draft tube included both theporous matrix and the non—porous glass tubes. Differential pressure drops acrossthe porous draft tubes increased with overall draft tube length. For a given risergas superficial velocity, the amount of liquid perfused through the porous drafttube depends on the vertical location of the porous section in relation to the rest ofthe draft tube. A draft tube with a short lower glass support has a greater overall30 PPI, 5L PBS, thick wall•^—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)73inward perfusion rate which results in greater riser liquid superficial velocity andhence lower riser gas holdup.The effect of serum addition on the differential pressure drop across theporous draft tube was determined in the absence of antifoam. Generally, thedifferential pressure drops decreased with serum addition. The magnitude of thedifferences between the measured differential pressure drops of airlift reactorscontaining serum and those without serum increased with increasing riser gassuperficial velocity. Visual observation suggested that more bubbles were presentin the downcomer with increasing serum addition. The stagnant gas bubbles in thedowncomer reduced the differential pressure drop across the draft tube. There waslittle or no difference in the measured differential pressure drops when the serumlevel was increased from 2 to 5 % (Figures 20 to 23).1.75OE 1.50EtZac▪ 1.25..oLI 1.000co2O 0.75tL22o 0.5020.250.000.000 0.004^0.008^0.012^0.016Riser Gas Superficial Velocity (m/s)Figure 18. Effect of porous draft tube length on the pressure drop across the thickwall 30 PPI porous draft tube. Pressure drop was measured at the base of thereactor.is6 pieces, 100 PPI, thin wall withshort supporting pieces (260 mm total),5L liquid volume—A— PBS alone■ 2% serum in PBS--v-- 5% serum in PBS3.074...••••• 100 PPI, 6 pieces in 5 LPBS, thick wallfE• 2 '50ai-0 2.0 -=O0 0.5 -(0a.• 0.0• Short supporting pieces (260 mm)• Long supporting pieces (360 mm)0.000^0.004 0.008 0.012^0.016Riser Gas Superficial Velocity (m/s)Figure 19. Effect of porous draft tube length on the pressure drop across the thickwall 100 PPI porous draft tube. Pressure drop was measured at the base of thereactor.0.000^0.004^0.008^0.012^0.016Riser Gas Superficial Velocity (m/s)Figure 20. Effect of serum concentration on differential pressure drops across thethin wall 100 PPI porous draft tube, measured at the base of the reactor, at variousriser superficial velocities.750.8O-EEa_▪ 0.6 -6 Pieces, 30 PPI, thin wall withshort supporting pieces (260 mm),5L Liquid volumesaO 0.4 -=000_OOOO2<▪ 0.2-o_0O ^ PBS aloneO 2% Serum in PBSA 5% Serum in PBS0.00_ 0.000^0.004^0.008^0.012^0.016Riser Gas Superficial Velocity (m/s)Figure 21. Effect of serum concentration on differential pressure drops across thethin wall 30 PPI porous draft tube, measured at the base of the reactor, at variousriser superficial velocites.02.5 -EE• 2.0 -(Li.0• 1.50`01.0 -O00.2 0.5 -0• 0.0a_0.0006 pieces, thick wall, 100 PPI, .5L volume I^Ilong supporting pieces (360 mm total)^ ___I,--,-1-- -,0.004^0.008^ 0.012^0.016■ PBS alone• 2% Serum in PBS• 5% Serum in PBSRiser Gas Superficial Velocity (m/s)Figure 22. Effect of serum concentration on differential pressure drops across thethick wall 100 PPI porous draft tube, measured at the base of the reactor, atvarious riser superficial 1.0 -000Co0< 0.5 -006 pieces, 30 PPI, thick walllong supporting pieces (360 mm total)5L PBS■ PBS alone• 2% Serum in PBS• 5% Serum in PBS762.0EE°- 1.5 -a6_c)0.00.000 0.004Riser Gas0.008^0.012^0.016Superficial Velocity (m/s)Figure 23. Effect of serum concentration on differential pressure drops across thethick wall 30 PPI porous draft tube, measured at the base of the reactor, at variousriser superficial velocities.5.2 Mathematical ModelingThe most critical fluid dynamic parameters in an airlift reactor are the gasholdup and the liquid circulation velocity. It is necessary to establish therelationship between these parameters and gas flow, liquid and gas properties andreactor geometry in order to be able to model this type of reactor. The liquid flowvelocity, and gas holdup can be estimated from an overall energy balance over theairlift loop, similar to that given by Eq (10):Power input = rate of, energy dissipationi.e. Ei = ER + ED + EB + ET + EF + Ep^ (27)whereEi = energy input as per Eq. (1)ER = energy dissipation due to wakes behind bubbles in the riser;77ED = energy loss in the downcomer due to upflow motion of bubbles as perEq (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 otherfactors (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 theporous draft tube due to the rough draft tube surface. The gas holdup in thedowncomer is assumed to be zero. Equation (27) is therefore, reduced toEi = 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. + Ugr) (hD / dR) (ULr AR)^(28)' Cf P/ ULr3 n hD (dR/2 )wheredR = riser diameterCf.= Fanning friction factor, 0.018 for turbulent flow in a rough pipe;ER = plg hD AR Ub Egr^(29)whereUb = terminal bubble rise velocity .The terminal bubble rise velocity was set as an arbitrary, but realistic value of 0.25m/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 writeEp = (2 7C r hp Up) (k Lp ,u Up)^ (30)wherehp = 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 byUp (h)= pi g (hD — h) egr / (,u k L)for Reynolds numbers, Rep, less than 30 (Dullien, 1975),whereh = vertical distance measured from the base of the airlifthp = dispersion height measured from the base of the airliftThe Reynolds number is defined as17) U pRep = 1 3 /I Pwhere DP = effective average pore diameterThe energy loss due to fluid turn around at the bottom of reactor, EB can beestimated (Chisti et al., 1988) by:EB = 0.5 p1 ULd3 AD KB^ (33)whereKB = the dimensionless frictional loss coefficient as per Eq (12).AD = downcomer cross—sectional areaULd = downcomer superficial liquid velocityIn this case, the riser gas holdup was not known a priori. The well—known gasholdup correlation of Hills (1976; Eq. 16) was used. Because of the short—circuiting flow through the porous draft tube, Egr, ULr, and Up all varied withheight along the reactor. ULr increased as the flow moved upwards, while Egr and78(31)(32)79Up decreased. For ease of modeling, the airlift was divided vertically into 5 equalsections. Values of the riser gas holdup, riser superficial liquid velocity and theliquid superficial velocity through the porous draft tube element were calculatedfrom the base up. Simulation runs were also performed for an airlift with 3 or 9equal divisions. For the range of riser superficial velocities investigated, theresulting simulation values differed less than 1 percent when the airlift wasdivided into 9 instead of 5 equal sections. However, this difference increased to 6percent when 3 equal sections, instead of 5 equal sections, were used for themodel simulation. The iterative Levenberg—Marquardt method, a quasi—Newtonmethod, was used to solve for the listed constraints simultaneously. Appendix 3contains the program listing and a simulation example can be found in Appendix4. The pressure drop across the draft tube can be calculated from the riser gasholdup with the assumption of no gas holdup in the downcomer.Figures 24 and 25 show the calculated and measured pressure drops atvarious injected rise air superficial velocities. The Darcy resistance coefficient ofthe porous matrix, k, in Eq. (30) was determined based on ceramic foam cylindersof equivalent porosity using the equation:OP= k•g•L•Up (34)For fixed Ups, the pressure drop across the foam cylinder was measured and k wascalculated from the slope of AP vs. Up. k was found to be 4.3 x 10 8 , 6.7 x 107 and3.3 x 107 m-2 for the 100, 50 and 30 PPI foam cylinders (for water with Reynoldsnumber, Re p , less than 40), respectively. However, the actual Darcy resistancecoefficient could vary considerably from the foam cylinders to the hollowcylinders of the same porosity. The porous hollow cylindrical elements werestacked end—to—end to form the draft tube. Some gaps between the hollow ceramiccylindrical elements and between the glass supporting tube and the hollow80cylindrical element were inevitable. Those gaps decreased the flow resistance andreduced the k value. The tendency for the proposed model to over—estimate thedifferential pressure drop across the draft tube can be explained, at least in part, bythe differences between the actual Darcy resistance coefficient and the values,based on the cylinders, used for calculation. However, according to Figure 24, theapparent k value for the thick—wall 100 PPI porous draft tube should be about 6.7x 107 instead of the value used, i.e. 4.3 x 10 8 m-2 . The 6.4 fold differencebetween 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 porousdraft tube at the base of the reactor also contributed to the difference between the0.000^0.004^0.008^0.012^0.016Riser Air Superficial Velocity (m/s)Figure 24. Pressure drop across the thick—wall porous draft tube, measured at thebase 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 cylindricalceramic elements end—to—end. Each point indicates a measured value, while thelines indicate theoretical calculations based on the model.81predicted and measured values. An attempt was made to measure the differentialpressure drop across the porous draft at various axial positions. The small scale ofthe airlift used and the swirling motion of liquid in the airlift made suchmeasurement very difficult. In fact, the magnitude of the measurement error wastypically 2 to 3 times greater than the collected data. In some instances, themeasured differential pressure drop across the porous draft tube was greater in thetop section than measured values at the base of the draft tube.The model also predicted values of superficial liquid flow velocity forcorresponding riser gas flow values. An attempt was made to compare thepredicted superficial liquid velocities in the downcomer with measured values (seeRiser Air Superficial Velocity (m/s)Figure 25. Pressure drops across the thin—wall porous draft tube at the base of thereactor versus riser superficial air velocity for different pore spacings. The airliftcontained 5 L water and the porous draft tube consisted of 6 cylindrical ceramicelements end—to—end. Points indicate measured values, while lines indicatetheoretical calculations based on the model.82Figure 26). Liquid circulation velocities were calculated by monitoring the timeneeded for a small colored neutrally buoyant polyurethane particle of 3 mmdiameter to travel a known distance along the downcomer. Because of the liquidflow through the porous draft tube, the colored particle moved almost randomlydown the downcomer instead of in a straight line. Moreover, the particlesometimes travelled too close to the draft tube wall and the frictional force eitherslowed or stopped the particle movement. Therefore, large errors were associatedwith the measured downcomer liquid superficial velocities. The model generallyunder—estimated the values of downcomer superficial velocity.0.75 0.60>CD 0.5-crTs▪ 0.4 -• 0.3 -COE0c 0.2-3OCDD 0.1 -cp0.00.00^30PPI, dick wall----50  PPI,thick wall^100 PPI, thick wall^30 PPI, thin wall50 PPI, thin wall^100 PPI, thin wall0.41^0.82^1.23 1.83Riser Superficial Gas Velocity (m/s)Figure 26. Average downcomer liquid superficial velocity at various riser gassuperficial velocities. Lines indicate the theoretical calculation. Each pointindicates measured values for the airlift with a 100 PPI, thick wall porous drafttube of 360 mm. Error estimation was calculated based on 10 measurements.I—a— 30 PPI, short supports, thick wall- -o-- 30 PPI, short supports, thin wallA 30 PPI, long supports, thick wall- -7-- 100 PPI, short supports, thick wall- -100 PPI, long supports, thick wall--+-- 100 PPI, short supports, thin wallx non-porous glass draft tube835.3 Mass Transfer CoefficientSince the porous matrix draft tube occupied only 5% to 11% of the current5L reactor volume, an air flow rate of 25 cm 3/s, which corresponded to Ugr =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 108 cells/mL matrix. Mass transfercoefficients of a similar magnitude were observed by Murakami et aL (1991) foran airlift fiber bed bioreactor. For example, a kLa of 0.02 s -1 was determined at ariser air superficial velocity of 0.0086 m/s for an external airlift with a draft tubediameter of 50 mm. However, to support a BHK culture with a cell concentrationof 108 cells/mL medium, a kLa value of at least 0.027 s -1 is needed (seeAppendix 4 for calculation). The current system could only provide a sufficientkLa to support 4.4 x 107 cells / mL medium.0.020^0.018-0.016-0.014-0.012-0.010-••■•••co,0.008-0.006-0.004-0.002-o.0.000^0.004^0.008^0.012 0.016Riser Gas Superficial Velocity (m/s)Figure 27. Volumetric mass transfer coefficient at various riser gas superficialvelocities. Two standard deviations are represented by the error bar indicated.84PBS containing 5% serum was used to determine the effect of serumaddition on the mass transfer coefficient. It was found that the addition of serumhad a minimal effect on mass transfer as shown in Figure 28. The downcomer gasholdup increases with the addition of serum. However, since stagnant gas bubblesin the downcomer contribute little to the overall mass transfer, the riser gas holdupis likely not affected by the presence of serum.The measured mass transfer coefficients are compared with those predictedby Equation (21) in Figure 29. The empirical equation proposed by Stejskal andPotucek (1985) underpredicted kLa by a considerable margin. Equations (18)through (20) could not be used since the riser gas superficial velocity used in thecurrent system was too low. kLa values, therefore, need to be empiricallydetermined for the current system.0.0150.010-0.005-is;0.000 1 0.000^0.004^0.008^0.012^0.016Riser Gas Superficial Velocity (m/s)Figure 28. Effect of 5% serum addition on mass transfer coefficient. The porousthin— wall 30 PPI draft tube was supported by short glass tubes. Error barsindicate two standard deviations.--n-- glass draft tube in PBS with 5% serumo^ glass draft tube in PBS- -A - - 30 PPI in PBS with 5% serum- -v- 30 PPI in PBS850.014-0.012-0.010-0.006-0.004-0.0020.000 ^0.000 0.004^0.008^0.012 0.016Riser 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. Theairlift contains 5 L PBS without any serum addition.5.4 Scale-up Potential Assessed by Proposed ModelThe proposed airlift reactor could be scaled—up by either increasing thereactor 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—uppotential of the airlift reactor can be assessed using the proposed model. Althoughthe model predicts the correct hydrodynamic trends, it would be unrealistic todesign a large scale airlift reactor using the proposed model without furthervalidation. A pilot scale reactor should be constructed to test the model accuracy.86However, the proposed model can be used to optimize the draft—tube thicknessand maximize cell loading of the current 5—L airlift reactor.The permissive air flow rate must be defined before any scale—up evaluationcan proceed. Air flow rate is limited by foam formation due to presence of serumproteins in the medium. Antifoam is usually used to prevent foaming. Antifoamconcentration 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 inputgas per reactor volume per minute) was the maximum allowable air flow rate forDMEM/F-12 medium with 1.2% fetal calf serum. Since the medium used in thepresent study contained 5% fetal calf serum, the allowable maximum air flow rateshould be lower than 1 vvm. A value of 0.4 vvm is used for all the followingscale—up calculations because the reactor has tested and operated at an air flowrate of 0.4 vvm with very little antifoam C addition (less than 800 ppm).The medium perfusion velocity through the porous draft tube can becalculated 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 PPIporous 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 areshown in Figure 30. The maximum shear stress associated with the perfusion flowcan be estimated using (Perry and Wang, 1989):dFr = g • UP k 4 . (1—em ) (35)whereI" = shear stress (N/m2);d F = diameter of matrix fibers plus cell layer;CM = void fraction of the porous matrix.10.09.01-1)8.0C\17.080 6.0U)Sta)^AR = 0.0038 m2- AR = 0.0019 r112^ AR = 0.0007 m2- --  AR = 0.0003 m28710^15^20^25^30^35^40Porous Draft Tube Wall Thickness (mm)Figure 30. Calculated average perfusion velocity through the porous matrix versusselected 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 exceed0.04 m/s to keep the maxium shear stress below 2 N/m 2. The assumptions used toevaulate the maximum shear stress are (1) em = 0.7, (2) k = 6.7 x 107 m-2 and (3)dF = 1 mm. Most of the predicted perfusion velocities shown in Figure 30 arearound 0.04 m/s. Because of the reactor configuration, the draft tube thicknessusually can not exceed 50 mm, it is necessary to reduce the input gas flow rate todecrease the shear stress on cells. The perfusion rates through the porous matrixpredicted by the model can supply sufficient oxygen to support cell concentrationsof greater than 1x10 9 cells/mL matrix (or 10 12 cells total), if oxygenconcentration in the bulk fluid can be maintained at air saturation. The porous88matrix can occupy as much as 37% of the current reactor volume. For larger airliftreactors, the draft tube thickness could be increased to the extent that the mediumperfusion rate through the porous matrix could not sustain high levels of cellconcentration within the matrix without exposing the cells to excessive shearforces. The maximum theoretical allowable draft tube thickness is about 690 mmto support 108 cells/mL matrix at a shear stress of 2 N/m 2 or 345 mm to support acell concentration of 2 x 10 8 cells/mL at the same shear stress level of 2 N/m2 .The porous matrix volume to total reactor volume ratio would thus decrease as thereactor volume increases. However, the ultimate factor which controls the cellloading, as in the case of the current 5 L reactor is not the thickness of the porousdraft tube but the mass transfer characteristics. The current airlift system couldonly support a total BHK cell population of 2.5 x 10 11 cells at an air flow of 0.4vvm.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 inthe downcomer, (2) a high aspect ratio (from 6:1 to 12:1) to increase volumetricmass transfer rate of oxygen and (3) a porous draft tube of uniform thickness butwith a controlled pore size distribution (decreasing in size from the top to thebottom) to compensate for the increased differential pressure drop across the drafttube at the bottom and, therefore, maintain an uniform medium perfuse fluxthroughout the entire draft tube length.5.5 Long Term CultureThe porous draft tube airlift system was for animal cell culture. The drafttube of the tested system consisted of two 30 PPI thick—wall tubular ceramic89pieces mounted between two supporting glass tubes. The total length of the drafttube 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 mLDMEM + 10% NCS. It was difficult to achieve an even inoculation of cellsthroughout the porous matrix. Intermittent gas sparging (sparging for 1 min out ofevery 20 min) to provide enough liquid circulation to distribute cells throughoutthe 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 aposition 50 mm above the top of the draft—tube, i.e. there was 1.7 L medium intotal. The medium was changed whenever the glucose concentration droppedbelow 0.5 g/L. The total medium volume was eventually increased to 4 L as thecell 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 reactorvolume. 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 theceramic 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 steadystate operation (see Figure 31). With a similar ratio of inoculating cells to porousmatrix volume, the fixed—bed ceramic perfusion system took about 360 h to reachsteady state. Figure 31 shows the cumulative amount of glucose consumed and thetotal lactate produced by the BHT( cells. The glucose—to—lactate conversion ratioin this case was 0.79, indicating that glucose was used less efficiently than in theperfusion system (see Section 4.4). The use of the single hole sparger did notprovide sufficient mass transfer of oxygen and resulted in higher lactate yield. Thenumber of immobilized cells within the matrix was also somewhat lower than for350rn0ow 300-a0250a)0 100cp5000oo°oo°C8000OAOA00AAEo GlucoseA Lactate 110 1^•^I.^t^.90the fixed bed perfusion system (i.e. 9.7 x 10 7 vs. 1.3 x 108 cells / mL matrix).Nevertheless, the airlift system was successfully operated for 62 days, despite theless—than—optimum oxygen supply.0^200 400 600 800 1000 1200 1400 1600Time (h)Figure 31. Cumulative glucose used and lactate produced by BHK cells in theporous draft—tube airlift bioreactor. Inoculum of 4 x 10 8 cells was introduced tothe DMEM + 10% NCS medium.916.0 RESULTS AND DISCUSSION — POROUS MICROCARRIERSAlthough it is possible to scale up the previously described airlift reactorbeyond 100 L in volume, the large cell number required as inoculum makes theprocess difficult to start up. Moreover, representative cell samples from the airliftreactor can not be obtained without sacrificing reactor. On the other hand, it isrelatively easy to initiate and scale—up microcarrier cultures. Cell culture usingconventional microcarriers in a conventional stirrer bioreactor is common. Cellgrowth occurs on the particle surface only. The use of porous particles providesadditional surface area for cell growth and protects cells from the shear forces inthe bulk fluid. Suspended bead immobilization systems can be used in a number ofdifferent reactor configurations including suspended beds or stirred tankbioreactors. However, most of the commercially available macroporous beads arecollagen based (collagen, gelatin, or collagen—glycosaminoglycan). Thesematerials are obtained from undesired complex biological materials such as skin,tendons, or ligaments which are difficult to standardize, regulate and are prone tocontamination if not sterilized properly. Ceramic particles could, in principle, beused as microcarriers. However, the abrasive nature of the small ceramic particlesand their high specific gravity precluded their use as suitable microcarriers. Siranporous glass beads are also abrasive in nature and have high specific gravity. Thecell loading on Siran porous glass particles was reduced drastically when theseparticles were used in stirred reactors instead of fixed bed bioreactors (Kratje etal., 1991).Polystyrene carriers have a number of potential advantages over the porousmicrocarriers currently available. Polystyrene microcarriers are made from aninexpensive, chemically well defined material, which is amenable to a variety ofsurface modifications. The material is sterilizable and is already accepted as a cell92culture support in the form of plates and flasks. Modified polystyrene porousparticles were therefore tested in this study as a new type of microcarrier.6.1 Cell Attachment Rate6.1.1 Effect of surface chemistry group modificationsCell growth depends on the microcarrier surface characteristics such aswettability, and chemical group expression. Previous data showed that cells grownon dextran carriers with a moderate degree of substitution with DEAE anion—exchange groups (1.5 meq/g) surpass those grown on beads with highersubstitution level (Levine et al., 1979). Polystyrene dishes have been used for cellculture for more than two decades; they are also commonly used to manufacturemicrocarriers. Cells do not attach readily to untreated polystyrene (e.g.bacteriological grade petri plates). Modifications of the properties of thepolystyrene surfaces makes them more suitable for cell culture. Thesemodification 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. Someresearchers have suggested the importance of carboxyl groups (Ramsey et al.,1984), while others attribute the improved performances to the presence ofhydroxyl group (Curtis et al., 1983). The importance of surface carbonyl groupshas also been suggested (Eriel et al., 1991).It might be expected that the surface properties of a support would be criticalfor cell adherence. In the initial investigation, the rate of cell attachment orentrapment was, therefore, compared for the various Polyhipe surfacemodifications in this study. A carrier loading of 9 g /L medium was used to93provide excess carrier surface area. 10 mL aliquots from a single batch wereadded to 100 mL siliconized glass roller bottles. A 0.1 mL sample of freelysuspended cells was taken from each roller periodically and enumerated using ahemocytometer and a cell counter. The cell viability was greater than 95% for allruns. The glass bottles were agitated once every 30 min. Cellattachment/entrapment was calculated from the decrease in suspension cellnumber. Error bars indicate two standard deviations, calculated using data fromduplicate runs, on the figures which follow. Cells suspended in medium withoutadded particles in siliconized glass rollers were used as controls. Control rollerbottles provided a measure of cell attachment to the glass roller surface and clumpformation. Figures 32 and 33 show that most of the cell attachment took place inthe first 120 to 150 min and that the suspended cell number remained roughlyconstant thereafter. For both BHK and Vero cells Polyhipe S80 had the fastestattachment rate and attained the highest attached/entrapped cell number for thesurface modifications tested. In Figure 33, due to that a large number of the Verocells in the control cultures attached to the wall of -the glass bottle, the totalattached Vero cells includes cells attached to the microcarrier as well as thoseattached to the bottle.944x1073x107 -3x107 -co 2x107C.)Y 2x1071:3 1 X1 07 -a)_cas 1x10 '0 5x10"-- 0- Cortrol --0-- P80- CM80 —V—S80—0--- Flex-CM80100^200^300^4()0Time (min)1 .6x107 -1.4x1071.2x107 -2 toxid -a)-o.c• 8.0x10-as< 6.0x1CP0- 4.0x1CP -2.0x1CP -0^100^200'Time (min)0.0-^- - Control--0--  P80A--- CM80-^Flex-CM80—0-- S80300 400Figure 32. Effect of surface modifications on BHK cell attachment rate. Eachroller has a starting suspension cell concentration of 7.5x10 6 cells /mL andcontains 10 mL DMEM with 10% newborn calf serum.Figure 33. Effect of surface modifications on Vero cell attachment rate. Eachroller has a starting suspension cell concentration of 1.9x10 6 cells /mL andcontains 10 mL DMEM with 10% newborn calf serum.95The influence of surface modification on the rate of attachment of Vero cellswas less dramatic. However, S80 beads still had the highest cellattachment/entrapment rate among all Polyhipe particles tested (Figure 33).Similar experiments were performed on polystyrene microcarriers with other typesof surface chemistry modifications. They included PER80, DEA80, and DEA-LG80 microcarriers. Vero cells attached equally well to the positively chargedDEA80 and the negatively charged S80 microcarriers (see Figure 34). However,BHK cells attached less well to the DEA80 beads than to the S80 microcarriers asshown in Figure 35. Enhanced cell adhesion to polystyrene surfaces treated withsulphuric acid has also been observed by several other investigators (Thomas etal., 1986; Maroudas, 1976; Curtis et al., 1983; Lydon and Foulger, 1988). It shouldbe noted that the treatment by concentrated sulphuric acid at ambient temperature,as in this case, produces mainly hydroxylation and limited sulphonationl. Theresulting S particles were hydrophilic in nature and wetted easily. All other typesof Polyhipe could not be wetted without the use of 70% ethanol.1 Gregory, D. Microporous Materials Ltd., Braunston, Deventry, Northants, UK, personalcommunication (1992).100 20CTime (min)Figure 35. Attachment of BHK cells to treated polystyrene microcarriers. Eachroller contains 10 mL DMEM medium supplemented with 10% newborn calfserum and has a starting suspension cell concentration of 1.6x10 6 cells /mL.100Time (min)Figure 34. Effect of surface modification on Vero cell attachment to polystyrenemicrocarriers. Each roller bottle contains 10 mL DMEM with 10% newborn calfserum and has a starting suspension cell concentration of 2.0x10 6 cells /mL.3.0x1072.5x107 -co2.0x107COva) 1.5x107csvsis- 1 .0x107 -oF-5.0x106 -2.00x1071.50x107 -w00tooxid-0fiSaso 5.00x106400.0000.0 mo0- ^ Control- --A. PER80- p- DEA80- O DEA-LG8030096- -0-•- Control- -0--  D EA80DEA-LG80—V— S80- -0- - PER80976.1.2 Cell attachment/entrapment rate for various microcarriersOf all the modified polystyrene surfaces tested, the S80 microcarriers werefound to be most suitable for animal cell culture. The performance of the S80microcarriers in terms of cell attachment rate was, therefore, compared to that ofthe commercial microcarriers, Cytodex-1, Cultispher—G and collagen—coated P80supplied by Solo Hill. Because of the smaller particle diameters and relatively lowparticle densities, Cytodex-1 and Cultispher—G microcarriers had more "exposed"outer surface area than an equal mass of Polyhipe (with the particle diametersranged from 500 to 1000 pm.) However, cell attachment/entrapment rates wereessentially identical for all tested microcarriers (Figures 36 and 37).6x1& ^5x1CP -t-i; 4x1& -0a)3x1CP -_cC.)2x1CP1x1CP -^ - Control-0-- Cytodex-I ^A CG—V— S80 ----0- - Collegen-Coated^0 ip^I^ I^0 50 100^150^200^250Time (min)Figure 36. Vero cell attachment to various microcarriers. Equal masses of eachtype of microcarriers (9 WL) were used. Each 10 mL cell inoculum had aconcentration of 7 x 105 cells/mL DMEM.2.5x107--0---Collegen-Goated-O-S80A Cytodex-1- v- CG -^-Control98400100 200Time (min)Figure 37. BHK cell attachment to various microcarriers. Each roller contains 10mL DMEM with a initial suspension cell concentration of 2 x 10 6 cells /mL and 9g/L of carrier.6.1.3 Sulphuric acid treatmentThe degree of surface hydroxylation varies with the duration of treatmentwith sulphuric acid. The effect of this treatment on cell attachment/entrapment ratewas investigated using S40 beads for treatment time of 15, 60, and 240 min andfor S80 beads, treated with sulphuric acid for 60 min, Only small differences interms of surface chemistry were found (Table 5, data supplied by themanufacturer).Table 5. Surface characteristics of S microcarriersTreatmentdurationZeta potential(mV)C (%) 0 (%) S (%) Cl (%) N (%) Na (%)S80 60 min -43.77 ± 5.24 64.44 25 5.14 - 3.66 1.76S40-15 15 min -57.85 ± 2.56 78.85 15.21 2.52 - 2.27 1.15S40-60 60 min -50.81 ± 3.32 79.98 14.10 2.53 0.07 1.34 1.98S40-240 240 min -39.04 ± 3.84 70.71 21.13 3.75 0.07 1.60 2.7499As is evident from Figures 38 and 39, cell attachment/entrapment rates weresimilar for all the S beads, despite differences in pore sizes (i.e. 80 um vs. 40 /um)and treatment time.0.03.0x106U)0 2.5x106-c02.0x106U)1.5x106-0ar4E'a)0 1 . CX1 06 -C.)0 5.0x105 -YCO—0— S80--0-- S40-15A S40-60- -p- - S40-240-..-0---  Control0^100 ^300ITime (min)Figure 38. Attachment of BHK cells to sulphuric acid treated Polyhipemicrocarriers as a function of time following treatment length of 15, 60 and 240min. BHK cells were suspended in the 10 mL DMEM / 10% NCS medium.6.1.4 Influence of inoculation procedureThe addition of a concentrated cell inoculum to a dry microcarrierpreparation might enhance the initial entrapment and cell penetration into thecarrier particles. However, the entrapped air within the pores of the dry S Polyhipeprevented the uptake of medium. Most of the particles were not wettedimmediately upon the addition of the medium and floated on the medium surface.4.0x1&E 3.5x1CP3.0x1CP0a) 2.5x105o_.c 2.0x1CP -c01- .5x10 -c0O 1.0x1of -a)02 5.0x105 -a)0.0—0— S80—0-- S40-15A S40-60- S40-240- -0- - Control100Non—wetted S Polyhipe, like other non—wetted polystyrene surfaces, did notsupport cell attachment/entrapment (Figure 40). The difference in porosity of theparticles had little effect on cell attachment rate. The decrease in suspension Verocell concentration in the medium was due to cells attached to the bottom of theglass roller. Autoclaving of the S particles suspended in PBS released most of theentrapped air. The particles formed a suspension upon the addition of DMEM/10% NCS medium. The S beads prepared according to this protocol exhibitedenhanced cell adhesion properties, as one can see from Figures 38 and 39.0^100^200^300Time (min)Figure 39. Attachment of Vero cells to sulphuric acid treated Polyhipemicrocarriers as a function of time following treatment length of 15, 60 and 240min. Vero cells were suspended in the 10 mL DMEM / 10% NCS medium.101—0— S80-Wet--0-- S80-DryA S40-WetS40-Dry-^- Control3.0x10E 2.5x10U)Tf)02.0x1rYc 1.5x10002 1.0x10a)0(7)c 5.0x1050.U)0.00^50^100^150^200^250Time (min)Figure 40. Cell attachment to dry particles. Vero cell suspensions were addeddirectly to the sterile dry S40 and S80 microcarriers.6.1.5 Effect of particle diameterThe S40 (15 min concentrated sulphuric acid treatment) were sieved anddivided into three different particle diameter ranges: less than 0.3 mm, between0.3-0.5 mm and between 0.5-1.0 mm. The purpose of the experiment was todetermine the importance of outer "exposed" surface area. High microcarrierloading (9 g/L) of all three diameter ranges S40 microcarriers provided excesstotal surface area even with complete cell attachment. However, the area of theparticle surface exposed to the cell suspension was significantly higher for theparticles with the smallest diameter. If the cells penetrate quickly into the innerpores of the S40 particles at the start, then the cell attachment/entrapment rateshould not be affected by the particle diameter. The number of Vero cells attached102to the smaller diameter particles was 50% higher than the corresponding numberattached to the larger particles during the first 60 min (i.e. 1.5 x 10 7 vs. 1 x 107total attached cells, as shown in Figure 41) suggesting that initial attachment waspredominantly to the outer exposed particle surface.Ec o750cOcTs.Ea)0CO0a)00a)>c0_caca)a_ca=CI)I^ I0^100 200^300Time (min)Figure 41. Effect of particle diameter on the attachment rate of Vero cells toPolyhipe particles treated with sulphuric acid. Each roller conatins 10 mL DMEMmedium with 10% newborn calf serum and 9 g S40 carrier /L.6.1.6 Influence of the inoculum cell concentrationThe inoculum concentration might be expected to influence the cellattachment rate. However, for the range of Vero inoculum concentrations used,there was little or no difference among the normalized attachment rates (Figure42). The high carrier loading of 9 g/L provided sufficient surface area for initial103attachment up to 2.8 x 106 Vero cells. The saturation cell number was about 3.5 x108 BHK and 6 x 108 Vero cells per gram of S40 microcarriers, respectively.90zr) 80-(.2• 50-a)0_cas 40-as1^1^1^•^I^IfVero cells with 9 g/L S40 beads—1:1-5.65x16 cells/mL--0--2.82x16 cells/mLA 1.41x1CP cells/mL- -v- -0.71x16 cells/mL- -0- -controlr0^20^40^60^80^100 120 140 160Time (min)Figure 42. Inoculum concentration effect on the Vero cell attachment rate. Thecontrol was inoculated with 3 x 106 cells/mL. Each roller conatins 10 mL DMEMmedium with 10% newborn calf serum and 9 g S40 carrier /L.6.1.7 Effect of medium composition on cell attachment rateCell attachment rate can be strongly influenced by the medium composition(Hu et al., 1985). The presence of serum was shown to inhibit attachment andentrapment of 2E11 hybridoma cells (See Section 6.2.7). For Vero cells thereappeared to be no significant effect of medium composition on attachment rate(Figure 43) for the range of conditions investigated.104JE 5.0x1d-5c 4.0x10-iT55a)0 3.0x10 -0a)2.0x1& -Ea)1.0x10-—o—DMEM + 10% NCS—0—MEWF-12 + 10% NCSA^DMEM/F-12- -v- -PBS----'-DMEM + 10% NCS (Control)I0^50^100^150^200^2500.0300Time (min)Figure 43. Effect of medium composition on Vero cell attachment rates to S40microcarriers. The initial pH of the various media was 7.2. Cell viability wasgreater than 90% for all cells except those in PBS. Viability of the Vero cells inPBS decreased from 95% to 60% after 4 hours. The control contained nomicrocarrier and cells were suspended in DMEM with 10% NCS.6.2 Long Term Microcarrier CulturesPrevious work showed that polystyrene particles treated with sulphuric acidhad the highest cell attachment rates. However, this does not guarantee that thelong term cell culture performance of S Polyhipe will be superior to carriers withdifferent surface modifications. Long term cell culture experiments werenecessary to determine the effect of particle surface chemistry modification on cellgrowth characteristics.1056.2.1 Biomass evaluationThe sponge—like open pore structure of the styrene based microcarriers isillustrated in Figure 7. Direct evaluation of biomass within these microcarriers isdifficult. Cell growth could not be monitored by light microscopy due to theopaque nature of the Polyhipe particles. Glucose uptake rate (GUR) was used asan indicator of cell growth for the fixed bed ceramic perfusion system. Forcultures in exponential growth, under constant operating conditions, the rate ofglucose consumption is proportional to the cell concentration. However, the use ofGUR could be misleading depending on changes, for instance, in glucosedepletion 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 therelationship between GUR and cell number changed as illustrated by Figure 44.For porous bead cell loading calculations, the GUR was, therefore, determined inthe glucose concentration range between 4.5 and 2 g/L after each mediumexchange. 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 cellnumbers, several methods of enumeration were compared. Direct cell numberenumeration techniques were investigated. Treatment of Polyhipe microcarrierswith trypsin released about 10% of the estimated total cell number, based on GUR.Alternatively, the particles were resuspended in a solution of crystal violet (0.1g/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 estimatedcell number based on GUR could be accounted for. Presumably the pore structureof the macroporous beads entraps cells and prevents complete recovery. Nuclei—0— 1 x 108 cells/ mL--o-- 5 x 107 cells / mLI^•^•a1^•^I I^I I^•^I^•^I106counts recovered from cells grown on Cytodex microcarriers were within 10% ofthe cell number estimated from GUR. 1.2as^•CC1.0cur).0.8a)In0 0.5^1.0^1.5^2.0^2.5^3.0^3.5^4 0Medium Glucose Concentration (g/L)Figure 44 Effect of medium glucose concentration on BHK cell glucose uptakerate. Cells were allowed to grow to confluency in parellel T—flasks and themedium 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 massover the course of suspension hybridoma 2E11 batch cultures (Sonderhoff et al.,1992). In the present work, known amounts of the porous microcarriers wereextracted with trichloracetic acid (2.5%) and the ATP released was assayed. Thetotal cell number immobilized in the microcarrier was then calculated based on theaverage value of ATP obtained from BHK suspension cells. The cell numberestimation based on the ATP assay was close to the estimate from the glucoseuptake rate. Table 6 summarizes the results for steady state cell number estimation107on CM80 microcarriers (0.5 g/L) based on three different methods. The pattern ofglucose utilization for the CM80 BHK culture is shown in Figure 45; the mediumwas changed whenever the glucose concentration in the medium dropped below 1gil, . The culture reached steady state after 250 h.Table 6. Comparison of estimated cell number (cells / mL beads) by alternativemethodsMethods Nuclei Count ATPConcentrationGlucose UptakeRatets^.4 fmol/cell 0.46 x 10-9g/cell/dayEstimated cellnumber permL carriers1.2x108 2.1x108 2.5 x 108001C0I..3a'000=(7900C000 2-3-4-00^50^100^150^200^250•^1^•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 theglucose concentration dropped below 1 g/L. 5 x 10 7 cells were introduced asinoculum.1086.2.2 Cell growth on Polyhipe in roller bottles6.2.2.1 Vero cell growth, effect of carrier surface modificationThe performance of the porous polystyrene particles with the followingsurface chemical group modifications was tested using Vero cells. Themodifications included CM, DEA, S, 0, and CS. The pattern of glucose utilizationfor Vero cells growing on CM Polyhipe is shown in Figure 46. Figure 47 showsscanning 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 attachedto the laminae are not easily distinguished in this micrograph. Similar cultureconditions were used for all the Polyhipe particles. Figure 48 gives the cumulativeglucose utilization profile for these cultures. It shows that the glucose consumptionis significantly less for the cells grown on the Q Polyhipe and CS Polyhipe. The5a 4 -oC0""6 30UCO00^100^200^300^400=^I^=^ITime (h)Figure 46. Glucose concentrations in semi-continuous perfusion of Vero cells onCM80. The roller culture was inoculated with 1 x 10 8 cells and contained 0.5 gCM80 beads/L DMEM with 10% Fetal calf serum.0500109profiles for the other cultures appear to be virtually identical.The glucose utilization rate (from the slope of the glucose concentrationbetween 2 and 4.5 g/L vs. time plot, e.g. Figure 46) of vero cells on particles withdifferent modifications was used to estimate cell number as a function of culturetime for the Polyhipe cultures. In most cases the cell number increased rapidly forabout 200 h and then stabilized at a value of approximately 1.25 x 10 8 cells/mL ofmatrix. Different behaviour was found for the vero cells on Q80 and CS80polystyrene particles (see Figure 49). Figure 49 showed that polystyrenemicrocarriers modified by addition of chloromethyl, diethylamine group, andsulphonation can support vero cell growth equally well. Surprisingly, vero cellsgrowth on the modified polystyrene particles (P40 and P80) was comparable tothat on the charged DEA80, S80 and CM80 particles. It should be noted that thenon—wetted Polyhipe polystyrene particles do not support any cell growth.Figure 47. Vero cells on DEA80 particles.1103.0 —2.5 —0.5 —If^•0^100^200 300Time (h)0.0500400o S80^ CM80xFigure 48. Cumulative glucose utilization by Vero on particles with differentmodifications. Operating conditions are outlined in Figure 46.1.5001.251.00 —•0"5 0.75-JE(I) 0.50—CD2>• 0.25 —0.006 6^ ^6 ^ 60^XO x xxCbO Ff A^ ^ 6 pX6 4 + o DEA80A P400 0 + +v P80+ 080x CS800^100^200^300^400Time (h).e? o A^x500Figure 49. Calculated Vero cell densities on various Polyhipe particles based onglucose utilization rates. Operating conditions are outlined in Figure 46.30.00 -13^25.00 -^ aOO^20.00 -^ O8810.00 -OOT200^300Time (h)400^5005.00 - 80.000^100^^ Cytodex-1 Io DEA801116.2.2.2 BHK cell growth, effect of carrier surface modificationThe growth characteristics of BHK cells on Polyhipe were similar to thoseof Vero cells. In roller bottles, the cells grew predominantly on and within theparticles. The performance of Polyhipe compares well to that of Cytodex 1 inthese cultures as shown in Figure 50. DEA80 Polyhipe was used here as itrepresents the surface modification most comparable to DEAE dextran (Cytodex1). The DEA80 beads had an exchange capacity of 1.9 meq/g, similar to thereported optimum exchange capacity range (1.5 — 2.5) for DEAE Sephadexmicrocarriers (Levine et al., 1979). The optimal exchange capacity depends onboth the cell type and the microcarrier material (Himes and Hu, 1987); hence, theoptimal exchange capacity of the DEA80 beads may not be 1.9 meq/g, particularlybecause the charge groups are on the cell growth surface, in contrast to the DEA80Figure 50. Cumulative glucose used by BHK cells grown on 5 g/L ofmicrocarriers in DMEM/10% fetal calf serum with 10 ,uM zinc. 2 x 10 8 cells wereintroduced to the roller cultures as inoculum.20.00a-6) 15.00--oa)ci)Da)eno 10.00-c)=5Tu1-2 5.00-BHK cells on 2 g/L carriers^4i1is. °c; ooo Cytodex-1^ Cultispher-GO CM80^ P80I^I^10^100 200 300Time (h)0.00400 500112particles where the exchange capacity is uniformly distributed throughout theparticle. 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 wereapproximately 6.1 x 107 cells mL-1 of Cytodex-1 and 8.5 x 107 cells mL-1 ofDEA80. It should be noted that the microcarrier volumes were over—estimatedsince the settled volumes of the microcarriers, suspended in PBS, were used forthe cell concentration calculation. The actual microcarier cell concentrations couldbe 30% higher. The particle diameter of DEA80 was greater than that of Cytodex-1 (i.e. 500 to 1000 gm vs. 131 gm). Therefore, the outer 'exposed' surface area tovolume ratio of the DEA80 beads was significantly lower than that of theCytodex-1. BHK cell growth within the pores of the DEA80 particles presumablycompensates for the lower outer surface area.Figure 51. Cumulative glucose uptake of BHK cells on various types ofmicrocarriers with a concentration of 2 g/L. Operating conditions as per Figure50.113Growth of BHK cells on CM80, Cytodex I and Cultispher—G particles (2g/L) in roller bottles was also investigated. Figure 51 shows that the growthcharacteristics of BHK on the three types of microcarriers were similar based onthe glucose utilization rates. Final cell concentrations were estimated to be 9.4 x107 cell mL-1 of Cytodex, 1.4 x 108 cells mL-1 of CM80 and 1.0 x 108 cellsmL-1 of Cultispher—G. Cultispher—G particles have a lower void volume withinthe particles compared to Polyhipe particles (i.e. 50% vs. 90%). This likelyaccounts for the lower cell concentration obtained with the Cultispher—G cultures.BHK cells grew equally well on Cytodex-1 and CM80 particles atmicrocarrier concentrations of 2 or 5 g/L. However, the increase in microcarrierconcentration 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 foldincrease in the glucose uptake rate (see the slopes of Figures 51 and 52). Adecrease in the cell number per unit volume of microcarrier with increased carrierloading was also noted by Smiley et al. (1989) for CHO cells grown on Cytodex-1microcarrier. Croughan et al. (1988) found that increasing the microcarrierconcentration by a factor of two resulted in a 25% reduction in average growthrate of FS-4 cells. They attributed this to an increased frequency of collisionsbetween the microcarriers. They also found that the growth kinetics of FS-4 cellswere strongly influenced by the cell concentration. Nutrient limitations caused byinsufficient medium exchange, uncontrolled medium pH, and high oxygenconsumption rate are all factors that may also prevent the cell concentration fromincreasing in proportion to the microcarrier concentration.I^I^.^1BHK cells on 5 g/L carriers t 0go08^_oo0o®^-0 OOO88o g0 g o ° 8 o Cytodex-1O^ur v Cultispher-GoO^iI1cr uO CM8011430.0025.00-a, 20.00goa)U)=15.00U)O0=(5 10.00ccoI-5.000.000^100^200^300^400^500Time (h)Figure 52. Microcarrier (5g/L) cumulative glucose consumption of BHK rollercultures in 100 mL DMEM medium. Each roller culture was inoculated with2x108 cells.6.2.23 Transferrin productionThe cumulative transferrin produced versus time for various microcarriers isshown in Figure 53. The transferrin production rate of BHK cells grown onCytodex was greater than for those grown on CM80 or Cultispher-Gmicrocarriers. At steady state (after 100 h), the transferrin production rates were115, 101 and 90 gg/10 9 cells/h for BHK cells grown on 2 g/L of Cytodex, CM80and Cultispher-G microcarriers, respectively. Mignot et al. (1990) reported lowercell specific productivity of CHO cells (i.e. ,ug/cells/h) when grown on Cultispher-G cultures compared to Cytodex-3. Protein secretion can be limited to theexposed cell surface area available for protein transport. Observed under themicroscope, cells grown on the Cytodex-1 microcarriers generally formed amonolayer of cells or a multilayer less than 5 cells thick. The Cytodex-1115microcarrier also has a much smaller particle diameter. The total exposed cellsurface area to cell number ratio is higher for cells on Cytodex-1 than for theother two types of microcarriers. The low level of agitation used in this study mayalso introduce a diffusion limitation which is unfavorable to the cells grown withinthe pores of the CM80 and Cultispher—G cultures. Either of these factors mayexplain the observed reduction in the transferrin production rate.Time (h)Figure 53. Cumulative transferrin produced in BHK roller cultures containing 2 gmicrocarrier/L DMEM medium, supplemented with 10 fiM zinc.6.2.3 Particle pore size effectThe particle pore size had little or no effect on cell attachment rate as onecan 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, aseries of 100 mL glass rollers with 2 g/L particles suspended in 30 mL medium0is500^60010023.3320.0016.6713.3310.006.673.330.000 200^300^400Time (h)0I^•••• ^▪ P5•• o • P25•♦^A • P40• • o P80a ° 1 g Iarj •i Is0i1i i e**• • ••• •00g0.0 •O 0Og0 •0 a •a tt0 •Oft^• •a 2^ ^ t a •116was used. Each roller was inoculated with 107 BHK cells. BHK cell growth wassimilar on polystyrene particles with different pore size (Figure 54). One of the P5cultures failed to grow and had to be re—inoculated at 200 h. All cultureseventually reached similar steady state glucose uptake rates corresponding to cellconcentrations of 2.2 — 2.4 x 10 8 cells/ mL particles.Figure 54. Total glucose used by BHK cells on polystyrene particles of differentpore sizes. Each roller was inoculated with 10 7 BHK cells and contained 2 gmicrocarrier /L medium.Failure to observe differences in cell growth with particles of different poresizes might be caused by mass transfer limitations which restrict growth within theparticle. A second series of experiments using duplicate 250 mL spinner flaskswith low particle loading (0.5 g/L) was performed. It was reasoned that the highagitation introduced by the use of the spinner, low particle loading and frequentmedium exchange (i.e. medium was changed whenever glucose concentrationdropped below 2 g/L) would reduce any diffusion limitation. Before cell culturing,17.50 •^1Effect of pore size on the immobilized BHK cellsgrowth on polystyrene beads15.00-12.5010.00-ocr,▪ 7.50 -CO=Cu1 o 5.00 -Og X2.50-1^•^1^•^1^'^1^•^1^•0^50 100 150 200 250 300^ o P5A v P25P40x x P800.00exX+^OOOV v-VA A A .350 400 450^ -X^3K ^0+ 6 ^ 0O15 ^OO0x_ g X117all particles were sieved to ensure that each type of microcarrier tested had aparticle diameter between 250 and 500 ,um. P25 particles were not wetted andmost of the particles were washed out after the first medium change. Cell growthwas similar for all particles except for P25 cultures (Figure 55). The glucoseuptake rates stabilized at 0.018 g/h per L medium (except for the P25 BHKcultures) equivalent to a cell concentration of 1.56 x 10 8 cells/mL carriers.Time (h)Figure 55. Cumulative glucose used for BHK polystyrene microcarrier spinnerculture. Each spinner contains 0.5 g particle/L DMEM / 10% NCS medium.In these experiments, the BHK cells formed cell clumps with diameters up to0.5 mm. These clumps were difficult to separate from the similar sizedmicrocarriers. Thus, the measured BHK glucose uptake rates reflected the totalcell population within the spinners instead of those on the porous beads. Toovercome this problem the experiments were repeated using Vero cells. Vero cells17.5I0^100^200^300^400Culture Time (h)Figure 56. Cumulative glucose consumed by Vero cells on P microcarriers. Eachspinner was inoculated with 7.8 x 10 7 Vero cells and contained 0.5 g microcarrierper L DMEM medium..b)-FD:1 10.0-co(i)7.5 -=Tses 5.0 -1—12.5-15.0-0.02.5-Ifla135A v P25o + P40x * P80500118grow preferentially on surfaces and do not usually form cell clumps. The glucoseuptake rate of Vero cultures should therefore indicate the attached cell populationonly. However, no differences in total glucose uptake rates or lactate productionrates were evident for any of cultures (see the slopes of Figures 56 and 57).These results for both BHK and Vero cells indicated that pore size did notinfluence cell growth. Since the P5 pore size of 5 yin (nominal) is smaller than thesize of either cell type, it appears that cell growth was mainly on the outer surfaceof the particles.11912.5to88X 0o + P40x x P8010.0 -0.0^04000000^ P5A v P252.5-0 100^200^300^400^500Time (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 cellpenetration into the particles. However, the laser beams did not penetrate thepolystyrene structures resulting in dark shadows in the microscope images (Figure58). 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 ofthe P5 particles was too small for cell entry; most of the Vero cells grew on theouter surface of the particles. Little difference in pore sizes between the P40 andP25 particles was found. Cells appeared to grow equally well on both types ofparticle. P80 particles had many large void spaces throughout the particles,indicated by the empty spaces in the middle of the thin sectioned microscopeimage. The true effect of pore size on cell loading might be masked by the120inconsistencies found with the particle structures and size distributions. Accordingto the manufacturer, it was difficult to manufacture particles of 80 aum pore sizewith the same structural consistency as particles with 40 ,um pore size. The largervoid space reduced the total available space for cell attachment and entrapment.Therefore, particles with 40 ,um pore size were used for subsequent long termculture studies.Figure 58. Confocal images of BHK cells growing on a P80 Polyhipe particle. Theparticle has been stained with fluroscein diacetate (green) to indicate viable cellsand propidium iodide (red) to indicate dead cells. Overlayed green and red imagesare shown in white ( = 100 ,um).(A)121(B)(D) r•11..**^-Figure 59. Thin sectioned microscopy images of Vero cells on polystyreneparticles 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.(C)1221236.2.4 Minimum cell inoculum requirementA minimum inoculum concentration is needed to initiate cell attachment andsubsequent cell growth for animal cell microcarrier cultures (Hu et al., 1985;Forestell et al., 1992). This cell number depends on the medium composition, themedium pH level, the microcarrier type and the cell type. An inoculum of at least1-2 cells per bead is generally recommended to minimize the number ofunpopulated microcarriers (Forestell et al., 1992). The effect of initial inoculumconcentration on Vero cell grown on S40 and Cytodex-1 was examined usingeither 100 or 250 mL spinner cultures containing 2 g beads per L DMEM medium.Duplicate spinners were inoculated with 1.36 x 105, 2.72x 105, 4.08x 105 and5.44x 105 cells per mL medium — corresponding to 1, 2, 3, and 4 Vero cells perCytodex-1 microcarrier, respectively. The volume of the larger S40 particles wasup to 20 times greater than that of Cytodex-1. Therefore, the approximate S40microcarrier number per g dry weight could be 20 times less for Cytodex-1. Dueto the irregular shapes of the particles, the exact number of S40 carrier particlesper g dry weight was difficult to estimate. Figures 60 to 63 illustrate that cellsreached confluency faster on Cytodex-1 than on S40 beads, especially at lowinoculation concentration, even though the S40 microcarriers were inoculated withhigher cell—to—bead ratios. With the lowest inoculation concentration (i.e. level #1of Figure 60 to 63), the glucose uptake rate of the Cytodex-1 cultures reachedsteady state after 200 h. However, it took 350 h for the corresponding S40 culturesto reach steady state. Despite the lower steady state glucose uptake rate for S40Vero cultures (i.e. 0.075 vs. 0.1 g/L/h), the final estimated cell concentration washigher (2.7 x 108 and 2.5 x 108 cells / mL carrier for S40 and Cytodex,respectively). The less dense Cytodex-1 carriers had more carrier volume on thebasis of equal mass. The high Vero cell loading on the S40 particle is confirmedai vVero Cylodex-1 cultures with differentinoculating cell concentrations10-1X?AM (5I w 8I a0i 115a0411--1,----1--1--11--,^0^100^200^300^400^500^600Time (h)8aa 0 x 8^At ^X s3 .^Tic xx0A3K 6 X gx 8* Ar 80^ o Level #1o + Level #2• x Level #3^ X Level #404o-30-VDto-6cn§ 20-E140-35-..-.. 30-V,0 25-D00 20-z10-5-002 + ^ 0124by the scanning electron micrographs (Figure 64). The cut—open particles hadVero cells growing throughout the entire particles.Figure 60. Cumulative glucose used by Vero cells on Cytodex-1 microcarriers.45Vero S40 cultures with differentinoculating cell concentrationsxxXxi too0+xLevel #1Level #2Level #3V Kt Level #4 1 +X^X ^^0 A +iiiVX ii A +^^^X i I +^^ 0x ii + + ^ 02+A^^ 0^x a + t,^a ox ik, AE A^^ ^^oxiI ° 0 0 0 0 0ODD °o o °1888 00A 1 *100^200^300^400^500^600Time (h)00Figure 61. Cumulative glucose used by Vero cells on S40 microcarriers.X80ill^00x^0^ 0 Level #1o^+ Level #2A^x Level #3v^I( Level #40805 10 -n5-0 •0^^ 0^ 0^ 00O• wlt I100),(v0 Ixg i6^ ^ 4s1x ^a ^^118  0 g 0200^300^4000500^60040125Vero Cytodex-1 cultures with differentinoculating cell concentrations0 .*O1 I0^100^200 400 500Time (h)Figure 62. Cumulative lactate produced by Vero cells on Cytodex-1 microcarriers.Vero S40 cultures with differentinoculating cell concentrations060030-30-a2 20-fU_J10-x^ o Level #1o +xYiLevel #2Level #3Level #4 xxx 0^)K+t35-30-'LT-6 25-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 hculture time. (A) and (B) are two views of the same particles.60-oco-6) 40-S 30-010-50- 8 A -8 A _5AOA0A^ A0AOARR -RA^_ oROA^R a02 RR^-RP IR .01R^RR ^R OOgl A aR-n nnn^ 840o CytodexA Cultispher-G01276.2.5 Cell growth on S40 Polyhipe in spinnersBHK Cell growth of the S40 Polyhipe was compared with that of Cytodex-1and Cultispher—G microcarriers. Carrier concentration was 2.0 g/L. The culturemedium was changed whenever the glucose concentration dropped below 2 g/L.The diffusion limitations and the severity of the concentration gradients within theparticles should be reduced by frequent replacement of medium. 250 mL spinnercultures were initiated with a high inoculum concentration (i.e. 10 8 BHK cellseach). As indicated in Figure 65, little difference was noted in the cumulativeglucose used or the lactate produced in the three microcarrier cultures over the 600h culture time.807060O50 r-40 a)30 CoaCD0-20 (4,r-1000^100^200^300^400^500^600Time (h)Figure 65. Cumulative glucose used and lactate produced by BHK cells onmicrocarriers (2 g/L) in 200 mL medium. Each 250 mL spinner culture wasinoculated with 10 8 cells.The glucose uptake rates of all three cultures reached steady state values ofapproximately 0.1 WL/h after 200 h as shown in Figure 66. The estimated total cell0.12o0 066OAA0600^ 2 ORCIOR0.10..7.1......^ 0 A `-' Og^0^0.09 6,06 °s H o 8 flo A 4^EtA8as 0.08^oCCa ^2...= 0.07al^AI s I 0".=D 0.06^^O Ari)O 0c.) 0.05^A=C7o S40^ CytodexA Cultispher-GI^'^I^r^.^I^,^i1 00 200 300 400 500^6000.110.040.03 ..r0.00 1 0128number in each of the 250 mL spinners was 1.04 x 10 9 cells (or 5.21 x 106 BHKcells/mL medium). Because S40 particles are more dense than Cytodex-1 and CGparticles, the carrier volume based on unit mass is reduced. Since the process canbe scaled up in both process intensity (i.e. carrier loading) and volume, it is moreappropriate to calculate the cell loading based on the carrier volume instead of thetotal 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 toaccommodate cell growth compared to Cultispher—G (90% vs. 50%). Theestimated BHK cell density on the S40 was 2.0 x 10 8 cells/ mL carrier after a 600h culture period, in contrast to 1.3 x 108 cells / mL carrier for both the Cytodex-1and Cultispher—G.Figure 66. Glucose utilization rates of various BHK microcarrier spinner cultures.Operating conditions as outlined in Figure 65.129000.PP^AA'0^PP^A 0"P" o'EI)3^s oA ' o 'P'^goOp^ts, ,si0.05 -200^300^400^500^600^700Time (h)Figure 67. Unit carrier cumulative glucose used by BHK cells on microcarriers.Operating conditions as outlined in Figure Cell penetration depthFigure 68 shows the SEM of S40 microcarrier particles after being culturedfor 600 h. The beads shown are near spherical and covered with BHK cells. Thesectioned particle shows that BHK cells grew throughout the center of the particle.Mass transfer limitation was evident as indicated by the cell concentrationgradient; with the highest cell density at the outer perimeter of the particle. Thetheoretical oxygen penetration depth could be used to estimate the maximumdistance for cells to infiltrate into the particle. It should be noted that cells onlyneed to penetrate 30% of the diameter to utilize 70% of the sphere volume. Thefollowing assumptions were made to characterize oxygen diffusion andconsumption:0.500.45 -0.40 -•6- 0.35 -c.)-JE 0.30 -cn^.Fp 0.25 -U)f,,D) 0.20 -0(-3- 0.15 -To• 0.10 -° ^0.00 s0 1 Oo6o'o'^6o'^,ffn 11'1—0— 340—0-- Cytodex-I—A— Cultispher-G130(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 :Deff (I--ir2dCs )=Q X = Qr2 dr^dr^02Where Deff = effective diffusivity;r = radial coordinate;Q = total oxygen consumtion rate;X = biomass;002 = specific oxygen uptake rate.The boundary conditions are:(1) at r = ri (outer perimeter)Cs=Cb *(2) at r = rd (radial position at which oxygen concentration is zero)Cs=0dC s _ 0dr —Equation (36) can be solved with these boundary conditions to give:C^[0.2 _ rd 2 ) _ 2 rd 2 in (___r )]s = 6DQeff^rd(37)The penetration depth of oxygen (PD = ri — rd) for the case of no externaldiffusion limitation can be estimated to be*1p _ [ 6 Defff b  ]4D^Q(36)(38)131The theoretical cell penetration depth of 262 ,um into a particle agreed well withthe 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 inwater)(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 thebulk medium or a reduction of effective diffusivity from 2.65x10 -5 cm2/s to about1.3 x 10-5 cm2/s).(3) Biomass of 2 x 10 8 cells/mL immobilized within the particle (i.e. Q = 40mmoles/L/h).Combinations of other dissolved oxygen levels and various appropriate voidfractions could also give rise to a PD of 262 ,um. The measured dissolved oxygenconcentrations varied from 100% air saturation after the medium had just beenreplaced 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 fractionnear the outer perimeter of the particle is low, while the interior of the particle isalmost completely free of cells.132Figure 68. (A) Scanning electron micrograph of S4() particles covered with BEAKcells (B) A cut—opened S40 particle revealing the central portion of themicrocarrier.1336.2.7 Hybridoma cell growth in Polyhipe particlesCollagen—based macroporous beads are often used to grow hybridoma cells(Almgren, 1991). Hybridoma cells only attach loosely to surfaces. Thus physicalentrapment within the three dimensional structure of the beads is thought to be themain mechanism of immobilization (Almgren, 1991). The polystyrenemacroporous 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 protocolas for the Vero and BHK roller cultures. Under these seeding conditions the cellsdid not grow selectively within the pores of the particle. Most of the cells in theculture failed to be trapped within the carrier. Himes and Hu (1985) measuredincreased rates of CHO cell attachment to Cytodex (Pharmacia) microcarrierswhen DMEM/serum was replaced by Ca++ and Mg++ free PBS. A second seriesof experiments was therefore performed in which the cells were incubated with themacroporous beads in PBS (without Ca++ and Mg++) for 1 h without agitation.The particles were then washed once with serum containing medium and finallyresuspended in medium containing 10% fetal calf serum. This technique promotedattachment and growth of 2E11 cells within chloromethyl beads (but was noteffective for the unmodified polystyrene or those treated with sulphuric acid).Proteins in the culture medium might bind preferentially to the particle surfacesand reduce the mammalian cell attachment. Thus, in the absence of serum, theremay be an opportunity for direct binding of the hybridomas to the surfaces of thebeads.Figures 69 and 70 show scanning electron micrographs of the surface of theCM80 beads containing hybridoma cells. These confirm that cells were entrappedin the pores, even near the surface of the particles. Figure 71 shows the changes ofcell concentration freely suspended in the culture fluid and the glucoseconcentration as a function of time. It can be seen that the cells grow rapidly134outside the particles and failed to be trapped within the carrier after each mediumchange. The glucose uptake rate increased (see the slope of the Figure 72) despitesuspension cell removal during each medium change, indicating an increase in thenumber of cells within the particles. Release of hybridomas from these surfaceregions are suspected to be responsible for the continued presence of suspensioncells, particularly during the startup periods of these cultures. The loss of cells tothe surrounding medium prolonged the period required for the hybridomas toestablish a high steady state immobilized cell density (400 h compared with 150 hfor BHK cells under similar inoculation and growth conditions). The maximumcell concentration in the macroporous bead culture was about 3 x 10 7 cells/mL ofcarrier, about 20% of the concentration achieved using the Vero cells. Understeady state conditions, the antibody concentration of the 2E11 cells grown onCM80 macroporous beads reached 40 mg/L of monoclonal antibody every 2 days(when the culture medium was changed). Antibody production rate increased from0.63 to 0.89 mg/L/h while the glucose uptake rate remained relatively stable (seeFigure 72). This indicated that the CM80 Polyhipe particles retained most of thestarting hybridoma cell population, since the antibody production rate is related tocell growth rate. Antibody productivity increased as the cell growth ratedecreased.135Figure 69. Hybridoma cells growing within the pores of chloromethyl PolyhipeparticleFigure 70. Enlarged view of hybridoma cells growing in a pore of a chloromethylPolyhipe particle.1361004-801O80E40 TDC0C20 0a_050^100^150^200^300Time (h)O00)c 3-o•t;Ca)UCO2-•••■I1Steady-state GURa▪ )O00.8rn0.6CDCC00.4 =20_0020.0Figure 71. Growth of hybridoma cells with chloromethyl Polyhipe (5 g/L) in aroller bottle. 2 x 107 cells suspended in PBS without serum were used asinoculum; PBS solution was later replaced by 100 mL DMEM with 10% fetal calfserum.1.0• Antibodyo Glucose •CaO/— to-0400Time (h)Figure 72. Cumulative glucose utilization and antibody production rate ofhybridoma cells entrapped in 5 g/L of CM80 macroporous beads.0^200i ,6008001377.0 RESULTS AND DISCUSSION —EFFECT OF CULTURE SYSTEMSON CELL GROWTH AND CELL PRODUCTIVITY7.1 Comparison of Cell Specific Transferrin Productivity of Different CultureSystemsBHK cells were also grown in standard suspension cultures to compare therate of transferrin production with cells grown in the fixed bed ceramic perfusionsystem. 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 cellscultivated in suspension. The cell concentration was about 5.5 x 10 6 cells/mLduring the stationary phase in the spinner flask. The transferrin production rate ofthe suspended BHK cells during the exponential phase of growth in the spinnerflask (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 matrices253.0 -0..,^ lactate2.^\- ^ 0 0-0-0-0 0-6 \ 0 (V ,A—A-•_••co^- 0 ,..._....._,o^13^/ - -- Aglucose \ 0 •^transferrintill 0- -c.'o 2.0 -^ ,^/^ /ic^ 0 A^O 0a) 1.5-co ,)s ,♦^cell concentrationas^ o 0^ -0 \^.•• •.-J^1 .0 '-'^ A'8a)v)o^o/PPP^ 1.;\..0.0--,^ACD=/0 4^\ET 0_0 r.)--^.^.__...t ii-0.0-^• 4I _ t•—/A. --^0^40^80^120^160^200Process Time (h)I^'^I 020—_1 .__I• u3"-bo• 7.<15=ca) Ca100 oC_)a)co^(1)c5 EsI-Figure 73. Transferrin production by BHK cells in spinner suspension culture withDMEM/F-12 medium supplemented with 10 ,uM zinc and 5% fetal calf serum.500400 -—0— S40—0— Cytodex—A— Cultispher-G(73 300 --0a)C.)2200100 - DMEM00^100^200^300^400^500^600Time (h)OodPinDMEM F-12 0 OWav-•,o ,e1,o n000 4y—o,^DMEM w 5OL M zinco":ro;fr138(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 alsocompared with cells on Cultispher—G and Cytodex-1 microcarriers. Figure 74shows that the transferrin productivity per g of carrier was similar for the threemicrocarriers tested. However, because of their higher density, the transferrinproductivity per unit carrier volume for the S40 beads was 33% greater than forCytodex-1 and CG particles; this is shown in Figure 75.Transferring the cultures from DMEM to DMEM/F-12 did not induceincreased transferrin production rate (see slopes of Figure 74). In fact, thetransferrin production rate decreased from 3.46 mg/109 cells/day to 2.30 mg/109cells/day (or 7.5 to 5 mg HTF/ g—glucose), suggesting that the low glucoseconcentration of DMEM/F-12 might have resulted in diffusion limitations. It islikely that the induction due to the small amount of zinc present in F-12 (about 3Figure 74. Total transferrin produced by BHK on microcarriers (2 g/L) in 200 mLmedium. Each 250 mL spinner culture was inoculated with 108 cells.10/—0— S40—0— Cytodex-I—A— Cultispher-Gp' 46crEro .4 45^ JA/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).3-a)Ue2 .r...a)ca 2 -.= 0•1r)Ens cr)E1 -I-0d I^I^ I0^100^200^300^400 _ 500^600Time (h)Figure 75. Transferrin production per unit carrier volume by BHK cells on variousmicrocarriers. Operating conditions as outlined in Figure 74.The S40 microcarrier spinner cultures gave a maximum specific transferrinproductivity of 4.4 mg/10 9 cells/day (i.e. yield of 9.6 mg HTF / g—glucose). Thisis 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 withinthe two batches of cells. The effect of the culture system on cell productivitycannot be fully determined unless all the inocula comes from a single batch. InFigure 76, BHK cells harvested from a single roller bottle were divided equally to425- ^ S40O RollerA SuspensionOOOO ^A8 ^ AA5-^ AO p A A A AA• I^I^ 1^•^I0^50 100^150^200^2500300^350140inoculate roller, spinner and S40 (0.5 g/L) microcarrier cultures. Each culturesystem contained 200 mL DMEM medium. DMEM/F-12 medium was used afterthe cultures reached steady state (at 200 h). Prior to the formation of large cellclumps in the spinner suspension culture, a large proportion of the cells in thesuspension culture were lost when the medium was changed; this retarded thespinner suspension growth. The estimated final cell concentrations were 5.2 x 106,2.55 x 106, 2.51 x 106 cells/mL for the roller, the microcarrier and the spinnercultures, respectively.Time (h)Figure 76. Cumulative glucose used by BHK cells.Each culture system contained200 mL DMEM medium. DMEM/F-12 medium was used after the culturesreached 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 productionof transferrin (shown in Figure 77). However, the increase in the product could notbe fully explained by the increase in cell number. The maximum cell specific141transferrin productivity rates were 10.3, 4.5 and 4.7 mg/10 9 cells/day (ormaximum yields of 21.7 9.8, 10.2 mg HTF/ g—glucose) for the roller, themicrocarrier and the suspension culture, respectively. The reason for the high cellspecific transferrin productivity of the roller system is unknown. However, theBHK cells in the roller culture utilized the glucose more efficiently and producedless L—lactate (see Figure 78). The glucose—to—lactate conversion ratio decreasedfrom 0.83 to 0.55 for the BHK roller culture, and more transferrin was producedper unit mass glucose used (Figure 79). It should be noted that the reduction ofglucose—to—lactate conversion ratio occured only during multilayer cell growth.The conversion ratio remained constant around 0.8 when cell number in the rollerwas kept low. The lower glucose—to—lactate conversion ratio in the roller cultureeither suggests better oxygenation despite the potential mass transfer limitation tocells in the inner layer or reduction of available glucose to cells in the inner layer.The measured dissolved oxygen concentration of the roller culture wasconsistently higher than that of the spinner culture (i.e. 60% vs. 30% air saturationafter the medium was replaced for 24 h). However, the _optimum dissolved oxygenconcentration for cellular production was not determined. Spier and Griffiths(1984) found that significant increases in cell yield and multiplication rate forvarious mammalian cells including BHK cells, when the oxygen partial pressurewas 9-10% compare to the normal 21%. A stirred system with controlleddissolved oxygen concentration is needed to eliminate the mass transfer limitationencountered in the current spinner flask systems and to determine the optimumoxygen concentration.142—0— S40 (0.5 g/L)—A- Suspension^ Roller cultureTime (h)Figure 77. Cumulative transferrin produced by BHK cells from various culturesystems.^1.00^V^V^V V V—V—V—V—V—V—V-ocozN 0.75-00)0)0020.50-Tocr,^ ^ ^ ^ ^ ^ ^ 0 - 0 ^ ^__AA—A—A —A—A—A—0— S40—A— Roller—p— Suspension0.250•^I^•^I^•50^150^200^250100Time (h)Figure 78. Glucose—to—lactate conversion ratios for various BHK cultures.300^350242213 20o• 128■2317,C 64E209,0^,//0 ,-05 0AA1430 A S40O Roller^ Suspension0 ,' oOA0^100^200^300^400Time (h)Figure 79. Transferrin produced per g glucose utilized for various BHK culturesystems.7.1 Comparison of Cell Loading and Large Scale CelLCulture SuitablityThe steady state cell concentration in all three porous carrier systems testedreached 1 x 108 cells/ mL matrix. The three—dimensional configuration of theporous matrix mimics tissue structure more closely than the conventional two—dimensional configuration and favours multilayer cell growth. The fixed bedceramic perfusion system offers a significant increase in cell density compared tothe commercial Opticell system (i.e. 1.25x10 8 vs. 2x107 cells/cm3) for BHK cellsdespite its lower surface area to volume ratio when compare to Opticell (i.e. 12versus 32 cm 2 surface area per cm 3 volume). The cell loading within the porousmatrix of current fixed bed perfusion system is superior to those of other fixed bedreactor proposed by other researchers. i.e. 3.2 x 106 CHO cells/mL matrix for thefiber bed reactor used by Perry and Wang (1989), 1 x 10 8 hybridoma cells/mL144inner pores space for the ceramic perfusion system employed by Applegate andStephanopoulos (1990), 1 x 10 7 hybridoma HB32 cell/mL void volume of theglass bead packed bed reactor used by Ramirez and Mutharasan (1989), and 1.4 x107 Vero cells/mL matrix for the fixed bed glass "Porospher" tested by Looby andGriffiths (1988). The advantages of the current fixed bed system are: (1) thestructure consists of multiple interconnected channels instead of the straightparallel channels found in the Opticell systems; this greatly reduces the possibilityof channel blockage due to over—grown cells. (2) The non—toxic ceramic elementcan be easily cleaned, sterilized and reused. We have used the same ceramicelement for more than ten different culture experiments. (3) the start—up procedureis relatively easy and the inoculating volume requirement is low. However, thefixed bed ceramic system can only be scaled up to a limited extend (i.e. maximumporous matrix volume of 10 L) without suffering from reduction in cellconcentration.The airlift system with a porous draft tube also permits greater cell loadingthan those reported by other researchers. For instances,. the maximum cell loadingin the external airlift packed bed reactor used by Murdin et al. (1989) reached only2.5 x 106 hybridoma cells/mL polyester foam. The fiber—bed airlift bioreactorused by Chiou et al. (1991) achieved a maximum cell concentration of 6.8 x 107CHO cells/ mL in a packed glass fiber bed while the maximum BHK cellconcentration 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 themedium through the porous matrix and simplifies reactor construction and scale-up. However, this system is somewhat difficult to initiate. The difficultiesassociated with initiating the airlift system (i.e. high inoculum volumerequirement) and the inability of obtaining representative samples from the culturereactor may hinder the system from being used as a suitable large scale cell145culture system. The airlift system could be used for the cultivation of suspensioncells since large scale airlift reactors are commonly used for hybridoma cellgrowth. The porous draft tube could entrap suspension cells and potentiallyincrease the cell concentration in the reactor. Inoculation procedure for thecultivation of suspension cells is also much simpler when compared to adherentcells; the inoculum volume does not need to cover the entire draft tube. With theinput gas flow, the convective flow from the bulk fluid through the porous drafttube exists regardless of the liquid height. As cell density in the bulk fluidincreases, medium can then be added to the reactor to increase fluid volume.The maximum cell concentrations obtained from the porous microcarriersystem were similar to those reported for various commercial porous microcarrierculture 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 utilizingVerax, and Cultispher—G beads. Since these values were supplied by themanufacturers, they usually represented the best results obtained from optimizedprocesses. In our comparative studies, cell loading per volume of carrier was 20%higher in S40 Polyhipe microcarrier than in Cultispher—G particles (see Section6.2.5). Cell density attained in Polyhipe also exceeds that reported for other non-commercial porous particles, such as the polyurethane foam inveatigated byMatsushita et al. (1990) or the reticulated polyvinyl formal resin foam used byYamaji and Fukuda (1991) and Yamaji et al. (1989).Although the fixed bed perfusion system can only be scaled up to a verylimited extend, its scale is well suited for use as a cell propagator to supply therequired inoculating cells for other type of reactors. Both the airlift system and theporous microcarriers system can be scaled up by both increasing the volume andthe carrier loading of the system. The airlift system was somewhat difficult toinitate when compared to the microcarrier system. However, once the cell146population was established within the airlift, the reactor is simple to operate andeasy to maintain for extended periods. Cell recovery from the microcarrier systemis difficult, complicating the process of cell propagation. The inaccessibility ofcells within the matrix is not a disadvantage as long as the product of interest issecreted from the cells to the bulk medium. With its ease of scaling—up, simpleinoculation procedure, and the ability to obtain representative samples from thereactor, the porous microcarrrier system is the best general choice for large—scalehigh cell density animal cell culture.1478.0 CONCLUSION & RECOMMENDATIONSThe feasibility of using open—porous matrices as support materials for largescale animal cell cultivation has been demonstrated. The following majorconclusions can be drawn from the experimental data and modelling resultspresented in the previous chapters:Fixed bed perfusion system• The current fixed bed ceramic perfusion system offers a superior cell loadingcompared to the Opticell system (i.e. 1.25x10 8 vs. 2x107 cells/cm 3) for BHKcells. The advantages of the current ceramic foam are: (1) reduction of channelblockage due to over—grown cells, and (2) re—usable non—toxic ceramic element.• The ceramic foam perfusion system offers the benefits of an immobilized culturesystem and ease of product removal with the additional benefit of continuousculture monitoring by oxygen consumption rate monitoring.• The ceramic perfusion system can be used as a cell propagator. Vero cells couldbe grown and harvested from the fixed bed ceramic foam repeatedly providing astable source of cells.Airlift reactor• The airlift system supported long term cell culture at high cell densitycomparable to that produced in the externally circulated fixed bed ceramicperfusion system. BHK cells were grown successfully in the airlift system formore than 8 weeks.• The differential pressure drop across the porous draft tube was greatlyinfluenced by the porosity of the tubular ceramic elements. The differentialpressure 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, themass transfer coefficient for oxygen was not affected.• The model proposed for the airlift reactor adequately described the overallhydrodynamic trends. It allows the prediction of the riser gas holdup, superficialliquid velocity and liquid perfusion rate through the porous draft tube.• Based on the mass transfer characteristics, the current airlift system couldsupport a cell density of at least 3.6 x 10 8 cells / mL matrix or 5 x 107 cells/mL oftotal reactor volume.Porous microcarrier system• Attachment rates of BHK and Vero cells were more rapid on sulphonatedpolystyrene microcarriers (S80, S40) than on Polyhipe with other surfacemodifications and were similar to the attachment rates for Cytodex andCultispher—G microcarriers.• Vero and BHK cells grew as well on the unmodified polystyrene as on thechloromethyl, diethyl amino or sulphonated polystyrene surfaces. At steady—statein batch—fed perfusion cultures, Vero and BHK cell densities of greater than 1 x108 cells/mL carrier on glucose uptake rates and ATP recovery were estimated.• Cell recovery from the porous Polyhipe was difficult. Only about 10 % of theestimated total cell population could be harvested following trypsinization. Thusthe recovery of viable cells from Polyhipe for seeding large—scale bioreactorsappears to be a problem. About 60 % of the cells could be accounted for on thebasis of nuclei released from the Polyhipe particles by digestion with citric acid.• Hybridoma cells were not effectively entrapped by Polyhipe particles when theparticles were seeded in the presence of serum. Chloromethyl Polyhipe did bindand retain a significant fraction of hybridoma cells when the initial contacting was149done without serum. Maximum cell concentrations were about 3 x 10 7 cells/ mLcarrier, 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 andCultispher-G microcarriers. Performance based on carrier weight was equivalent.• Scanning electron microscopy confirmed the attachment and entrapment of Veroand BHK cells within both types of porous matrices. Scanning electronmicrographs indicated possible mass transfer limitations within the porous S40microcarriers.The following recommendations are made for future studies to expand thescope of this present study.• The effect of using serum-free culture medium on cell loading within the porousmatrix and cellular productivity in the proposed bioreactors should be investigated.• The effect of serum concentration on cell recovery in the perfusion propagatorshould be examined. Higher cell recovery rate might be obtained by reducing theserum 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 bedetermined, which could then be used to validate the proposed model. The overallgas holdup of the airlift could also be determined by measuring the differencebetween the height of the liquid and the height of dispersion. The scale of thecurrent airlift is too small and does not permit such measurements to be takenwithout 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 dueto cell attachment and growth should be monitored and the porosity changebecause of cell growth within the porous matrix should be investigated.150• The proposed airlift system was difficult to initiate for the cultivation ofanchorage—dependent cells. However, large—scale airlift reactors (i.e. over 1000L) are commonly used for suspension cell culture. The proposed airlift systemshould be tested for the cultivation of suspension cells, such as hybridoma cells.By adjusting the porosity of the porous draft tube, entrapment of suspension cellsmay take place and possibly increase the cell concentration within the reactor.• Dissolved oxygen was not controlled in the spinner microcarrier cultures anddiffusion limitations within the microcarrier were evident. The oxygen limitation ismost likely caused by pore blockage near the outer surface of the microcarrier dueto over—grown cells and the low mass transfer coefficient of the spinner flasksused. A better oxygenated stirred tank system should be used to increase thedissolved oxygen concentration in the bulk medium and hence, reduce the masstransfer limitations within the porous beads.• Increased bead to bead collision and high shear stress resulting from increasedparticle 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 microcarriersat high concentration (i.e. greater than 5 g /L) and in stirred bioreactors with highagitation should be evaluated and compared with the performance of non—porousmicrocarriers.• The nature of the observed enhanced cellular transferrin productivity in rollercultures may be due to the dissolved oxygen concentration being optimum. Theeffect of dissolved oxygen concentration on cell productivity should to be furtherinvestigated.NOMENCLATUREal) = 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 theCT airlift reactor (m2)AD = cross—sectional area of downcomer (m2)AR = cross—sectional area of riser (m2)Cf = Fanning friction factorCs = dissolved oxygen concentrationCb * = dissolved oxygen concentration in the bulk mediumCbi = dissolved oxygen concentration at the outer perimeter of the shperedB = average bubble diameter (m)DC = Column diameter (m)Da =Column inner diameter (m)Deff = effective diffusivity (m2s-1)d F= diameter of matrix fibers plus cell layer (m)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)151152hp = 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 sphererd = radial position at which oxygen concentration is zeroR = gas constantRep = Reynolds numberT = 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 spargerand 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)Eg = gas hold—upEgr = gas hold—up in the riserEg d = gas hold—up in the downcomerEM = void fraction of the porous matrixp1 = liquid density (kg/m3)9 = sparger efficiency (typically around 0.06)1.t = liquid viscosity (Pa s)T = shear stress (N/m2)AP = differential pressure drop (Pa)153154REFERENCESAdema, E., Shnek, D., Cahn, F., and Sinsky, A.J., " Use of porous microcarriers inagitated 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 ofBiologicals 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—passceramic matrix bioreactor for large—scale mammalian cell culture", AIChE Fallmeeting, 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 tofermentations", PhD Thesis, University of Waterloo, Ontario, (1981).Bello, R.A., Robinson, C.W. and Moo—Young, M., "Gas holdup and overallvolumetric oxygen coefficient in airlift contactors", Biotechnol. Bioeng., 28,369-381 (1985).Bello, R.A., Robinson, C.W., and Moo—Young, M., "Liquid circulation andmixing 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).155Bleim, B., and Katinger, H.W.D., "Scale-up engineering in animal celltechnology: 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 21cells using the Opticell culture system", J. Tissue Culture Methods, 8, (4), 147-154 (1983).Boraston, R., Thomson, P.W., Garland, S., Birch, J.B., "Growth and oxygenrequirements of antibody producing mouse hybridoma cells in suspensionculture", 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 ofmonoclonal 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 fermentationequipment", 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., "GasHold-up distribution in a gas-lift column", Biotechnol. Bioeng. Symp., 4, 363-378 (1973)Chiou, T.W., Murakami, S., Wang, D.I.C., and Wu, W.T., "A fiber-bed bioreactorfor anchorange-dependent animal cell cultures: part I. bioreactor design andoperations", Biotechnol. Bioeng., 37, 755-761 (1991).Chisti, M.Y., and Moo-Young, M., "Airlift reactors: characteristics, applicationsand design considerations", Chem. Eng. Commun., 60, 195-242 (1987).156Chisti, M.Y., Halard, B., and Moo—Young, M., "Liquid circulation in airliftreactor", Chem .Eng. Sci., 43 (3), 451-457 (1988).Chisti, M.Y., and Moo—Young, M., "Perdiction of liquid circulation velocity inairlift reactors with biological media", J. Chem. Technol. Biotechnol, 42, 211-219 (1988).Chisti, M.Y., Airlift Bioreactors, p.147-148, Elsevier Science Publisher, NewYork, NY, (1989).Cho, M.H., and Wang, S.S., "Enhancement of oxygen transfer in hybridoma cellculture 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 microcarrierconcentration 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 topolystyrene surfaces", J.Cell Biol., 97, 1500-1506 (1983).de Bruyne, N.A., "The design of bench scale reactors", in Spier ,R.E., andGriffiths, J.B., (eds), Animal Cell Biotechnology, -Vol. 3, p.159, AcademicPress, 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 cellculture 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 tallbubble 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).157Dussap, G., and Gros, J.B., "Energy consumption and interfacial mass transfer inan airlift fermentor", Chem. Eng. J., 25,151-162 (1982).Emery, A.N., Lavery, M., Williams, B., Handa, A., "Large—scale hybridomaculture" 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 growthon oxygen—containing films deposited by radio—frequency plasmas: the role ofsurface carbonyl groups", J. Biomater. Sci. Polymer Edn., 3, (2), 163-183(1991).Familletti, P.C., and Fredericks, I.E., "Techniques for mammalian cellimmobilization", Bio/Technol., 6 (1), 41-44 (1988).Feder, J., "The determination of large—scale animal cell perfusiono culturesystems", in Upstream processes: equipment and techniques, 125-152, AlanLiss, Inc., (1988).Foran, D.J., Calm, F., and Eikenberry, E.F., "Assessment of cell proliferation onporous microcarriers by means of image analysis", Analytical and QuantitativeCytology and Histology, 13, (3), 215-222 (1991).Forestell, S.P., Kalogerakis, K., and Behie, L.A., "Development of the optimalinoculation 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 ofprostacylin production by cultured human endothelial cells", Science, 227,1477(1985).Frangos, J.A., Mclntire, L.V., and Eskin, S.G., "Shear stress induced stimulationof mammalian cell metabolism", Biotechnol. Bioeng., 32,1053-1060 (1988).Freeman, W., and Davidson, J.F., "Holdup and liquid circulation in bubblecolumns", Trans. Inst. Chem. Eng., 47, T251—T262 (1969).158Funk, 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 serumtransferrin 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 oncultured hybridoma cells", Biotechnol. Bioeng., 35, 940-947 (1990).Griffiths, J.B., "Advances in animal cell immobilization technology" in AnimalCell 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 circulationand mixing characteristics of arilift contactors", Can. J. Chem. Eng., 64, 521-523 (1986).Handa, A., Emery, A.N., and Spier, R.E., "On the evaluation of gas-liquidinterfacial 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-liquidinterfaces on the growth of suspened mammalian cells: mechanisms of celldamage by bubbles", Enzyme Microb. TechnoL , 11 (4), 230-235 (1989).Hatch, R.T., Experimental and theoretical studies of oxygen transfer in an air-liftfermenter, PhD Thesis, Massachusetts Institute of Technology, Cambridge,Massachusetts (1973).Hills, J.H., "The operation of a bubble column at high throughputs I - Gas holdupmeasurements", Chem. Eng. J., 12, 89-99 (1976).159Himes, V.B. and W.-S. Hu., "Attachment and growth of mammalian cells onmicrocarriers 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 inmedical research, p 97, Pargamon, Oxford, (1980).Hu, W.S., Merier, J., and Wang, D.I.C., "A mechanistic analysis of the inoculumrequirement 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 growthof 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 animalcells 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 bya 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 masspropagation of animal cells in suspension culture", German Chem. Eng, 2,31-38 (1979).Katinger, H.W.D., and Scheirer, W., "Status and development of animal celltechnology using suspension culture techniques", Acta Biotechnol, 2:3-41(1982).160Kawase, Y., and Moo-Young, M., "Mixing and mass transfer in concentric-tubeairlift 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 bedreactor for the cultivation of animal cells", Proceedings of the 11th ESACTMeeting (1991).Kilburn, D.G., and Webb, F.C.,"The cultivation of animal cells at controlleddissolved oxygen partial pressure", Biotechnol. Bioeng., 10, 801-814 (1968).Klemperer, H.G. and Knox, P., "Attachment and growth of BHK and liver cells onpolystyrene: effect of surface groups introduced by treatment with chromicacid", Lab. Pract., 26, 179-180 (1977).Koide, K., Horibe, K., Kitaguchi, H., and Suzuki, N., "Contributions of annulusand 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 holdupand volumetric liquid-phase mass transfer coefficient in bubble column withdraft 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-phasemass transfer coefficient in bubble column with draft tube and with gasdispersion into annulus", J. Chem. Eng. Japan, 16, (5), 407-413 (1983b).Kratje, R.B., Jager, V., and Wagner, R., "Comparison of the production efficiencyof 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 humaninterleukin-2 in fluidized bed bioreactor", Biotechnol. Bioeng., 39, 233-242(1992).161Kriegel, E., Pilepp, E., Scheffler, U., "Entwicklung eines bioreaktors nach demumlaufprinzip", Tech. Mitt. Krupp. Forsch—Ber., 36, 41-43 (1978)Kunas, K.T., and Papoutsakis, E.T., "Damage mechanisms of suspended animalcells in agitated bioreactos with and without bubble entrainment", BiotechnoLBioeng., 36, 476-483 (1990).Kwong, C.W., Hsieh, J.H., Syu, M.J., and Chou, S.S., "Anti—hepatitis B surfaceantigen monoclonal antibody IgM production in suspension and immobilizedbioreaktors", Biotechnol. Lett., 11, (6), 377-382 (1989).Lambert, K.L., Boraston, R., Thompson, P.W., and Birch, J.R., "Production ofmonoclonal 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 polyurethanefoam matrix", European Society for Animal Cell Technology (ESACT) GeneralMeeting, 8th, Tiberias, Israel, (1987).Lee, C.H., Glasgow, L.A., Erickson, L.E., and Patel, S.A., "Liquid circulation inairlift fermentors", in Ho, C.S, and Oldshue, J.Y., (eds.), BiotechnologyProcesses 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 polymersurfaces 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 theirdevelopments for continuous culture with cell recycle", in Spier, R.E., andGriffiths, J.B., (eds.), Animal Cell Biotechnology, Vol. 3, 222-237, AcademicPress, Toronto, (1988).162Levine., D.W., Wnag., D.I.C., and Thilly, W.G., "Optimization of growth surfaceparameters in microcarrier cell culture", Biotechnol. Bioeng., 21, 821-845(1979).Lim, F. & A.M. Sun., " Microencapsulation islets as bioartificial endocrinepancreas", 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 carrierculture", TIBTECH, 8, 204-209 (1990).Ludwig , A., Kretzmer, G., and Schugerl, K., "Determination of a critical shearstress 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., "Ceramicmatrix 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 onceramic 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 spreadingfactor", Biomaterials, 9, 525-527 (1988).Margaritis, A., and Sheppard, J.D., "Mixing Time and oxygen transfercharacteristics of a double draft tube airlift fermentor", Biotechnol. Bioeng., 23,2117-2135 (1981).Maroudas, N.G., "Sulphonated polystyrene as an optimal substratum for theadhesion and spreading of mesenchynal cells in monovalent and divalent salinesolutions", 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 new163substrratum for cell attachment", Appl. Microbiol. Biotechnol., 33, 287-290(1990).McQueen, A., and Bailey, J.E., "Influence of serum level, cell lines, flow type andviscosity 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 biologicaltechnology", J. Chem. Technol. Biotechnol., 41, 105-120 (1988).Merchuk, J.C., "gas Hold—up and liquid velocity in a two—dimensional air liftreactor", Chem. Eng. Sci., 41, 11-16 (1986a)Merchuk, J.C., "Hydrodynamics and hold—up in airlift reactors", Encyclopedia ofFluid 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 culturedin macroporous microcarriers", Cytotechnology, 4, 163-171 (1990).Miltenburger, H.G., and David, P., "Mass production of insect cells insuspension", Develop. Biol. Stand., 46, 183-186 (1980).Murakami, S., Chiou, T.W., and Wang , D.I.C., "A fiber—bed bioreator foranchorange—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 andmetabolism of hybridomas immobilized in packed beds: comparison with staticand suspension cultures", Enzyme Microbiol. Technol., 11 (6), 341-346 (1989).Nikolai, T.J., and Hu, W.S., "Cultivation of mammalian cells on macroporousmicrocarriers", Enzyme Microbiol. Technol., 14, (3), 203-208 (1992).164Nikolai, T.J., Madhusudan, V.P., Geotghebeur, S., and Hu, W.S., "Improvedmicroscopic observation of mammalian cells on microcarriers by fluorescentstaining", Cytotechnology, 5, 141-146 (1991).Nilsson, K., F. Buzsaky & K. Mosbach, "Growth of anchorage-dependent cells onmacroporous microcarriers", Bio/Technol. 4, 989-990 (1986).Nollert, M.U., Diamond, S.L., and McIntire, L.V., "Hydrodynamic shear stressand 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 hollowfiber bioreactors", Adv. Biotechnol., 8, 763-783 (1990).Radlett, P.J., Telling, R.C., Whiteside, J.P., and Maskell, M.A., "The supply ofoxygen to submerged cultures of BHK 21 cells", Biotechnol. Bioeng., 14, 437-445 (1972).Ramsey, W.S., Hertl, W., Nowlan, E.D., and Binkowski, N.J., "Surface treatmentsand 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 geltaincarriers 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 cellculture", Adv. Appl. Microbiol. 31:139-179 (1986).165Robinson, C.W., "On power input and liquid phase mixing times inpneumatically—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 MedicalPublishing, 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 forbubble 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 separatorbehaviour, 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 YorkAcad. Sci., 369, 47-59 (1981).Smiley, AL., Hu, W—S, Wang, D.I.C., Production of human immune interferon byrecombinant 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 themeasurement of oxygen transfer in microbial systems", Tsao, G.T., ed. inAnnual Reports on Fermentation Processes, Acadamic Press, London, (1982).Sonderhoff, S.A., Kilburn, D.G., and Piret, J.M., "Analysis of mammalian cellbiomass based on cellular ATP", Biotechnol. Bioeng., 39, 859-864 (1992).166Spier, R.E., and Griffths, B., "An examination of the data and concepts germane tothe 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 virusfrom 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 embryonickidney 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 largescale cuilture of mammalian cells", Biotechnol. Bioeng., 28, 1064-1071 (1986).Thomas, K.D., Tighe, B.J., and Lydon, M.J., "Cellular interactions with syntheticpolymer surfaces II. surface modification of polystyrene for cell culture", inAdvances 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 airliftbioreactors", 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 perfusionsystem 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-sensitiveinsect cells", in Webb C. and Mavituna F., eds: Plant and animal cell culture:processes possibilities, Chichester: Ellis Horwood, 125-136 (1987).167Tramper, J.D., Joustra, D., and Vlak, J.M., "Bioreactor design for growth ofshear-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 forgrowth of fragile insect cells", Bioprocess Eng. , 3:37-41 (1988b).Van Wazel, A.L., "Cultivation of anchorage-dependent cells and theirapplications", J. Chem. Technol. Biotechnol., 32, 318-323 (1982).Van Wazel, A.L., and Van der Velden de Groot, C.A.M., "Large scale cultivationof animal cells in microcarrier culture", Process Biochem., 13, 6-8 (1978).Vournakis, J.N., and Runstadler, P.W., "Microenvironment: the key to improvecell 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 unitprocess system based on 3 mm diameter glass spheres for the production offour 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 FluidEngineering Center, Cranfield, Bedford, England), 157-172 (1986).Yamaji, H., Fukuda, H., Nojima, Y., and Webb, C., "Immobilization ofanchorage-independent animal cells using reticulated polyvinyl formal resinbiomass support particles", Appl. Microbiol. Biotechnol., 30, 609-613 (1989).Yamaji, H., and Fukuda, H., "Long-term cultivation of anchorage-independentanimal cells immobilized within reticulated biomass support particles in acirculating bed fermentor", Appl. Microbiol. Biotechnol, 34, 730-734 (1990).168Ziltner, 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).APPENDICESAppendix 1. Model PC-61 A/D board data logging programprogram mA_Card_data;uses crt, dos;constmA_Card = $330;number_of channels = 8;Gain = 0.68;Resistance = 270 {ohms};maxfreq = 30;maconvers = 185.60508;maoffset = 3;typeout_file = text;vari, j, k : integer;sint^: longint;portnum : integer;milliamps,sfreq^: real;ch^: char;outfile : out_file;filename : string;procedure initialise;beginclrscr;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] := 0end {initialise};procedure cursor_off;beginport[$3B4] := 10;port[$3D4] := 10;port[$3B5] := 13;port[$3D5] := 13169170end {cursor_off};procedure cursor_on;beginport[$3B4] := 10;port[S3D4] := 10;port[$3B5] := 11;port[$3D5] := 6end {cursor_on};procedure write_headings;begingotoxy(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;beginport[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;beginwrite(output, 'Name of file (DIR:\FILENAME) ');readln(input, filename);assign(outfile, filename);rewrite(outfile);end {openfile};procedure closefile;beginclose(outfile);end;procedure setup;beginwrite('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;vari, 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;begintakevery := 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');repeatrunsum := 0;for i := 1 to sint dobeginfor j :=1 to (takevery-1) dobeginvoltage := Read_Voltage;end;voltage := Read_Voltage;runsum := runsum + voltage;171172(* 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 thenbeginfirstsample := 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;beginclrscr;openfile;setup;initialise;survey;(* cursor off;write_ headings;repeatfor j := 1 to number of channels do beginport[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) thenwrite('Overflow')else beginwrite(milliamps:7:3, ")end { if };end { for }until keypressed;cursor on;gotoxy(1, 22); write(Test terminated normally');read(input, ch)*)closefile;end.Appendix 2. Anglican controller data logging program listingCLSPRINT "open"OPEN "com1:1200,n,8,2,rs,asc,ds0" FOR RANDOM AS #1 LEN = 256PRINT "done"LINE INPUT "Enter file name to store DO data: "; filename$OPEN filename$ FOR OUTPUT AS #2TIMER ONsend$ = "!01R1;"PRINT "press any key to begin , press any key to end"begin$ = INPUT$(1)timeini = TIMERDOPRINT #1, send$ret$ = INPUTS(LOC(1), #1)a = ABS(LEN(ret$) — 4)PRINT (TIMER — timeini), ","; LEFT$(ret$, a)PRINT #2, (TIMER — timeini), ","; LEFT$(ret$, a)FOR i = 1 TO 6000NEXT iLOOP UNTIL INKEY$ <> ""PRINT "stop"CLOSE #1CLOSE #2END173174Appendix 3. MathcadTm airlift model program listingEnergy input:0.79hp ADE^.7= Q-P y in 1 + K B := 11.40. (P T A bEnergy Dissipations:given1.)dissipations in the riser due to bubble wakes6 1'2+ 6 3+ 6 4+ 6 5^E R=A R.p-gh aV IR^52.) Energy dissipation in the downcomer due to upflow bubble motions withrespect to liquid: assumption: e in the downcomer is zero, so energy dissipated inthe dowmcomer is zero3.) Energy losses due to flow through porous draft tube9) I 6 511Ui=p-g.hp- (hp10 Lc 7II 2=p•g- h D -10 L k p.1)Ie3 113=p-gihr)hD - 2 L k, 3)1 6 2-1 -1 -U 4=frgilip - (n.10 L k p.6 1 1 1thp -^ 1^2 '10^L k p,where: U=flow velocity through porous draft tubeii=viscosityL=draft tube thicknessV LR2=V2^h PLR1U 5.—^ 5h pV LR3=V LR2 + 3 -Ur^52 -^h PV LR4 =V LR3 + -U^ 52^hV LR5'V LR4 + --U^ 52^h pV LR6=VI,R4 + --u^ 5QAR"30.24 + 1.35. Q +(^V LR1 + V 1,R2—A R^2QA R2 .1 (0.24 + 1.35 Q + V LR3 + V LR2 093A R^2QA R0930.24 + 1.35 Q(^V la4 + V LR3A R +^21753 /2E 3>0E 4?°E5.0Hills (1976)valid fortrgr greaterthan 3 miserQA RQ V iR5 +V LR4 /3930.24 + 1.35A R^2QA RH 5-^0.930.24 + 1.35^+Q V LR6 + V LR5)((1.1. 1 +U 2 +U 3 +13 4 +UTE ^5^• (2.7E-r)4.) Energy losses due flow through the top and the bottom sections of the airliftI KE^3 - (Bn= p. (Via l) —2- 0.5-A. R3 + 0.018-h D.w.rA DTotal energy balance yields:E 1 =E R + E BT EpE BT ->0 176A R^2177Appendix 4 Sample Calculation(a) Estimated kLa requirementLiterature reported oxygen consumption rate (OCR) for BHK cells is about 0.20mmoles/109 cells/h. For 108 cells/mL medium in the proposed reactors, OCR is 20mmoles/L/h.OCR = kLa (C*—C)whereC* = 0.2 mmoles/L at 37°CC = 0 to detemine the minimum kLa neededThe required kLa is, therefore, estimated to be 0.027 s -1 .(b) kLa determinationA mass transfer coefficient can be conveniently defined by a simple massbalance 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 = dCdt (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 theliquid—phase mass transfer coefficient.Depending on the type of flow pattern inside the reactor, Eq (39) can beincorporated in an overall oxygen balance in the liquid phase, and thus oxygensupply rates can be readily evaluated. kLa of a well—mixed batch process can beobtained by integrating equation (39):f  dC = kLa fdt^ (39a)c (C -C)^0Orln( C -  ) = (kLa) t^ (39b)C -CIf a fractional approach to equilibrium (E) is defined as the ratio of the masstransfer at any instant (i.e. C—00) to the maximum possible transfer (i.e. C *—00),orE— C - C°C * -C o^(40)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 dissolvedoxygen) and E=1 (saturation condition) are easier to set and monitor on chartrecorders and meters. The mass transfer coefficient, kLa, is the slope of a plot ofln(1/(1—E)) vs. t.(c) Scale—up of the fixed bed perfusion systemIn order to maintain a cell concentration of 1.3 x 108 cells/mL matrix, it isnecessary to supply sufficient oxygen to the immobilized cells. A greater perfusionrate is needed to supply oxygen as the total cell population increases with theincreased fixed bed length. However, increasing in the perfusion rate alsoincreased shear stress. According to Bleim and Katinger (1988), it would be ideal178179to keep the average shear stress within the fixed bed (calculated using Equation32) around 0.5 N/m 2. Assuming that (1) the Darcy resistance coefficient of the 30PPI porous draft tube with cell attached = 6.7 x 10 7 m -2 , (2) void fraction of thematrix with cells = 0.7, and dF = 1 mm, the maximum allowable perfusionvelocity is estimated to be 1 cm/s. The perfusion rate of 1 cm/s allowed the fixedbed system to be scaled—up 108 times in volume over the current system (i.e. to amaximum of 5.43 L matrix volume). The maximum size of the ceramic matrix isabout 10L even if the average shear stress was allowed to increase from 0.5 to 1.0N/m2 .(d) Airlift simulation exampleSeveral parameters regarding to the physical configuration of the bioreactor, suchas draft tube wall thickness, permeability of the draft tube and the reactoroperating conditions, such as air flow rate, dispersion height are needed for themodel simulation. The values of the input parameters are listed below:Q = 3.33 x 10-5 m3/s;^AR = 0.0022 m2;^L = 0.013 mhp = 0.55 m;^AD = 0.00665 m2; k = 6.7 x 107 nr2hp = 0.21 m; Ab = 0.004 m2;^r= 0.025 mThe resulting outputs are:egr varied from 0.8% to 1.7%Up varied from 35 to 105 mm/sUfr varied from 0.23 to 1.12 m/sRep = 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 theporous matrix while an average value of 1.8 Lis of medium flow down thedowncomer. The volume ratio of medium perfused through the matrix to mediumflow down the downcomer increases as the porosity of the draft tube matricesdecreases.180Appendix 5 Raw dataRaw data of Figure 8Time(day)Glucose Concentration (g/L)30 PPI 50 PPI 100 PPI0 4.02 3.9 4.021 3.74 3.6 3.882 3.3 3.3 3.753 2.5 2.95 3.724 1.1 2.6 3.684.5 0.6 - -5 - 225 3.656 - 1.9 -7 - 1.55 -8 - 13 -Raw data of Figure 10Time (day) oxygen uptake rate(it-moles/min)estimated cell number(x 10-8)0 - 0.022.92 0.96 0.13.92 1.74 0.1754.17 2.2 0.224.91 4.07 0.415.93 6.97 0.76.41 658 0.77.14 9.17 0.8287.92 9.18 0.822822 10.53 -Data for Fames 11 and 12Time (days) Total glucose used (g) Estimated^cellnumber (x 10 -8)Total lactate produced (g)0 0 0.210598 00.96 0372 0.251359 0.2581.92 0.888 0.29606 0.6182.92 1.44 0.30163 1.173.92 1.998 030163 1.9924.17 2.136 0.434957 2.1124.91 2.912 0.585473 2.745.93 4.04 0.520168 3.676.41 4.428 0.52862 3.977.14 5.258 0578962 4.727.92 6.033 0.419451 5.34822 6.198 0.507723 5.538.98 7.2 0.567299 6.329.24 7.4 0.468957 6.679.93 8.06 0.446745 7.15181Data for Figures 11 and 12 continued10.09 8.17 0.60737 7.19510.93 9.47 0.791103 8.0711.26 9.92 0.642293 83911.43 10.09 0.763946 8.5911.96 11.05 1.023332 9.4412.18 11.48 0.987647 9.6812.43 11.9 0.911391 10.0512.89 12.67 1.031141 10.6413.47 13.9 1.009761 11.6613.89 14.57 0.773163 12.1213.93 14.62 1.095815 12.1714.39 15.9 1.174207 13.1514.91 16.7 1.233277 13.7315.11 17.3 1.552793 14.0615.39 18.06 1.176543 14.6715.91 18.9 1.128761 15.2816.17 1956 1399342 15.761635 20.03 1.097739 16.1616.98 20.93 1.152467 16.7217.14 21.38 1.64219 17.0917.27 21.8 1387435 17.317.91 23 0.939766 18.1918.03 23.19 0.989348 18.3418.9 24.98 0.980293 19.6719.1 25.29 1.187005 19.9219.21 25.6 1.332582 20.1719.91 27.06 1.238571 21.2320.27 27.95 1.194049 21.9320.91 29.18 1.071163 22.7921.91 31.2 0.966973 24.2222.04 31.4 1.282685 243822.26 32.1 1369283 24.9122.89 3327 122106 25.8123.11 33.85 1.457511 26.223.22 34.15 1.245201 26.4623.91 35.43 1.047576 27.524.09 35.79 1.81856 27.8524.22 36.4 2.124266 28.324.46 37.15 1.621168 29.1124.9 38.4 1.641549 29.9525.27 39.8 2.228261 31.125.9 41.5 2.091962 32526.06 41.8 1.242777 32.8126.19 42.4 1.76369 33226.96 44.6 2.030576 34.9627.94 47.62 1.613799 37.528.9 50.82 1.743196 40.0429.21 51.83 1.791141 40.0929.91 53.2 1.417174 42182Data for Figures 11 and 12 continued30 53.5 1.43763 42.3130.2 53.9 1.449277 42.5830.9 55.3 1.086957 43.0831.2 55.96 1.141304 43.6431.9 57.5 1.195652 44.7632.2 58 1.050723 45.4532.9 59.6 1.074016 46.1133.2 60.32 1.273293 46.833.9 61.69 1.184005 47.8345 63.06 1.152304 49.235 63.86 1.055255 49.7435.45 65.36 1.340582 50.8735.9 663 1.473429 51.5636.47 67.97 1363783 52.9536.91 68.5 1.123473 53.3737.21 69.62 1.341815 54.2637.9 70.9 1.518587 55.0638.22 72.2 1.608038 55.9538.9 73.3 1543516 56.6739.19 7433 1.404723 57.4839.88 75.62 1.473174 58.02Data for Figure 13BHK Cell Concentration (x 10 -4) LDH (IU/L)4 52220 6040 74.290 97.4^-170 139.2Data for Figure 14Time (day)5% serumTotal transferinproduced (mg)Time (day)2.5% serumTotal transferinproduced (mg)0 0 0 00.84 7.4 0.2 1.51.17 15.6 0.48 12134 24 1 20.11.87 35.5 1.26 22.22.09 37.81 1.44 242.34 3838 1.99 34.92.8 53.82 2.23 40.8338 65.12 2.36 43.53.84 87.22 3 61.94.3 92.02 3.12 634.82 107.32 3.99 73.24.19 73.74.3 76.75 83183Data for Figure 15DMEM^ DMEM/F-12Time (day) Total Transferrinproduced (mg)Time (day) Total Transferrinproduced (mg)0 0 0 00.729 1.06 0.96 0.6311.729 2.94 1.92 2.7722.729 3.99 2.92 4.443.729 5.23 3.92 8.1844.719 6.46 4.17 15.965.948 11.2 4.91 18.2166.792 13.6 5.93 23.2847.715 15.3 6.41 29.2248.715 18.2 7.14 33.6329.729 23.9 7.92 45.52310.72 27.9 8.22 53.92311.69 33.8 8.98 59.8639.24 66.9639.93 74.66310.09 79.26310.93 86.66311.26 94.86311.43 103.26Data for Figure 16Time (11) Total glucose used (g) Total lactate produced (g)0 0 018 037 0.152^_26 0.38 03442 0.51 0.39966 0.9 0.80890 1.45 1.156114 2.24 1.914142 2.9 2.764144 2.9 2.764162 3.65 3.507186 4.9 4.248210 5.9 5.367212 5.9 5.367234 7.27 5.517258 8.64 8.43261 8.64 8.43282 10.55 10.02306 11.91 11.039308 11.91 11.039330 14.16 12.966354 15.43 13.84355 15.43 13.84Data for Figure 16 continued378 17.67 15.666403 18.96 16.446404 18.96 16.446426 21.14 . 18.078429 21.27 18.028455 22.5 18.708455 22.5 18.708480 24.91 21.117500 26.03 21.223501 26.03 21.223523 28.7 23.188546 30.11 24.088549 30.11 24.088552 30.29 24.262571 32.69 26.321595 33.95 27.355596 33.95 27355619 34.72 28.193642 35.63 28.864666 36.3 29.469667 36.3 29.469691 36.81 29.888715 3732 30341715 37.32 30341739 37.72 30.826763 38.6 31.513787 39.25 32.129815 40.07 33.1835 4055 33.348835 40.55 33.348859 41 33.784882 41.87 34.545906 42.54 35.074930 43.26 35.588931 43.26 35.588955 44.47 36.388980 45.48 37.3311002 46.21 37.9031009 46.45 37.8451009 46.45 37.8451026 47.46 38.5221051 48.85 39.61075 49.88 40321076 49.88 40.321099 51.19 41.2561123 52.17 41.792184185Data for Figure 17Non-porousgalss drafttubeNon-porousthick wall 100PPINon-porousthick wall 100PPIUpy (m/s) AP (mm F170) ±AP AP (mm f170) ±AP AP (mm 1110) ±AP0.00E+00 0.00 0.00 0.00 0.00 0.00 0.001.63E-03 0.82 0.04 0.45 0.03 0.57 0.054.90E-03 2.18 0.07 126 0.09 1.50 0.098.16E-03 3.63 0.08 2.26 0.13 2.26 0.151.22E-02 5.54 0.12 3.25 0.15 3.13 0.121.63E-02 6.60 0.21 3.92 0.16 3.75 0.12Thick wall 100 PPI 50 PPI 30 PPIUty. (m/s) AP (mm I-190) ±AP AP (mm I-1,0) ±AP AP (mm H70) ±AP0.00E+00 0.00 0.00 0.00 0.00 0.00 0.001.63E-03 0.52 0.06 0.37 0.04 0.29 0.034.90E-03 1.11 0.06 0.78 0.07 0.67 0.048.16E-03 156 0.10 1.11 0.04 1.08 0.051.22E-02 2.19 0.15 1.43 0.07 1.54 0.071.63E-02 2.48 0.17 1.80 0.07 1.75 0.11Thin wall 100 PPI 50 PPI 30 PPIUEr (m/s) AP (mm I-170) ±AP AP (mm 1-1,0) ±AP AP (mm F1 ,70) ±AP0.00E+00 0.00 0.00 0.00 0.00 0.00 0.001.63E-03 0.57 0.04 0.10 0.00 0.09 0.014.90E-03 1.62 0.07 0.32 0.03 0.21 0.038.16E-03 2.17 0.24 0.44 0.00 0.38 0.041.22E-02 3.33 0.29 0.57 0.00 0.49 0.041.63E-02 3.86 0.17 0.74 0.08 059 0.09non-porous50 PPIthin wallUgr (m/s) AP (mm H2O) ±AP0.00E+00 0.00 0.001.63E-03 0.62 0.064.90E-03 1.47 0.178.16E-03 2.54 0.081.22E-02 352 0.241.63E-02 4.17 0.05186Data for Fieure 18draft tubeconfigurationLong^topshort^bottomLong^toplong^bottomShort^toplong^bottomShort^topshort^bottomIlex (m/s) AP(mmH70)±AP AP(mmH70)±AP AP(mmH70)±AP AP(mm1190)±AP0.00E+00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.001.63E-03 0.10 0.01 0.20 0.02 0.15 0.03 0.09 0.004.90E-03 0.35 0.03 056 0.02 0.45 0.03 0.27 0.058.16E-03 0.58 0.03 0.80 0.04 0.69 0.00 0.49 0.061.22E-02 0.77 0.05 1.25 0.04 1.05 0.04 0.62 0.041.63E-02 0.88 0.08 1.48 0.04 1.26 0.00 0.77 0.09Data for Finure 19draft tubeconfigurationlong top and^long bottom Short top and^short bottomUar (m/s) AP (mm 1170) ±AP AP (mm H90) ±AP0.00E+00 0.00 0.00 0.00 0.001.63E-03 0.52 0.06 0.43 0.034.90E-03 1.11 0.06 0.88 0.078.16E-03 1.56 0.10 1.13 0.071.22E-02 2.19 0.15 1.66 0.091.63E-02 2.48 0.17 1.73 0.15Data for Finure 20MediumtypePBS PBS +2% FCSPBS +5% FCSUgr (m/s) AP (mmH20)±AP AP (mm1190)±AP AP (mm11,0)±AP0.00E+00 0 0 0 0 0 01.63E-03 0.81447 0.04906 0.77453 0.0291 0.81874 0.044354.90E-03 1.84029 0.04676 1.57774 0.03361 1.75501 0.045358.16E-03 2.61897 0.0287 2.47252 0.04021 2.35886 0.048051.22E-02 3.62619 0.09903 3.03661 0.07778 3.31138 0.060891.63E-02 4.26652 0.18381 3.71127 0.08947 3.79387 0.12408Data for Finure 21MediumtypePBS PBS +2% FCSPBS +5% FCSUgr (m/s) AP (mmH20)±AP AP (mm1-170)±AP AP (mmH70)±AP0.00E+00 0.000 0.000 0.000 0.000 0.000 0.0001.63E-03 0.147 0.013 0.107 0.007 0.090 0.0084.90E-03 0.299 0.022 0.226 0.010 0.228 0.0088.16E-03 0.408 0.030 0.315 0.012 0.314 0.0111.22E-02 0.580 0.039 0.392 0.018 0.381 0.0181.63E-02 0.687 0.041 0.511 0.032 0.470 0.014187Data for Figure 22MediumtypePBS PBS +2% FCSPBS +5% FCSUgr (m/s) AP (mmH70)±AP AP (mmH70)±AP AP (mmH70)±AP0.00E+00 0.000 0.000 0.000 0.000 0.000 0.0001.63E-03 0.523 0.056 0.547 0.036 0563 0.0504.90E-03 1.106 0.060 0.926 0.043 0.949 0.0598.16E-03 1.558 0.105 1.156 0.064 1.317 0.0721.22E-02 2.186 0.145 1.642 0.077 1.701 0.0691.63E-02 2.484 0.171 1.924 0.082 2.053 0.106Data for Figure 23MediumtypePBS PBS +2% FCSPBS +5% FCSUgr (m/s) AP (mmH70)±AP AP (mmH2O)±AP AP (mmH2O)±AP0.00E+00 0.000 0.000 0.000 0.000 0.000 0.0001.63E-03 0.294 0.026 0.309 0.017 0.299 0.0194.90E-03 0.666 0.045 0.542 0.025 0.582 0.0548.16E-03 1.081 0.048 0.774 0.033 0.769 0.0381.22E-02 1.537 0.073 1.071 0.048 1.074 0.0651.63E-02 1.751 0.109 1.278 0.071 1305 0.075Data for Figures 24 and 25Model prediction100 PPI(thick)SOPPI(thick)30 PPI(thick)100 PPI(thin)SOPPI(thin)30 PPI(thin)U^(m/s)gr AP (mmH2O)AP(mmH70)AP(mm11,0)AP (mmH2O)AP(mmI-170)AP(mmI-170)0.00E+00 0.00 0.00 0.00 0.00 0.00 0.001.63E-03 0.88 0.60 0.48 0.46 056 0.754.90E-03 1.94 1.20 0.93 0.91 1.14 1.708.16E-03 2.74 1.63 125 122 156 2.481.22E-02 3.58 2.06 1.58 1.52 2.00 3.211.63E-02 4.32 2.43 1.86 1.77 2.38 3.90Data for Figure 26Model prediction measured^data100PPI(thick)5OPPI(thick)30 PPI(thick)100PPI(thin)SOPPI(thin)30 PPI(thin)100PPI(thick)Ugr (m/s) ULd(m/s)Uhl(m/s)ULd(m/s)Uj(m/s)Uu(m/s)U1(m/s)U1(m/s)±ULd(m/s)0.00E+00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.001.63E-03 0.06 0.11 0.15 0.05 0.05 0.10 0.08 0.024.90E-03 0.10 0.20 0.27 0.08 0.13 0.17 0.14 0.038.16E-03 0.13 0.25 0.35 0.10 0.17 0.23 0.18 0.07122E-02 0.16 0.31 0.43 0.11 0.21 0.28 0.25 0.121.63E-02 0.18 0.36. 0.50 0.13 0.24 0.32 0.28 0.11188Data for Figure 27reactorconfig.3OPPI^thick wallwith short^supports3OPPI^thin wallwith short^supports30PPI^thick wallwith long^supportsglass draft tubeUpy (m/s) kT a (1/s) std. dev. kr a (1/s) std. dev. kT a (1/s) std. dev. ki. a (1/s) std. dev.0.60E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+001.63E-03 3.44E-03 1.15E-04 2.81E-03 1.68E-04 2.98E-03 4.13E-05 3.15E-03 1.29E-044.90E-03 7.60E-03 5.04E-04 7.65E-03 2.04E-04 7.72E-03 4.47E-04 6.48E-03 2.29E-048.16E-03 1.29E-02 9.50E-04 9.85E-03 2.84E-04 9.92E-03 7.80E-04 1.00E-02 3.84E-04122E-02 1.46E-02 9.06E-04 1.20E-02 432E-04 1.70E-02 3.13E-04 1.14E-02 3.21E-041.63E-02 _ 137E-02 1.23E-03 1.37E-02 3.97E-04 1.70E-02 1.23E-03 1.40E-02 4.51E-04reactorconfig.10OPPI^thick wallwith short^supports^100 PPI^thin wallwith short^supports100 PPI^thick wallwith long^supportsU97 (m/s) 14 a (1/s) std. dev. kT a (1/s) std. dev. kT a (1/s) std. dev.0.60E+00 0.00E+00 0.00E4-00 0.00E+00 0.00E+00 0.00E+00 0.00E+001.63E-03 2.92E-03 2.33E-04 3.21E-03 1.96E-04 3.97E-03 6.85E-044.90E-03 7.32E-03 1.82E-04 8.97E-03 6.80E-04 6.83E-03 2.60E-048.16E-03 8.65E-03 2.64E-04 1.16E-02 2.42E-04 1.22E-02 7.40E-041.22E-02 1.18E-02 4.12E-04 1.48E-02 5.32E-04 1.21E-02 4.68E-041.63E-02 1.21E-02 7.12E-04 1.63E-02 3.23E-04 1.36E-02 1.06E-03Data for Figure 28reactorconfig.glassdrafttube withpbsglass drafttube withserumthin 3OPPIwiht pbsthin 30PPI withserumUpy (m/s) 14 a (1/s) std. dev. kr a (1/s) std. dev. kT a (1/s) std. dev. kT a (Vs) std. dev.0.60E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+001.63E-03 3.15E-03 1.29E-04 3.67E-03 1.41E-04 2.81E-03 1.68E-04 3.10E-03 5.05E-054.90E-03 6.48E-03 2.29E-04 8.67E-03 1.60E-04 7.65E-03 2.04E-04 7.80E-03 1.84E-048.16E-03 1.00E-02 3.84E-04 1.09E-02 9.51E-05 9.85E-03 2.84E-04 1.02E-02 4.02E-041.22E-02 1.14E-02 3.21E-04 136E-02 2.12E-04 1.20E-02 4.32E-04 1.26E-02 5.58E-041.63E-02 1.40E-02 4.51E-04 1.42E-02 2.28E-04 137E-02 3.97E-04 138E-02 7.05E-04Data for Figure 29reactorconfiguration100PPI^with thick wall andlong supportsModelpredictionsUpy (m/s) kT a (1/s) std. dev. 14 a (1/s)0.60E+00 0.00E+00 0.00E+00 1.84E-031.63E-03 3.97E-03 6.85E-04 2.41E-034.90E-03 6.83E-03 2.60E-04 3.56E-038.16E-03 1.22E-02 7.40E-04 4.71E-031.22E-02 1.21E-02 4.68E-04 6.14E-031.63E-02 1.36E-02 1.06E-03 7.58E-03189Data For Figure 30Wall^Thickness(mm)AR = 0.038 m 2 AR = 0.0019 m2 AR = 0.0007 m 2 AR = 0.0003 m 210 0.0517 0.07867 0.0856 0.09515 0.0408 0.06061 0.069 0.07820 0.0347 0.0493 0.059 0.065930 0.0398 0.04712 0.05340 0.04025 0.045Data for Figure 31Time (h) Total glucose used (g) Total lactate produced (g)0 0.00 0.0027 0.20 0.2747 0.44 0.5449 0.44 1.0671 0.99 1.3277 1.10 1.4095 1.51 1.7196 1.29 1.71126 1.29 2.13119 1.33 2.23120 133 2.23126 137 2.30144 2.22 2.86171 2.80 3.45190 3.18 3.91214 3.74 4.73215 3.74 4.73225 3.93 5.03238 4.34 5.29246 4.34 5.29263 4.63 5.51264 4.63 5.51270 5.06 5.84287 6.28 732294 6.83 7.32312 7.53 10.48337 8.48 11.48359 9.13 11.38360 9.13 1138365 9.38 1135383 10.65 12.70391 11.64 13.23407 13.04 14.33414 13.56 14.72430 14.71 15.64438 15.20 16.39455 16.00 16.67462 16.54 17.05481 17.87 18.01482 17.87 18.01Data for Figure 31 continued504 19.23 19.30511 19.65 19.55527 20.99 20.36527 20.99 20.36535 21.70 21.18551 24.02 22.80553 24.15 22.54558 24.86 22.93558 24.86 22.93575 27.41 24.90581 28.35 25.72581 28.35 2532599 31.28 26.94604 31.73 27.72604 31.73 27.72623 35.26 30.08626 35.80 29.87626 35.80 29.87647 39.43 32.09648 39.43 32.09674 44.05 35.60676 44.05 35.60695 47.94 38.97701 48.80 38.93701 48.80 38.93719 52.62 42.09725 53.97 43.59727 53.97 43.59743 59.05 47.91750 60.49 48.91750 60.49 48.91767 66.65 54.13772 67.81 55.02772 67.81 55.02791 73.81 60.61794 74.13 58.90794 74.13 58.90815 82.01 67.45816 82.01 67.45839 90.09 74.33839 90.09 74.33863 98.01 81.25868 98.81 81.47868 98.81 81.47887 105.93 87.21892 106.85 87.96892 106.85 87.96911 113.97 95.37916 114.89 94.57190Data for Figure 31 continued916 114.89 94.57936 122.01 100.96940 122.69 101.34941 122.89 101.77941 122.89 101.77959 131.73 109.08965 133.01 110.28965 133.01 110.28987 143.65 118.71988 143.65 118.711013 154.13 127.581013 154.13 127.581033 163.65 135.221033 163.65 135.221056 174.69 139.661058 174.69 139.661085 175.93 140.991079 182.21 146.161082 182.21 146.161085 182.77 146.951104 190.01 152.781128 195.65 156.981128 195.65 156.981152 203.05 163.141152 203.05 163.141175 21133 169.711175 211.33 169.711199 219.61 176.721200 219.61 176.721223 228.17 183.651224 228.17 183.651248 236.85 190.771248 236.85 190.771271 245.61 197331274 245.61 197331295 254.65 204.931300 255.17 205381300 255.17 205381319 263.33 212.771320 26333 212.771347 272.69 220.421347 272.69 220.421367 281.57 226.931367 281.57 226.931391 291.21 235.011398 291.93 235.181398 291.93 235.181415 296.77 238.631415 296.77 238.63191192Data for Figure 31 continued1439 302.21 242.891439 302.21 242.891463 307.57 246.791463 307.57 246.79Data for Figure 32Time(min)Control std. dev. P80 std. dev. CM80 std. dev. S80 std. dev. Flex-CM80std. dev.0 0 -- 0 -- 0 -- 0 -- 0 --60 211000 405172 365000 410829 1.202E6 345421 2.517E6 132228 1.098E6 595030150 220500 311834 1.188E6 -- 774500 1.466E6 3.272E6 811405 1.739E6 53386240 -- -- 14500 -- 651500 544472 3.169E6 348839 2.551E6 40305330 332750 232991 1.074E6 85913 1.241E6 126218 3.307E6 856306 2.612E6 432749420 1.08E6 119501 825000 137178 2.084E6 568867 3.108E6 426031 2.94E6 37830Data for Figure 33Time(min)Control std. dev. P80 std. dev. CM80 std. dev. Flex-CM80std. dev. S80 std. dev.0 0 0 0 0 0 0 0 0 0 060 3.208E6 3.277E6 8.098E6 3.495E6 4.55E6 2.199E6 4.518E6 152027 6.38E6 3.125E6150 8.758E6 1.011E6 1.187E7 3.46E6 8.076E6 602808 9.19E6 1.513E6 1.223E7 2.383E6240 1.037E7 1.771E6 1.196E7 2.597E6 1.076E7 1.777E6 1.048E7 572756 1.354E7 2.721E6330 1.187E7 1.575E6 _1.383E7 1.452E6 1.199E7 . 841457 1.142E7 1.694E6 1.53E7 2.238E6Data for Figure 34Time(min)Control std. dev. DEA80 std. dev. DEA-LG80std. dev. S80 std. dev. PER80 std. dev.0 0 0 0 0 0 0 0 0 0 060 265000 18296 760000 206475 447000 362392 659000 25043 98100 313778150 275000 367412 139E6 424264 1.28E6 713735 1.51E6 57540 954000 327832240 419000 431688 1.59E6 375473 1.18E6 936916 1.64E6 3977 1.07E6 233610330 388000 439113 1.55E6 381395 1.16E6 773663 1.52E6 29256 987000 184643Data for Figure 35Time(min)Control std. dev. S80 std. dev. PER80 std. dev. DEA80 std. dev. DEA-LG80std. dev.0 0 0 0 0 0 0 0 0 0 060 2.349E6 2.955E6 1.65E7 1.361E6 7.191E6 620368 6.993E6 395979 1.142E7 3.935E6150 1.173E7 7.213E6 2.305E7 4.838E6 1.189E7 7.56E6 1.601E7 2.445E6 1.569E7 272236240 1.496E7 282135 2.473E7 3.608E6 1.434E7 5.4E6 1.72E7 3.665E6 1.532E7 3.435E6330 1.212E7 485075 2.647E7 2.868E6 1.75E7 4.503E6 1.979E7 252437 1.878E7 3.656E6Data for Figure 36Time(min)Control std. dev. Cytodex-1std. dev. Culti-spherstd. dev. S80 std. dev. CollegenCoatedstd. dev.0 0 0 0 0 0 0 0 0 0 060 156000 4680 340000 151000 80500 87000 228000 12400 190000 89600150 135000 17400 373000 33000 389000 44500 424000 25900 441000 4240240 161000 . 74600 432000 50100 387000 93300 460000 30100 534000 3010193Data for Figure 37Time(min)CollegenCoatedstd. dev. S80 std. dev. Cytodex-1std. dev. Culti-spherstd. dev. Control std. dev.0 0 0 0 0 0 0 0 0 0 060 654000 297161 541000 42544 693000 131757 1.07E6 13199 339000 289029150 1.4E6 130932 1.38E6 59072 1.97E6 171355 1.71E6 171055 186000 510148240 1.53E6 144780 1.6E6 40010 1.89E6 225838 1.69E6 125022 468000 710170330 1.67E6 68872 1.79E6 51453 1.99E6 161927 . 1.75E6 73173 116000 414600Data for Figure 38Time(min)S80 std. dev. S40-15 std. dev. S40-60 std. dev. S40-240 std. dev. Control std. dev.0 3E6 0 3E6 0 3E6 0 3E6 0 3E6 060 1.77E6 106000 1.57E6 99000 1.55E6 49500 1.81E6 170000 2.84E6 332000150 848000 53000 946000 67200 795000 2120 846000 48100 2.78E6 177000240 521000 55200 488000 81300 415000 99700 444000 15600 2.59E6 311000330 315000 22600 466000 58700 400000 10600 355000 9900 1.96E6 283000Data for Figure 39Time(min)S80 std. dev. S40-15 std. dev. S40-60 std. dev. S40-240 std. dev. Control std. dev.0 3.58E6 0 3.58E6 0 3.58E6 0 3.58E6 0 3.61E6 3890060 2E6 117000 2.1E6 220000 2.74E6 166000 231E6 258000 3.51E6 203000150 1.52E6 184000 132E6 14100 1.64E6 38900 1.81E6 148000 3.02E6 431000240 910000 14100 660000 184000 1000000 138000 1.2E6 108000 2.91E6 108000330 598000 190000 580000 89600 653000 95500 605000 87800 2.75E6 0Data for Figure 40Time(min)S80-WETstd. dev. S80-DRYstd. dev. S40-WETstd. dev. S40-DRYstd. dev. Control std. dev.0 2.744E6 0 2.744E6 0 2.744E6 0 2.744E6 0 2.744E6 030 1.644E6 95812 2.687E6 112076 1.925E6 37830 2.846E6 258565 2.483E6 183140120 691777 5342 1.98E6 51618 904833 37476 2324E6 148963 1.876E6 73539240 376333 31112 1318E6 178190 522666 16027 1.825E6 180233 1.241E6 . 94045Data for Figure 41Time(min)Control std. dev. <300 pm std. dev. 300 - 500pmstd. dev. > 500 pin std. dev.0 2.04E6 0 2.04E6 0 2.04E6 0 2.04E6 060 1.71E6 299000 544000 26800 1.04E6 158000 1.17E6 71400150 1.56E6 89600 81200 98800 532000 65100 624000 39600240 1.45E6 94300 76000 11300 295000 49500 340000 31100330 1.72E6 38700 19000 1410 115000 14100 186000 5660194Data for Figure 42Time(min)5.6x106 std. dev. 2.8x106 std. dev. 1.4x106 std. dev. 0.7x 06 std. dev. Control std. dev.0 0 0 0 0 0 0 0 0 0 060 21.2 11.8 25.9 1.7 30.5 1.8 23.5 12.1 -1.29 0.0390 53.2 2.1 58.9 5.4 64.6 3.8 72.8.-- 2.0 10.0 3.5150 58.9 1.9 79.7 1.1 84.2 0.8 81.7 1.3 28.2 2.8Data for Figure 43Time(min)DMEMserumstd. dev. DMEM/F-12serumstd. dev. DMEM std. dev. PBS std. dev. Control(DMEMserum)std. dev.0 5.77E6 0 5.8E6 0 6.09E6 42400 6.12E6 0 6.13E6 3540035 4.32E6 76400 4.69E6 890000 3.97E6 381000 4.03E6 76700 6.15E6 8450095 3.78E6 187000 4.48E6 1.08E6 3.55E6 607000 4.09E6 85200 6.08E6 47000185 2.11E6 101000 2.45E6 233000 3.11E6 37500 2.91E6 320000 5.62E6 427000275 1.35E6 112000 _ 2.39E6 418000 2.78E6 294000 3.03E6 770000 4.83E6 28600Data for Figure 44Medium Glucose Conc.WOUnit GUR (g/h/L) Unit GUR (g/h/L)1 0.8 0.51.5 1.3 0.92.2 1.7 13.4 1.8 0.954 1195Data for Figures 45 and 46Time (h)BILK cellsGlucoseConcentrationWO0 3.327 1.6751.5 0.8851.5 4.1472 2.8396 1.5120 0.83120 4.1144 2.62168 1.3168 4.17192.75 2.1217 0.95217 4.17240 2.1264 0.88264 4.08288 2312 0.86Time (h)Vero cellsGlucoseConcentrationWO0 452 25872 1.5196 0.796 3.11120 2.03144 1.08144 4.23168 2.28197 0.82197 4.34218 2.09240 0.76240 4.23264 1.7288 0.46288 4.26312 1.7336 0.45336 4.24363 1.63388 0.47388 4.12408 2432 0.68432 4.12456 1.79196Data for Figure 48Time(h)CM80 DEA80 P40 Time(1)P80 S80 Time(h)Q80 Time(h)CS800 0 0 0 0 0 0 0 0 0 052 0.142 0.086 0.086 21 0.029 0.029 27 0.05 24 0.03272 0.249 0.158 0.158 43 0.071 0.071 51.5 0.106 48.5 0.09896 0.33 0.264 0.264 67 0.156 0.156 72 0.165 74 0.176120 0.438 0.348 0.348 91 0.284 0.284 96 0.222 98 0.261144 0.533 0.463 0.463 115 0.362 0.362 120 0.274 120 0.347168 0.728 0.561 0.561 139 0.508 0.508 144 0.29 144 0.46197 0.874 0.798 0.798 166 0.649 0.649 168 0.35 168 0.51218 1.099 0.926 0.926 190.5 0.853 0.853 193 0.393 192 0.602240 1.232 1.156 1.156 211 0.97 0.97 217 0.427 216 0.745264 1.485 1.297 1.297 235 1.181 1.181 240 0.46 242.5 0.836288 1.609 1.591 1.591 259 1.306 1.306 264 0.497 266.5 0.996312 1.865 1.688 1.688 283 1.507 1.507 288 0.531 290 1.087336 1.99 1.907 1.907 307 1.632 1.632 312 0.57 311.5 1.224363 2.251 2.048 2.048 322 1.857 1.857 336.5 0.614 335.5 1.325388 2367 2.262 2.262 356 1.983 1.983 362 0.642 359 1.54408 2579 2.38 2.38 386 0.718 383 1.646432 2.711 2.583 2.583 408 0.788 410 1.894456 2.944 2.717 2.717 432 0.858Data for Figure 49 (cell concentration x 10 -8Time(h)CM80 DEA80 P40 Time(h)P80 S80 Time(h)080 Time(h)CS800 0.60 039 0.39 0.0 0.24 0.24 0.0 0.31 0.0 03052 0.60 039 0.39 21.0 0.24 0.24 27.0 0.31 24.0 0.3072 0.64 0.59 059 43.0 0.40 0.40 51.5 038 48.5 0.4296 0.58 0.58 67.0 0.65 0.65 72.0 039 74.0 0.49120 0.62 91.0 0.63 0.63 96.0 0.33 120.0 0.64144 0.65 0.65 139.0 0.83 0.83 120.0 0.21 168.0 0.44168 0.97 190.5 1.04 1.04 168.0 031 216.0 0.69197 1.05 1.05 235.0 1.03 1.03 193.0 0.23 266.5 0.82218 1.24 283.0 1.00 1.00 217.0 0.21 290.2 0.84240 1.20 1.20 322.0 1.38 1.38 240.0 0.22 3115 0.56264 1.16 264.0 0.22 335.5 0.87288 1.20 1.20 288.0 0.22 383.0 1.00312 1.17 312.0 0.25336 1.06 1.06 336.5 0.21363 1.06 386.0 0.47388 1.07 1.07 408.0 0.45408 1.17432 1.04 1.04Data for Figure 50Time (h) Cytodex DEA800 0 024 0.43 0.545.5 0.97 1.1270 1.8 2.0294 2.37 2.6894 2.37 2.68118 3.76 4.01142 4.92 5.42142 4.92 5.42169 6.78 7.49193 7.78 8.52193 7.78 8.52214 9.92 10.25238 11.12 11.56238 11.12 1156262 13.3 13.31262 13.3 13.31285.5 15.45 15.24285.5 15.45 15.24309.5 17.75 17.25309.5 17.75 17.25337 20.26 19.6337 20.26 19.63595 22.61 21.653595 22.61 21.65382 24.96 23.88382 24.96 23.88406 26.21 26406 26.21 26430 28.53 283430 2853 28.3Data for Figure 51Time (h) Cytodex CG Time (h) CM80 Time (h) P800 0.00 0.00 0 0.00 0 0.006.25 0.23 0.29 6 0.13 24.5 0.5624 1.26 1.28 24 0.61 50.5 15830 1.53 1.58 30 0.99 74.75 2.85485 2.52 2.57 48 1.99 97 3.5272 3.12 3.19 54 2.32 120.5 5.2672 3.12 3.19 72 3.38 1455 6.2297.5 4.53 4.73 78 3.78 169 8.38123 535 5.48 96 532 192.25 9.71123 5.35 5.48 97 5.32 218 11.56144 6.50 6.81 102 5.74 243.5 12.29150 6.87 7.14 1215 7.37 265 13.84168 7.55 7.79 146.5 8.07 288.5 14.71168 7.55 7.79 146.75 8.07 312.5 16.52174 7.97 8.16 170 10.13 336 17.22197Data for Figure 51 continued192 9.28 9.62 192 11.32 360.5 19.16198 9.58 9.91 216 12.21 387.25 20.01216 9.98 10.40 240 13.80 412 21.53216 9.98 10.40 264 14.66240 11.63 11.44 290.5 16.22245.5 11.77 11.62 315 16.98267 12.52 12.54 336 18.11267 12.52 12.54 360 19.06291 13.83 13.66 366.25 19.36313 14.61 14.53 384 20.55313 14.61 14.53 390 20.85319 14.97 14.89 408.5 21.47336.75 16.19 15.94343 16.52 16.26361 17.26 16.95361 17.26 16.95367 17.61 17.22385 18.80 18.34391 19.11 18.65409 19.70 19.27Data for Figure 52Time (h) Cytodex DEA80 Time (h) CM800 0 0 0 024 0.43 0.5 27 1.6345.5 0.97 1.12 51.5 2.4270 1.8 2.02 51.5 2.4294 237 2.68 72 3.7394 2.37 2.68 96 5.06118 3.76 4.01 120 5.73142 4.92 5.42 120 5.73142 4.92 5.42 144 7.21169 6.78 7.49 168 8.53193 7.78 8.52 168 8.53193 7.78 8.52 192.75 10.6214 9.92 10.25 217 11.75238 11.12 11.56 217 11.75238 11.12 11.56 240 13.82262 13.3 1331 264 15.04262 133 1331 264 15.04285.5 15.45 15.24 288 17.12285.5 15.45 15.24 312 18.26309.5 17.75 17.25 312 18.26309.5 17.75 17.25 336.5 20.63337 20.26 19.6 362 21.86337 20.26 19.6 362 21.86359.5 22.61 21.65 386 24.37359.5 22.61 21.65 408 25.46382 24.96 23.88 408^, 25.46198Data for Figure 52 continued382 24.96 23.88 432 27.71406 26.21 26406 26.21 26430 28.53 28.3430 28.53 28.3Data for Figure 53Time (h) CM80 Time (h) Cytodex Time (h) CG0 0 0 0 0 048 16.929 72 9.9324 6 2.254 21.629 123 19.5638 24 4.772 22.829 168 30.1068 24 4.772 22.829 216 433213 30 6.22978 24.039 267 63.2213 72 9.932496 30.699 313 75.7213 123 19.563897 30.699 361 87.2213 168 33.9233102 32.434 409 106.656 216 47.1378121.5 38.914 267 55.7378146.5 41514 313 65.7378146.75 41.514 361 73.2378170 47.694 409 813378170 47.694192 60.904216 64.914216 64.914240 74.303264 79.094264 79.094290.5 87.046315 89.774315 89.774336 96.739360 99.703360 99.703366.25 101.021384 106.656390 108.128408.5 109.153408.5 109.153199Data for Figure 54Time(h)P5 P5 P25 P25 Time(h)P40 P40 P80 P800 0.000 0.000 0.000 0.000 0 0.000 0.000 0.000 0.00023.5 0.785 0.725 0.375 0.285 23.5 0.560 0.565 0.140 0.35547 1.595 1.445 0.725 0.645 47 1.200 1.225 0.235 0.78068.5 2.230 2.060 1.165 1.075 68.5 1.735 1.815 0.480 1.33592.25 2.680 2.410 1.575 1.490 92.25 2.115 2.235 0.570 1.745116.5 3.023 2.695 2.050 1.910 116.5 2.480 2.520 0.700 2.125117 3.023 2.695 2.050 1.910 117 2.480 2.520 0.700 2.125141.5 3.858 3.380 2.650 2.460 141.5 3.100 3.080 0.950 2.640165.5 4.793 4.113 3.445 3.273 165.5 3.900 3.893 0.950 3.255191 5.873 5.050 4.460 4.315 191 4.950 4.710 0.920 4.215213 6375 5.645 5.105 4.915 213 5.530 5.335 0.950 5.010214 6.375 5.645 5.105 4.915 214 5.530 5335 0.950 5.010236.5 7.030 6.282 5.760 5.645 236.5 6.220 5.900 1.050 5.595260.75 7.825 7.087 6.680 6.475 260.75 6.970 6.630 1.520 6.395285 8.425 7.557 7.250 285 7.575 7.105 2.025 6.940309 8.788 7.987 7.710 309 7.965 7.500 2.450 7.395309.5 8.788 7.987 7.710 309.5 7.965 7.500 2.450 7.395334 9342 8.467 8.415 334 8.550 7.960 3.050 7.850359.5 9.998 9.107 9.515 359.5 9.205 8.660 3.810 8.745381 10.748 9.982 10.410 381 10.030 9.400 4.510 9.705405 11.573 10.702 11.965 405 10.748 10.095 5.157 10590429 12.793 11.977 13.400 411 10.975 10370 5.345 10.847453.5 13.843 13.067 15.212 429 11.845 11.290 6305 11.792477.5 14.853 14.092 16350 453.5 12.925 12340 7.280 12.972503.5 16.908 16.317 18.215 477.5 13.800 13.205 8.120 14.107528.75 18.233 17.697 19.535 503.5 15.330 14.720 9.735 15.837548.5 20.103 19347 21.155 528.5 16.650 15.980 10.995 17.257573 21.523 20.772 22535 548.5 17.955 17.485 12.415 18.802573 19395 18.875 13.860 20.317Data for Figure 55Time(h)PS P5 P25 P25 P40 P40 P80 P800 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.00022 0.290 0320 0355 0360 0.320 0.280 0.345 0350475 0.455 0.730 0.835 0.830 0.630 0.710 0.805 0.76071.5 0.745 0.990 1.085 1.160 0.930 1.060 1.255 1.22095.5 1.250 1.420 1.535 1.470 1.170 1.470 1.685 1.650118 1515 1.790 1.785 1.720 1.270 1.860 2.020 2.070118 1.515 1.790 1.785 1.720 1.270 1.860 2.020 2.070142 1.930 2.280 2.135 2.000 1.710 2.390 2.885 2.480142 1.930 2.280 2.135 2.000 1.710 2390 2.885 2.480166.5 2355 2.910 2.425 2.160 2.070 3.100 3.645 3.130166.5 2355 2.910 2.425 2.160 2.070 3.100 3.645 3.130190 2.735 3510 2.555 2.470 2310 3.890 4325 3.820190 2.735 3.510 2.555 2.470 2.310 3.890 4.325 3.820216 3.415 4.240 2.755 2.650 2.510 4.510 5.205 4.400216 3.415 4.240 2.755 2.650 2.510 4.510 5.205 4.400240.5 4.245 5.240 2.945 3.020 3.180 5.760 6.215 5.310240.5 4.245 5.240 2.945 3.020 3.180 5.760 6.215 5.310264 5.295 6.100 3.065 3.260 3.750 6.770 7.415 6.170200Data for Figure 55 continuedTime(h)P5 PS P25 P25 P40 P40 P80 P80264 5.295 6.100 3.065 3.260 3.750 6.770 7.415 6.170287 6.095 6.690 3.225 3.520 3.980 7.580. 8.235 6.810287 6.095 6.690 3.225 3.520 3.980 7.580 8.235 6.810312 6.915 7.730 3.325 3.710 4.710 8.840 9.725 7.810312 6.915 7.730 3.325 3.710 4.710 8.840 9.725 7.810336 8.005 8.640 3.475 3.780 5.410 10.250 11.255 8.910336 8.005 8.640 3.475 3.780 5.410 10.250 11.255 8.910360 8.835 9520 3.565 4.010 6.120 11.200 12.115 9.890360 8.835 9.520 3.565 4.010 6.120 11.200 12.115 9.890387.5 9.865 10.890 3.705 4.290 7.100 12.760 13.695 11.310387.5 9.865 10.890 3.705 4.290 7.100 12.760 13.695 11310412.5 10.075 11.890 3.855 4.710 8.400 13.600 14.745 12.240412.5 10.075 11.890 3.855 4.710 8.400 13.600 14.745 12.240431.5 11.185 12.690 3.945 4.950 9.060 14.410 15.625 13.020Data for Figure 56Time(h)P5 P5 P25 P25 P40 P40 P80 P800 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.00023 0.390 0.390 0.285 0.440 0510 0.460 0.485 0.600485 0.620 0.740 0585 0.750 0.820 0.860 0.915 0.99071 1.110 1.060 1.095 1390 1.460 1.410 1565 1.56095 1.650 1.450 1.525 2.130 2.090 2.090 2.095 2.180121 2340 2.010 2205 2.910 2.860 2.770 2.685 2.750121 2340 2.010 2.205 2.910 2.860 2.770 2.685 2.750148 3.050 2.420 2.965 4.150 3.940 3.800 3.655 3.780167 3.710 3.040 3.555 4.850 4.660 4.480 4295 4500191 4.290 3.720 3.985 5.310 5.160 5.020 4.855 5.000191 4.290 3.720 3.985 5.310 5.160 5.020 4.855 5.000215 5.030 4.570 4.785 6.420 6.080 5.930 5.775 5.990239 5.740 5.160 5.575 7.090 6.800 6.660 6.605 6.750239 5.740 5.160 5.575 7.090 6.800 6.660 6.605 6.750263 6390 5.440 6.275 7.760 7.640 7.400 7.375 7550290 7.180 5.770 7.115 8590 8.430 8.170 8315 8.440290 7.180 5.770 7.115 8.590 8.430 8.170 8315 8.440314 7.960 6.080 7.935 9.510 9.030 8.850 9.075 9.310335 8580 6.400 8.555 10.080 9.730 9570 9.805 10.100335 8.580 6.400 8.555 10.080 9.730 9.570 9.805 10.100359 9.310 6.670 9.215 10.940 10.400 10.230 10.715 11.000383.5 9.970 6.980 9.965 11590 11.080 10.880 11565 11.8803835 9.970 6.980 9.965 11.590 11.080 10.880 11.565 11.880407 10.810 7.440 10.805 12520 11.870 11.650 12.645 12.890431 11.420 7.830 11.505 13.170 12.470 12.290 13.345 13.630431 11.420 7.830 11.505 13.170 12.470 12290 13345 13.630455.5 12.290 8.280 12335 14.150 13.380 13.180 14.355 14.770481.5 13.110 8.870 13.265 15.020 14.220 14.070 15.435 15.7704815 13.110 8.870 13265 15.020 14.220 14.070 15.435 15.770503 13.790 9.250 13.915 15.860 14.880 14.860 16.405 16.760201Data for Figure 57Time01)P5 P5 P25 P25 P40 P40 P80 P800 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.00023 0.216 0.238 0.250 0.248 0.249 0.293 0.264 0.30448.5 0.455 0.501 0.526 0.585 0.601 0.664 0.706 0.78471 0.826 0.781 0.865 1.058 1.011 1.129 1.129 1.14795 1.243 1.066 1.254 1.608 1.551 1.649 1.483 1.530121 1.639 1.340 1.698 2.108 2.025 2.085 1.801 1.743121 1.639 1.340 1.698 2.108 2.025 2.085 1.801 1.743148 2.269 1.740 2.378 3.137 2.882 2.939 2.553 2.507167 2.740 2.152 2.782 3.625 3.453 3.423 3.053 3.075191 3.035 2.547 3.261 3.922 3.755 3.865 3.458 3.456191 3.035 2.547 3.261 3.922 3.755 3.865 3.458 3.456215 3.544 3.058 3.933 4.731 4.509 4.561 4.140 4.251239 4.064 3.496 4.605 5.236 5.139 5.028 4.744 4.704239 4.064 3.496 4.605 5.236 5.139 5.028 4.744 4.704263 4.550 3.672 5.146 5.885 5.691 5.594 5.318 5.288290 5.019 3.916 5.721 6.368 6.167 6.045 5.919 5.858290 5.019 3.916 5.721 6.368 6.167 6.045 5.919 5.858314 5.598 4.165 6.357 7.009 6.735 6.616 6.564 6.566335 6.019 4372 6.807 7.425 7.081 7.013 6.999 7.014335 6.019 4.372 6.807 7.425 7.081 7.013 6.999 7.014359 6.596 4.601 7.370 8.117 7.696 7.595 7.759 7.7953835 7.150 4.888 7.938 8.605 8.137 8.043 8344 8.4063835 7.150 4.888 7.938 8.605 8.137 8.043 8.344 8.406407 7.728 5.222 8.568 9.332 8.687 8.613 9.173 9.178431 8.238 5.545 9.108 9.852 9.111 9.083 9.701 9.703431 8.238 5.545 9.108 9.852 9.111 9.083 9.701 9.703455.5 8.906 5.884 9.738 10.617 9.723 9.712 10.463 105774815 9.417 6.269 10.279 11.104 10.224 10.205 11.132 11.1454815 9.417 6.269 10.279 11.104 10.224 10.205 11.132 11.145503 9.955 6.604 10.818 11.791 10.757 10.806 11.958 11.964Data for Figure 60Time Level#1Level#1Level#2Level#2Level#3Level#3Level#4Level#40 0 0 0 0 0 0 0 021 0.13 0.09 0.3 0.22 0.37 0.16 0.26 0.1746.75 0.16 0.25 0.57 0.42 0.78 0.46 0.64 0.569.5 0.44 0.6 1.03 0.99 1.5 1.09 1.29 1.3292 0.44 0.83 1.62 1.52 2.24 1.55 1.96 2.192 0.76 0.83 1.62 1.52 2.24 1.55 1.96 2.1116.75 124 1.41 2.41 234 331 2.7 3.18 3.68141.5 2.01 2.07 3.45 3.31 4.49 3.65 4.07 4.69141.5 2.01 2.07 3.45 3.31 4.49 3.65 4.07 4.69165.5 2.59 2.88 4.82 4.64 5.86 4.68 5.47 6.1165.5 2.59 2.88 4.82 4.64 5.86 4.68 5.47 6.1189 3.62 3.75 6.21 6.11 7.47 6.14 6.91 7.92189 3.62 3.75 6.21 6.11 7.47 6.14 6.91 7.92214 4.77 5.1 7.67 7.62 9.17 7.91 8.29 9.94214 4.77 5.1 7.67 7.62 9.17 7.91 8.29 9.94238.5 6.03 5.78 9.34 9.27 10.94 9.36 8.7 11.98202203Data for Figure 60 continuedTime Level Level Level Level Level Level Level Level(h) #1 #1 #2 #2 #3 #3 #4 #4238.5 6.03 5.78 9.34 9.27 10.94 9.36 8.7 11.98261 7.68 7.36 11.4 10.88 12.96 10.9 10.41 14.05261 7.68 7.36 11.4 10.88 12.96 10.9 10.41 14.05285 9.4 9.01 13.9 12.63 14.91 12.58 12.31 16.24285 9.4 9.01 13.9 12.63 14.91 12.58 12.31 16.24309 11.36 11.24 15.79 1456 16.77 14.4 13.95 18.62309 11.36 11.24 15.79 14.56 16.77 14.4 13.95 18.62333.5 12.89 12.77 17.2 15.99 1839 16.88 15.24 20.6333.5 12.89 12.77 17.2 15.99 18.39 16.88 15.24 20.6356.5 14.56 14.54 19.14 17.98 20.57 18.25 17.15 22.66356.5 14.56 14.54 19.14 17.98 20.57 18.25 17.15 22.66381 16.33 16.4 20.76 19.89 22.87 19.64 19.01 24.48381 1633 16.4 20.76 19.89 22.87 19.64 19.01 24.48404.5 17.79 18.23 22.44 21.8 24.88 21.2 20.55 26.34404.5 17.79 18.23 22.44 21.8 24.88 21.2 20.55 26.34428 19.77 20.21 24.35 23.88 26.98 22.71 22.56 28.16428 19.77 20.21 2435 23.88 26.98 22.71 22.56 28.16452 21.65 22.29 26.01 26.39 28.85 24.36 25.89 30.05452 21.65 22.29 26.01 28.85 24.36 25.89 30.05476 23.52 24.29 27.86 31.79 26.66 27.86 31.97476 23.52 24.29 27.86 31.79 26.66 27.86 31.97500 25.51 25.97 29.96 33.46 29.03 30.36 33.52500 25.51 25.97 29.96 33.46 29.03 30.36 33.52522.5 27.41 27.47 31.8 35.37 30.49 32 34.9522.5 27.41 27.47 31.8 35.37 30.49 32 34.9550.5 29.57 2937 34.12 37.29 31.93 34.07 36.95550.5 29.57 29.37 34.12 37.29 31.93 34.07 36.95574.5 31.86 31.2 36.42 39.09 33.61 36.26 38.71574.5 31.86 31.2 36.42 39.09 33.61 36.26_ 38.71597 33.53 32.81 38.23 40.58 35 38.24 40.18Data for Figure 61Time Level Level Level Level Level Level Level Level(h) #1 #1 #2 #2 #3 #3 #4 #40 0 0 0 0 0 0 0 021 0.26 0.11 0.25 0.13 0.26 0.23 0.25 0.346.75 0.2 0.2 031 0.21 0.46 0.46 0.42 0.569.5 0.4 0.32 057 0.5 0.8 0.86 0.68 0.8192 0.41 0.24 0.82 0.62 1.04 1.09 0.89 1.02116.75 0.71 033 133 1.07 1.37 1.61 1.33 2.06141.5 1.02 0.68 1.95 1.61 1.89 2.4 1.96 2.13141.5 1.02 0.68 1.95 1.61 1.89 2.4 1.96 2.13165.5 1.22 0.71 2.45 2 2.06 3.15 2.21 2.62165.5 122 0.71 2.45 2 2.06 3.15 2.21 2.62189 1.64 1.06 3.56 2.62 2.74 4.32 3.1 3.64189 1.64 1.06 3.56 2.62 2.74 4.32 3.1 3.64214 1.8 1.17 4.67 3.46 3.31 5.63 3.81 4.39214 1.8 1.17 4.67 3.46 3.31 5.63 3.81 4.39238.5 2.46 1.48 5.88 4.73 4.34 7.51 5.08 5.37238.5 2.46 1.48 5.88 4.73 4.34 7.51 5.08 5.37Data for Figure 61 continuedTime(It)Level#1Level#1Level#2Level#2Level#3Level#3Level#4Level#4261 2.94 1.83 7.07 5.81 4.86 9.15 6.41 6.52261 2.94 1.83 7.07 5.81 4.86 9.15 6.41 6.52285 3.83 2.44 8.58 7.46 5.94 11.3 8.16 7.65285 3.83 2.44 8.58 7.46 5.94 11.3 8.16 7.65309 4.67 2.97 10.72 8.52 7.13 14.09 10.42 9.71309 4.67 2.97 10.72 8.52 7.13 14.09 10.42 9.71333.5 5.69 3.58 12.41 9.93 8.68 16.24 12.04 11.55333.5 5.69 3.58 12.41 9.93 8.68 1624 12.04 11.55356.5 6.86 4.75 14.17 12.03 10.61 18.36 13.83 13.32356.5 6.86 4.75 14.17 12.03 10.61 18.36 13.83 13.32381 8.06 538 16.02 13.95 12.92 21.88 15.44 15.46381 8.06 538 16.02 12.17 12.92 21.88 15.44 15.46404.5 9.26 6.27 17.7 13.94 153 24.13 16.92 17.24404.5 9.26 6.27 17.7 13.94 153 24.13 16.92 17.24428 10.95 7.69 19.6 16.05 17.84 26.36 19.17 19.1428 10.95 7.69 19.6 16.05 17.84 2636 19.17 19.1452 12.33 9.25 21.61 18.28 20.22 28.29 21.13 21.25452 12.33 9.25 21.61 18.28 20.22 28.29 21.13 21.25476 14.46 10.9 23.54 21.77 22.8 30.33 23.45 24.4476 14.46 10.9 23.54 21.77 22.8 3033 23.45 24.4500 16.52 12.53 25.82 23.25 25.06 33.27 25.53 26.34500 16.52 12.53 25.82 23.25 25.06 33.27 25.53 26345225 18.36 14.04 27.97 24.85 27.14 35.49 27.54 28.25522.5 1836 14.04 27.97 24.85 27.14 35.49 27.54 28.25550.5 20.59 15.73 3031 26.54 29.43 3734 29.73 30.11550.5 20.59 15.73 3031 26.54 29.43 37.54 29.73 30.11574.5 22.95 17.78 32.61 28.26 31.46 39.85 32.15 31.875743 22.95 17.78 32.61 28.26 31.46 39.85 32.15 31.87597 24.92 19.74 34.24 29.8 33.28 41.41 34.13 33.17Data for Figure 62Time Level Level Level Level Level Level Level Level#1 #1 #2 #2 #3 #3 #4 #40 0 0 0 0 0 0 0 021 0.034 0.104 0.198 0.149 0.052 0.12 0.131 0.14246.75 0.134 0.347 0.617 0.485 0.16 0.339 0.381 0.4669.5 0.33 0.717 1.153 0.967 0364 0.729 0.779 1.03292 0.753 1.457 2.015 1.832 0.779 1.416 1.44 1.97792 0.753 1.457 2.015 1.832 0.779 1.416 1.44 1.977116.75 1.06 2.071 2.623 2.609 1.08 2.021 2302 3.186141.5 1.69 2.907 3.706 3.448 1.715 2.834 3.038 3.982141.5 1.69 2.907 3.706 3.448 1.715 2.834 3.038 3.982165.5 2.31 4.067 4.89 4.674 2.454 3.977 3.888 5.174165.5 2.31 4.067 4.89 4.674 2.454 3.977 3.888 5.174189 3.01 5.127 6.22 5.884 3.06 5.156 5.09 5.283189 3.01 5.127 6.22 5.884 3.06 5.156 5.09 6.647214 4.01 6.325 7.619 7.17 4.063 6.243 65 6.735214 4.01 6.325 7.619 7.17 4.063 6.243 6.5 6.735238.5 4.686 7.348 8.86 8.497 4.433 7.469 7.651 8.323204Data for Figure 62 continuedTime Level Level Level Level Level Level Level Level(II) #1 #1 #2 #2 #3 #3 #4 #4238.5 4.686 7.348 8.86 8.497 4.433 7.469 7.651 8.323261 6.067 9.074 10.614 10.079 5.81 8.987 9.053 10.245261 6.067 9.074 10.614 10.079 5.81 8.987 9.053 10.245285 7.565 11.41 12.456 11.837 7.305 10.597 10.598 12.22285 7.565 11.41 12.456 11.837 7.305 10.597 10.598 12.22309 9.964 12.898 14.03 13.145 9.045 12.092 11.933 14.185309 9.964 12.898 14.03 13.145 9.045 12.092 11.933 14.185333.5 11.303 14.195 15.557 14.448 10.527 13.423 14.065 16.088333.5 11.303 14.195 15.557 14.448 10.527 13.423 14.065 16.088356.5 12.772 15.982 17.756 16359 12293 15376 15346 18.058356.5 12.772 15.982 17.756 16.359 12.293 15376 15346 18.058381 14.481 17.094 19.245 17.772 13.899 17.546 16.744 19.35381 14.481 17.094 19.245 17.772 13.899 17.546 16.744 1935404.5 15.829 18.848 21.396 19.257 15.818 19.62 18.292 21.325404.5 15.829 18.848 21.396 19.257 15.818 19.62 18.292 21.325428 17.504 20.658 23.705 21.367 17.639 21.458 19.575 22.838428 17.504 20.658 23.705 21367 17.639 21.458 19.575 22.838452 19.19 21.879 25.098 23.715 19.278 23.594 20.802 24.296452 19.19 21.879 25.098 23.715 19.278 20.802 24.296476 20.764 23.241 27.344 25.329 20.877 22.731 25.973476 20.764 23.241 27344 25.329 20.877 22.731 25.973500 22.594 25.11 28.738 27.438 22.482 24543 27.488500 22.594 25.11 28.738 27.438 22.482 24543 27.4885225 24.713 27273 30.946 29.038 24.016 25.912 28.925522.5 24.713 27.273 30.946 29.038 24.016 25.912 28.925550.5 26.457 .29.035 32.526 30.717 25.451 27.205 303135505 26.457 29.035 32526 30.717 25.451 27.205 30313574.5 28.466 31.073 33.744 32.666 27.118 28.66 31.977574.5 28.466 31.073 33.744 32.666 27.118 28.66 31.977597 29.864 32585 35.157 34222 28398 29.844 33217Data for Figure 63Time Level#1Level#1Level#2Level#2Level#3Level#3Level#4Level#40 0 0 0 0 0 0 0 021 0.059 0.024 0.064 0.057 0.124 0.116 0.072 0.09446.75 0.091 0.104 0.191 0.199 0.352 0.332 0.227 0.25369S 0.198 0.158 0.404 037 0.619 0397 0388 051992 0.366 0.259 0.724 0.659 0.964 1.067 0.802 0.881116.75 0.498 0.304 1.005 0.849 1.17 1.235 1.008 1.0761413 0.77 0.513 1.55 1.306 1.671 1.977 1.621 1.619141.5 0.77 0.513 1.55 1.306 1.671 1.977 1.621 1.619165.5 0.99 0.671 2.13 1.546 2.032 2.759 1.997 2.161165.5 0.99 0.671 2.13 1.546 2.032 2.759 1.997 2.161189 1.33 0.868 2.9 2.097 2.432 3.567 2.574 2.761189 1.33 0.868 2.9 2.097 2.432 3.567 2.574 2.761214 1.68 0.945 4.2 2.683 2.975 4.578 3.164 3.36214 1.678 0.945 4.196 2.683 2.975 4.578 3.164 3.36238.5 1.994 1.129 4.666 3.557 3.669 5.788 4.075 3.934205Data for Figure 63 continued...Time Level Level Level Level Level Level Level Level(11) #1 #1 #2 #2 #3 #3 #4 #4238.5 1.994 1.129 4.666 3.557 3.669 5.788 4.075 3.934261 2.544 1.461 5.712 4.381 4.336 7.122 5.173 4.917261 2.544 1.461 5.712 4.381 4.336 7.122 5.173 4.917285 3.15 1.876 7.115 5.823 5.385 9.016 6.723 5.575285 3.15 1.876 7.115 5.823 5.385 9.016 6.723 5.575309 3.803 2.375 8.754 6.768 6.476 11.216 8.488 7.181309 3.803 2.375 8.754 6.768 6.476 11.216 8.488 7.181333.5 4.669 2.691 10346 8.068 7.868 13.057 9.916 8.818333.5 4.669 2.691 10.346 8.068 7.868 13.057 9.916 8.818356.5 5.451 4.667 11.844 10 9.563 15.15 11.717 10.422356.5 5.451 4.667 11.844 10 9.563 15.15 11.717 10.422381 6.407 6.054 13.33 11.511 11.369 17.497 13.055 12.121381 6.407 6.054 13.33 11.511 11369 17.497 13.055 12.121404.5 7.58 6.951 14.889 13.152 13.461 19.508 14.43 13.633404.5 7.58 6.951 14.889 13.152 13.461 19.508 14.43 13.633428 8.796 8.016 16.337 14.767 15.581 21.403 16.215 15.124428 8.796 8.016 16.337 14.767 15.581 21.403 16.215 15.124452 9.984 9.275 18.002 16.484 17.701 23.045 17.796 16.798452 9.984 9.275 18.002 16.484 17.701 23.045 17.796 16.798476 11.596 10.439 19.578 18.818 19.816 24.539 19546 18.818476 11.596 10.439 19.578 18.818 19.816 24.539 19.546 18.818500 13.233 12.087 21.581 20.197 21.937 26.728 21.596 20.561500 13.233 12.087 21.581 20.197 21.937 26.728 21.596 20.561522.5 14.778 13.648 23.617 21.922 24.149 28.993 23.728 22.6335225 14.778 13.648 23.617 21.922 24.149 28.993 23.728 22.633550.5 16.379 15.039 25.534 23.199 26.037 30.402 25.715 23.939550.5 16.379 15.039 25.534 23.199 26.037 30.402 25.715 23.939574.5 18.471 16.65 27.631 24531 27.98 32.128 27.984 25309574.5 18.471 16.65 27.631 24531 27.98 32.128 27.984- 25.309597 20.151 18.343 29.256 26.049 29.602 33.789 29.669 26.809206Data for Figures 65-67 & 74-75I^S40^Cytodex^ 1^CGTime 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. TF0 0.00 0.00 8.21E-03 0.00 0.00 0.00 0.00 0.00 8.21E-03 0.00 0.00 0.00 0.00 0.00 8.21E-03 0.00 0.00 0.0049 1.83 1.23 8.21E-03 1.53 17.41 14.51 1.69 1.22 8.21E-03 1.53 17.30 14.51 1.70 1.23 8.21E-03 1.53 17.80 14.5193 3.80 2.57 1.05E-02 3.17 37.81 31.51 3.83 2.58 1.05E-02 3.17 4730 31.51 3.56 2.50 1.05E-02 3.17 42.80 3151117 5.25 3.65 1.30E-02 4.38 51.40 42.83 5.57 3.79 1.30E-02 4.38 67.00 42.83 4.92 3.55 1.30E-02 4.38 58.30 42.83142 7.00 5.10 1.62E-02 5.83 66.10 55.08 7.42 5.34 1.62E-02 5.83 86.10 55.08 636 4.92 1.62E-02 5.83 73.60 55.08165.5 9.15 7.05 1.92E-02 7.63 82.20 68.50 9.70 7.42 1.92E-02 7.63 105.60 68.50 8.20 6.52 1.92E-02 7.63 88.80 68.50189 11.50 9.23 2.14E-02 9.58 99.51 82.93 12.03 9.68 2.14E-02 9.58 125.00 82.93 10.29 8.41 2.14E-02 9.58 105.60 82.93214 14.35 11.49 1.94E-02 11.96 119.31 99.43 14.85 11.96 1.94E-02 11.96 142.40 99.43 12.73 10.40 1.94E-02 11.96 125.30 99.43240 16.42 13.18 1.76E-02 13.68 136.21 113.51 17.00 13.86 1.76E-02 13.68 160.60 113.51 14.80 12.13 1.76E-02 13.68 140.70 113.51261.5 18.49 14.75 1.91E-02 15.41 151.91 126.59 18.80 15.21 1.91E-02 15.41 172.80 126.59 16.74 13.60 1.91E-02 15.41 156.00 126.59285 20.72 16.51 1.90E-02 17.27 165.61 138.01 21.07 17.05 1.90E-02 17.27 187.90 138.01 18.93 15.36 1.90E-02 17.27 170.40 138.01309 22.99 18.32 1.79E-02 19.16 179.71 149.76 23.44 18.89 1.79E-02 19.16 201.90 149.76 21.28 17.20 1.79E-02 19.16 183.10 149.76334 25.10 20.18 1.78E-02 20.92 19751 164.59 25.67 20.75 1.78E-02 20.92 216.60 16459 23.46 19.06 1.78E-02 20.92 195.10 16459358 27.35 22.01 1.86E-02 22.79 207.31 172.76 27.98 22.58 1.86E-02 22.79 228.10 172.76 25.69 20.81 1.86E-02 22.79 208.23 172.76382 2956 23.75 1.78E-02 24.63 220.81 184.01 30.06 24.32 1.78E-02 24.63 242.30 184.01 27.74 22.48 1.78E-02 24.63 220.43 184.01407.5 31.74 25.39 1.79E-02 26.45 239.61 199.68 32.20 26.01 1.79E-02 26.45 257.40 199.68 29.86 24.16 1.79E-02 26.45 232.83 199.68429 33.76 26.89 1.89E-02 28.13 246.84 205.70 34.17 27.53 1.89E-02 28.13 269.60 205.70 31.72 25.60 1.89E-02 28.13 247.53 205.70453 36.04 28.43 1.96E-02 30.03 267.04 222.53 36.11 28.95 1.96E-02 30.03 286.90 222.53 33.70 27.04 1.96E-02 30.03 266.43 222.53477 38.47 30.31 2.11E-02 32.06 290.24 241.87 38.45 30.74 2.11E-02 32.06 310.70 241.87 35.94 28.80 2.11E-02 32.06 288.83 241.87501 41.10 32.24 2.25E-02 34.25 310.74 258.95 40.84 32.59 2.25E-02 34.25 335.60 258.95 38.36 30.71 2.25E-02 34.25 311.83 258.95525 43.87 34.13 2.19E-02 36.56 334.54 278.78 43.22 34.32 2.19E-02 36.56 356.10 278.78 40.94 32.52 2.19E-02 36.56 331.53 278.78549.75 46.43 36.49 2.06E-02 38.69 354.04 295.03 45.64 36.42 2.06E-02 38.69 377.30 295.03 43.35 34.50 2.06E-02 38.69 350.43 295.03575 49.02 39.08 2.04E-02 40.85 371.74 309.78 48.25 39.03 2.04E-02 40.85 397.00 309.78 45.89 37.05 2.04E-02 40.85 366.93 309.78598 51.36 41.45 2.10E-02 42.80 391.24 326.03 50.80 41.71 2.10E-02 42.80 417.70 326.03 48.35 39.58 2.10E-02 42.80 387.93 326.03621 53.85 44.08 2.10E-02 44.88 413.14 344.28 53.31 44.27 2.10E-02 44.88 439.90 34428 51.00 42.22 2.10E-02 44.88 406.93 344.28207Data for Figure 71Time (h) Cell Concentration (x 10 -7)123.5 0.51144 1.95168 2.31241 2.34336 2.41384 3.21480 3.02576.5 2.83672 1.91744 2.37840 2.68Data for Figure 71Time (h) Glucose ConcentrationWOCell leakage (x 10-4)0 3.71 4.324 3.58 1248.5 2.84 3574 1.54 8698 0.82 7798.5 4.11 77120 3.73 6144 3.17 22168 1.91 62192 1.37 10.1192.5 4.21 10.1216 3.22 17.42423 1.68 70266.5 0.69 50Data for Figure 72Time (h) Total^glucoseused (g/L)Antibodyproduction rateOng,a1100 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 --208209Data for Figure 72 continued336^8.82384 9.07389^9.07408 11.09408.1^11.09435 13.66^0.63197435.1^13.66458.5 16.24^0.68376458.6^16.24480 18.3504.5^19.4^0.54466504.6 19.4528.5^19.92552.5 21.08576.5^22.42^0.38943576.6 22.42603.5^23.83627.5 25.28^0.64833627.6^25.28648.5 27672.5^28.49^0.89087672.6 28.49696.5^30.53720.5 31.73^0.73069720.6^31.73744 33.56771.5^34.73^0.82515771.6 34.73794^36.8Data for Figure 73Time (h)^Glucose^LactateconcentratioiEgaCszjTransferrin^Cell concentration/......_ ji^concentratiointratioin m L^x10-5/ mL0 2.98 0.21 0.00^11.517 2.76 0.37 0.74 13.324 2.54 0.43 1.03 13.641 2.49 0.54 1.19 15.648 2.41 0.81 1.21 21.165 2.06 1.09 2.55 24.172 1.9 1.12 5.30 33.190 1.43 1.59 8.80 40.6102 1.13 1.81 10.90 50116 0.76 2.21 13.60 65124 0.53 2.31 1530 71137 0.3 2.59 18.20 103144 0.22 2.63 19.2 102161 0.19 2.81 19.5 101168 0.16 2.82 20.3 96185 0.15 2.82 20.2 49.6193 0.14 2.82 20 72.8209 0.14 2.83 20 24216 0.13 2.83 20 25Data for Figure 76Time (10 S40 Suspension roller.0 0 0 024 0.775 0.78125 0.5312524 0.775 0.78125 0.5312548 1.985 1.83125 1.2512572 2.865 2.74125 1.6712572 2.865 2.74125 1.6712595 4.045 4.24125 2.40125117.5 5.235 5.43125 3.01125117.5 5.235 5.43125 3.01125142 7.195 7.76125 3.01125142 7.195 7.76125 3.01125166.5 8.325 10.25125 3.63125166.5 8.325 10.25125 3.63125190.5 9.885 12.40125 4.29125190.5 9.885 12.40125 4.29125216 10.655 14.75125 5.04125216 10.655 14.75125 5.04125240 12.335 16.86125 6.19125240 12.335 16.86125 6.19125262 12.905 18.94125 6.99125264 12.905 18.94125 6.99125285.5 14.615 21.51125 8.13125285.5 14.615 21.51125 8.13125310 15.245 23.90125 8.88125310 15.245 23.90125 8.88125Data for Figure 77Time (h) S40 Suspension roller0 0 0 024 9.9 5.4 6.872 22.55 16.5 24.2117.5 41.45 47.8142 68.05166.5 61 37.95 90.1190.5 116.1216 77.3 51.6 151.05240 191.45262 89.6 71.6 230.65285.5 276.15310 103.05 95.25 325.4334 378.4357.5 117.8 119.95 436.5210Data for Figure 78Time S40 Suspension roller0 0.848 0.836 0.94024 0.848 0.821 0.94072 0.848 0.767 0.940117.5 0.848 0.691 0.940142 0.848 0.629 0.940166.5 0.848 0.577 0.940190.5 0.848 0.555 0.940216 0.848 0.555 0.940240 0.848 0.555 0.940262 0.848 0.555 0.940285.5 0.848 0.555 0.940310 0.848 0.555 0.940Data for Figure 79Time (It) S40 Suspension roller0 10.29 10.13 8.714B/24 6.57 9.86 8.8372 5.91 10.55 8.82117.5 8.74142 8.78166.5 7.89 10.45 10.32190.5 13.57216 1025 10.03 16.86240 18.99262 10.93 11.36 18.20285.5 19.09310 11.21 11.57 21.17334 22.133575 10.43 10.82 2251211


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



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"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items