CELL DAMAGE MECHANISMS AND STRESS RESPONSE IN ANIMAL CELL CULTURE
DISSERTATION
Presented in Partial Fulfillment of the Requirements for
the Degree Doctor of Philosophy in the Graduate
School of the Ohio State University
By
Claudia Berdugo, Ch. Eng. M. Sc.
Graduate Program in Chemical and Biomolecular Engineering
*****
The Ohio State University 2010
Dissertation Committee:
Dr. Jeffrey J. Chalmers, Adviser
Dr. Jessica Winter
Dr. Andre Palmer
ABSTRACT
Animal cell culture is a widely used technology for producing recombinant proteins. The
ability to make post translational modifications and secrete the active forms of the protein
into the culture medium represents major advantages over other processes. The growing
market demand for pharmaceuticals has created a need for increased production capacity;
however, achieving productivity gains in both the upstream stage and downstream processes can subject cells to aggressive environments such as those involving
hydrodynamic stresses. Although numerous studies have explored the consequences of
cell damage due to hydrodynamic stress, there has been a lack of understanding of the
mechanism of such damage at a cellular level. Cell damage can also influence biomedical
applications. Cells manipulated in instruments such as diagnosis and analysis devices can
experience hydrodynamic forces.
The level of cell damage is influenced by the hydrodynamic conditions in the bioprocess
or biomedical equipment as well as the cell line sensitivity. To evaluate and compare cell
sensitivity among different cell lines, a flow contraction device, previously designed by
our group was used. Cells were exposed to well defined and controlled hydrodynamic
forces and cell damage was estimated as a function of energy dissipation rate (EDR).
EDR is a scalar value that represents the rate of dissipation of kinetic energy per unit of
ii mass or volume. Using this methodology we found human cell lines highly sensitive to
hydrodynamic forces.
Hydrodynamic evaluations were performed in ten different bioreactor configurations
Impeller Sparger. The best configurations were chosen based on kLa response surface
model for testing in cell culture experiments. The configurations chosen were used to
evaluate the expression of stress proteins under moderate hydrodynamic stress in bioreactors as well as cell cycle profile and its relationship to recombinant protein production. The results suggest that for a clonal cell line evaluated G1 phase of the cell
cycle may be more conducive to producing the recombinant protein. In addition, a
relationship between hydrodynamic stress and expression of stress proteins was observed.
The type of stress protein and the level of expression seem to be dependent on cell type
and differences could even be observed between clones of the same cell line.
Cell damage was also evaluated in a fluorescent activated cell sorter (FACS) models
Vantage and Aria. Cells can be exposed to very high hydrodynamic forces when flowing
through channels and nozzle in the sorting process. Results indicate that not only are cells
damaged in a flow cytometer, but that this damage can vary from cell line to cell line as
well as from specific conditions/type of flow cytometer and flow conditions. In addition,
studies were conducted to evaluate cell growth behavior after stress as well as the effect
of sorting on cell cycle. Extended growth lag phase was observed in cells exposed to
hydrodynamic stress, and the sensitivity of any specific cell line can be a function of the
growth phase of the cell.
iii
Dedicated to my sister
iv
ACKNOWLEDGMENTS
I want to thank God for giving me the courage to accomplish this goal, and reminding me how vast the world is and how little we know.
I would like to thank my advisor, Professor Jeffrey J. Chalmers, for giving me the opportunity to join his research group, for his financial support and his insights in my research. I am also grateful for his support in getting me opportunities to go to the industry and getting involved in training and internship programs.
Special thanks to my committee members Dr. Jessica Winter and Dr. Andre Palmer for their comments, questions and directions during my qualified, candidacy, and defense exams. I want to thank also Dr. Andrea Doseff for her guidance during my candidacy exam and my research.
I would like to show my gratitude to GlaxoSmithKline. It was an honor for me to work at their research facilities at King of Prussia, PA. I learnt a great deal from that experience and I gained perspective into the special requirements of process development in the
v pharmaceutical industry, which helped me later in my research. In particular I would like
to gratefully acknowledge the enthusiastic supervision and friendship of
Dr. Oscar Lara Velasco, his inspiration and efforts to explain things clearly and simply
helped me to take advantage of that great opportunity. I wish to thank Dr. Ilse
Blumentals, a person with great charisma and wisdom, who believe in people and knows
how to bring the best of them. I also thank Dr. Prem Patel for his kindness and support in my internship. Thanks to everyone in the group of Process development.
I am grateful to all my colleagues in Dr. Chalmers group, lab mates in Dr. Yang’s group
and office mates from Dr. Bakshi’s group. I appreciate the company, talks, and
stimulating environment to learn and grow. I also appreciate the support of Paul Green,
Leigh Evrard, David Cade, Angela Benett, Susan Tesfai, Lynn Flanagan and Bill Cory, in
the department of Chemical Engineering, and the technical support in the FACS project
from Bryan, Nicole, Priya, Katrina, Serra, Arup and Adeline.
I am forever in debt to Ruben. I admire your gifted mind and I am thankful to have you as
my mentor. Your teaching and support helped me to defeat my fears when I started this journey, I wouldn’t have made it without your help. Thanks from the bottom of my heart.
It is a pleasure to thank so many friends that helped me in many ways to make it through
this work. Please forgive me if I forget to mention someone but, many precious moments
might escape at 2 am while I am writing this section. Nacho, thanks for being there every
time I needed a friend and thanks for going along with me on crazy ideas, even running
vi 10K under the rain. Oliver, a truly giving person with a big heart, thanks for taking care
of me. Sharing your roof has been like having a brother (and you are a good cook). Elba,
thanks for listening and for having always a word of support. Laura thanks for sharing
with me all the goodness of your heart. My runner mates: Nicole, Elsa and Jake, fun and
healthy moments we shared. Priya: thanks for the treasure of your friendship. Thanks to
all my dear friends for the emotional support, camaraderie, entertainment and caring they provided: Bryan, Katrina, Yadira, Daniel, Alejandra, Courtney, Leo, Adeline, Glenn,
Toño, Pamela, Miryam.
My friends and surrogate family in USA: Charlene, you are a blessing in my life, you made me feel I have family far from my land in a moment where I needed the most.
Sabrina and Shane, thanks for welcoming me in your house and giving me a home. So much love I felt in that family, I am really lucky for I am still there.
With immense joy I want to thank Jeremy. The peace and happiness that I found with you helped me to accomplish this goal. It is easier to bear the challenges because I know you are always at the finish line. Thanks for your constant help, support, and patience. You have drawn a smile in my heart and in my life.
I wish to thank my entire family for their love and support:
Inesita: Gracias por contagiarnos con tu entusiasmo por la vida. Claudia Imelda:
Gracias por estar pendiente de mi, por tus palabras y espiritualidad. Angela Maria: Ha sido hermoso conocerte y tenerte cerca, tu y yo sabemos cuánto REALMENTE cuesta
vii este reto y me siento orgullosa de como lo estás afrontando. Adri siempre muy cerca de
mi corazón por lo que compartimos creciendo juntas. Todos mis tíos y tías en Colombia,
mis primitos y toda mi familia. Sé que han estado pendientes de mi y les agradezco con gran amor desde mi corazón
Mami: Tu constante deseo de superación ha sido mi ejemplo en el logro de esta meta. Tu inmensa capacidad de amor es el símbolo de bondad que jamas he visto en nadie. Papi:
Gracias por todas las enseñanzas de amor a tus hijos y familia, por tu dedicación y esfuerzo. Ricardo: Hermanito este triunfo comenzo contigo, gracias por tu empeño en mostrarnos hasta donde podemos llegar. Mi admiración a mi brillante hermanito. Javi:
Gracias por tu gran corazón, te admiro muchísimo desde lejos por todo lo que has logrado. Tu carisma, dedicación y éxito son ejemplo para todos nosotros. Sandra:
Hermanita eres un hermoso regalo que la vida me dio. Agradezco inmensamente todo tu
apoyo. Sin ti no hubiera sido capaz de superar tantos tropiezos grandes y pequeños
durante estos años de lucha. Tus palabras siempre me iluminaron. Admiro tu tenacidad,
tu integridad y amor a tu familia. Dios te bendiga siempre. Mis cuñis: Felipe y Lina.
Gracias por el amor y cuidado a nuestra familia y sobrinitos, ustedes son dos hermanos
mas. Mis sobrinitos y sobrinita: Dany, Migue, Martín, Tomás y Manuela: Ustedes son la
alegría de mi corazón. A mi Hermosa familia: Los llevo en mi corazón y cada pequeño
logro en mi vida es reflejo de lo que sembramos con esfuerzo en nuestra familia.
A la memoria de Rociito, abuelita y Juan Camilo, vivos en el corazón de quienes los
amamos tanto.
viii
VITA
March 30, 1969….…….………… Born Duitama, Boyacá, Colombia 1995...... Chemical Engineering. Universidad Nacional de Colombia. Bogotá, Colombia. 2000……………………………... Master of Science in Biotechnology. Universidad Nacional Autónoma de México. México. 2000 2004……………………… Research Engineer. ECOPETROL. Colombia
2005 2006……………………… Departamental Fellowship. The Ohio State University 2006 2007……………………… Graduate Research Associate, The Ohio State University 2007 2008……………………… COOP GlaxoSmithKline, King of Prussia, PA
2008 Present…………………….. Graduate Research Associate, The Ohio State University.
PUBLICATIONS
Beltran, L.; Moreno, N.; Berdugo, C.; Zamora, A.; and Buitrago, G. (1998). “Estrategia para el diseno de un medio de cultivo para la fermentacion de B. thuringiensis”. Revista Colombiana de Biotecnologia. 1(1): 28 34
Berdugo, C.; Mena, J.; Acero, J.; and Mogollon, L. (2001). “Increasing the production of Desulfurizing Biocatalyst by means of Fed Batch Culture”. CT&F Ciencia, Tecnologia y Futuro. 1(2): 5 9
Berdugo, C.; Caballero, R. and Godoy, R. D. (2002) “Aqueous organic phase separation by membrane reactors in biodesulfurization reactions”. CT&F Ciencia, Tecnologia y Futuro. 2(3): 97 112
Acero, J.; Berdugo, C.; and Mogollon, L. (2003). “Biodesulfurization process evaluation with a Gordona rubropertinctus strain”. CT&F Ciencia, Tecnologia y Futuro. 2(4): 43 54 ix
Mollet, M.; Godoy Silva, R.; Berdugo, C. and Chalmers, J. J. (2007). Acute Hydrodynamic Forces and Apoptosis: A Complex Question. Biotechnology and Bioengineering. 98 (4): 772 788.
Mollet, M; Godoy Silva, R; Berdugo, C; and Chalmers, J. J. “Computer Simulations of the Energy Dissipation Rate in the Fluorescence Activated Cell Sorter”. Biotechnology and Bioengineering (2008). 100(2): 260 272
x
FIELDS OF STUDY
Major Field: Chemical Engineering
Minor Field: Biotechnology, Biochemical Engineering, Cell Culture.
xi
TABLE OF CONTENTS
Page Abstract…………………………………………………………………………. ii Acknowledgments…………………………………………………………….... v Vita……………………………………………………………………………... ix List of tables……………………………………………………………………. xv List of figures………………………………………………………………….... xvii
Chapters: 1. Introduction……………………………………………………………….. 1 1.1 Animal Cell culture……………………..……………………………..... 1 1.2 Objectives………………………………………………………………. 3 1.2.1 Goal…………………………………………………………………. 3 1.2.2 Methodology………………………………………………………… 3 1.3 Scopes of this study…………………………………………………….. 4 1.3.1 Cell damage in a fluorescent activated cell sorter (FACS) (chapter 3)…………………………………………………………...... 4 1.3.2 Cell damage in a fluorescence activated cell sorter (chapter 4)……... 6 1.3.3 Effect of hydrodynamic conditions on cell cycle, stress proteins and recombinant protein productivity (chapter 5)……………………..… 7 1.4 References……………………………………………………………… 9 2. Literature review: mixing, aeration and hydrodynamics in bioreactors….. 11 2.1 Abstract………………………………………………………………… 11 2.2 Introduction……………………………………………………………... 13 2.3 Aeration………………………………………………………………… 16 2.3.1 surface aeration……………………………………………………… 22 2.3.2 perfluorocarbons…………………………………………………….. 24 2.3.3 oxygen carriers………………………………………………………. 26 2.3.4 Membrane aeration………………………………………………….. 26 2.3.5 Sparging……………………………………………………………... 27 2.3.6 Sparger design………………………………………………………. 28 2.3.7 CO2 accumulation and removal……………………………………... 30 2.4 Mixing………………………………………………………………….. 34 2.4.1 Stirred tank reactors…………………………………………………. 36 2.4.2 Geometry……………………………………………………………. 37 2.4.3 Impeller……………………………………………………………… 37 2.4.4 Baffles……………………………………………………………….. 40 xii 2.4.5 Mixing times and energy dissipation rates………………………….. 40 2.5 The relationship of animal cells to hydrodynamic forces……………… 46 2.5.1 Molecular mechanism involved in the response of animal cells to mechanical forces……………….…………………………………… 48 2.5.2 Cell damage concept………………………………………………… 52 2.5.3 Hydrodynamical cell damage……………………………………….. 54 2.5.4 Quantification of cell damage……………………………………….. 57 2.5.5 Detrimental effects of sparging……………………………………... 63 2.6 Summary……………………………………………………………….. 69 2.7 References……………………………………………………………… 71 3. Cell damage in a fluorescent activated cell sorter……………..…………. 118 3.1 Abstract………………………………………………………………… 119 3.2 Introduction…………………………………………………………….. 120 3.3 Materials and methods…………………………………………………. 123 3.3.1 Cell culture……………..…………………………………………… 123 3.3.2 Single shear stress studies……………………….………………….. 124 3.3.3 Cell sorting…….………………………………………………….… 124 3.3.4 Cell damage analysis……………………………...………………… 125 3.3.5 Cell cycle analysis….…………………………………………….…. 126 3.3.6 Cell arrest in G2 phase…………………………………………..…. 127 3.3.7 Computational Fluid Dynamics (CFD) simulations……..…………. 127 3.4 Results………………………………………………………………….. 130 3.4.1 Single shear stress studies……….…………………………………... 130 3.4.2 Growth after hydrodynamic stress exposure………………………... 132 3.4.3 FACS sorting studies at different cell densities…………………….. 135 3.4.4 FBS protective effect in FACS sorting……………………………… 135 3.4.5 Effect of sorting on cell cycle……………………………………….. 137 3.4.6 FACS Flow Rate Measurements……………………………………. 139 3.4.7 CFD Simulations……………………………………………………. 140 3.5 Discussion………………………………………………………………. 141 3.6 References……………………………………………………………… 145 4. Effect of impeller sparger configurations on mass transfer capabilities and cell culture performance…………………………………. 169 4.1. Abstract…………………………………………………………………. 169 4.2. Introduction ……………………………………………………………. 170 4.3. Materials and methods………………………………………………….. 173 4.3.1 Methodology to evaluate kLa……………………………………….. 173 4.3.2 Experimental design ………………………………………………... 174 4.3.3 Volumetric oxygen mass transfer coefficient calculation kLa……… 175 4.3.4 Power number and Energy Dissipation rate calculation…………….. 177 4.3.5 Cell culture experiments……………………………………………. 178 4.4. Results and discussion…………………………………………………. 179 4.4.1 Estimation of kLa at 2L scale.……………………………………… 179 4.4.2 Data analysis for different configurations……….…………………… 180 4.4.3 Surface Response Analysis…………………………………………… 183
xiii 4.4.3.1 Effect of sparger position on mass transfer coefficient………………. 183 4.4.3.2 Effect of antifoam on mass transfer coefficient……………………… 184 4.4.3.3 Analysis of reactor configuration that include perforated tube………. 184 4.4.3.4 Analysis of reactor configuration that include sintered sparger……… 184 4.4.3.5 Analysis of reactor configuration that include PBT down pumping and rushton impeller in the bottom………………………………………………. 185 4.4.4 EDR calculation in the configurations evaluated…………………….. 185 4.4.5 Cell Culture…………………………………………………………… 186 4.5. Conclusions………………………………………………………………. 187 4.6. References……………………………………………………………… 189 5. Effect of hydrodynamic conditions on cell cycle stress proteins and recombinant protein productivity………………………………………. 209 5.1 Abstract…………………………………………………………………. 209 5.2 Introduction…………………………………………………………….. 210 5.2.1 Cell Damage Mechanisms………………………………………………. 210 5.2.2 Mammalian Stress Response……………………………………………. 212 5.2.2.1 Structure and function of Stress Proteins………………………………. 213 5.2.2.2 Induction of response………………………..………………………… 215 5.2.2.3 Stress proteins and hydrodynamic stress…………………………….. 216 5.3 Materials and methods………………………………………………….. 218 5.3.1 Cell line……………………………………………………………… 218 5.3.2 Static Cell Cultures……………………………………………….…. 219 5.3.3 Bioreactors……………………………………………………………. 220 5.3.4 Continuous stress……………………………………………………... 221 5.3.5 Stress Proteins analysis……………………………………………….. 222 5.3.5.1 Cell fixation………………………………………………………….. 222 5.3.5.2 Staining and analysis………………………………………………… 222 5.3.6 Analytical methods………………………………………………….. 223 5.3.7 Cell cycle analysis…………………………………………………… 224 5.3.8 Parameter calculation……….………………………………………. 224 5.3.8.1 Growth rate…………………………………………………………. 224 5.3.8.2 Integral of viable cell concentration (IVC)…………………………. 225 5.3.8.3 Specific Productivity……………………………………………….. 225 5.4 Results and discussion…………………………………………………... 226 5.4.1 Effect of configuration on cell cycle profile…………………….…... 226 5.4.2 Effect of configuration on expression of stress proteins..…………… 229 5.4.3 Expression of stress proteins in Reactors Versus TFlask……………. 232 5.4.4 Expression of Stress proteins in a highly sensitive cell line………… 235 5.5 Conclusions……………………………………………………………… 237 5.6 References……………………………………………………………… 241 6. Conclusions and recommendations…………………….…………………. 266 Bibliography……………………………………………………………………. 271
xiv LIST OF TABLES
Table Page
2.1 Specific oxygen uptake rates (q ) reported for animal cell lines……… O2 93
2.2 Examples of empirically derived equations for kLa estimation found in literature………………………………………………………………... 94 2.3 Power number for selected impellers for turbulent regime…………….. 95 2.4 Relative Oxygen carrying capacities (k) of water, perfluorochemicals and gas vesicles (Modified from Sundararajan and Ju, 2006)………….. 96 2.5 Benefits and drawbacks of PFCs (modified from Lowe et al., 1998)…. 97 2.6 Correlations derived by Cui et al. (1996) for power drawn in systems with multiple Rushton impellers………………………………………... 98 2.7 Methodologies reported in literature for measurement of mechanical properties of animal cells and/or determination effect of hydrodynamical forces on animal cells………………………………… 99 2.8 Some parameters reported in literature to correlate the effect of hydrodynamical forces on cells………………………………………… 100 3.1 Median and range of energy dissipation rate that cells are exposed to in the Torture Chamber at different volumetric flow rates……...………… 149 3.2 Grid size and type of injection in the calculation of EDR……………… 150 3.3 Highest level of EDR that a particle would experience in a BD FACSVantage for different operating conditions in a 70 and 100 m nozzle ……………………………………………...... 151 3.4 Effect of sorting on cell cycle profile and fraction of cells in G2 phase.. 152 3.5 Highest level of EDR that a particle would experience in a BD FACS Aria for different conditions in a 70 m nozzle……………………….. 153 4.1 Experimental design to evaluate effect of sparger position. Side: Sparger is located at side of the impeller near to the wall of the tank. Edge: Sparger is located below the edge of the impeller. Centered: Sparger is located under the impeller in the center. High (h): Distance of the sparger with respect to the lower impeller, negative ( ) indicates negative value in the axes with respect to the impeller located in the center (0, 0, 0). Clearance (C): distance between impellers……………. 192 4.2 Configurations impeller sparger evaluated ……………………………. 193 xv 4.3 O2 Experimental table to test K L a capabilities of the 2L bioreactors for configurations that include perforated tube. (00) indicates center points, ( + or similar pattern ) indicate corner points…………………………... 193
4.4 O2 Experimental table to test K L a capabilities of the 2L bioreactors for configurations that include sintered sparger. (00) indicates center points, ( +, or similar pattern) indicate corner points…………………. 194
4.5 O2 Experimental table to test K L a capabilities of the 2L bioreactors for configurations that include sintered sparger of 50 m size pore. (00) indicates center points, ( +, or similar pattern) indicate corner points…. 194
4.6 O2 Data treatment to calculate K L a , Configuration C2 ………………… 195 4.7 Summary of Fit…………………………………………………………. 196 4.8 Parameter Estimates……………………………………………………. 196
4.9 O2 Prediction correlations to calculate K L a for the configurations evaluated……………………………………………………………….. 197
4.10 O2 K L a range reached with configurations including perforated tube and a combination of impellers……………………………………………… 198
4.11 O2 K L a range reached with configurations including sintered sparger and a combination of impellers……………………………………………… 198
4.12 O2 K L a range reached with configurations including sintered sparger and a combination of impellers……………………………………………… 198 4.13 Input data and power number and constants calculations using
O2 measured K L a for configuration C2 (Dual impeller and perforated tube)…………………………………………………………………….. 199 4.14 Input data and power number and constants calculations using
O2 measured K L a for configuration C5 (PBT impeller and sintered sparger)…………………………………………………………………. 199 5.1 Stress agents reported that induce stress proteins expression ……...… 245 5.2 State of art on relationship between protein expression and cell cycle phases …………………………………………………………………... 246
xvi LIST OF FIGURES
Figure Page 1.1 Aspects considered in the methodology to evaluate the effect of hydrodynamic forces on mammalian cell cultures…….………….… 10 2.1 Geometrical configuration, flow patterns and total non gassed power drawn into the liquid as a function of impeller spacing for a mixing system with multiple Rushton turbines (Adapted from Hudcova et al., 1989)……………………………………………………….……... 101 2.2 Impeller configurations commonly employed in animal cell culture.... 102 2.3 Regions of highest shear rate and highest energy dissipation rate behind the blades of a Rushton turbine……………………….……… 103 2.4 Effect of the gas flow rate on the power drop under gassed conditions for two impeller geometries in Newtonian, water like fluids. (Adapted from Galindo and Nienow, 1993)……………………….…. 104 2.5 Effect of the number and size of the baffles on the power drawn by an impeller in a cylindrical stirred tank reactor (Adapted from Oldshue, 1983)…………………………………………………….…. 105 2.6 Change in the tank diameter and impeller diameter as the volume of the vessel increases from 0.5 to 10,000 liters keeping constant the geometrical ratios H/T = 1 and T/D = 3………………………….…... 106 2.7 Lines of constant, maximum EDR in a vessel as a function of impeller rotational speed and diameter for Rushton turbine in water. H/T = 1 and T/D = 3……………………………………………….…. 107 2.8 Average EDR for the whole vessel as a function of impeller diameter and RPM using a Rushton turbine in water. H/T = 1 and T/D = 3………………………………………………………….…………… 108 2.9 Calculated maximum and average energy dissipation rate as a function of RPM for an Applikon bioreactor containing four baffles. The single points correspond to experimental measurements without baffles (Adapted from Mollet et al., 2004)……………………….….. 109 2.10 Molecular signaling and response cascade in endothelial and smooth muscle cells (A) before and (B) after stimulation by hydrodynamic forces. …………………………………………………………….….. 110
xvii 2.11 Effect of agitation rate on cell concentration, viability and aggregate diameter of murine NSC in batch suspension in a 125 mL spinner flask. Data points are average of duplicate runs. (a) Cell 1 1 concentration: ( ) 60 rev·min ; ( ) 100 rev·min . Viability: ( ) 60 rev·min 1; ( ) 100 rev·min 1.(b) Average aggregate diameter:( 1 1 ) 60 rev·min ; ( ) 100 rev·min . Standard deviation:( ) 60 rev·min 1; ( ) 100 rev·min 1. (Adapted from Sen et al. 2001)….…. 112 2.12 Effect of stirring speed on cell concentration after 7 days of culture of Vero cells on Cytodex microcarriers on 250 mL spinner vessels (Data from Hirtenstein and Clark, 1980)………………………….…. 113 2.13 Experimental curves for the percentage of damage experienced by cells in a custom design microfluidic device for single abuse experiments. Adapted from Ma et al. (2002), Mollet et al. (In Press) and Mollet et al., (submitted)…………………………………….…... 114 2.14 Summary of the reported energy dissipation rate at which cells are affected as well as the reported levels of energy dissipation rate in various bioprocess environments. Adapted from Ma et al. (2002) and Mollet et al. (2004).……………………………….………………….. 115 2.15 A three dimensional plot of the number of cells associated with each bubble as a function of cell concentration (cell·mL 1) and Pluronic F 68 concentration. The dots indicated experimental data and the surface is the plot of a multiple variable regression. (Adapted from Ma et al., 2004)…………………………………………………….… 117 3.1 Photograph (a), top view (b), and a perspective view (c) of the flow contraction device …………………………………………..……….. 154 3.2 Single Pass set up: Cells are centrifuged and resuspended into fresh media. They pass once through the Torture chamber at different flow rates ………………………………………………….…………….…. 155 3.3 Simplified sketch flow cytometer ……………...………………….…. 156 3.4 Exit sheath flow rate as a function of flow rate scale and sheath pressure for the 70 m nozzle (a), and sample flow rate as a function of flow rate scale and sheath pressure for a 70 m nozzle (b)……….. 157 3.5 Cell damage estimated from single pass experiments in the shear stress device ……………………………………..……………….….. 158 3.6 Growth kinetic of THP1 cells after stress exposure in shear stress device (TC), cells were exposed at 90 mL/min ……...…………….… 158 3.7 Growth kinetic of THP1 cells after sorting in FACS Aria, cells were grown in media with 10% FBS ………..…………………………..… 159
xviii 3.8 Growth kinetic of THP1 cells after sorting in FACS Aria, cells were grown in media with 30% FBS ……………………………………… 159 3.9 Growth kinetic of THP1 cells after sorting in FACS Aria, cells were grown in media with 10% FBS, 30%FBS and conditioned media…. 160 3.10 Cells suspensions at different cell concentrations were sorted under the same conditions. Cell damage was calculated based on the amount of LDH in the supernatant after sorting ………………….… 160 3.11 THP1 cells sorted in media with 0% FBS. Cells were counted with hematocytometer before and after sorting. Aria reports number of events …………………..…………………………….……………… 161 3.12 THP1 cells sorted in media with 10% FBS. Cells were counted with hematocytometer before and after sorting. Aria reports number of events ……………………………………………………………….. 161 3.13 Statistical analysis of THP1 cells sorted in media with 10% FBS versus cells sorted in media with 0%FBS …..……………………… 162 3.14 Cell cycle profile before and after stress exposure in flow cytometer.. 163 3.15 Cell cycle profile of cells arrested in G2 phase before and after stress exposure in flow cytometer…………………………………………... 164 3.16 Statistical analysis of fraction of THP1 cells in G2 phase before and after sorting………………………………………………………….. 165 3.17 Photograph of BD FACS Aria Flow cell components………………. 166 3.18 View of the nozzle geometry and mesh used for the simulation. Geometry and mesh were built in Gambit……………………………. 167 3.19 Fluent output of the simulations of particles flowing through the nozzle The color coded figures correspond to the levels of EDR…… 168 4.1 Type of impellers evaluated: Rushton (a), Pitch Balde Turbine (b)………………………………………………………...…………… 200 4.2 Geometrical configuration to evaluate effect of impeller and sparger
O2 location on K L a . h: distance between lower impeller and sparger, C: distance between impellers, sparger location: side, center, edge…. 200
4.3 O2 Sample calculation of K L a from collected data. DO profile……….. 201
4.4 O2 Sample calculation of K L a from collected data. Obtaining slope….. 201
4.5 O2 Actual vs. Predicted K L a after fitting a response surface to data
O2 collected during the K L a assay for configuration C2………………. 202
xix 4.6 O2 Effect of location of sparger on K L a Sparger was located in three different positions: At side of the impeller, near to the wall of the tank, edge of the impeller and centered under the impeller……..……. 202
4.7 O2 Surface response results on effect of antifoam on K L a ……………. 203
4.8 O2 Effect of configuration impeller/sparger on K L a . Perforated tube and combination of impellers………………………………………… 203
4.9 O2 Effect of configuration impeller/sparger on K L a . Sintered sparger and combination of impellers………………………………………… 204
4.10 O2 Effect of configuration impeller/sparger on K L a . Perforated tube and combination of impellers including dual impeller (PBT down – pumping and rushton impeller)……………………………………….. 204 4.11 Summary of the reported energy dissipation rate at which cells are affected as well as the reported levels of energy dissipation rate in various bioprocess environments. Adapted from Ma et al. (2002) and Mollet et al. (2004)…………………………………………………… 205 4.12 Cell growth in bioreactors with best configurations…………………. 207
4.13 CO2 profile in cell culture evaluation………………………………… 207 4.14 Recombinant protein profile in cell culture evaluation……………… 208 5.1 Summary of the reported energy dissipation rate at which cells are affected as well as the reported levels of energy dissipation rate in various bioprocess environments. Adapted and improved from Ma et al. (2002) and Mollet et al. (2004)…………….……………….…..... 247 5.2 Diagram of the experimental setup for continuous, chronic exposure of suspended animal cells to high levels of hydrodynamic forces. Adapted with permission from Godoy silva et al (2009)……………. 249 5.3 Growth kinetic (a) and viability (b) for four cell cultures performed in bioreactor with two different impeller/sparger configurations. One configuration operated with perforated tube sparger and dual impeller (PBT in the top and Rushton bottom), the second configuration operated with sintered sparger (50 m) and PBT impeller. Two strains of the CHO clonal cell line were used strain A and strain B were seeded in each bioreactor as is indicated in the label………….. 250
xx 5.4 Concentration of glucose (a) and lactate (b) for four cell cultures performed in bioreactor with two different impeller/sparger configurations. One configuration operated with perforated tube sparger and dual impeller (PBT in the top and Rushton bottom), the second configuration operated with sintered sparger (50 m) and PBT impeller. Two strains of the CHO clonal cell line were used strain A and strain B were seeded in each bioreactor as is indicated in the label………………………………………………………………. 251 5.5 Cell cycle diagram (A) Typical cell cycle histogram obtained in flow cytometer (B). Content of DNA increases along the x axis. First peak correspond to cells in G1 phase, small area in the center with lines correspond to S phase, peak further right correspond to cells in G2 phase…………………………………………………………………. 252 5.6 Cell cycle profile (A), Relationship between cell cycle and titer: subpopulation in G1 phase (B), subpopulation in G2 phase (C), and subpopulation in S phase (D)………………………………………… 253 5.7 Cell cycle profile comparison between reactors with different hydrodynamic conditions. Two impeller/sparger configurations were evaluated: A. Open tube sparger and dual impeller Rushton bottom – PBT top. B. Sintered sparger and PBT impeller……………………… 254 5.8 Relationship between cell cycle and titer subpopulation in G1 phase for two different reactor configurations. 1. Open tube sparger and dual impeller Rushton bottom – PBT top. 2. Sintered sparger and PBT impeller………………………………………………………….. 255 5.9 Typical plots for stress proteins analysis in flow cytometry. Dot plot (a) indicates gated populations for live cells, dead cells, debri and aggregates. Histogram to estimate mean fluorescence intensity of samples as indicative of level of expression of stress proteins (b)…… 256 5.10 Analysis of relationship between population expressing HSP70 and specific productivity …………………….………………………….. 257 5.11 Analysis of relationship between population expressing HSP90 and specific productivity…………………………………………………. 258 5.12 Profile of stress proteins expression. Two different strains of a clonal CHO cell line where seeded in two bioreactors with the configuration sintered sparger and PBT Impeller. Strain A (a), and Strain B (b)…... 259 5.13 Profile of stress proteins expression. Two different strains of a clonal CHO cell line where seeded in two bioreactors with the configuration open tube sparger and dual impeller (PBT top, Rushton bottom). Strain A (a), and Strain B (b)………………………………………… 260
xxi 5.14 Growth kinetic of cell line CHO 6E6. (a), and viability, (b), for cell culture in agitated bioreactor (2 L working volume) and TFlasks, to compare profile of stress protein expression…………………………. 261 5.15 Concentration of glucose, (a), and lactate, (b), as a function of time of two batchs of cells growing in TFlasks Vs. cells growing in agitated bioreactor (2L working volume). Cell line CHO 6E6……………….. 262 5.16 Comparison on expression of stress proteins between reactor and TFlasks. Cell line CHO 6E6…………………………………………. 263 5.17 THP1 cell suspension was subjected to recycle through the TC at 50 mL·min 1, cells passed 10 times through TC. Samples were taken 1, 6, 8 10 hr in recovery to evaluate HSP70 and HSP90 expression. Purple line corresponds to unstained control, green line corresponds
to control (non stressed cells), red line correspond to sample cells stressed at 50 mL/min……………………………………………….. 264 5.18 THP1 cell suspension was subjected to recycle through the TC at 90 mL·min 1, corresponding to 1.1×108 W·m 3, cells passed 10 times through TC. After stress, cells were incubated at 37 °C in a 5% CO2 atmosphere. Samples were taken 2 and 7 hr in recovery to evaluate HSP70 and HSP90 expression. Purple line corresponds to unstained control, green line corresponds to control (non stressed cells), red line correspond to sample cells stressed at 90 mL/min…………………… 265
xxii
CHAPTER 1
INTRODUCTION
1.1. ANIMAL CELL CULTURE
Cultures of animal cells are able to reproduce clinically active copies of proteins used by the human body. These proteins have been used in a recent generation of drugs for the replacement or reinforcement of deficient or defective proteins or component characteristics of different diseases. (Cartwright, 1994). Although recombinant proteins can also be obtained from bacteria, animal cell culture has emerged as an attractive technology for producing recombinant proteins. These cells have the machinery to make post-translational modifications as well as regulation mechanisms to assure proper folding and also secrete these active forms into the suspending media. These unique properties allow the production of proteins in their functional configuration, which can
represent significant savings in the overall process. Other technologies that use bacteria
typically require significant downstream processes that include lysis of cells in order to
release the recombinant protein. They also require additional steps like solubilization and
refolding in order to obtain the bioactive conformation of the protein.
1 There are a variety of pharmaceutical products obtained through the use of cell cultures.
Cells Hamster Ovary (CHO cells) was the cell line used for the production of tPA (tissue plasminogen activator), a protein used in the treatment of cardiac disease. Since its
approval in 1986, this cell line has been used to obtain other proteins such as β-interferon used in multiple sclerosis, Factor IX used in hemophilia B and erythropoietin used in anemia among others. Over the last decade, a large number of monoclonal antibodies, produced in CHO cells have became approved, or are in various stages of approval. This success of antibodies is primarily the result of high specificity of the antibodies that can be preserved in this CHO derived products as well as the “humanized” characteristics of these antibodies that prevent human immune response.
The advantages of cell culture technology are reflected in the growing market of recombinant proteins, whose sales were estimated at $54.5 billion in 2007 and are expected to increase up to $75.8 billion by 2012 (Research and markets, 2008). Of course, the impact of the development of animal cell cultures should not be assessed purely as an issue of economics. The key benefit of the technology is the potential for developing treatments to previously incurable diseases and providing an improved quality of life to those who receive treatments thus derived. Due to the popular demand for monoclonal antibody and recombinant protein production, and continued economic pressures, the pharmaceutical industry continues working to increase production capacity, however, such increasing demands for productivity not only in the upstream stage but also in the downstream processes can subject cells to aggressive environments including
2 hydrodynamic stress. More broadly, the effect of hydrodynamic forces on animal cells
used for research purposes is also of interest.
1.2. OBJECTIVES
1.2.1. Goal
This work’s intent is to improve the understanding of the effect of hydrodynamic stress on mammalian cells. Cell damage due to hydrodynamic stress occurs in biomedical devices such as Flow cytometer as well as industrial equipment. We focused on the quantification of cell damage of both industrially relevant cell lines and human cell lines used for research purposes.
1.2.2. Methodology
Two cell lines were chosen to evaluate the stress response in this work. Chinese Hamster
Ovary (CHO) and THP1. CHO is a cell line widely used in the pharmaceutical industry for the production of recombinant proteins. CHO cells were exposed to different hydrodynamics conditions in order to evaluate the stress response as well as the effect on viability and productivity of the cell culture. THP1 is a human cell line, which is routinely studied in screening assays to assess cytotoxicity induced by potential agents.
THP1, being from human origin, also allows for many other types of analysis and represents a model of study of human cells used in flow cytometry analysis. In addition,
3 other human leukemia cell lines as well as primary cells were used in screening of cell
sensitivity.
A summary of aspects considered in the methodology is presented in Figure 1.1. The
methodology will be discussed in detail in every chapter, briefly, the analysis include:
characterization of cell cultures in static cultures, cell sensitivity screening using single pass assays, cell damage in flow cytometer (THP1 cells), cell damage in bioreactors
(CHO), stress response, and hydrodynamic characterization of bioreactors and flow
cytometer.
1.3. SCOPES OF THIS STUDY
1.3.1. Cell damage in a fluorescent activated cell sorter (FACS). (Chapter 3)
The first part of the work developed to study cell damage in FACS was published in the paper:
Mollet, M., Godoy-Silva, R., Berdugo, C, Chalmers, J.J. (2008). Computer Simulations
of the Energy Dissipation Rate in a Fluorescence-Activated Cell Sorter: Implications to
Cells. Biotechnology and Bioengineering, Vol. 100, No. 2, June 1, 2008.
The work reported in this paper focused on the BectonDickenson flow cytometer with a
model name of FACS Vantage. Chinese hamster ovary cells (CHO) and a human
leukemia cell line (THP1) were used as model to evaluate integrity in cell sorting.
4 Claudia Berdugo’s contribution to this work includes:
- Experiments of cell sorting in FACS Vantage with THP1 cells
- Data and results included in Figure 4.7 as well as data from cell sorting include in
Tables 4.3 and 4.4
- Editing of the document, bibliographical review, analysis and discussion.
The second part of the work is presented in this chapter and focus in hydrodynamics studies and cell damage on FACS Aria. This instrument is equipped with three lasers
(488nm, 633nm and 405 nm), 13 fluorescent channels and 2 scatter channels. FACS Aria offers expanded detection of fluorescence with respect to the Vantage since the last one does not have 405 nm laser. Another important difference between Vantage and Aria is a fundamentally different design in the fluidics: the point of interrogation of cells by the laser is “upstream” of the point of the drop creation (unlike the FACS Vantage in which the cells in drops are interrogated), and the actual nozzle where the “sheath fluid: and fluid containing the cells converges is significantly different. Both instruments are widely used in Biomedical research, we focused on the quantification of cell damage, based upon practitioners reported damage.
The effect of hydrodynamic stress induced by sorting these cells in these two instruments was quantified both experimentally as well as theoretically using computational fluid dynamics packages. The results indicate that not only are cells damaged in a flow cytometer, but that this damage can vary from cell line to cell line as well as specific 5 conditions/type of flow cytometer and flow conditions in rapid contractions (the torture
chamber). In addition, the sensitivity of any specific cell line can be a function of the
growth phase of the cell.
1.3.2. Hydrodynamic studies in bioreactors. (Chapter 4)
The work presented in this chapter was developed as part of an internship at
GlaxoSmithKline.
Large-scale cell cultures require efficient mass transfer systems for economically viable production. Mass transfer capabilities are determined by the bioreactor’s agitation and aeration configuration. The aeration and agitation configuration also dictates the level of shear stress that cells will experience during culture. A balance should be accomplished to obtain the best mass transfer capabilities and low shear stress during bioreactor design.
In this work, the performance of ten different impeller-sparger configurations was evaluated. A full factorial experimental design with two center points was used to characterize the mass transfer capabilities for every configuration. The best impeller-
O2 sparger configuration was chosen based on the K L a response surface model for testing
in cell culture experiments.
The evaluation included different types of sparger and impeller as well as relative
location of impeller along the shaft and clearance between impellers. Results indicated 6 that the volumetric oxygen mass transfer coefficient is affected by the relative sparger
location and is less affected by clearance between impellers.
Once the best impeller sparger configuration was identified a cell culture experiment was carried out to assess its effects on culture performance. The proposed configuration supported high cell density cultures through improved gas dispersion, acceptable shear
rates and low foam formation.
1.3.3. Effect of hydrodynamic conditions on cell cycle, stress proteins and
recombinant protein productivity. (Chapter 5).
It has been proposed that the dependence of protein expression on cell cycle phase can be used quantitatively in animal cell bioreactor optimisation. Protein synthesis has been associated on particular cell cycle phases for different recombinant genes, however there is not a consensus in the literature regarding to the cell cycle phase where the cells are the most productive.
In this work, a potential dependence of recombinant proteins productivity on cell cycle was studied. In addition, we have evaluated the expression of stress proteins under different hydrodynamic conditions and we analyzed the relationship between the expression of stress proteins and recombinant protein productivity. Also, the expression of stress proteins in highly sensitive cell lines will be discussed. Stress proteins investigated have been studied in the context of heat and nutritional stress but it has not
7 been determined the expression of stress proteins in response to hydrodynamic
conditions.
The results strongly suggest that secretion occurs primarily during the G1 cell cycle phase. Some differences were observed depending on the feeding strategy and the
hydrodynamic conditions. On the other hand, the expression of stress proteins follows a
characteristic profile that is related with the state of the culture.
8
1.4. REFERENCES
Cartwright, 1994. Animal Cells as bioreactors. Cambridge University Press.
Godoy-silva, R., Berdugo, C. and Chalmers, J. (Peer review). Literature Review: Mixing, Aeration and hydrodynamics in bioreactors. Wiley Encyclopedia of Industrial Biotechnology.
Ma, N.; Koelling, K. and Chalmers, J. J. (2002). The fabrication and use of a transient contractional flow device to quantify the sensitivity of mammalian and insect cells to hydrodynamic forces. Biotechnology and Bioengineering. 80:428-437.
Mollet, M., Godoy-Silva, R., Berdugo, C, Chalmers, J.J. (2008). Computer simulations of the energy dissipation rate in a fluorescence activated cell sorter: implications to cells. Biotechnology and Bioengineering. 100(2): 260-272.
Mollet, M.; Godoy-Silva, R.; Berdugo, C. and Chalmers, J. J. (2007). Acute Hydrodynamic Forces and Apoptosis: A Complex Question. Biotechnology and Bioengineering. 98 (4): 772-788.
Research and markets, 2008. Biologics Pipelines: Who has the next promising recombinant proteins? http://www.researchandmarkets.com/reports/603343
9
THP1 CHOCHO cell line CHO-K1CHO-6E6 (Cell(ATCC) line ATCC) (GlaxoSmithKline) (Cell(ATCC) line ATCC)
Static Cultures Flow Cytometer Reactor Studies Static Cultures Reactor Studies
CellCell culture: cycle, Single Pass LDH Different Single Pass configurations to StressCell cycle, Proteins: evaluate Mass HSPStress 70, proteins HSP 90 Transfer RecombinantHSP70, HSP90 Protein LDH Stress proteins capabilities LDH kLa, EDR Molecular markers Molecular markers Continuous Stress Stress proteins Hydrodynamics Stress proteins Stains Stains characterization CellCell culture: cycle, Stress Proteins: Molecular markers Cell cycle, Molecular markers HSP 70, HSP 90 Stains Stress proteins Stains RecombinantHSP70, HSP90 Protein Recombinant protein
Figure 1.1. Aspects considered in the methodology to evaluate the effect of hydrodynamic forces on mammalian cell cultures.
10
CHAPTER 2
LITERATURE REVIEW: MIXING, AERATION AND HYDRODYNAMICS IN
BIOREACTORS
The content of this chapter is currently under the process of peer review as a chapter contribution for the Wiley Encyclopedia of Industrial Biotechnology. The chapter was written in conjunction with Dr. Ruben Godoy, a former PhD student at Dr. Chalmers’ group.
2.1. ABSTRACT
Animal cell culture has been used extensively for the production of a wide variety of substances with biological and therapeutic activity. In spite of being a mature area of development, no universally accepted method of design or scale up of animal cell culture has been developed.
When designing or scaling up a new process, mixing and aeration are among the most important operations to be considered since they achieve some of the most fundamental
11 objectives carried out in a bioreactor such us keeping certain level of homogeneity in the physicochemical parameters of the culture and providing oxygen to the cells. On one
side, oxygen is a crucial nutrient involved in growth and energy production in animal cell
culture; however, the low solubility of oxygen in water based media forces its continuous
supply to the culture, usually through direct sparging. Foam and cell bubble interactions
resulting from gas sparging may be problematic or even catastrophic for the culture but
they are usually controlled through the use of antifoaming substances and protective
surfactants.
Mixing, on the other hand, is absolutely necessary in animal cell culture to keep cells
suspended and improving mass and heat transfer. While some degree of mixing is
obtained through air sparging, particularly in the case of air lift bioreactors, most bioreactors rely on some kind of mechanical device (impeller) to draw kinetic energy into
the broth. Even though the choice of type and geometric characteristics of the impeller
are highly subjective, the optimum stirring speed is commonly evaluated experimentally
in bench scale bioreactors; even so, a large proportion of under mixed, large scale animal
cell culture bioreactors are currently in operation in many facilities as a result of a
wrongfully perceived “shear sensitivity” of animal cells. Although it has been
conclusively shown that certain culture conditions can lead to cell damage and death,
such damage is very dependent on the particular cell line, the culture characteristics such
as surface dependent vs. free suspension growth and the presence of any gas medium
interface where bubbles can rupture, especially when protective surfactants are not present. While experimental evaluation is advisable, for common industrial cell lines and
12 culture media “shear” sensitivity is not an issue and therefore scaling up conditions
should be focused on improving mass transfer limitations.
2.2. INTRODUCTION
Animal cells have an innate ability for proper folding and post translational processing of proteins that makes them preferable as host for producing biological components of
therapeutic and diagnostic interest; as a result, the culture of mammalian cell have been
used extensively for the production of biologics, including virus vaccines, monoclonal
antibodies (MAbs), hormones, enzymes, growth and blood factors. Although the
industrial exploitation of animal cell cultures started over five decades ago with the production of Salk polio virus vaccine in primary monkey kidney cells (Griffiths, 2000), it has been during the last 21 year period when there has been a rapid increase in the number of FDA approved products produced in mammalian cell culture, starting in 1986 with the production of recombinant tissue plasminogen activator (tPA) with genetically engineered CHO cells and the simultaneous introduction of the first Mab, ORTOCLONE
OKT3 (muromonab CD3) as a treatment for solid transplant rejection. Since then, 29
MAbs and many other therapeutic agents have been approved for marketing in the USA for the diagnosis or treatment of diverse diseases, a majority of which are produced in animal cell cultures (Ozturk, 2006; Thomson Centerwatch, 2007). By 2005, one estimate of the world market for antibodies produced in animal cell culture was 14 billion dollars and this number was expected to grow to more than 16 billion by 2006 (Research and
Markets, 2007). While the specific, and or volumetric productivity continues to improve,
13 i.e. greater than 5 g/L of product is now considered the norm, this rapid growth in the
number and demand of biologics has driven the industry to expand the size and number
of production facilities and lead to statements that shortfalls in manufacturing capacity exist (Molowa and Mazanet, 2003). All of this positive growth puts significant pressure to continually improve the size and productivity of commercial animal cell culture systems.
Despite the fact that reports exist showing that animal cell culture has been conducted for over a hundred years (Harrison, 1907; Carrel, 1912), no universally accepted method of design or scale up of animal cell culture has been developed. Diverse culture methodologies including static cultures (T flask), roller bottles, cultures on microcarriers and freely suspended cells in batch, fed batch or perfusion systems, among others, have evolved. While freely suspended, fed batch processes have emerged as the predominate ones, much of this evolution has been guided by the particularities of the process, the cell line and, most importantly, the in house expertise of the researchers specific to each organization. Although there are reports of large scale (1000 to 2000 liter) airlift bioreactors for protein/antibody production (Birch et al., 1985; Varley and Birch, 1999;
Hesse et al., 2003), at least 70% of the licensed processes for recombinant proteins, antibodies and vaccines using microcarriers or freely suspended cultures use traditional stirred tank bioreactors with reported capacities up to 20,000L (Chu and Robinson, 2001;
Butler, 2005; Meier, 2005). Several reasons account for this preference including the vast empirical knowledge accumulated for the design, scale up and operation of this type of reactor in the chemical and biochemical industries over the last century, the versatility of
14 this reactor type allowing its adaptation to several different processes with minor or no
modifications and, finally, the relative simplicity. As a result, most guidelines for bioprocesses design and scale up are based on stirred tank reactors and as such will be
the primary focus for the remainder of this review.
All animal cell processes are aerobic; oxygen is a crucial nutrient involved in growth and
energy production in such cells. Unfortunately, because of its low solubility in water based media, dissolved oxygen is consumed quickly, requiring a continuous supply in
order to keep cells alive. The easiest way of providing oxygen to a stirred cultured is
through surface aeration; however, surface aeration is not sufficient for cultures beyond
~100 L. Other aeration methods (i.e. perfluorocarbons, membrane aeration, oxygen
carriers) have been used successfully at scales up to 500 L but their use is limited by high
cost and downstream concerns (perfluorocarbons) or by limited design data and
difficulties in maintenance (membrane aeration). These limitations have placed air and
oxygen enriched air sparging as the most common and simplest method for continuously providing oxygen in bioreactors.
Agitation, on the other hand, is essential for satisfactory mass transfer and homogeneity
in a bioreactor with or without sparging; this homogeneity is crucial for process control.
Although sparging provides some degree of agitation by itself, most animal cell cultures
rely on some sort of mechanical device to impart enough kinetic energy to the fluid so a
certain degree of homogeneity is reached.
15 Both agitation and sparging are essential for the success of industrial cultures of animal
cells; yet, both are associated with cell damage as a result of cell bubble, liquid cell
and/or solid cell interactions. Significant progress has been made over the last two
decades in understanding of cell damage mechanisms and in this chapter we will attempt
to present an overview of the current practices and protocols for aeration and mixing in bioreactor design and operation as well as their connection to hydrodynamical damage in
animal cell processes.
2.3. AERATION
Oxygen is a key substrate in animal cell cultures; unfortunately, its sparing solubility in
water based media (7.8 mg·l 1 when bubbling air in water at 760 mm Hg and 25°C)
requires a continuous supply throughout the culture for most larger scale situations. At
high cell concentrations, the rate of oxygen consumption may exceed the rate of oxygen
supply; in those conditions, the dissolved oxygen (DO) concentration falls down to a point (the critical oxygen concentration) where it becomes limiting. Reported critical
oxygen concentration for animal cells ranges typically between 1 and 10% of air
saturation (Gotoh et al., 2001; Zeng and Bi, 2006). Below such range, typically the
oxygen consumption rate as well as the mithocondrial activity decrease, while the
specific glucose and glutamine consumption rates increase (to compensate for ATP production) with the corresponding increase in the lactate and ammonium formation rates
(Miller et al., 1987; Ozturk and Palsson, 1990; Lin and Miller, 1992). Together with a reduced mithocondrial activity, there is a reduction in the specific secretion rate of
16 proteins and volumetric accumulation of product which translates into higher cell viability, delayed death and lower chromosomal damage but also into low product yield
(Ozturk and Palsson, 1990; Packer and Fuehr, 1977).
On the other hand, elevated oxygen concentration (~100% of air saturation) may stimulate the generation of reactive oxygen species (ROS) that can alter cellular macromolecules such as DNA, proteins and lipids, impairing cell growth or even causing death. (Zeng and Bi, 2006).
As a result, cell culture should be constrained within an optimal range of dissolved oxygen that must be determined experimentally, taking into account that optimal DO values for cell growth may be different than optimal values for protein production. Oh et al. (1989) reported satisfactory cultivation of mammalian cells from approximately 5 to
100% of air saturation. Jan et al. (1997) studied the effect of dissolved oxygen from 10%
to 150% of air saturation on the growth of a murine hybridoma cells at steady state in
serum free continuous culture; while no significant effect on cell viability, growth rate or
specific antibody production rate was found, an increase in the amount of glucose utilized
at higher oxygen concentrations was detected. It was speculated that this increased
glucose consumption rate was associated with an increased activity of antioxidant
enzymes needed to reduce the cytotoxic effect of ROS. Using the same system, Kunkel et
al. (1998) described an increase in the level of galactosylation of the mAb chains with
increased dissolved oxygen.
17 For common animal cells lines under favorable oxygen supply, reported specific oxygen
18 16 1 1 uptake rates (qO2) range between 6.4×10 and 4.5×10 mol oxygen·cell ·s (See Table
2.1). Given that currently cell densities of up to 107 cells·ml 1 are common place at the commercial scale (deZengotita et al. (2000); Wurm, 2004; Nienow, 2006), the actual
oxygen uptake rate in a typical culture may range from 6.4×10 5 to 4.5×10 3 mol
oxygen·m 3·s 1; this value is at least one or two orders of magnitude lower compared to
yeast (i.e. 2.2 ×10 2 mol oxygen·m 3·s 1 for Candida guilliermondii (Gimenes et al.,
1 3 1 2002)) and bacteria (i.e. 3.3×10 mol oxygen·m ·s for E. coli, corresponding to qO2=
12 to 21 mmol ·gDW·h 1 (Fong et al., 2005) and up to 190 gDW·l 1 (Shiloach and Fass,
2005)). This low oxygen requirement may help to explain why even with an agitation
intensity for animal cell cultures about 100 fold lower than for microbial or fungi cultures
(average power drawn per unit volume ~1000 W·m 3 vs. ~10 W·m 3 respectively) current bioreactors are still able to provide enough oxygen to met animal cell needs (Varley and
Birch, 1999; Ma et al., 2006). However, with the advent of perfusion cultures, fortified
media and improved fed batch culture strategies, it is expected that cell concentrations
will increase by an order of magnitude (Kompala and Ozturk, 2006). Therefore, an
equivalent increase in oxygen transfer rate (OTR) must be achieved to supply enough
oxygen to the culture.
1 The OTR is determined from the product of the mass transfer coefficient, kLa [s ], and
the driving force, which is the difference between the saturated dissolved oxygen
18 concentration in the medium, C * , and the actual dissolved oxygen concentration in the O2 broth, C : O2
OTR = k a ⋅(C * − C ) (Eq. 2.1). L O2 O2
The saturating dissolved oxygen concentration in the medium in equilibrium with the gas phase, is determined by:
y ⋅ P C * = O2 T (Eq. 2.2) O2 H
where y is the molar fraction of oxygen in the gas (0.21 for air), P is the total pressure O2 T
of the gas and H is the Henry’s constant which is a function of temperature and the composition of the medium. A simple inspection of Eq. (2.1) reveals that in order to increase OTR it is necessary to increase either k a or C * since C can only been L O2 O2 reduced to a minimum threshold to avoid productivity issues as explained before. An increment of C * can be achieved by an increase in pressure of the gas (Eq. 2.2), but this O2 action also results in an increment in the concentration of dissolved CO2 and this may
result in inhibitory effects. OTR also may be increased up to 5 times by increasing the
molar fraction of oxygen in the gas using oxygen enriched air or even a separate stream
of pure oxygen into the broth. While this strategy is commonly used in bench bioreactors,
it will lead to increased levels of dissolved CO2 originating additional concerns about pH control and osmolality (Nienow, 2006). Therefore, probably the most prudent way to further increase the OTR is by increasing kLa.
19 Although kLa can be affected by several different factors, most proposed correlations in
literature (some of them presented in Table 2.2) adopt the form of an equation dependent
on two parameters that are independent of the impeller type, media composition, and
scale:
α β k L a = A ⋅[(ε T )g ] ⋅[Vs ] (Eq. 2.3)
where α and β are approximately 0.5 ± 0.1, respectively (Nienow, 2006). Vs is the surface
velocity of the gas, defined by Eq. (2.6), and (ε T )g is the average total energy dissipation
rate drawn into the liquid by both the impeller (ε T )Ig and the gas sparging (ε T )S :
(ε T )g = (ε T )Ig + (ε T )S (Eq. 2.4)
The energy delivered to the system from air sparging can be calculated from:
(ε T )S ≈ Vs ⋅ g ⋅ ρ (Eq. 2.5)
2 where g is the acceleration due to gravity (9.81 m·s ), ρ is the liquid density and Vs is the
superficial velocity, given by:
20 ϕ ⋅V 4 ⋅ R Q 4 ⋅ Q 60 V = = = (Eq. 2.6) s A π ⋅T 2 π ⋅T 2
3 1 where Q is the air flow rate [m ·s ], T is the vessel diameter [m], VR is the volume of medium in the reactor [m3] and φ is the air flow rate expressed as v.v.m. Since cell
damage as a result of bubble bursting (see cell damage section) is typically minimized at
low air flow rates, (ε T )S can often be neglected for stirred tank reactors used for animal cells.
The energy drawn into the liquid by one impeller under gassed conditions, (ε T )Ig can be
estimated as:
ρ ⋅ N 3 ⋅ D 5 ()ε = P (Eq. 2.7) T Ig 0g VR
1 where N is the impeller speed [s ], D is the impeller diameter [m] and P0g is the power
number under aerated conditions, which depends on the Reynolds number, the impeller
type and the gas flow rate. For typical culture conditions, the properties of the fluid are
2 quite similar to those of water, so the flow is essentially turbulent ( Re = ρ L ⋅ N ⋅ D
5 ~10 ); in that case, P0g becomes independent of the Reynolds number; additionally, as the
gas flow rate (φ) is very low, typically ranging from 0.005 to 0.01 v.v.m. (Nienow et al.,
1996), P0g is essentially equal to the Power number under non gassed conditions (P0). In
21 spite of that, the relationship between P0g and P0 is typically important in impeller
selection. Values for P0 are presented in Table 2.3 for impeller typically used in animal
cell bioreactors (Nienow, 2006). Further details about P0 and P0g are presented in the
mixing section.
The term A in Equation (2.3), and therefore kLa, are very sensitive to composition. For
example, by introducing ions in water up to a limiting concentration of ~10 g/L NaCl,
Van’t Riet (1979) showed that kLa can easily increase by a factor of 100% respect to the
value for distilled or tap water at the same conditions of (ε T )Ig and VS; the presence of
ions strongly changes the coalescent characteristic of pure water, increasing drastically
the interfacial area. In contrast, Metz et al. (1979) showed that in a stirred bioreactor with
1 otherwise constant conditions (400 rpm, Vs = 30.6 m·h ) the addition of 0.5 ppm of a
1 surfactant reduced the kLa from ~180 to ~115 h . Similar results were presented by
Morāo et al. (1999) for three different types of antifoaming substances. Addition of
Pluronic F 68 also lowers kLa (Lavery and Nienow, 1987); therefore, caution should be taken when choosing a particular correlation for designing purposes.
2.3.1. Surface aeration
Oxygen transfer through the headspace is, obviously, the most common and easy method
of gas exchange used for small vessels at the laboratory scale, including T flasks,
spinners, roller bottles and small bench scale reactors. The maximum volume at which surface aeration can be used alone depends on a series of factors such as cell 22 concentration, whether or not the headspace gas is enriched with oxygen and whether or
not bubble entrainment is allowed (if the system is agitated). However, a general principle which holds for scale ups is that the area per unit volume for mass transfer
decreases with the reciprocal of the diameter of the vessel. To alleviate this problem at
moderate scale, the use of surface impellers (an impeller close to the liquid gas interface)
can significantly increase the OTR as a result of an increased renewal of the liquid
directly in contact with the gas. If bubble entrainment occurs, a further increase in OTR
can be achieved as a result of the significant increase in the interfacial surface. Working
with a 500 ml spinner vessel, Aunis et al. (1989) reported a 50% increase in the OTR by
moving the impeller within 0.13 times the impeller diameter of the liquid surface. Hu et
al., (1986) reported an increase in the oxygen supply up to 4 times in their reactor by
using an additional surface impeller. Another approach for high values of surface OTR
involves the use of special impeller types inducing bubble entrainment such as helical
and double helical ribbons together with surface baffles (Kamen et al., 1992).
A variation in the surface aeration approach is the disposable bag reactors (i.e. the
Wave® bioreactor), which can range in size from 100 ml to 300 L. This system achieves sufficient OTR by means of a rocking motion that induces a wave within the bag that is sufficient to suspend the cells, mix the medium and provides a virtually bubble free oxygen transfer through the rapid turnover of the medium surface. Singh (1999) presents data for Wave reactors from 100 ml up to 100 liter, with corresponding kLa values
1 ranging from 1.6 to 4 h . While for small scale these values of kLa are in the same range as for other surface aerated reactors, it is in the lower end of the range for sparged,
23 mechanically stirred reactors (Lavery and Nienow, 1987). However, for volumes larger
than 10 liter the kLa in a Wave reactor can still up to one order of magnitude larger than for a spinner or reactor of comparable size; consequently, it is reported to be able to support cell growth up to concentrations of 7×106 cell ml 1 (Singh, 1999).
2.3.2. Perfluorocarbons
Perfluorocarbons (PFCs) are hydrocarbons whose hydrogen atoms are partly or totally
replaced by fluorine. They are hydrophobic and immiscible in water based media;
therefore, the addition of PFCs yields a four phase (two liquid, one solid and one gas phase) mixture in sparged bioreactors. Because of the much higher solubility of oxygen
in PFCs than in water (see Table 2.4), there has been an active interest in using them as
gas vectors to improve oxygen transfer and CO2 removal in bioreactors. The benefits of
using PFCs are summarized by Lowe et al. (1998) and presented in Table 2.5, together
with their main disadvantages.
Several studies have been conducted with animal cell cultures in which the medium was
supplemented with perfluorocarbon emulsions. Cho and Wang (1988) employed perfluoromethyldecalin to oxygenate mouse hybridoma cells, resulting in increased cell
density and rate of Mab synthesis. Using the PFC FC40™, Ju and Armiger (1992) found
that the exponential growth phase of mouse hybridoma cells was prolonged, along with a
more than six fold increase in viable cell concentration, which was consistent with
theoretical predictions based on the enhanced oxygen transfer. Although the high
24 concentration of surfactants needed to keep PFCs in suspension might be deleterious for
cells or for downstream operations, such effects have not been reported. It has also been
suggested that the formation of emulsions itself might be problematic for recycling and
re oxygenation. Gotoh et al. (2001a; 2001b) proposed two interesting alternatives to solve this potential problem with emulsions: they developed a loop column reactor having a wetted wall of PFCs to culture insect cells (Sf9). With this arrangement, the oxygen was transferred by direct contact of the broth with the liquid film created on the annular wall of the column. They reported a two fold increase in the viable cell density and specific growth rate of the cells as well as a five fold reduction in the specific lactate yield. In the second permutation, the same authors grew cell suspensions using a planar layer of PFC at the medium PFC interface in a spinner flask and the PFC was recirculated and re oxygenated in an outer aeration unit. With this arrangement, the maximum cell density attained in the PFC mediated aeration culture was higher than that in a surface aeration control culture; also, the recombinant protein production by virus infected insect cells was significantly increased. The PFC itself did not seem to cause any negative effects on insect cell growth, viral infection, or recombinant protein production.
However, in spite of all the previous advantages, PFCs introduce additional cost and the potential problems in downstream processing have kept PFC so far as only a research curiosity (Varley and Birch, 1999).
25 2.3.3. Oxygen carriers
At least two other oxygen carriers, in addition to PFC, have been suggested for use to
enhance oxygen supply in cell culture: cyanobacterial gas vesicles (GVs) and hemoglobin
(Sundararajan and Ju, 2006). While the use of hemoglobin as an oxygen carrier has only been tried in microbial cultures and it has been limited by its high cost and by its high affinity for oxygen (i.e. it requires long residence times in the reactor to effectively release the bound oxygen), GVs have actually been tested in animal cell culture. GVs in cultures of Vero cells on microcarriers have been reported to increase by 30% the glucose uptake rate although no effect on cell concentration was described. GVs can be produced at a relatively low cost from simple cyanobacterial cultures, they do not require any suspension stabilizing agent (unlike PFCs), and they have more than twice the oxygen carrying capacity as PFCs (See Table 2.4). However, they can not be autoclaved; consequently, sterilization is cumbersome and time consuming (Sundararajan and Ju,
2006). It remains to be seen if a commercial application will be found for GVs.
2.3.4. Membrane aeration
In membrane aeration, the membrane provides enough interfacial area for oxygen
diffusion while it acts as a physical barrier that keeps cells away from gas liquid
interfaces, eliminating any cell damage related to bubble bursting phenomena or cell
entrapment in foam. This methodology has been reported to give high oxygen transfer
1 1 rates (i.e.~0.5 g O2·l ·h ) that are comparable to sparging, although they require higher
26 gas pressure and flow rates (Moreira et al., 1995; Lütz et al., 2006). Ma et al. (2006) categorize the different types of membranes into two groups: a microporous type and a diffusional type. The microporous type typically is a thin membrane made up of nylon, polyvinylidene fluoride, polyester, etc., with pore sizes ranging from 0.01 m to about 20
m and where the medium is in direct contact with air in the micropores of the membrane. The liquid gas interface is held stationary via both hydrophobic forces and gas pressure. In the diffusional type of membrane, the oxygen diffuses through the actual membrane material. Some of the most common examples are silicon tubing and silicon hollow fiber membrane oxygenators.
There are reports of the use of membranes for oxygen transfer up to 150 L (Lehmann et al., 1988; Vorlop et al., 1989); however, their use at larger scales is problematic. While there is always the risk of rupture, clogging, and fouling of the membranes, these risks significantly increase during scale up since one is faced with classical surface to volume ratio problems, i.e. as the volume of the vessel increases, one needs to increase the surface area of the membranes, typically tubing, even faster. At some point, the amount of tubing required becomes unmanageable, especially when one considers the challenge of cleaning and sterilizing large vessels (Kretzmer, 2002; Ma et al., 2006).
2.3.5. Sparging
Despite all of the alternatives reviewed above, the most commonly used way to introduce
and remove gaseous nutrients and byproducts from animal cell culture, after simple head 27 space mass transfer, is by sparging the gas directly into the medium. While potentially catastrophic effects can occur (i.e. cell death of the whole culture in a few hours), through the use of specific additives the destruction of a majority of the cells can be prevented.
More on both the mechanism(s) of death and prevention will be presented later.
2.3.6. Sparger design
Proper selection of a gas sparger involves a compromise between maximizing gas
transfer (O2 and CO2), minimizing cell damage and foam formation, and complying with
GMP considerations (easily cleanable, steam sterilizable and free draining). Ideally, spargers should be designed to disperse bubbles sufficiently to avoid coalescence into larger bubbles as they rise within the vessel (Marks, 2003); however, excessive pressure drop across the orifice may create sufficient hydrodynamic forces to inhibit net cell growth (Murhammer and Goochee, 1990) and therefore should be avoided.
There are three commonly used designs in cell culture: the point sparger, ring sparger and
frit sparger (Marks, 2003). Point sparger is the simplest design consisting of a pipe
ending in a nozzle through which air is introduced into the vessel; its simplicity makes it
ideal for cleaning and sterilization. However, the moderate control of the bubble size that
can be achieved by changing the nozzle diameter, gas flow rates and position (depth) of
the nozzle (Kulkarni and Joshi, 2005) results in the formation of large bubbles. Although
Orton and Wang (1991) suggested the use of macrospargers producing large bubbles
(diameter in the range of 6 8 mm) for minimizing foaming and bubble related cell death,
28 the low surface per unit volume created with these spargers results in the use of higher flow rates in order to provide enough oxygen to the culture.
To address the high flow rates needed when a single point sparger is used, yet still follow
GMP guidelines, one might add a number of air nozzles drilled at the bottom of a ring, allowing for free drainage from the tube. While an improvement with respect to a reduction in air flow through a single hole, it has been argued that this enhancement is not sufficient to justify the added expense and cleaning difficulties (Marks, 2003).
Recently, Harris et al. (2005) presented a new approach for a multi point sparger design called the multiple slot disperser which consists of an assembly of flat, parallel arrays of slot nozzles. In water it produced a dense, three dimensional plume of air bubbles with a consistently narrow bubble size distribution and a relatively sharply defined median bubble diameter. Most importantly, the plume attributes may be modified within wide limits by changing the slot width, plate and array dimensions as well as the gas flow rate.
Additionally, the method of construction is highly adaptable, avoiding the need for pores or drilled holes and it is relatively easy to clean and autoclave.
A significant alternative to the single or multiple hole disperser, is the well known and used frit sparger which utilizes porous materials (ceramic, sintered stainless steel or
Teflon) to create an almost unmeasureable amount of very small bubbles within the bioreactor. These bubble are nearly ideal for gas transfer because of the large interfacial gas liquid area they provide and the extended time of contact as a result of lower rising velocities of the smaller bubbles. However, disadvantages of these spargers include
29 cleaning difficulty, more stable foam and reduced CO2 stripping capability, forcing in
some cases the use of both a frit and a large bubble sparger and the use of antifoaming
chemicals (Marks, 2003). The cleaning challenge can be so significant that the authors
know of at least one commercial process in which the frit sparger is replaced after every batch. Documenting some of these reported advantages and disadvantages, Nehring et al.
(2004) tested both a stainless steel porous sparger and a porous ceramic (TiO2) microsparging system for the cultivation of MDCK cells from a 2 to 100 liter scale. The ceramic system produced bubble diameters of 100 – 500 m, and values of kLa up to 50
h 1, 3 times higher than those produced by the stainless steel sparger at the same gas flow
rate; however, formation of compact, stable foam and pH problems were observed.
Similarly, Chisti (1993), found the use of a multi hole sparger to be preferable over a porous metal sparger for scale up of a bioreactor as a result of foaming problems caused by the smaller bubble size generated by the fritted sparger.
2.3.7. CO2 accumulation and removal
Somewhat ironically, as animal cell culture progressed from a research scale to relatively
large scale, commercial process, the addition of oxygen became less of a concern and the
removal, and in some cases the initial addition, of CO2 became more problematic. This
results from a number of issues as a consequence of the complex role CO2 plays in cell culture. Carbon dioxide is a nutrient needed for synthesis of nucleic acid components such as purines and pyrimidines (Smith et al., 2005); therefore, low levels of this
component may inhibit cell growth. Alternatively, CO2 is also a byproduct of aerobic
30 metabolism and high levels of dissolved CO2 are inhibitory for cells and have been
reported to strongly affect cell growth and metabolism (deZengotita et al., 1998; Mostafa
and Gu, 2003; Zhu et al., 2005). To further complicate the situation, carbon dioxide
forms part of a complex buffer system given by the next set of equilibrium reactions:
pCO CO ↔ CO ; []CO = 2 (Eq. 2.8) 2 ( g ) 2 (dissolved ) 2 dissolved H
[H 2CO3 ] −3 CO2 (dissolved ) + H 2O ↔ H 2CO3 ; K h = = .1 70×10 @ 25°C (Eq. 2.9) []CO2 dissolved
− + + − [HCO3 ]⋅ [H ] −4 −1 H 2CO3 ↔ H + HCO3 ; K a1 = = 5.2 ×10 mol ⋅l @ 25°C (Eq. 2.10) []H 2CO3
2− + − + 2− [CO3 ]⋅[H ] −11 −1 HCO3 ↔ H + CO3 ; K a2 = − = .5 61×10 mol ⋅l @ 25°C (Eq. 2.11) []HCO3
Since sodium bicarbonate is typically used as a pH buffer in the culture medium, from the previous set of reactions it is easy to deduce that a change in the dissolved CO2 concentration could lead to a change in the pH through the following relationship:
10 −14 K ⋅ K ⋅ pCO ⋅ K ⋅ K ⋅ pCO K ⋅ K ⋅ pCO H + = + h a1 2 + 2 a1 a2 2 ≈ 10 −14 + h a1 2 (Eq. 2.12) [ ] 2 H []H + []H + ⋅ H []H + ⋅ H ⋅
31 where pCO2 is the partial pressure of CO2 in the exit gas in equilibrium with the medium.
A rise of pCO2 at a controlled pH typically leads to an associated increase in the medium osmolality as a result of both an increased dissolved CO2 concentration and the addition
of a base, such as sodium hydroxide, to maintain a constant pH. The combined influence
of high osmolality and pCO2 might conduce to an even more significant change in the metabolism of the cell. Finally, to add a final level of complexity to the situation, CO2 is
much more soluble than O2, and this CO2 solubility is sufficiently sensitive to pressure such that a significant gradient in CO2 can exist between the top and bottom of a 1000 L,
weakly mixed, animal cell culture bioreactor.
Several published examples of the effect of the concentration of dissolved CO2 on a cell culture are reported next. In 1990, Drapeau et al. studied the effect of increasing dissolved CO2 concentration from 53 to 165 mm Hg on recombinant CHO cells for the production of hM CSF in a 2500 L reactor. At the highest pCO2, the specific cell growth rate decreased by 52% and the specific production rate of hM CSF dropped by 56% relative to the culture with the lower pCO2. While the dissolved oxygen and pH were constant and equal in both cultures, it is unclear if the osmolality was kept constant. In
1996, Gray et al. also described an inhibitory effect in CHO cell density and specific production rate with reported reductions of 33% and 44%, respectively, in a system in which pure oxygen microbubbles using a fritted sparger or silicone diffusion aerators were used to transfer oxygen to the culture. Mostafa and Gu (2003) also reported a 40% drop in the specific production rate of a therapeutical glycoprotein produced by CHO cells when pCO2 increased from 68 mm Hg at bench (1.5 l) scale reactors to 179 mm Hg
32 in pilot plant (1000 l) bioreactors. In the last two cases, osmolality was not controlled so
the real cause of the dramatic effects observed is potentially a combination of the effects
of increased pCO2 and osmolality.
To isolate the impact of pCO2 and osmolality on CHO cell growth, Zhu et al. (2005) performed experiments in bench scale bioreactors where they changed only one variable at a time. Raising pCO2, from 50 to 150 mm Hg, under a controlled osmolality of approximately 350 mOsm·kg 1 resulted in a 9% reduction in specific cell growth rate.
Increasing osmolality from 316 to 450 mOsm·kg 1 resulted in a linear reduction in
specific cell growth rate (0.008 h 1 per 100mOsm·kg 1) up to 60%. The combined effects
1 of high pCO2 (140 160 mm Hg) and osmolality (400 450 mOsm·kg ) caused a 20% drop
in viable cell density.
In contrast to CO2 buildup as a result of low levels of gas medium exchange, the high rates of aeration in bubble columns or airlift reactors may lead to excessive CO2 stripping, which can have equally inhibitory effects (Birch et al., 1987). To solve this problem, the gas for aeration usually includes a certain ammount of CO2 to both keep the pH in a proper value and avoid a reduction of pCO2 below 40 50 mm Hg, range usually
considered optimal for animal cell culture (Gódia and Cairó, 2006).
33 2.4. MIXING
Mixing can be defined as “the operation in which two or more materials (gas, liquid
and/or solid) are distributed throughout a mass in varying degrees of uniformity
dependent upon the change and state to be accomplished” (Uhl and Gray, 1967). In
animal cell bioreactors, this mixing is typically accomplished by the addition of energy to
the broth in the form of mechanical agitation and gas sparging. Ideally, this energy should be delivered to the culture broth in a completely homogenous manner such that any fluid
element, or in this case any cell, will experience a similar distribution of energy in a
similar period of time. Also ideally, this energy delivery should be sufficient to distribute both the cells, and molecularly sized nutrients and by products, in such a manner that no
“process effecting concentration gradients” exist in the vessel. The term, “process
effecting concentrations” is used since a concentration gradient is also a time dependent phenomena in a dynamic environment such as a bioreactor and the characteristic time
scales need to also be considered. As it will be discussed below, far from ideal situations
exist in most animal cell bioreactors.
In general, the hydrodynamic environment experienced by a particular cell in a culture is
a function of its spatial position in the interior of the bioreactor, of the characteristics of
the mixing system, and of time. With the exception of fungal and some microbial
cultures, the rheological characteristics of the broth do not change dramatically with time,
remaining essentially Newtonian and very similar to water. Depending on the final cell
concentration and the potential release of nucleic acids into the culture, a slight increase
34 of the viscosity at the end of the culture may become apparent, but typically this is not
considered important. However, in spite of the low viscosity, a poor design of the mixing
system may lead to catastrophic consequences. If the mixing is not intense enough, cells
will settle at the bottom of the reactor creating zones with limitations of mass and energy
transfer and unsatisfactory homogeneity, rendering the control systems of the reactor
inefficient and ending up with low viability and productivity. Even if agitation is
sufficient to maintain a suspended culture, a number of reports exist of undermixed
commercial animal cell culture process in which significant dissolved CO2 concentration
exist between the top and bottom of the bioreactor, and “zones” of relatively unmixed,
high pH “clouds” exist in the vessel after base addition for considerable periods of time.
Langheinrich and Nienow (1999) measured pH in three different zones of an 8 m3 bioreactor equipped with a 0.225T Rushton turbine intended for the culture of mammalian cells; they found pH gradients close to 1 unit between the alkali addition zone and the impeller discharge when addition of the Na2CO3 was done on the liquid
surface. Ozturk (1996) observed “snow ball” aggregation of mammalian cells,
corresponding to alkaline cell lysis, close to the headspace addition point of base in a poorly agitated bioreactor: the concept of an animal cell suspended in a “pool” of pH 9 or
10 is not a desirable thought!
On the other hand, it has long been feared, but substantiated with little data, that if
suspended animal cell cultures are mixed too aggressively in typically designed bioreactors, the hydrodynamic forces experienced by the cells may inhibit cell growth, productivity, or have other undesirable effects. Unfortunately, historically, no universally
35 accepted algorithm or protocol allowing for the proper design of mixing systems for
animal cell cultures has existed; therefore, an empirical, case by case process has
typically been conducted at each organization (company) as an animal cell process was
scaled up. Also, many (majority?) of these established processes are under mixed. While this approach has resulted in a number of organizations having significant “in house”
expertise, well documented, fundamentally based, peer reviewed scale up protocols are
still lacking. However, a sufficient amount of experimental data is beginning to
accumulate which allows for the development of such scale up algorithms.
The importance of mixing to animal cell culture is far from unique; in fact, specific professional societies, such as the North American Mixing Forum, are dedicated solely to understanding and optimizing mixing in the chemical process industry. Taken in a proper perspective, the challenge of mixing in animal cell culture is not nearly as complex as mixing in many other industries. Consequently, much can be learned from this community.
2.4.1. Stirred tank reactors
As mentioned before, for a number of reasons, not the least of which is the relative
simplicity and considerable knowledge/data base, the classical stirred tank reactor is the predominate method to culture animal cells from bench scale up through large scale,
commercial systems. While a complete review of stirred tank bioreactors is beyond the
scope of this chapter, a quick summary will be presented. The interested practitioner is
36 encouraged to read more complete discussions on the topic (Oldshue, 1983; Hempel,
1988; Tatterson, 1991).
2.4.2. Geometry
A typical, stirred tank, animal cell bioreactor is a mixing vessel with an aspect ratio of
tank height (H) to tank diameter (T) of approximately (H/T) = 2 3 (See Figure 2.1). This
tank is penetrated by a shaft which extends down the center line of the tank; the outside
end of this shaft is attached to a motor while the other end to one or more impellers
which are suspended in the culture broth. Typically, the impeller diameter (D) is about ⅓
to ½ of the tank diameter. The tank is filled with broth up to a height (HL) of ~ 0.7 H.
The off bottom clearance (C), is usually around ~ 0.1 0.3 T. In the case of multiple
impeller systems, the separation between impellers ( C) is highly variable as it will be
discussed next, but normally is set around one impeller diameter.
2.4.3. Impeller
A number of different impeller designs exist and are used (see Figure 2.2), ranging from
radial flow impellers such as the traditional Rushton turbine used extensively in the
mixing community to provide a high level of intense mixing by way of a high local shear rates, to axial flow impellers like the marine propeller which provide better bulk
movement of the broth, relative to the Rushton. Impeller selection is based on a series of
considerations including its power number (P0) and its pumping number (Nq). Impellers
37 with larger ratio Nq/P0 are superior in producing better bulk mixing at lower shear rates.
At the same average energy per unit volume drawn into the liquid (ε T )Ig , impellers with a larger diameter have a higher pumping capacity and therefore may have a better mixing performance. Impeller designs generating lower shear rates are thought to be superior in reducing alleged cell damage as a result of hydrodynamic forces. In this sense, hydrofoils are superior to pitched blade turbines (PBT) and PBTs are superior to Rushton turbines
(Ma et al., 2006). Chen et al. (2003) presented a novel methodology to evaluate impellers based on dinoflagellate: when stimulated by hydrodynamic forces, these organisms generate light; at speeds corresponding to the same oxygen transfer coefficient, a Rushton turbine (150 rpm) generated a much higher luminescence (~20000 counts in 10 ms) compared to marine propellers (170 rpm; ~2000 counts in 10ms) indicating that much higher forces are created in the bioreactor with the first impeller.
In non gassed conditions, the power drawn by an impeller is a function of Reynolds number, impeller type and its geometry. For impeller Reynolds numbers over 104 (normal
conditions in culture), the power number becomes independent of Reynolds and depends
only on variables such as blade angle, number and width, baffle number and width, and
clearance ratio (C/D).
Equation (2.7) governs the power drawn by a single impeller under gassed conditions. In such circumstances, gas reaching the impeller is driven to the back of the blades (see
Figure 2.3) of the impeller generating air cavities that are stabilized by the lower pressure in such zone and reduce the amount of power the liquid is able to transmit to the broth; 38 depending on the amount of gas, the size of such cavities may grow to a point where they
surround the whole impeller; this phenomena is known as flooding and reduces
drastically the amount of power the impeller is able to draw into the liquid because now
the impeller is practically rotating in air. Characteristic curves of P0g/P0 are presented in
Figure 2.4; observe that impellers like the Scaba, are much less prone to flooding than the
Rushton turbine and the reduction in power drawn is limited or non existant as a result of their design.
When the tank is sufficiently large, multiple impellers can be attached to a single shaft, with the potential to use a combination of impellers (i.e. both a Rushton and a marine type) to improve the mixing performance. Under non sparged conditions, the power drawn into the liquid by the different impellers depends on the distance between them.
Hudcova et al. (1989) established that in the case of Newtonian, low viscosity media, the power drawn in non gassed conditions by two Rushton turbines is maximized when the impellers are placed apart as a minimum ~1.8 times the impeller diameter because the flow patterns generated by each impeller practically do not interfere each other (see
Figure 2.1); in such case, the total power drawn into the system is simply the sum of the power drawn by each impeller. Similar results were obtained by Machon et al. (1985) for a diversity of combinations of impellers. Under gassed conditions, however, there is an uneven distribution of the gas between the impellers: the lower impeller draws less power than the upper one and usually works as a gas disperser. A similar behavior (uneven power drawn) is observed when the separation of the impellers is increased until a maximum is reached when the impellers are apart a distance greater than 2D; in this
39 situation, the prediction of the total power drawn into the liquid is much more complex.
Correlations proposed by Cui et al. (1996) to calculate the power drawn by individual impellers in gassed systems with multiple (1 3) Rushton turbines are presented in Table
2.6; in this case, the bottom and top impellers are presented as drawing the same power.
Linek et al. (1996) also developed gassed power correlations for a system with four
Rushton turbines; the interested reader is advised to consult their work.
2.4.4. Baffles
To further improve the effectiveness of the impeller, baffles are used, typically up to four, in which case they are placed perpendicular to the tank wall at 90 degree increments around the vessel. The width of the baffles influences the power drawn into the liquid and therefore the mixing performance, as presented in Figure 2.5; a typical width of the baffle is on the order of 10% of the tank diameter.
2.4.5. Mixing times and energy dissipation rates
As mentioned before, many bioreactors have been designed assuming the high “shear sensitivity” of the cells, which translates on poor mixing performance and a correspondingly lack of homogeneity. In turbulent systems under non gassed conditions, the time required for mixing the content of a bioreactor up to a certain degree of homogeneity (usually 95% of the final response), which is referred to as the mixing time,
40 θm, is independent of the impeller type and can be related to design and operational parameters by the next set of equations (Ma et al., 2006):
− 1 2 −1 ( 3 ) T θ m = 3.5 ⋅ N ⋅ P0 ⋅ ( ) for HL/T =1 (Eq. 2.13) D
− 1 .2 43 −1 ( 3 ) T θ m = 3.3 ⋅ N ⋅ P0 ⋅ ( ) for HL/T >1 (Eq. 2.14) D
where θm will be in the units of the inverse of the impeller rotational speed. Eq. (2.13) correlates very closely for a wide variety of impellers with an almost identical equation derived by Kresta et al. (2006), except the coefficient changes from 5.3 to 5.8. For clarity purposes, Eq. (2.13) can be rearranged by assuming a cylindrical tank
V = π ⋅T 2 ⋅ H and using Eq. (2.7) to give: ( R ( 4) L )
(1 ) 1 2 3 −( ) 3 θ = 57 4. ⋅ T ⋅ ()ε 3 ⋅T for HL/T =1 (Eq. 2.15) m ( D) [ T g ]
where θm will be in seconds when the remaining variables are in SI units. For the derivation of equation (2.15), it was assumed that the density of the fluid is very close to that of water (1000 kg·m 3) and that usual gas flow rates are so low in animal cell culture
that P0g ≈ P0 and (ε T )g ≈ (ε T )Ig . Equation (2.15) is identical to the obtained by Nienow
(1997) and indicates that during scale up (increase of T) for geometrically similar tanks
(T/D constant), the mixing time can be kept constant only if the power input per unit
volume is increased. In contrast, the concern of potential hydrodynamic cell damage has 41 resulted in a majority of large scale animal cell bioreactors to be designed with power per
unit volume significantly decreased relative to laboratory scale vessels, ending up in
increased mixing times and relatively poor mixing.
One concept commonly used in the mixing industry is the energy dissipation rate, or
EDR, which has units of power per unit volume (i.e. W·m 3). On a bulk, or average basis,
this is a very appealing parameter since the average energy dissipation rate, (ε T )g , is just
the rate of energy added to the system divided by the volume in which the energy is
delivered. From a fundamental point of view, the EDR, ε, can be determined using the
following, first principle, equation (Bird et al., 2006; Brodkey, 1995):
ε ≡ τ : ∇U (Eq. 2.16)
where τ is the stress tensor and ∇U is the velocity gradient tensor. For an incompressible
Newtonian fluid, Equation (2.16) becomes (Brodkey, 1995):
T T ε ≡ τ : ∇U = [∇U + (∇U ) ]: ∇U = ∑∑[∇U + (∇U ) ] ij∇U ji (Eq. 2.17) i j
In Eq. (2.17), is the viscosity of the medium, ∇UT is the transpose of ∇U and ∇U is the
gradient of the velocity vector given by Eq. 2.18.
42
∂U x ∂U y ∂U z ∂x ∂x ∂x ∂U ∂U ∂U ∇U = x y z ∂y ∂y ∂y ∂U x ∂U y ∂U z ∂ z ∂ z ∂ z (Eq. 2.18)
Equation (2.16) can be derived from first principles using the second law of thermodynamics. More specifically, Equation (2.16) emerges from the loss term in the energy balance equation. Much like the Navier Stokes equation, Equation (2.16) is valid for any flow regime, be it laminar, transitional, or fully turbulent.
It is this dual nature, a pragmatic value (the power of the mixing motor turning the impeller shaft going into the bioreactor divided by the number of liters in the vessel) and the fundamental, first principle definition, that has lead to the use of EDR to characterize the flow conditions acting on cells for over thirty years (Blustein and Mockros, 1969) and has been extensively used in the fluid mechanical/mixing community for even longer
(Kresta et al. 1998).
As it might be imagined, but not typically fully appreciated, the flow in a stirred tank reactor is highly inhomogeneous. Qualitatively, this was demonstrated in the 1970’s visually by demonstrating that very strong vortices emanate from the tips of the impeller blades; such vortices are responsible, for example, for gas dispersion in the impeller blades (See Figure 2.3). These qualitative observations have been quantified by experimental studies, using a variety of experimental techniques to demonstrate that the
43 local energy dissipation rate in the impeller region, and specifically in these vortices, has been reported to be over 103 times higher than the local EDR in the bulk fluid away from
the impeller (Cutter et. al. 1966; Costes and Couderc, 1988; Zhou and Kresta, 1996;
Mollet et al. 2004).
Given the importance that high, local EDR levels have on rapid, molecular mixing, Zhou
and Kresta (1996a, 1996b) conducted a significant number of quantitative studies on the
maximum and distribution of EDR as a function of a number of variables in stirred tank
vessels, including: tank volume, impeller type, and impeller rotational speed (RPM).
These studies were, in general, consistent with other published studies. Several important
conclusions that are relevant to animal cell bioreactors can be made:
1. The energy dissipation rate is always high in the impeller discharge stream.
Specifically, 43.5 % of mechanical energy added to the vessels is dissipated in
the impeller discharge region of a Rushton turbine, and 70.5% in a pitched
blade turbine (PBT). In addition, the “average” energy dissipation rate is over
an order of magnitude higher in the impeller region when compared to bulk
average for the whole vessel.
2. For a given impeller type, the maximum, local energy dissipation rate in the
impeller region can be approximated using a non dimensional constant and to
a specific, non dimensional location relative to the impeller. This non
dimensional constant, ψ, (our notation) is given by Eq. 2.19.
44 3 2 ψ =ε max / ρ ⋅ N ⋅ D (Eq. 2.19)
In. Eq. 2.19, D is the impeller diameter and N is the impeller speed with units of s 1. If
specific geometric ratios are held constant (including the type of impeller), ψ did not vary
significantly over a large range of impeller speeds. If the D/T (impeller diameter/vessel
diameter) or C/T (off bottom clearance/vessel diameter) ratio changed, within certain
limits, up to a 46% increase in the value of ψ was observed when working with a RT.
These observations also held for the PBT impeller.
Using this relationship for the maximum EDR in a stirred tank vessel, along with the classical P0 and P0g, one can approximate the mean and maximum, local EDR as one scales the stirred tank mixing, keeping the HL/T, D/T, and C/T ratios nearly the same and using the same type of impeller. Figures 2.6, 2.7, and 2.8 present predictions of vessel size and dimensions, maximum local EDR, and average EDR. Specifically, Figure 2.6 presents the change in the tank diameter and impeller diameter as the volume of the vessel increases from 0.5 to 10,000 liters keeping geometrical ratios constant; Figure 2.7 presents lines of constant, maximum EDR as a function of RPM and impeller diameter, and Figure 2.8 presents the average EDR for the whole vessel as a function of impeller diameter and RPM.
Alternatively, Figure 2.9 presents the mean and the maximum EDR, calculated respectively with the power number (Eq. 2.7) and the maximum EDR relationship (Eq.
2.19, using a value of ψ = 13.7), as a function of impeller rotational speed. This plot 45 corresponds, specifically to a 2L Applikon bioreactor which was mixed with a standard
Rushton Turbine impeller. The circle and square data points correspond to specific values of EDR measured with a particle tracking technology (Mollet et al. 2004).
While all of this discussion on EDR is helpful, a significant question should be raised with respect to how EDR relates to cells. This will be addressed in the next section.
2.5. THE RELATIONSHIP OF ANIMAL CELLS TO HYDRODYNAMIC
FORCES
There is probably no subject in animal cell culture cultivation that is more misunderstood and misquoted as being a “major limitation” to large scale animal cell culture than the
“shear sensitivity” of the cells. This misunderstanding, at least partially, arose from early
reports (Bryant et al., 1960; Telling and Elsworth, 1965) whose significance will be
discussed later on. It also seems logical to think that a sufficiently intense force could
disrupt the integrity of the cell membrane causing cell death; especially when animal
cells lack a “cell wall” compared to robust bacteria. To further reinforce this perception,
there are clear, well established studies which do demonstrate almost complete cell
destruction as a result of purely hydrodynamic forces and forces associated with gas
sparging.
This misunderstanding has had a significant, negative impact on the field in the past, and
its legacy continues to this day in both the perception of many cell culturists, as well as in
46 the under mixed, large scale animal cell culture bioreactors that are currently in operation in many facilities. To begin establishing some structure to the notion of the “shear sensitivity” of animal cells, a number of concepts need to be further defined and refined:
1. The term “shear stress” is misleading. In reality, except for specific studies in
specifically designed vessels, at the most fundamental level cells are subjected
to time varying shear and elogational forces in typical bioprocessing
equipment. In addition, it has been argued that elogational forces are more
damaging to cells than shear forces (Garcia Briones and Chalmers, 1994).
2. There is a big difference between the hydrodynamic forces needed to affect or
kill a cell attached to a surface relative to a cell in free suspension. This
difference will be quantified below.
3. There is big difference in the potential to damage cells when gas medium
interfaces are present relative to the absence of any gas medium interface, and
this is especially true when protective surfactants are not present. This also
will be discussed later on.
4. The sensitivity of animal cells to hydrodynamic forces can vary from a cell
line to another, and even within different clones of the same cell line; however
these differences have not been documented to prevent most cells of
commercial interest to be cultured in bioreactors.
5. Due to the complexity of hydrodynamics, and especially under turbulent
conditions, fundamental, first principle, mathematical models describing the
hydrodynamic forces operating on cells in most bioprocessing equipment do
47 not exist and probably never will. Consequently, we are left with continually
improving empirical models and experimental studies that allow us to develop
reasonable scale up methodologies.
With this introduction, what is known on the effect of hydrodynamic forces on cells, both attached and in free suspension as well as a scale up methodology will be presented next.
2.5.1. Molecular mechanism involved in the response of animal cells to mechanical
forces
Liquid cell interactions resulting from fluid flow have been extensively studied, predominately with cells that normally experience fluid flow, such as endothelial cells,
EC. Particularly interesting are models of the interactions of fluid forces with EC and
their subsequent response to small changes of shear stress in laminar flow. Related to the
EC response is that of smooth muscle cells (SMC) to cyclic stretch resulting from the pulsatile nature of blood flow. These studies provide models to begin understanding the physiological responses to mechanical forces inside the human and animal circulatory
and respiratory systems. While it is challenging to extrapolate the results from
experiments with EC/SMC cells in laminar flow chambers to non EC/SMC cells in the
much more complex nature of turbulent flow in bioreactors, the significant, non lethal
effects of hydrodynamic forces provide insight into potential mechanisms and responses
to investigate.
48 Several excellent reviews, including those of Malek and Izumo (1994), Li et al. (2005),
Lehoux et al. (2006), Tzima (2006), Li and Xu (2007), Chien (2007) and Haga et al.
(2007), document what is known about the response of vascular endothelial cells to sub lethal hydrodynamic forces. One of the hallmarks of the EC response appears to be the rearrangement of the cell’s cytoskeleton under the influence of fluid flow (Papadaki and
Eskin, 1997). The cytoskeleton is a three dimensional intracellular network of several different types of protein filaments (like actin) that determines the shape, mechanical properties and movements of the cell, such as those involved in muscular contraction or phagocytosis. The filaments interplay closely with the cellular and nuclear membranes by anchoring to several specific membrane proteins. When the cell is exposed to mechanical forces, the cytoskeleton attempts to resist them and undergoes deformation, specially if the cell is attached to a solid surface. Such deformation implies spatial reorganization of the cytoskeletal lattice; it is presumed that this reorganization is able to transmit the mechanical force along the whole cytoplasm and the nucleus. This mechanism suggests that the deformation of the cell brings together different cell components (i.e. proteins, enzymes, transcription factors and nucleic acids) in both the cytoplasm and the nucleus, originating a cascade of reactions that may lead to a wide diversity of cell responses.
A second proposed mechanism that allows cells to respond to fluid flow involves molecules at the luminal cell surface since they are in direct contact with the fluid. These molecules can be activated by direct conformational change as a result of flow (i.e. rupture or formation of disulfide bonds or bending/stretching of protein domains) or indirectly through mass transfer gradients. Such molecules include (Li et al., 2005):
49 1. Integrins: a family of more than 20 transmembrane heterodimers normally
connected to specific extracellular ligands such as fibronectin, vitronectin and
collagen. After the onset of flow, the integrin activation starts within 1 minute
and last for more than 6 hours, originating an intracellular cascade involving
kinases and cytoskeletal proteins.
2. Ion channels: shear stress increases Ca+2 influx; higher Ca+2 concentration
can lead to multiple Ca+2 dependent cellular responses, including regulation of
other channels (Cl and K+). The response of this receptor is fast (within a
minute) but subsides ~5 minutes after the beginning of the stress.
3. Receptor tyrosine kinases: A series of surface proteins (like Flk 1) transiently
activated by shear stress, resulting in their oligomerization, tyrosine
phosphorylation, association with other proteins (i.e. Shc) and a series of
consequent gene transcription events.
4. GPCRs and G proteins: G proteins are activated within 1 second after the
onset of shear stress and they are involved in shear induced Ras GTPase
activity that serves as molecular switches for a variety of cellular signaling
events
5. PECAM-1: Shear stress causes the transient phosphorylation of PECAM 1
within 1 minute, a glycoprotein expressed in EC, platelets and leucocytes that
plays an important role in leucocyte aggregation and EC leukocyte
interaction.
6. Membrane lipid bilayer: Application of shear stress to EC cells causes an
immediate (< 10 sec) increase in the fluidity of their membrane lipid bilayer in
50 the upstream direction followed by a secondary, larger increase reaching a
peak at 7 minutes. Alternatively, the fluidity of the membrane is reduced on
the downstream side of the cell, indicating that the EC membrane lipid bilayer
can sense the applied shear stress with spatial discrimination.
After sensing mechanical forces, the transduction of shear stress to the interior of the cell
ends up in a series of fast signaling cascades that include opening of K+ and Ca+2
channels, activation of heterotrimeric G proteins, production of NO , tyrosine phosphorylation of proteins such as Shc, c src, and focal adhesion kinase (FAK), activation of mitogen activated protein kinase (MAPK), protein kinase C (PKC), and C
Jun N terminal kinase (JNK), release of reactive oxygen species (ROS), and activation of transcriptional regulators such as c fos, c jun, c myc, and nuclear factor (NF) κB. Slower responses include increased expression of genes for intercellular cell adhesion molecule
(ICAM) 1, nitric oxide synthetase (NOS), platelet derived growth factor (PDGF), tissue factor, transforming growth factor (TGF) β, and monocyte chemoattractant protein
(MCP) 1 and decreased expression of the vasoconstrictor endothelin 1 (Et 1) (Tzima,
2006). Figures 2.10 A and B describe the signaling cascade in EC/SMC that ends up in
effects such as increased gene expression, proliferation or apoptosis, cell migration and
lignment, increased cell membrane permeability and changes in the mechanical properties of the cell.
51 While these results are interesting from a molecular biology point of view, few of these observations have been confirmed in suspended cells of biotechnological interest.
However, they do provide potential responses and mechanisms to investigate.
2.5.2. Cell damage concept
The ambiguity in defining the concept of damage may arise in the fact that conditions that generate biological responses in cells may not be the same that generate negative effects on the process from the point of view of yield or productivity. Consider, for example, the culture of murine neural stem cells (NSC) in 125 mL spinner flasks (Sen et al., 2001). Obviously, neural stem cells are not selected in nature to stand hydrodynamical forces as they grow in still environments and we might anticipate that increased hydrodynamical forces could affect cells’ behavior. However, Figure 2.11 shows that increasing agitation rate (starting from static culture in T flask) increases significantly the number of viable cells, probably as a result of improved mass transfer and control of environmental variables inside the spinner; at the same time, viability is notoriously increased at higher stirring speeds and the mean diameter of aggregates is reduced, which translates into avoiding limiting conditions (specially oxygen) in the core of aggregates that could conduce to apoptosis or necrosis. The combined effect of an improved mass transfer and a decreased resistance to the transport of nutrients to every cell overcomes any negative effect of cellular damage that may exist, resulting in an increment of cell productivity. Quite similar results were obtained by Moreira et al
52 (1995) for BHK cells grown as suspended natural aggregates, where again the maximum
stirrer speed evaluated was 100 rev·min 1.
On the other hand, several reports indicate that highly stirred environments can per se affect viability of suspended mammalian cell cultures. As an example, Hirtenstein and
Clark (1980) showed that while agitation has a beneficial effect in the growth of Vero cells on microcarriers in spinner vessels, a further increase in stirrer speed after the optimum affected dramatically the cell concentration in the system (see Figure 2.12).
Similarly, Kunas and Papoutsakis (1990) reported that stirrer speeds above 700 rev·min 1
reduced cells’ viability on cultures of Hybridoma CRL 8018 when no bubbles are present, while not damage was seen at speeds less than 600 rev·min 1. To further add complexity to the observations, Venkat et al. (1996) and Croughan et al. (1987) reported
that 150 210 rev·min 1 in spinner vessels can result in significant damage to CHO cells
attached to microcarriers.
From the two previous paragraphs the obvious conclusion is the existence of a maximum
threshold of stirring speed beyond which hydrodynamical stresses can have detrimental
effects. However, how stirring speed relates to more fundamental properties of the fluid,
the flow and the cells remains more of a experimental exercise than a theoretically one.
For practical purposes, detrimental effects can extend beyond conditions causing cell
death (reduced viability, increased necrosis or apoptosis) to poorer product quality or
lower productivity as compared with an appropriate control.
53 2.5.3. Hydrodynamical cell damage
Reports detailing cell damage or death are abundant; the first reports appeared
simultaneously with early attempts to scale up cultures of mammalian cells (Bryant et al.,
1960; Telling and Elsworth, 1965). Augenstein et al. (1971) passed cultures of HeLa S3 and mouse L929 through capillary tubes of different diameters. Cell death occurred during the residence time of cells in the capillary and it could be correlated to average wall shear stress (10 200 N·m 2) as well as the power dissipated. Midler and Finn (1966)
reported that damage to shear sensitive protozoa cells in a uniform shear device showed a
two phase behavior: a rapid primary damage followed by a slow decline of viable cells.
This indicates that not only the magnitude of the hydrodynamic force but also the
exposure time should be taken into account when considering cell damage.
A diversity of methodologies have been used in the literature to assess the effects of
mechanical forces on cells; the most representative are presented in the Table 2.7.
Rheometers and parallel plates laminar flow chambers have been often used to study cell
sensitivity to shear stress, as the stress tensor can be defined with great precision and can be kept constant for long periods in those devices; Joshi et al (1996) presents an
extensive review of the use of rheometers on animal cells and outline a number of
limitations of these devices, including:
Limited or null oxygenation capacity restricts experiments to short periods not
representative of long cell cultures.
54 Settling of cells forces continuous mixing and also restricts experiments to short
intervals.
Normal stresses (i.e. extensional flow) can not be applied to cells, even though it
has been suggested that they can be as harmful or even more damaging to cells
than shear (Taylor, 1934; Garcia and Chalmers, 1994; Gregoriades et al., 2000).
Hydrodynamic stress fluctuations, normally found in turbulent flows, may be
even more damaging that steady shear stresses. Although laminar flow chambers
are suitable to apply such fluctuations, the ability of rheometers to exert them is
limited.
it is difficult to apply the results obtained within these devices to figure out proper
conditions for the cells inside bioreactors, given the much greater complexity of
the turbulent flow and the wide range of hydrodynamic forces that a single cell
may experience along the culture time.
Because of the previous limitations, studies performed in stirred tanks are often preferred from a design point of view (see Table 2.7). Unfortunately, the wide diversity on geometrical configuration of bioreactors and conditions used for such studies makes it extremely difficult to draw conclusions about what levels of hydrodynamic forces cells are able to withstand. As it was described previously, the quantitative determination of meaningful values of hydrodynamic forces in turbulent systems such as stirred tanks requires the measurement of velocity profiles along the whole bioreactor, which is a hard task to do because it requires expensive equipment and the measurements are limited to a few points inside the vessel. For that reason, authors normally resort to try to explain
55 their results based on a series of parameters, some of them based on fluid dynamics
theory, some of them strictly empiric. Some of those parameters are presented in Table
2.8 and a rationalization for their use will be presented in the next section.
Harmful effects other than cell lysis or death caused by hydrodynamic forces are often referred as sub lethal; studies focusing on sub lethal effects in bioprocessing equipment
have been limited and sometimes contradictory. Al Rubeai et al. (1990) found increased
glucose consumption and mitochondrial activity under intense agitation and in the
absence of air bubbles. Al Rubeai et al. (1995) found that in addition to cell death under
conditions of intensive agitation (1500 rpm), sub lethal effects included lost of microvilli
on the cell’s surface and changes in cycle distribution of the cell population. Al Rubeai et
al (1993) and Lakhotia et al. (1992) reported changes in DNA synthesis rate at sub lethal
conditions; in contrast, Passini and Gooche (1989) did not find any effect of sublethal
stress on DNA synthesis. Mufti and Shuler (1995) observed that human hepatoma cells
attached to microcarriers and grown in 50 mL spinners, responded to moderate levels of
agitation by inducing a cytochrome P450 monooxygenase (CYP1A1) activity; CYP1A1
is involved in the oxidation of arachidonic acid, a substance whose metabolism has been
observed to be altered by hydrodynamic stress in endothelial cells cultured in parallel plate flow chambers (McIntire et al., 1987).
56 2.5.4. Quantification of Cell Damage
The often contradictory nature of sub lethal hydrodynamic effects may be the result of
confusion as several different observed events are classified within the same category. As
an example, cells growing attached to a surface (surface culture or microcarriers) could
experience similar phenomena as those described for endothelial cells; however, cells
growing freely in suspended culture should be less likely to be affected by shear stresses
as the cells now are free to rotate, reducing considerably the deformation of the
cytoskeleton and varying continuously the surface exposed to the hydrodynamical forces.
Consistent with this concept, Ma et al. (2002) experimentally demonstrated that CHO
cells in suspension are four to five orders of magnitude less sensitive to a measure of
extensional and shear stress, energy dissipation rate, EDR, than CHO cells attached to
200 m microcarriers. Cherry and Papoutsakis (1988; 1989) invoked collision between
microcarriers and between microcarriers and solid parts of the bioreactor (impeller and
walls) as the main factors behind the increased sensitivity of cells growing in
microcarriers; as a logical consequence, new parameters (turbulence collision severity
and impeller collision severity) were proposed by the authors to correlate cell damage, parameters that can not be used properly in freely suspended cells as the inertia of free cells is much lower than that of microcarriers.
The core of the confusion in methodologies and parameters to asses cell damage is threefold. First, the difficulty to relate results obtained with laminar flow devices with those results obtained at the turbulent conditions prevalent in industrial bioreactors as a
57 result of our inability to properly characterize and quantify the bulk and local
hydrodynamic in the vessel; second, the lack of a biological model explaining how the
cells are affected by hydrodynamical forces and how that effect changes depending on
characteristics such as cell type, age and mode of growth (i.e. anchorage dependent or
suspended). Third, the lack of a relationship or mathematical model relating the first two points; such relationship should link the lethal and sub lethal effects of hydrodynamical
forces to both a hydrodynamic descriptor or parameter and to the time of exposure to it.
Having a single parameter (or at least a reduced set of them) that can be related to overall
yield, productivity, cell growth, etc., is highly desirable from an engineering point of
view to facilitate design and scale up of new processes. Garcia and Chalmers (1994)
suggested that “ideal” parameters should be independent of the geometric characteristics
of the culture system as cells do not “sense” the average conditions of the bioreactor but
the characteristics of the microenvironment surrounding each one of them at each time.
Consequently, criteria as mean power input, agitation rate, impeller tip speed, impeller based Reynolds number or integrated shear factor should not be used as they represent
average values derived from the particular geometric configuration of every system and because it has been shown that it is not possible to hold constant more than one of these parameters at the same time during scale up (Kossen, 1994).
While the idea proposed by Garcia and Chalmers (1994) results theoretically appealing,
its real implementation is currently an impossible task; first, the microenvironment
surrounding every cell is not only dependent on the position (three dimensional) but also
58 is time dependent as most industrial cultures are batch or fed batch and therefore their properties change constantly; second, the turbulent nature of the fluid dynamics inside an stirred bioreactor further complicates the panorama, as its description requires a second order stress tensor consisting of nine distinct stress vectors. Third, the behavior of every cell is likely to be history dependent: if two cells obtained from the same culture were placed in exactly identical conditions they probably will behave different as their previous history is expected not have been the same in the past. Finally, a precise characterization of turbulent phenomena is out of reach: from a theoretical point of view, the number of unknown variables is greater than the number of equations, so turbulent flow can not be solved mathematically from first principles; from a practical point of view, available techniques are limited to a very few points at a time (laser Doppler anemometry, constant temperature wire anemometry); increasing the number of points evaluated makes the cost prohibitive or imposes limitations in the resolution of the velocity profiles (particle tracking velocimetry). Therefore, in the end it is necessary to rely on some kind of average (time average, spatial average) or on a very gross distribution of some factor to describe hydrodynamic phenomena inside a bioreactor.
The most often used parameter for analysis of mixing phenomena is the Kolmogorov’s mixing microscale, η, which is a function of the ratio between the turbulence intensity
(represented by the local energy dissipation rate, ε, and the kinematic viscosity of the liquid, ν, is given by equation 2.20.
59 1 ν 3 4 η = (Eq. 2.20) ε
Kolmogorov’s mixing microscale (also denominated eddy size) was derived from classic dimensional analysis of turbulence by Andrey Kolmogorov in 1941; his theory of isotropic turbulence proposed that energy introduced into the liquid by the impeller undergoes a cascade from large, energy containing scales to small, energy dissipative scales. In the smallest scale, the fluid forms eddies where kinetic energy is ultimately converted into heat dissipation. As the Kolmogorov microscale depends only on fluid’s and flow’s properties and not on the system’s geometry or large scale flow patterns, it is a good candidate for general parameter to correlate cell damage. In fact, it has been proposed that cell damage occurs whenever the eddy length η, reaches the size of a freely suspended cell or the size of a microcarrier, depending on the culture type. Good predictions have been made in small scale vessels; however, this parameter suffers from several weaknesses: it requires turbulent flow (so it can not be properly applied to laminar flow); additionally, it has been shown experimentally that turbulence is not really isotropic although one can always assume “local” isotropy. Finally, experimental determination of ε is somewhat difficult; this last requisite usually has been overlooked by assuming that the energy transferred by the impeller is drawn into a limited volume around the impeller, although the precise size of that volume tends to be “determined” by every author so it matches his own results; since experimental measurements indicate that the energy dissipation is more concentrated in the very small volumes associated with the trailing vortices emanating from the impellers, as it was discussed before, Aloi and
60 Cherry (1996) obtained a fairly good agreement between the results obtained in capillary
flow and those obtained in bioreactors.
In contrast to the Kolmogorov’s mixing microscale, τηε Εnergy Dissipation Rate (EDR)
is a scalar parameter that is independent of the flow regime (turbulent/laminar) and
accounts for both shear and extensional components of three dimensional flow. EDR
represents the rate at which work is done on a fluid element or, for the purpose of this
review, a cell (Clay, 1997). The EDR has been widely used to quantify local mixing performance in stirred tanks (Kresta, 1998) as well as laminar flow devices (Mollet et al.,
2004; Mollet et al., 2007); even Kolmogorov’s mixing microscale is based on the local energy dissipation rate. Therefore, it seems quite appealing to use this parameter to correlate the hydrodynamical cell damage in diverse devices. Although EDR can not be measured directly, it can be calculated easily from Equations (2.17) and (2.18) provided that the broth behaves like an incompressible, Newtonian fluid (common situation in animal cell culture where the medium has water like properties) and that the velocity vector has been accurately determined. This last requirement, however, limits the application of EDR to systems where the determination of the velocity vector U has been done experimentally or it can be determined analytically (laminar flow).
Ma et al. (2002), designed and manufactured a microfluidic device in which suspended cells could be exposed to acute, high levels of EDR in a laminar flow environment; the advantages of this device include very well defined hydrodynamical characteristics and the possibility of emphasize the potentially more damaging extensional flow, although
61 some degree of shear is present. Using such device, they subjected an insect cell line (Sf
9), a Chinese Hamster Ovary cell line (CHO), a mouse hybridoma cell line (HB 24) and a
human breast cancer cell line (MCF 7) to single exposures of EDR and measured the
release of the cytoplasmic enzyme lactate dehydrogenase (LDH) as an indication of cell
membrane rupture. Every one of the cell lines studied showed significant cell lysis beginning around an EDR of 107 W·m 3 as observed in Figure 2.13. This is in agreement with previous results from Zhang and Thomas (1993), who needed average energy dissipation rates of 108 – 109 W·m 3 to completely disrupt TB/C3 hybridoma and NS1
myeloma cells when passing in turbulent flow through capillary tubes.
To better express their results, Ma et al. (2002) developed a one dimensional plot whose
improved version is presented in Figure 2.14. As it can be appreciated, the values of EDR
required to cause significant lysis in industrial cells are at least two orders of magnitude
higher than the maximum EDR measured under normal conditions in bioreactors. Mollet
et al. (2007) developed an improved, autoclavable second generation microfluidic device
to further investigate if besides the evident levels of necrosis already reported, the effect
of single exposures to high levels of EDR could lead to apoptosis. In addition to being
consistent with the previous work by Ma et al. (2002) in the levels of necrosis obtained
after passing the cells once through the micro device, Mollet et al. (2007) observed that
only a small fraction of Chinese Hamster Ovary cells (CHO K1) became apoptotic when
exposed to sublethal levels of EDR.
62 In a related subject, Mollet et al. (In Press) used the improved microfluidic device as well
as computer fluid dynamics (CFD) commercial software to simulate the single passage of
cells through the nozzle of a commercial fluorescence activated flow sorter (FACS). In
the FACS the cells experience high levels of EDR as a result of hydrodynamic focusing
in the nozzle, with levels of cell necrosis on the order of 20 to 35% using CHO cells
under typical operating conditions. They also found that non adapted cells (THP1) may be much more sensitive to hydrodynamic stress than adapted, industrial cell lines (see
Figure 2.13).
2.5.5. Detrimental effects of sparging
It is a well known fact the dramatic reduction in cell number and viability of animal cells
caused by air bubbling in sparged reactors (Handa et al., 1987; Oh et al., 1989; Kunas and Papoutsakis, 1990a; Kioukia et al., 1996). Even under conditions of unusually high
stirring speeds, Oh et al. (1989) and Kunas and Papoutsakis (1990a, 1990b) showed that as long as there is no air introduction into the culture medium (i.e. sparging or bubble entrainment from the central vortex), hybridomas could withstand stirring speeds as high as 600 rpm without detectable cell damage.
The specific cause responsible for cell damage as a result of cell bubble interactions may be attributed to four different events: bubble formation and detachment at the sparger, bubble break up and coalescence in the impeller trailing vortices, bubble rising through
63 the fluid or bubble disengagement at the surface of the culture. In the next lines we present a brief summary of the results of research on each of these potential events.
Significant cell damage occurring near the sparger was suggested by Tramper et al.
(1987) who estimated that shear stress produced as a result of bubble formation and
detachment could be up to one order of magnitude higher than that of bursting bubbles.
Murhammer and Goochee (1990) noticed an inverse correlation between the pressure
drop across the sparger and the net growth of insect cells in airlift reactors, and suggested
that the damage could be concentrated in the sparger zone as a result of the turbulent
oscillation of the detaching bubbles or the fast liquid movement to fill in the space
vacated by them.
In a series of insightful experiments under identical aeration and agitation conditions, Oh
et al. (1992) showed that bubble break up as a result of the interaction of sparged air with
the impeller is significant for cell damage. By comparing the results of placing the
sparger in several positions, the authors showed that sparging air directly under the
impeller led to an increase in the number of smaller bubbles and also to an increment in
cell death, although it is difficult to say if it was caused by damage in the impeller region
or just because increased bubble bursting in the medium air interface.
Kioukia et al. (1992) grew hybridoma cells in a 2.5 L reactor stirred with a Rushton
turbine and oxygenated only by surface aeration. In experiments with baffles in the
reactor, succesful growth and simmilar kinetic results were obtained at all speeds up to
64 400 rpm. In contrast, in experiments without baffles the peak cell density and growth rate
decreased with increased speeds and cell death occurred from the start of the batch
culture even though the average energy dissipation rate was only 20% of that for baffled
conditions. The authors argued that the cause for this behavior was the entrainment of air bubbles in the bulk of the liquid from and their further disengagement through the vortex
surface; furthermore, since the bubbles never reached the impeller region, the damage
could not be related to bubble break up/coalescence in the impeller and only was
associated with the bubbles bursting at the medium air interface.
Tramper et al. (1987) and Glasgow et al. (1992) showed that values of shear stress
derived from rising bubbles are extremely low (~0.1 Pa) and should, therefore, be
neglected. To confirm these results, Handa Corrigan et al. (1989) and Jöbses et al. (1991) introduced air in bubble colums of different sizes. With all other variables being constant, increasing column height, and thereby increasing the distance a bubble had to travel to rise to the top of the column, did not result in a reduction of viability for several cell types tested (hybridoma, BHK 21, myeloma and lymphoblastoid cell lines).
Several researchers suggest that bubble induced cell damage takes place only when a bubble, at the gas liquid interface, ruptures (Handa et al.1987; Bavarian et al., 1991;
Trinh et al., 1994). Phenomenologically, a bubble at a gas liquid interface rises to a particular height above the flat liquid surface while the remaining part is submerged; the
extent of the submerged part depends on the diameter. During rupture, the liquid drains by gravity from the film covering the raised part of the bubble up to a point where the
65 film is so thin that a hole develops on the top and the bubble burst; then, the liquid
rapidly moves down the walls of the bubble cavity until the liquid reaches the bottom of
the cavity; the resulting impact creates simultaneously upward and downward jets.
Several studies indicate that this process creates a very high energy dissipation rate over the very short period of time in which the rupture occurs. This bursting process was qualitatively presented by MacIntyre (1972), in a series of cinephotography derived profiles of the liquid gas interface during the bubble break up event.. Subsequently,
Boulton Stone and Blake (1993) and Garcia Briones et al. (1994) reported the results of
computer simulations of bubble break up with simulated profiles very close to those photographed by MacIntyre (1972); their model indicates that the film at the top of the bubble retreats at speeds as high as ~8 m·s 1. Of more significance, the calculated
maximum energy dissipation rate is many orders of magnitude higher than those typically
generated by an impeller. Additionally, such maximum EDRs are a function of the bubble radius being larger for smaller bubbles, indicating that smaller bubbles are potentially more damaging to cells. Interestingly, this result has been observed
experimentally in several different studies (Handa Corrigan et al., 1989; Jöbses et al.,
1991).
The maximum energy dissipation rates (W·m 3) calculated by García Briones et al.
(1994) during bubble break up ranged from 9.52×107 and 1.66×107 for bubbles with diameters of 0.77 and 1.70 mm, respectively, to 9.4×104 for a 6.32 mm bubble.
Inspection of Figure 2.14 indicates that these higher EDR values are well within the range of EDR values that were able to kill cells in the flow contraction device. This consistency between the simulated values of EDR during a bubble rupture, the values of
66 EDR shown to kill cells in the flow contraction device and the significantly lower EDR
values in the impeller region provide an overall consistent picture that cell damage occurs
due to bubble rupture at the liquid surface and the foam layer. The remaining question is
whether enough cells are killed due to bubble bursting to account for all cell death one
can observe in a sparged culture.
To address this question a statistically significant study showed that on average ~103 cells
are killed per 2 mm bubble burst (Trinh et al., 1994) and the cell concentration collected from the upward jet is twice the bulk concentration in the culture. This result is in agreement with observations by Chalmers and Bavarian (1991) and García Briones and
Chalmers (1992), who using video microscopy detected cell attachment and accumulation on gas liquid interfaces, especially in rising bubbles and the foam layer in the surface of the liquid in the reactor.
Parker and Barsom (1970) reported that the chemical composition of the film surrounding air liquid interfaces contained more hydrophobic materials compared to the bulk water; therefore, the phenomena of cell attachment to such film (microlayer) could potentially be explained as a result of the hydrophobic characteristics of the cell membrane.
Chattopadhyay et al. (1995) attempted to explain thermodynamically the cell adhesion to the gas liquid interface and suggested a model for the variation of free energy as a function of the surface tension of the interfaces involved (gas liquid cells). According to their model, modifying the gas liquid surface tension via chemical additives may create a condition where the change in free energy becomes positive and so the adherence stops
67 being spontaneous. While a number of different surface active agents have been studied,
including serum, dextran, hydrodyethyl starch, polyethylene glycol or polyvinyl alcohol,
Pluronic F 68 continues to be the additive of choice. Pluronic F 68, is a block copolymer
of Poly(oxyethylene) and poly(oxypropilene) and it is added to the culture medium in
concentrations of 0.5 3 kg·m 3 (Chisti, 1999).
In addition to reduce or inhibit cell adhesion to the microlayer (Chalmers and Bavarian,
1991; García Briones and Chalmers, 1992), by changing either the bubble film (Jordan et
al., 1994; Michaels et al., 1995) or the cell membrane properties (Wu, 1996; Wu et al.,
1997), other mechanisms have been proposed to explain the reduction or virtual
elimination by surfactants agents of cell death as a result of bubbling; they include:
1. Interaction of the surfactant with the cell membrane, reducing its fluidity and
rendering it stronger to mechanical stresses. (Zhang et al., 1992 ; Ramirez and
Mutharasan, 1990b).
2. Nutritive effects of the surfactant on the cells. As mentioned by Ma et al. (2006),
the short time needed for detection of protective effects after addition of most
chemical substances suggests that this mechanism of protection is not important.
In spite of the success of Pluronic F 68, a word of caution is in order: as cell densities increase, the effectiveness of F 68, even at high concentration, diminishes. A recent study by Ma et al. (2004) demonstrated that even at high concentrations, a significant number
of cells can be retained in the foam layer. Figure 2.15 is a figure from that report and as it
68 can be observed, at cell concentrations greater than 107 cells·ml 1 (which can be routinely obtained in industrial cultures) and relatively high F 68 concentration, more than 200 cell/bubble are removed into the foam layer.
2.6. SUMMARY
As it has been reviewed above, much like many other technological advancements, the practical and/or commercial development of processes based on animal cells has preceded the more fundamental understanding. However, significant progress has been
made on not only how those processes work but also on how to design and operate them based on either fundamental understanding or well justified correlations. Unlike in the early 1980s when industrialist had to scale up based on “rules of thumb”, it is now possible to design and operate bioreactors based on a significant amount of published, peer reviewed studies that, in general, present a consistent message.
Because of the relatively low mass transfer demands of, at least, the majority of past and current processes, industrial cell cultures have been “forgiving” of poor mixing conditions. However, just as fungal processes to make antibiotics have made tremendous increases in productivity over the last 60 years through both strain development and high density culture systems, there is every reason to believe the same will happen with industrial animal cell cultures. Therefore, it is entirely possible that the “forgiving” process of today will not be good enough for processes of the future. At some point, a situation will be reached when Pluronic F 68 will lose its effectiveness as the cell density
69 increases and the purely hydrodynamic conditions detrimental to suspended cells will need to be determined in order to adjust mixing conditions so the culture can be carried out successfully.
70 2.7. REFERENCES
Aloi, L. E. and Cherry, R. S. (1996). Cellular response to agitation characterized by energy dissipation at the impeller tip. Chemical Engineering Science. 51: 1523 1529.
Al Rubeai, M.; Emery, A. N.; Chalder, S. and Goldman, M. H. (1993). A flow cytometric study of hydrodynamic damage to mammalian cells. Journal of Biotechnology. 31(2): 161 177.
Al Rubeai, M.; Oh, S. K. W.; Musaheb, R. and Emery, A. N. (1990). Modified cellular metabolism in hybridomas subjected to hydrodynamic and other stresses. Biotechnology Letters. 12: 323 328.
Al Rubeai, M.; Singh, R. P.; Goldman, M. H. and Emery, A. N. (1995). Death mechanisms of animal cells in conditions of intense agitation. Biotechnology and Bioengineering. 45: 463 472.
Amanullah, A.; Serrano Carreon, L.; Castro, B.; Galindo, E. and Nienow, A. W. (1998). The influence of impeller type in pilot scale xanthan fermentations. Biotechnology and Bioengineering. 57(1):95 108.
Asay, T. and Kono, T. (1982). Estimation of oxygen absorption coefficient and power consumption in a stirred tank fermentor. Journal of Fermentation Technology. 60(3):265 268.
Augenstein, D. C. ; Sinskey, A J. and Wang, D. I. C. (1971). Effect of shear on the death of two strains of mammalian tissue cells. Biotechnology and Bioengineering. 13:409 418.
Aunins, J. G.; Woodson, B. A. Jr.; Hale, T. K. and Wang, D. I. C. (1989). Effects of paddle impeller geometry on power input and mass transfer in small scale animal cell culture vessels. Biotechnology and Bioengineering. 34(9):1127 1132.
Backer, M. P.; Metzger, L. S.; Slaber, P. L.; Nevitt, K. L. and Boder, G. B. (1988). Large scale production of monoclonal antibodies in suspension culture. Biotechnology and Bioengineering. 32:993 1000.
71 Barnabé, N. and Butler, M. (2000). The effect of glucose and glutamine on the intracellular nucleotide pool and oxygen uptake rate of a murine hybridoma. Cytotechnology. 34: 47–57.
Bavarian, F.; Fan, L. S. and Chalmers, J. J. (1991). Microscopic visualization of insect cell bubble interactions. I: Rising bubbles, air medium interface and the foam layer. Biotechnology Progress. 7:140 150.
Birch, J.R.; Boraston, R. and Wood, L. (1985). Bulk production of monoclonal antibodies in fermenters. Trends in Biotechnology. 3:162–166.
Birch, J.R.; Thompson, P. W; Boraston, R.; Oliver, S. and Lambert, K. (1987). The large scale production of monoclonal antibodies in airlift fermenters. In: Plant and Animal Cells: Process Possibilities. Webb, C. and Mavituna, F. (Eds). Ellis Horwood. Chichester, UK.
Bird, R. B.; Stewart, W. E. and Lightfoot, E. N. (2006). Transport Phenomena. 2nd Ed. John Wiley & Sons, New York.
Bluestein, M. and Mockros, L. F. (1969). Hemolysis effects of energy dissipation in flowing blood. Medical and Biological Engineering and Computing. 7(1):1 6.
Boraston, R.; Thompson, P. W.; Garland, S. and Birch, J. R. (1983). Growth and oxygen requirement of antibody producing mouse hybridoma cells in suspension culture. Developments in Biological Standardization. 55:103 111.
Born, C.; Zhang, Z.; Al Rubeai, M. and Thomas, C. R. (1992). Estimation of disruption of animal cells by laminar shear stress. Biotechnology and Bioengineering. 40(9): 1004 1010.
Boulton Stone, J. M. and Blake, J. R. (1993). Gas bubbles bursting at a free surface. Journal of Fluid Mechanics. 254:103 111.
Brodkey, R. S. (1995). The Phenomena of Fluid Motions. Dover Publishing, New York.
72 Bryant, J. C.; Schilling, E. L.; Earle, W. R. and Evans, V. J. (1960). Massive fluid suspension cultures of certain mammalian tissue cells. III. Exploratory supplementation experiments with chemically defined medium NCTC 109. Journal of the National Cancer Institute. 24:859 871.
Butler, M. (2005). Animal cell cultures: recent achievements and perspectives in the production of biopharmaceuticals. Applied Microbiology and Biotechnology. 68: 283– 291.
Carrel, A. (1912). On the permanent life of tissues outside the organism. Journal of Experimental Medicine. 15(5):516 528.
Carvalhal, A. V.; Marcelino, I. and Carrondo, M. J. T. (2003). Metabolic changes during cell growth inhibition by p27 overexpression. Applied Microbiology and Biotechnology. 63:164–173.
Chalmers, J. J. and Bavarian, F. (1991). Microscopic visualization of insect cell bubble interactions II: The bubble film and bubble rupture. Biotechnology Progress. 7:151 158.
Chattopadhyay, D.; Rathman, J. and Chalmers, J. J. (1995a). The protective effect of specific medium additives with respect to bubble rupture. Biotechnology and Bioengineering. 45:473 480
Chattopadhyay, D.; Rathman, J. and Chalmers, J. J. (1995b). Thermodynamic approach to explain cell adhesion to gas liquid interfaces. Biotechnology and Bioengineering. 48:649 658.
Chen, A. K.; Latz, M. I. and Frangos, J. A. (2003). The use of dinoflagellate bioluminescence to characterize cell stimulation in bioreactors. Biotechnology and Bioengineering. 83(1):93 103.
Cherry, R. S. and Papoutsakis, E. T. (1989). Hydrodynamic effects on cells in agitated tissue culture reactors. Bioprocess Engineering. 1:29 41.
Cherry, R. S. and Papoutsakis, E. T. (1988). Physical mechanisms of cell damage in microcarrier bioreactors. Biotechnology and Bioengineering. 32:1001 1014.
73
Chien, S. (2007). Mechanotransduction and endothelial cell homeostasis: the wisdom of the cell. American Journal of Physiology Heart and Circulation Physiology. 292:1209 1224.
Chisti, Y. (1993). Animal cell culture in stirred bioreactors: observations in scale up. Bioprocess Engineering. 9:191 196.
Chisti, Y. (1999). Shear sensitivity. In: M.C. Flickinger and S.W. Drew, (eds). Encyclopedia of Bioprocess Technology: Fermentation, Biocatalysis, and Bioseparations. Vol. 5. John Wiley & Sons. New York. Pp. 2379 2406.
Chisti, Y. (2001). Hydrodynamic damage to animal cells. Critical reviews in biotechnology . 21(2): 67 110.
Cho, M. H. and Wang, S. S. (1988). Enhancement of oxygen transfer in hybridoma cell culture by using a perfluorocarbon as an oxygen carrier. Biotechnology Letters. 10:855 860.
Chu, L. and Robinson D. K. (2001). Industrial choices for protein production by large scale cell culture. Current Opinion in Biotechnology. 12:180–187.
Clay, J. D. (1997). Molecular degradation of polymer solutions in fast transient extensional flow. Ph.D. Thesis. Chemical Engineering Department. The Ohio State University.
Costes, J. and Couderc, J.P. (1988). Study by laser Doppler anemometry of the turbulent flow induced by a Rushton turbine in a stirred tank: Influence of the size of the units II. Spectral analysis and scales of turbulence. Chemical Engineering Science. 43(10):2765 2772.
Croughan, M. S.; Hamel, J. F. and Wang, D. I. C. (1987). Hydrodynamic effect on animal cells grown in microcarrier cultures. Biotechnology and Bioengineering. 29: 130 141.
74 Croughan, M. S.; Sayre, E. S. and Wang, D. I. C. (1989). Viscous reduction of turbulent damage in animal cell cultures. Biotechnology and Bioengineering. 33: 862 872.
Cui, Y. Q. ; van der Lans, R. G. J. M. and Luyben, K. C. A. M. (1996). Local power uptakes in gas liquid systems with single and multiple Rushton turbines. Chemical Engineering Science. 51(11): 2631 2636.
Cutter, L. A. (1966). Flow and Turbulence in a Stirred Tank. AIChE Journal. 12:35 45.
Deshpande, R. R. and Heinzle, E. (2004). On line oxygen uptake rate and culture viability measurement of animal cell culture using microplates with integrated oxygen sensors. Biotechnology Letters. 26: 763–767.
deZengotita, V. M.; Kimura, R. and Miller, W. M. (1998). Effects of CO2 and osmolality on hybridoma cells: growth, metabolism and monoclonal antibody production. Cytotechnology. 28:213 227.
deZengotita, V. M.; Miller, W. M., Aunins, J. G. and Zhou, W. (2000). Phosphate Feeding Improves High Cell Concentration NS0 Myeloma Culture Performance for Monoclonal Antibody Production. Biotechnology and Bioengineering. 69(5):566 576.
Discher, D. E.; Mohandas, N. and Evans, E. A. (1994). Molecular maps of red cell deformation: hidden elasticity and in situ connectivity. Science. 266(5187):1032 1035.
Dorresteijn, R. C.; Degooijer, C. D.; Tramper, J. and Beuvery, E. C. (1994). A method for simultaneous determination of solubility and transfer coefficient of oxygen in aqueous media using off gas mass spectrometry. Biotechnology and Bioengineering. 43:149 154.
Drapeau, D.; Luan, Y. T.; Whiteford, J. C.; Lavin, D. P. and Adamson, S. R. (1990). Cell culture scale up in stirred tank reactors. Presented at the Annual Meeting of Society of Industrial Microbiology, Orlando, FL. USA.
Ducommun, P.; Bolzonella, I.; Rhiel, M.; Pugeaud, P.; von Stockar, U. and Marison, I. W. (2001). On Line Determination of Animal Cell Concentration. Biotechnology and Bioengineering. 72(5): 515 522
75 Eyer K.; Oeggerli, A. and Heinzle, E. (1995). On line gas analysis in animal cell cultivation: II. Methods for oxygen uptake rate estimation and its application to controlled feeding of glutamine. Biotechnology and Bioengineering. 45: 54 62.
Fife, J. P.; Derksen, R. C.; Ozkan, H. E.; Grewal, P. S.; Chalmers, J. J. and Krause, C. R. (2004). Evaluation of a contraction flow field on hydrodynamic damage to enthomopathogenic nematodes – A biological pest control agent. Biotechnology and Bioengineering. 86(1):96 107.
Fong S. S.; Joyce A. R. and Palsson, B. Ø. (2005). Parallel adaptive evolution cultures of Escherichia coli lead to convergent growth phenotypes with different gene expression states. Genome Research. 15:1365 1372.
Galindo, E. and Nienow, A. W. (1993). Performance of the Scaba 6SRGT agitator in mixing of simulated xanthan gum broths. Chemical Engineering and Technology. 16(2):102 108.
Garcia, M. A. and Chalmers, J. J. (1994). Flow parameters associated with hydrodynamic cell injury. Biotechnology and Bioengineering. 44:1089 1098.
García Briones, M. A. and Chalmers, J. J. (1992). Cell bubble interactions: Mechanisms of suspended cell damage. Annals of the New York Academy of Sciences. 665:219 229.
Garcia Briones, M. A.; Brodkey, R. S. and Chalmers, J. J. (1994). Computer simulations of the rupture of a gas bubble at a gas liquid interface and its implications in animal cell damage. Chemical Engineering Science. 49(14):2301 2320.
Gimenes, M. A. P.; Carlos, L. C. S.; Faria L. F. F. and Pereira, N. Jr. (2002). Oxygen uptake rate in production of xylitol by Candida guilliermondii with different aeration rates and initial xylose concentrations. Applied Biochemistry and Biotechnology. 98 100(1 9):1049 1059.
Glasgow, L. A.; Hua, J.; Yiin, T. Y. and Erickson, L. E. (1992). Experiment studies of interfacial phenomena in sparged reactors. In: Tatterson, G B. and Calabrese, R. V. (Eds.). Process Mixing: chemical and biological applications. AIChE Symposium Series. Vol. 88, Issue 286. Pp.103 108. American Institute of Chemical Engineers. New York.
76 Gódia, F. and Cairó, J. J. (2006). Cell metabolism. In: S.S. Ozturk and W S. Hu (Eds). Cell culture technology for pharmaceutical and cell based therapies. Pp. 81 112. CRC Press. Boca Raton, FL., USA.
Goldblum, S.; Bae, Y. K.; Hink, W. F. and Chalmers, J. J. (1990). Protective effect of methylcellulose and other polymers on insect cells subjected to laminar shear stress. Biotechnology Progress. 6:383 390.
Goldman, W. H. (2000). Mechanical manipulation of animal cells: cell indentation. Biotechnology letters. 22: 431 435.
Goto, T.; Mochizuki, G. and Kikuchi, K. I. (2001a). Perfluorocarbon mediated aeration applied to recombinant protein production by virus infected insect cells. Biochemical Engineering Journal. 7 (1):69 78.
Goto, T.; Mochizuki, G. and Kikuchi, K. I. (2001b). A novel column fermentor having a wetted wall of perfluorocarbon as an oxygen carrier. Biochemical Engineering Journal. 8(2):165 169.
Graf, R.; Apenberg, S.; Freyberg, M. and Friedl, P. (2003). A common mechanism for the mechanosensitive regulation of apoptosis in different cell types and for different mechanical stimuli. Apoptosis. 8: 531 538.
Gray, D. R.; Chen, S.; Howarth, W.; Inlow, D. and Maiorella, L. (1996). CO2 in large scale and high density CHO cell perfusion culture. Cytotechnology. 22:65 78.
Gregoriades, N.; Clay, J.; Ma, N.; Koelling, K. and Chalmers, J. J. (2000). Cell damage of microcarrier cultures as a function of local energy dissipation created by a rapid extensional flow. Biotechnology and Bioengineering. 69:171 182.
Griffiths, J. B. (2000). Animal cell products, overview. In: Spier RE (ed). Encyclopedia of cell technology, Vol. 1. Pp 71–76. Wiley, New York, USA.
Guck, J.; Ananthakrishnan, R.; Mahmood, H.; Moon, T. J.; Cunningham, C. C. and Käs, J. (2001). The optical stretcher: a novel laser tool to micromanipulate cells. Biophysical Journal. 81: 767 784.
77
Haga, J. H.; Li, Y. S. J. and Chien, S. (2007). Molecular basis of the effects of mechanical stretch on vascular smooth muscle cells. Journal of biomechanics. 40: 947 960.
Handa, A.; Emery, A. N. and Spier, P. E. (1987). On the evaluation of gas liquid interfacial effects on hybridoma viability in bubble column bioreactors. Developments in Biological Standardization. 66: 241 253.
Handa Corrigan, A.; Emery, A. N. and Spier, R. E. (1989). Effects of gas liquid interfaces on the growth of suspended mammalian cells: mechanisms of cell damage by bubbles. Enzyme and Microbial Technology. 11: 230 235.
Harris, R.; Ng, K. W. and Wraith, A. E. (2005). Spargers for controlled bubble size by means of the multiple slot disperser (MSD). Chemical Engineering Science. 60: 3111 – 3115.
Harrison, R.G. (1907). Observations on the living developing nerve fiber. Proceedings Proceedings of the Society for Experimental Biology and Medicine. 4:140–3.
Hempel, D. C. (1988). Fundamentals of scale up for biotechnological processes in stirred fermentors. In: Finn, R. K.; Präve, P.; Schlingmann, M.; Crueger, W.; Esser, K.; Thaver, R. and Wagner, F. (eds.). Biotechnology Focus 1: Fundamentals, applications, information. Hanser Publishers. Munich. Pp. 51 94.
Hensler, W.T. and Agathos, S.N. (1994). Evaluation of monitoring approaches and effects of culture conditions on recombinant protein production in baculovirus infected insect cells. Cytotechnology. 15:177–186.
Hesse, F.; Ebel, M.; Konisch, N.; Sterlinski, R.; Kessler, W. and Wagner, R. (2003) Comparison of a Production Process in a Membrane Aerated Stirred Tank and up to 1000 L Airlift Bioreactors Using BHK 21 Cells and Chemically Defined Protein Free Medium. Biotechnology Progress. 19: 833 843.
78 Hiller, G. W.; Aeschlimann, A. D.; Clark, D. S. and Blanch, H. W.(1991). A kinetic analysis of hybridoma growth and metabolism in continuous suspension culture on serum free medium. Biotechnology and Bioengineering. 38:733 741.
Hirtenstein, M. D. and Clark, J. M. (1980). Critical parameters in the microcarrier culture of animal cells. In: Richards, R. and Rajan, K. (eds). Tissue Culture in medical research. Vol. 2. Pergamon Press, Oxford.
Hoh, J. H. and Schoenenberger, C. A. (1994). Surface morphology and mechanical properties of MDCK monolayers by atomic force microscopy. Journal of Cell Science. 107(5): 1105 1114.
Hu, W.S.; Meier, J. and Wang, D.I.C. (1986). Use of Surface Aerator to Improve Oxygen Transfer in Cell Culture. Biotechnology and Bioengineering. 28 : 122 125.
Hu, W.; Gladue, R.; Hansen, J.; Wojnar, C. and Chalmers, J. J. (2007). The Sensitivity of the Dinoflagellate Crypthecodinium cohnii to Transient Hydrodynamic Forces and Cell Bubble Interactions. Biotechnology Progress. 23:1355 1362.
Hudcova, V.; Machon, V. and Nienow, A. W. (1989). Gas liquid dispersion with dual Rushton turbine impellers. Biotechnology and Bioengineering. 34:617 628.
Jan, D. C. H.; Petch, D. A.; Huzel, N. and Butler, M. (1997). The effect of dissolved oxygen on the metabolic profile of a murine hybridoma grown in serum free medium in continuous culture. Biotechnology and Bioengineering. 54: 153–164.
Jöbses, I.; Martens, D. and Tramper, J. J. (1991). Lethal events during gas sparging in animal cell culture. Biotechnology and Bioengineering. 37: 484 490.
Jones, W. R.; Ting Beall, H. P.; Lee, G. M.; Kelley, S. S.; Hochmuth, R. M. and Guilak, F. (1999). Alterations in the Young’s modulus and volumetric properties of chondrocytes isolated from normal and osteoarthritic human cartilage. Journal of Biomechanics. 32:119 127.
Jordan, M.; Sucker, H.; Einsele, A.; Widmer, F. and Eppenberger, H. M. (1994). Interactions between animal cells and gas bubbles: the influence of serum and Pluronic F
79 68 on the physical properties of the bubble surface. Biotechnology and Bioengineering. 43(6):446 454.
Joshi, J. B. ; Elias, C. B. and Patole, M. S. (1996). Role of hydrodynamic shear in the cultivation of animal, plant and microbial cells. Chemical Engineering Journal. 62:121 141.
Ju, L. K. and Armiger, W. B. (1992). Use of perfluorocarbon emulsion in cell culture. BioTechniques. 12:258 263.
Jüsten, P; Paul, G. C., Nienow, A. W. and Thomas, C. R. (1998) Dependence of Penicillium chrysogenum growth, morphology, vacuolation and productivity in fed batch fermentations on impeller type and agitation intensity. Biotechnology and Bioengineering. 59 (6): 762 775.
Kamen, A.A.; Chavarie, C.; Andre, G. and Archambault, J. (1992). Design parameters and performance of a surface baffled helical ribbon impeller bioreactor for the culture of shear sensitive cells. Chemical Engineering Science. 27: 2375 2380.
Keane, J. T.; Ryan, D. and Gray, P. P. (2003). Effect of shear stress on expression of a recombinant protein by Chinese hamster ovary cells. Biotechnology and Bioengineering. 81(2):211 220.
Kioukia, N.; Nienow, A. W.; Al Rubeai, M. and Emery, A. N. (1992). The impact of fluid dynamics on the biological performance of free suspension animal cell culture: further studies. Food and Bioproducts Procesing (Transactions of the Institution of Chemical Engineers. Part C.). 70:143 148.
Kioukia, N.; Nienow, A. W.; Al Rubeai, M. and Emery, A. N. (1996). The influence of agitation and sparging on the growth rate and infection of insect cells in bioreactors and a comparison with hybridoma culture. Biotechnology Progress. 12: 779 785.
Kompala, D. S. and Ozturk, S. S. (2006). Optimization of high cell density perfusion reactors. In: S.S. Ozturk and W S. Hu (Eds). Cell culture technology for pharmaceutical and cell based therapies. Pp. 387 416. CRC Press. Boca Raton, FL., USA.
80 Kossen, N. W. F. (1994). Scale up. In: Galindo, E. and Ramirez, O. T. (eds.). Advances in Bioprocess Engineering. Kluwer Academic Publishers. Dordrecht. Pp. 53 65.
Kresta, S. (1998). Turbulence in stirred tanks: anisotropic, approximate and applied. Canadian Journal of Chemical Engineering. 76:563 575.
Kresta, S. M., Mao, D. and Roussinova, V. (2006). Batch blend time in square stirred tanks. Chemical Engineering Science. 61(9):2823 2825.
Kretzmer, G. (2002). Industrial processes with animal cells. Applied Microbiology and Biotechnology. 59:135–142.
Kulkarni, A. A. and Joshi, J. B. (2005). Bubble Formation and Bubble Rise Velocity in Gas Liquid Systems: A Review. Industrial & Engineering Chemistry Research. 44: 5873 5931.
Kunas, K. T. and Papoutsakis, E. T. (1990a). Damage mechanisms of suspended animal cells in agitated bioreactors with and without bubble entrainment. Biotechnology and Bioengineering. 36:476 483.
Kunas, K. T. and Papoutsakis, E. T. (1990b). The protective effect of serum against hydrodynamic damage of hybridoma cells in agitated and surface aerated bioreactors. Journal of Biotechnology. 15:57 70.
Kunkel, J. P.; Jan, D. C. H.; Jamieson, J. C. and Butler, M. (1998). Dissolved oxygen concentration in serum free continuous culture affects N linked glycosylation of a monoclonal antibody. Journal of Biotechnology. 62:55–71.
Lakhotia, S. K. ; Bauer, K. D. and Papoutsakis, E. T. (1992). Damaging agitation intensities increase DNA synthesis rate and alter cell cycle phase distributions of CHO cells. Biotechnology and Bioengineering. 40: 978 990.
Langheinrich, C. and Nienow, A. W. (1999). Control of pH in large scale, free suspension animal cell bioreactors: alkali addition and pH excursions. Biotechnology and Bioengineering. 66: 171–179.
81 Langheinrich, C.; Nienow, A. W.; Eddleston, T.; Stevenson, N. C.; Emery, A. N.; Clayton, T. M.; Slater, N. K. H. (2002). Oxygen transfer in stirred bioreactors under animal cell culture conditions. Food and bioproducts processing. 80(1): 39 44.
Lavery, M. and Nienow, A.W. (1987). Oxygen Transfer in Animal Cell Culture Medium. Biotechnology and Bioengineering. 30(3): 368 – 373.
Lehmann, J.; Vorlop, J. and Buntemayer, H. (1988). Bubble free reactors and their development for continuous culture with cell recycle. In: Spier R.E. and Griffiths J.B. (eds), Animal Cell Biotechnology. Vol. 3. Academic Press Ltd., London, UK, pp. 221– 237.
Lehoux, S.; Castier, Y. and Tedgui, A. (2006). Molecular mechanisms of the vascular responses to haemodynamic forces. Journal of Internal Medicine. 259: 381 392.
Li, C. and Xu, Q. (2007). Mechanical stress initiated signal transduction in vascular smooth muscle cells in vitro and in vivo. Cellular Signaling. 19:881 891.
Li, Y. S. J.; Haga, J. H. and Chien, S. (2005). Molecular basis of the effects of shear stress on vascular endothelial cells. Journal of biomechanics. 38: 1949 1971.
Lin, A. A. and Miller, W. M. (1992). Modulation of glutathione level in CHO cells. Effects of oxygen concentration and previous exposure to hypoxia. Annals of the New York Academy of Sciences. 665:117 126.
Linek, V.; Moucha, T. and Sinkule, J. (1996). Gas–liquid mass transfer in vessel stirred with multiple impellers I: Gas–liquid mass transfer characteristics in individual stages. Chemical Engineering Science. 51(12): 3203 3212.
Lowe, K. C.; Davey, M. R. and Power, J. B. (1998). Perfluorochemicals: their applications and benefits to cell culture. Trends in Biotechnology. 16:272 277.
Lütz, S.; Rao, N. N. and Wandrey, C. (2006). Membranes in biotechnology. Chemical Engineering and Technology. 29(12):1404–1415.
82 Ma, N.; Chalmers, J. J.; Aunins, J. G.; Zhou, W. and Xie, L. (2004). Quantitative Studies of Cell Damage in Sparged Bioreactors Using A Bubble Collector to Segregate Bubble Rupture Events. Biotechnology Progress. 20(4):1183 – 1191.
Ma, N.; Koelling, K. and Chalmers, J. J. (2002). The fabrication and use of a transient contraction flow device to quantify the sensitivity of mammalian and insect cells to hydrodynamic forces. Biotechnology and Bioengineering. 80:428 437.
Ma, N.; Mollet, M. and Chalmers, J. J. (2006). Aeration, Mixing and Hydrodynamics in Bioreactors. In: S.S. Ozturk and W S. Hu (Eds). Cell culture technology for pharmaceutical and cell based therapies. Pp. 225 248. CRC Press. Boca Raton, FL., USA.
Machon, V.; Vlcek, J.; Skrivanek, J. (1985). Dual Impeller Systems for Aeration of Liquids: An Experimental Study. In: Proc. 5th European Conference on Mixing. Würzburg, Germany. Pp. 155 169.
MacIntyre, F. (1972). Flow patterns in breaking bubbles. Journal of Fluid Mechanics. 77:5211 5228.
Mahaffy, R. E.; Shih, C. K.; MacKintosh, F. C and Kas, J. (2000). Scanning probe based frequency dependent microrheology of polymer gels and biological cells. Physical Review Letters. 85:880–883.
Malek, A. M. and Izumo, S. (1994). Molecular aspects of signal transduction of shear stress in the endothelial cell. Journal of Hypertension. 12(9):989 999.
Mardikar, S. H. and Niranjan, K. (2000). Observations of the shear damage to different animal cells in a concentric cylinder viscometer. Biotechnology and Bioengineering. 68(6): 697 704.
Marks, D. M. (2003). Equipment design considerations for large scale cell culture. Cytotechnology. 42: 21–33.
83 McIntire, L. V.; Frangos, J. A.; Rhee, B. G.; Eskin, S. G. and Hall, E. R. (1987). The effect of fluid mechanical stress on cellular arachidonic acid metabolism. Annals of the New York Academy of Sciences. 516:513 524.
McQueen, A. and Bailey, J. E. (1989). Influence of serum level, cell line, flow type and viscosity on flow induced lysis of suspended mammalian cells. Biotechnology Letters. 11:531 536.
McQueen, A.; Meilhoc, E. and Bailey, J. E. (1987). Flow effects on the viability and lysis of suspended mammalian cells. Biotechnology Letters. 9(12):831 836.
Meier, S. J. (2005). Cell culture scale up: mixing, mass transfer, and use of appropriate scale down models. Biochemical engineering XIV. Harrison Hot Springs, Canada.
Menisher, T.; Metghalchi, M. and Gutoff, E. B. (2000). Mixing studies in bioreactors. Bioprocess engineering. 22(2):115 120.
Metz, B.; Kossen, N. W. P. and van Suijdam, J. C. (1979). The rheology of mold suspensions. Advances in Biochemical Engineering. 11:103 156.
Michaels, J. D.; Nowak, J. E. ; Mallik, A. K.; Koczo, K.; Wasan, D. T. and Papoutsakis, E. T. (1995). Interfacial properties of cell culture media with cell protecting additives. Biotechnology and Bioengineering. 47(4): 420 430.
Midler, J. Jr. and Finn, R. K. (1966). A model system for evaluating shear in the design of stirred fermentors. Biotechnology and Bioengineering. 8:71 84.
Miller, W. M.; Wilke, C. R. and Blanch, H. W. (1987). Effect of dissolved oxygen concentration on hybridoma growth and metabolism in continuous culture. Journal of Cell Physiology. 132:524 530.
Miller, W. M.; Wilke, C. R. and Blanch, H. W. (1988). Transient responses of hybridoma metabolism to changes in the oxygen supply rate in continuous culture. Bioprocess Engineering. 3:103 113.
84 Mollet, M.; Ma, N.; Zhao, Y.; Brodkey, R.; Taticek, R. and Chalmers, J. J. (2004). Bioprocess Equipment: Characterization of Energy Dissipation Rate and its potential to Damage Cells. Biotechnology Progress. 20:1437 1448.
Mollet, M.; Godoy Silva, R.; Berdugo, C. and Chalmers, J. J. (2007). Acute Hydrodynamic Forces and Apoptosis: A Complex Question. Biotechnology and Bioengineering. 98 (4): 772 788.
Mollet, M.; Godoy Silva, R.; Berdugo, C. and Chalmers, J. J. (in Press). Computer simulations of the energy dissipation rate in a fluorescence activated cell sorter: implications to cells. Biotechnology and Bioengineering.
Molowa, D. T. and Mazanet, R. (2003). The state of biopharmaceutical manufacturing. Biotechnology Annual Review. 9:285–302.
Morão, A.; Maia, C. I.; Fonseca, M. M. R.; Vasconcelos, J. M. T. and Alves, S. S. (1999). Effect of antifoam addition on gas liquid mass transfer in stirred fermenters. Bioprocess and Biosystems Engineering, 20(2): 165 – 172.
Moreira, J. L.; Cruz, P. E.; Santana, P. C. and Feliciano, A. S. (1995). Influence of power input and aeration method on mass transfer in a laboratory animal cell culture vessel. Journal of Chemical Technology and Biotechnology. 62:118 131.
Mostafa, S. S. and Gu, X. (2003). Strategies for improved dCO2 removal in large scale fed batch cultures. Biotechnology Progress. 19(1):45 51.
Motobu, M.; Wang, P. C. and Matsumura, M. (1998). Effect of shear stress on recombinant Chinese hamster ovary cells. Journal of Fermentation and Bioengineering. 85(2): 190 195.
Mufti, N. A. and Shuler, M. L. (1995). Induction of cytochrome P 450IA1 activity in response to sublethal stresses in microcarrier attached Hep G2 cells. Biotechnology Progress. 11(6): 659 663.
85 Murhammer D.W. and Goochee C. F. (1990). Sparged animal cell bioreactors: mechanism of cell damage and Pluronic F 68 protection. Biotechnology Progress. 6: 391–397.
Murhammer, D. W. and Goochee, C. F. (1990). Sparged animal cell bioreactors: mechanism of cell damage and Pluronic F 68 protection. Biotechnology Progress. 6: 391–397.
Nehring, D.; Czermak, P.; Vorlop, J. and Lübben, H. (2004). Experimental Study of a Ceramic Microsparging Aeration System in a Pilot Scale Animal Cell Culture. Biotechnology Progress. 20:1710 1717.
Nienow, A. W. (1997). On impeller circulation and mixing effectiveness in the turbulent flow regime. Chemical Engineering Science. 52(15):2557 2565.
Nienow, A. W. (2006). Reactor engineering in large scale animal cell culture. Cytotechnology. 50:9 33.
Nienow, A. W.; Langheinrich, C.; Stevenson, N. C.; Emery, A. N.; Clayton, T. M. and Slater, N. K. H. (1996). Homogenization and oxygen transfer rates in large agitated and sparged animal cell bioreactors: Some implications for growth and production. Cytotechnology. 22: 87 94.
Oh, S. K. W.; Nienow, A. W.; Al Rubeai, M. and Emery, A. N. (1989). The effect of agitation intensity with and without continuous sparging on the growth and antibody production of hybridoma cells. Journal of Biotechnology. 12:45 62.
Oh, S. K. W.; Nienow, A. W.; Al Rubeai, M. and Emery, A. N. (1992). Further studies of the culture of mouse hybridoma in an agitated bioreactor with and without continuous sparging. Journal of Biotechnology. 22:245 270.
Oldshue, J. Y. (1983). Fluid Mixing Technology. McGraw Hill Publications Co. New York. USA.
86 Orton, D. and Wang, D.I.C. (1991). Fluorescent visualization of cell death in bubble aerated bioreactors. Cell Culture Engineering III, United Engineering Foundation. February 2 7, 1991.
Ozturk, S. S. (1996) Engineering challenges in high cell density cell culture systems. Cytotechnology. 22: 3–16.
Ozturk, S. S. (2006). Cell culture technology – An Overview. In: S.S. Ozturk and W. S. Hu (eds). Cell culture technology for pharmaceutical and cell based therapies. Pp. 1 13. CRC Press. Boca Raton, FL., USA.
Ozturk, S. S. and Palsson, B. O. (1990). Effects of dissolved oxygen on hybridoma cell growth, metabolism and antibody production kinetics in continuous culture. Biotechnology Progress. 6:437 446.
Packer, L. and Fuehr, K. (1977). Low oxygen concentration extends the lifespan of cultured human diploid cells. Nature. 267:423 425.
Palomares, L. A. ; Lopez, S. ; Ramirez, O. T. (2004). Utilization of oxygen uptake rate to assess the role of glucose and glutamine in the metabolism of infected insect cell cultures. Biochemical Engineering Journal. 19(1): 87 93.
Papadaki, M. and Eskin, S.G. (1997). Effects of fluid shear stress on gene regulation of vascular cells. Biotechnology Progress. 13(3):209 221.
Parker, B. and Barsom, G. (1970). Biological and chemical significance of surface microlayers in aquatic ecosystems. BioScience. 20:87 93.
Passini, C. A. and Goochee, C. F. (1989). Response of a mouse hybridoma cell line to heat shock, agitation and sparging. Biotechnology Progress. 5: 175 188.
Ramirez, O. T. and Mutharasan, R. (1990a). Cell cycle and growth phase dependent variations in size distribution, antibody productivity and oxygen demand in hybridoma cultures. Biotechnology and Bioengineering. 36(8): 839 848.
87 Ramirez, O. T. and Mutharasan, R. (1990b). The role of the plasma membrane fluidity on the shear sensitivity of hybridomas grown under hydrodynamic stress. Biotechnology and Bioengineering. 36(9):911 920.
Research and Market (2007). Monoclonal Antibodies 2007: Competitive Landscape & Pipeline Insight. http://www.researchandmarkets.com/reportinfo.asp?report_id=367599
Rhiel, M.; Mitchell Logean, C. M. and Murhammer, D. W. (1997). Comparison of Trichoplusia ni BTI Tn 5B1 4 (High Five TM) and Spodoptera frugiperda Sf 9 insect cell line metabolism in suspension cultures. Biotechnology and Bioengineering. 55:909–920.
Rocha Valadez, J. A.; Hassan, M.; Corkidi, G.; Flores, C.; Galindo, E. and Serrano Carreon, L. (2005) 6 pentyl α pyrone production by Trichoderma harzianum: the influence of energy dissipation rate and its implications on fungal physiology. Biotechnology and Bioengineering. 91 (1):54 61.
Sauer, P. W.; Burky, J. E.; Wesson, M. C.; Sternard, H. D. and Qu, L. (2000). A High Yielding, Generic Fed Batch Cell Culture Process for Production of Recombinant Antibodies. Biotechnology and Bioengineering. 67(5): 585 597.
Schürch, U.; Kramer, H.; Einsele, A.; Widmer, F. and Eppenberger, H. M. (1988). Experimental evaluation of laminar shear stress on the behaviour of hybridoma mass cell cultures, producing monoclonal antibodies against mitochondrial creatine kinase. Journal of Biotechnology. 7(3): 179 184.
Sen, A.; Kallos, M. S. and Behie, L. A. (2001). Effects of hydrodynamics on cultures of mammalian neural stem cell aggregates in suspension bioreactors : Novel reactor engineering for the new millennium. Industrial and Engineering Chemistry Research. 40(23): 5350 5357.
Shiloach, J. and Fass, R. (2005). Growing E. coli to high cell density—A historical perspective on method development. Biotechnology Advances. 23:345–357.
Singh, V. (1999). Disposable bioreactor for cell culture using wave induced agitation. Cytotechnology. 30: 149–158.
88 Sinskey, A. J.; Fleischaker, R. J.; Tyo, M. A.; Giard, D. J. and Wang D. I. C. (1981). Production of cell derived products: virus and interferon. Annals of the New York Academy of Sciences. 369:47 59.
Smith, C. M.; Marks, A. D. and Lieberman M. A. (2005). Marks' Basic Medical Biochemistry: A Clinical Approach. 2nd Ed. Lippincott Williams & Wilkins. Philadelphia. USA.
Sundararajan, A. and Ju, L. K. (2006). Use of cyanobacterial gas vesicles as oxygen carriers in cell culture. Cytotechnology. 52:139–149.
Tatterson, G. B. (1991). Fluid mixing and gas dispersion in agitated tanks. New York. Mc Graw Hill.
Taylor, G. I. (1934). The formation of emulsions in definable fields of flow. Proceedings of the Royal Society of London. Series A. 146:501 523.
Telling, R.C. and Elsworth, R. (1965). Submerged culture of hamster kidney cells in a stainless steel vessel. Biotechnology and Bioengineering. 7(3):417 434.
Thomas, C. R.; Al Rubeai, M. and Zhang, Z. (1994). Prediction of mechanical damage to animal cells in turbulence. Cytotechnology. 15: 329 335.
Thomson Centerwatch (2007). Drugs approved by the FDA. http://www.centerwatch.com/patient/drugs/druglist.html
Tomos, D. (2000). The plant cell pressure probe. Biotechnology Letters. 22:437 442.
Tramper, J. J.; Joustra, D. and Vlak, J. M. (1987). Bioreactor design for growth of shear sensitive insect cells. In: Webb, C. and Mavituna, F. (Eds). Plant and animal cell cultures: process possibilities. Ellis Horwood. Chichester, U. K. Pp 125 136.
Trinh, K.; García Briones, M. A.; Hink, F. H. and Chalmers, J. J. (1994). Quantification of damage to suspend insect cells as a result of bubble rupture. Biotechnology and Bioengineering. 43:37 45.
89
Tzima, E. (2006). Role of Small GTPases in Endothelial Cytoskeletal Dynamics and the Shear Stress Response. Circulation Research. 98:176 185.
Uhl, V. W. and Gray, J. B. (1967). Mixing: theory and practice. Vol. II. Academic Press. San Diego, USA. p. 254.
Ujcová, E.; Fencl, Z.; Musílková, M. and Seichert, L. (1980). Dependence of release of nucleotides from fungi on fermentor turbine speed. Biotechnology and Bioengineering. 22: 237 241.
Van’t Riet, K. (1979). Review of measuring methods and results in non viscous gas liquid mass transfer in stirred vessels. Industrial and Engineering Chemistry Process Design and Development. 18(3):357 363.
Varley, J. and Birch, J. (1999). Reactor design for large scale suspension animal cell culture. Cytotechnology. 29:177 205.
Venkat, R.; Stock, R. and Chalmers, J. J. (1996). Study of hydrodynamics in microcarrier culture spinner vessels: A particle tracking velocimetry approach. Biotechnology and Bioengineering. 49:456 466.
Vorlop, J. and Lehmann, J. (1989). Oxygen transfer and carrier mixing in large scale membrane stirred cell culture reactors. In: Spier R.E., Griffiths J.B., Stephenne J. and Crooy P.J. (Eds). Advances in Animal Cell Biology and Technology for Bioprocesses. Butterworths, Svenoaks, UK, pp. 366–369.
Wang, N; Butler, J. P. and Ingber, D. E. (1993). Mechanotransduction across the cell surface and through the cytoskeleton. Science. 260(5111): 1124 1127.
Wernersson, E. S. and Trägårdh, C. (1999). Scale up of Rushton turbine agitated tanks.. Chemical Engineering Science. 54(19): 4245 4256.
90 Wong, T. K. K.; Nielsen, L. K.; Greenfield, P. F. and Reid, S. (1994). Relationship between oxygen uptake rate and time of infection of Sf9 insect cells infected with a recombinant baculovirus. Cytotechnology. 15:157–167.
Wu, J. (1996). Insights into protective effects of medium additives on animal cells under fluid stresses: the hydrophobic interactions. Cytotechnology. 22(1 3):103 109.
Wu, J. ; Ruan, Q. and Lam, H. Y. P. (1997). Effect of surface active medium additives on insect cell surface hydrophobicity relating to cell protection against bubble damage. Enzyme and Microbial Technology. 21(5):341 348.
Wurm, F. M. (2004). Production of recombinant protein therapeutics in cultivated mammalian cells. Nature Biotechnology. 22(11): 1393 1398.
Yoon, S. and Konstantinov, K. B. (1994). Continuous, real time monitoring of the oxygen uptake rate (OUR) in animal cell bioreactors. Biotechnology and Bioengineering. 44:983 990.
Zeng, A. P. and Bi, J. X. (2006). Cell culture kinetics and modeling. In: S.S. Ozturk and W S. Hu (eds). Cell culture technology for pharmaceutical and cell based therapies. Pp. 299 348. CRC Press. Boca Raton, FL., USA.
Zhang, Z. and Thomas, C. R. (1993). Modeling of animal cell damage in turbulent flows. In: Nienow A. W. (Ed). Proceedings of the 3rd. International conference on bioreactor and bioprocess fluid dynamics. Mechanical Engineering Publications Ltd. London. Pp. 475 482.
Zhang, S.; Handa Corrigan, A. and Spier, R. E. (1992). Oxygen transfer properties of bubbles in animal cell culture media. Biotechnology and Bioengineering. 40(2):252 259.
Zhang, Z.; Al Rubeai, M. and Thomas, C. R. (1993). Estimation of disruption of animal cells by turbulent capillary flow. Biotechnology and Bioengineering. 42: 987–993.
Zhang, Z.; Ferenczi, M. A. and Thomas, C. R. (1992). A micromanipulation technique with a theoretical cell model for determining mechanical properties of single mammalian cells. Chemical Engineering Science. 47 (6): 1347 1354.
91
Zhou, G. and Kresta, S. M. (1996a). Distribution of Energy Dissipation Between Convective and Turbulent Flow For Three Frequently Used Impellers. Chemical Engineering Research and Design. (Transactions of the Institution of Chemical Engineers, part A). 74:379 389.
Zhou, G. and Kresta, S. M. (1996b). Impact of Tank Geometry on the Maximum Turbulence Energy Dissipation Rate for Impellers. AIChE Journal. 42:2476 2490.
Zhou, W. and Hu, W. S. (1994). On line characterization of a hybridoma cell culture process. Biotechnology and Bioengineering. 44: 170 177.
Zhu, M. M.; Goyal, A.; Rank, D. L.; Gupta, S. K.; VandenBoom, T. and Lee, S. S. (2005), Effects of Elevated pCO2 and Osmolality on Growth of CHO Cells and Production of Antibody Fusion Protein B1: A Case Study. Biotechnology Progress. 21(1): 70 77.
92
qO 2 Cell line Reference (mol oxygen·cell 1·s 1) 5.8×10 17 to 6.9×10 17 Hybridoma NB1 Boraston et al. (1983) Hybridoma Miller et al. (1987); Miller et al. 5.3×10 17 to 1.1×10 16 AB20143.2 (1988) 4.2×10 17 to 1.0×10 16 Hybridoma KS1/4 Backer et al. (1988) 9.2×10 17 to 1.0×10 16 Hybridoma HB 32 Ramirez and Mutharasan (1990) 6.4×10 18 to 2.4×10 17 Hybridoma 167.4G5.3 Ozturk.and Palsson (1990) 6.4×10 17 to 1.2×10 16 Hybridoma X D Hiller et al. (1991) 1.67×10 16 Sf 9 Hensler and Agathos (1994) 8.6×10 17 to 1.0×10 16 Sf 9 Wong et al. (1994) 1.28 ×10 16 Hybridoma MAK Zhou and Hu (1994) 6.08×10 17 to 1.13×10 16 NS0 Yoon and Konstantinov (1994) 6.5×10 17 Hybridoma C1a Dorresteijn et al. (1994) 5.6×10 17 Hybridoma HFN7.1 Eyer et al. (1995) 5.53×10 17 CHO Gray et al. (1996) Trichoplusia ni BTI 8.0×10 17 to 1.6×10 16 Rhiel et al. (1997) Tn 5B1 4 2.5×10 17 to 4.5×10 16 Sf 9 Rhiel et al. (1997) Sp2/0 Ag14 derived 3.3×10 17 to 5.3×10 17 cells Sauer et al. (2000) (ATCC CRL 1581) B lymphocyte 3.3×10 17 to 1.2×10 16 Barnabé and Buttler (2000) hybridoma (CC9C10) 2.3×10 17 to 1.7×10 16 GS NS0 deZengotita et al. (2000) 9.0×10 17 CHO SSF3 Ducommun et al. (2001) CHO K1 5.6×10 17 to 2.2×10 16 Carvalhal et al. (2003) (ATCC CCL 61) 5.3×10 17 to 8.9×10 17 T CHO ATIII Deshpande and Heinzle (2004) 1.1×10 16 to 2.6×10 16 Sf9 (ATCC 1711) Palomares et al. (2004)
Table 2.1. Specific oxygen uptake rates (q ) reported for animal cell lines. O2
93
Correlation Units System information Reference 1 [kLa] : h Water 4.0 0.5 [ (ε T ) ] : W·m k a = 93 6. ⋅ (ε ) ⋅ (V ) Ig 2 < VR < 4400 L ( T Ig ) s 3 Van’t Riet 5.0×102 < (ε ) < 1.0×104 1 T Ig (1979) [Vs] : m·s 2 2 0.5 ×10 < Vs < 4 ×10 [VR] : liter 1 [kLa] : h Water plus ions. [ (ε T )Ig ] : W·m 7.0 2.0 2 < VR < 4400 Van’t Riet k a = 2.7 ⋅ (ε ) ⋅ (V ) 3 2 4 L ( T Ig ) s 5.0×10 < (ε ) < 1.0×10 1 T Ig (1979) [Vs] : m·s 2 2 0.5 ×10 < Vs < 4 ×10 [VR] : liter 1 [kLa] : h 2.0 kLa = .0 255 ⋅ (T ) ⋅ Λ ⋅ Γ [T] : m Water plus ions. − 3.1 5.1 [D] : m 100 liter < VR <100000 Asay and Λ = ()()H L ⋅ N [HL] : m liter Kono (1982) 3.2 6.0 Γ = ()()D ⋅ VS [N] : RPM 1 [Vs] : m·h 2000 < VR < 8000 1 [kLa] : h 0.3 < H < 1.3 L [VR] : liter T = 2 m
.0 72 7.0 [T] : m D = 0.45 m Langheinrich kLa = 125⋅ ((εT ) ) ⋅ (Vs ) 4 4 g [HL] : m 1 ×10 < Vs < 5 ×10 et al. (2002) (See Note) 3 2 4 [(ε T )g ] : W·m 5.0×10 < (ε T )g < 1.0×10
[V ] : m·s 1 Culture of CHO 320 cells s in serum free medium Note: This correlation was derived by the authors of this review exclusively from data in Figures 3 and 5 of Langheinrich et al. (2002)
Table 2.2. Examples of empirically derived equations for kLa estimation found in literature.
94
Impeller Pot Reference 45° pitched blade turbine (4 blades, 0.88 Tatterson (1991) w/D=0.1. Baffled tank) 45° pitched blade turbine (4 blades, 1.69 Tatterson (1991) w/D=0.3. Baffled tank) Marine Propeller (Square pitch, 3 0.32 Chisti (2001) bladed) Lightnin A310 hydrofoil 0.31 Chisti (2001) Rushton (6 blades, central disk, 10% 5.0 Tatterson (1991) baffled tank). Paddle (2 blades) 1.70 Chisti (2001) Scaba 6SRGT. (D/T=0.44) 1.6 Amanullah et al. (1998) “Elephant ear” impeller (3 blades, Menisher et al. (2000). 0.006 unbaffled) (calculated) Prochem Maxflo T. (6 blades, 1.3 Amanullah et al. (1998) D/T=0.44)
Table 2.3. Power number for selected impellers for turbulent regime.
95
k (at 37°C) Substance –3 –1 k/kwater [mol O2 · m · MPa ] Water 9.7 1.0 Perfluorodecaline 164 16.9 Perfluorotripropylamine 176 18.1 Perfluorotributylamine 156 16.1 Gas vesicles 388 40
Table 2.4. Relative Oxygen carrying capacities (k) of water, perfluorochemicals and gas vesicles (Modified from Sundararajan and Ju, 2006).
96
Benefits Drawbacks Dispersion is unstable requiring high High oxygen carrying capacity. amounts of surfactants and energy. Ease of sterilization (e.g., by autoclave). Surfactants may become toxic to cells. Provide a two phase (perfluorocarbon liquid and aqueous medium) interface Once suspension is formed it is difficult and physical support for cell division (in to separate PFCs for reoxygenation. stationary cell culture). Chemically and biologically inert Downstream concerns. scavengers of gaseous cellular products Expensive. (e.g. ethylene) Recoverable and recyclable.
Table 2.5. Benefits and drawbacks of PFCs (modified from Lowe et al., 1998).
97
Mean Condition Impeller Equation deviation P Q ⋅ N .0 25 ⋅ D −2 ≤ .0 055 Bottom 1− 0g = 9.9 ⋅ (Q ⋅ N .0 25 ⋅ D −2 ) <6 % P0 P Q ⋅ N .0 25 ⋅ D −2 > .0 055 Bottom 1− 0g = .0 52 + .0 62 ⋅ (Q ⋅ N .0 25 ⋅ D −2 ) <6 % P0 Middle P Q ⋅ N ≤ .0 013 1− 0g = 37 6. ⋅ ()Q ⋅ N 2.3 % and Top P0 Middle P Q ⋅ N > .0 013 1− 0g = .0 375 + 8⋅ ()Q ⋅ N 1.4 % and Top P0 Q: Gas flow rate (m3·s 1); N: Impeller rotational speed (s 1). Conditions for correlated data: 0.238 Table 2.6. Correlations derived by Cui et al. (1996) for power drawn in systems with multiple Rushton impellers. 98 Method Selected references Laminar Flow in capillary tubes Thomas et al. (1994) Turbulent flow in capillary tubes Augenstein et al. (1971) Magnetic twisters Wang et al. (1993) Optical deformation Guck et al. (2001) Micromanipulation (poking) devices Zhang et al. (1992); Goldman (2000) Stretching devices Graf et al. (2003) Aspiration devices Discher et al. (1994); Jones et al. (1999) Goldblum et al. (1990); Born et al. (1992) Cone and plate rheometers Graf et al. (2003) Schürch et al. (1988); Joshi et al. (1996) Concentric cylinder rheometers Mardikar and Niranjan (2000) Laminar flow between parallel Motobu et al. (1998); Keane et al. (2003) plates Contracting flow devices Gregoriades et al. (2000); Ma et al. (2002) Telling and Elsworth (1965) Hirtenstein and Clark (1980) Croughan et al. (1987); Oh et al. (1989) Stirred tanks Kunas and Papoutsakis (1990a,b) Oh et al. (1992); Thomas et al. (1994) Sen et al. (2001) Free jets McQueen et al. (1987) Bioluminescence Chen et al. (2003) Hoh and Schoenenberger (1994) Atomic force microscopy Mahaffy et al. (2000) Pressure probe Tomos (2000) Table 2.7. Methodologies reported in literature for measurement of mechanical properties of animal cells and/or determination effect of hydrodynamical forces on animal cells. 99 Parameter Definition Selected references Kunas and Papoutsakis Agitation rate N (1990a) Moreira et al. (1995) Middler and Finn (1966) Impeller tip speed πND Croughan et al. (1987) P N ρN 3 D 5 Mean power input ε = = p Ujcová et al. (1980) V V Sinskey et al. (1981) Integrated shear 2πND ISF = Croughan et al. (1987) factor T − D Chen et al. (2003) Volumetric integrated 2πND V VISF = s Chen et al. (2003) shear factor T − D V Croughan et al. (1987) 1 Cherry and Papoutsakis Kolmogorov eddy ν 3 4 (1988) size η = ε Croughan et al. (1989) Oh et al. (1989) Bluestein and Mockros Local energy (1969); ε = τ : ∇U dissipation rate Ma et al. (2002) Mollet et al., (2004) Shear stress tensor τ Keane et al. (2003) 2 Turbulent collision 3 4 π ⋅ ρb ⋅α ⋅ db Cherry and Papoutsakis TCS = ()εν severity 72 (1988) 9π 4 ⋅ ρ ⋅ n ⋅ n3α ⋅ D ⋅ d 4 Impeller collision b b b Cherry and Papoutsakis ICS = severity 512 ⋅V (1988) Jüsten et al. (1998) Energy dissipation ε EDCF = Rocha Valadez et al. circulation function t c (2005) Table 2.8. Some parameters reported in literature to correlate the effect of hydrodynamical forces on cells. 100 W b H HL C C D T a b c 2.2 2.0 1.8 a 01 1.6 c /P 02 P 1.4 T D/T H /T b L 1.2 0.56 0.33 1 and 2 0.29 0.33 2 One impeller 0.45 0.33 1 1.0 0.64 0.40 2 0.8 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 C/D Figure 2.1. Geometrical configuration, flow patterns and total non gassed power drawn into the liquid as a function of impeller spacing for a mixing system with multiple Rushton turbines (Adapted from Hudcova et al., 1989). 101 Paddle Rushton turbine “Elephant ear” Marine propeller Lightnin A 315 impeller Scaba 6SRGT Prochem hydrofoil 6SRGT Figure 2.2. Impeller configurations commonly employed in animal cell culture. 102 Air bubbles Impeller motion Figure 2.3. Regions of highest shear rate and highest energy dissipation rate behind the blades of a Rushton turbine. 103 −3 Flooding ()ε =160W ⋅m 1.2 T g Water 1.0 Scaba 6 SRGT. P ~1.45, N=225 rpm 0 0 /P 0g 0.8 P Flooding 0.6 Rushton Turbine. P ~5.2, N=150 rpm 0 0.4 0.0 0.5 1.0 1.5 2.0 Air flow rate (v.v.m.) Figure 2.4. Effect of the gas flow rate on the power drop under gassed conditions for two impeller geometries in Newtonian, water like fluids. (Adapted from Galindo and Nienow, 1993). 104 6 5 4 100 3 2 80 60 1 Number of baffles 40 RelativePower Drawn (Inwater) 20 2 4 6 8 10 Baffle width (% tank diameter) Figure 2.5. Effect of the number and size of the baffles on the power drawn by an impeller in a cylindrical stirred tank reactor (Adapted from Oldshue, 1983). 105 2.5 0.8 2.0 0.6 1.5 0.4 T (m) T (m) D 1.0 0.2 0.5 0.0 0.0 1 0 1 2 3 4 10 10 10 10 10 10 Volume (L) Figure 2.6. Change in the tank diameter and impeller diameter as the volume of the vessel increases from 0.5 to 10,000 liters keeping constant the geometrical ratios H/T = 1 and T/D = 3. 106 103 102 103 Wm-3 104 Wm-3 105 Wm-3 6 -3 Impeller rotational speed (RPM) speed rotational Impeller 10 Wm 101 0.01 0.1 1 Impeller Diameter (m) Figure 2.7. Lines of constant, maximum EDR in a vessel as a function of impeller rotational speed and diameter for Rushton turbine in water. H/T = 1 and T/D = 3. 107 6 50 RPM 10 100 RPM 250 RPM 105 500 RPM 750 RPM ) 4 3 10 103 102 101 0 Average EDR(Wm Average 10 10 1 10 2 0.01 0.1 1 Impeller Diameter (m) Figure 2.8. Average EDR for the whole vessel as a function of impeller diameter and RPM using a Rushton turbine in water. H/T = 1 and T/D = 3. 108 106 ) 3 105 104 103 102 101 Calculated average EDR Calculated maximum EDR Col 4 vs EDR max exp 0 10 Col 4 vs Col 6 Energy DisipationEnergy Rate (Wm 10 1 0 100 200 300 400 500 600 700 800 900 1000 RPM Figure 2.9. Calculated maximum and average energy dissipation rate as a function of RPM for an Applikon bioreactor containing four baffles. The single points correspond to experimental measurements without baffles (Adapted from Mollet et al., 2004). 109 (A) 110 Figure 2.10. Molecular signaling and response cascade in endothelial and smooth muscle cells (A) before and (B) after stimulation by hydrodynamic forces. 110 Figure 2.10. (Continued). (B) 111 111 20 100 80 15 1 60 10 cell ml 6 40 Viability (%) Viability 10 5 20 Viablecellconcentration 0 0 m) 400 0 50 100 150 200 Time (h) 300 200 100 0 Averageaggregate diameter ( 0 50 100 150 200 Time (h) Figure 2.11. Effect of agitation rate on cell concentration, viability and aggregate diameter of murine NSC in batch suspension in a 125 mL spinner flask. Data points are average of duplicate runs. (a) Cell concentration: ( ) 60 rev·min 1; ( ) 100 rev·min 1. Viability: ( ) 60 rev·min 1; ( ) 100 rev·min 1.(b) Average aggregate diameter:( ) 60 rev·min 1; ( ) 100 rev·min 1. Standard deviation:( ) 60 rev·min 1; ( ) 100 rev·min 1. (Adapted from Sen et al. 2001). 112 ] 7 1 10 Damage zone 106 105 104 Cell concentration [cellmL 20 40 60 80 100 120 Stirring speed [revmin 1] Figure 2.12. Effect of stirring speed on cell concentration after 7 days of culture of Vero cells on Cytodex microcarriers on 250 mL spinner vessels (Data from Hirtenstein and Clark, 1980). 113 100 CHO-K1 (ATCC CCL-61) (Ma et al., 2002) et al. 80 MCF7 (ATCC HTB-22) (Ma , 2002) CHO6E6 (Mollet et al., Submitted)2007) Sf9 (Ma et al., 2002) Hybridoma HB-24 (Ma et al., 2002) THP1 1st assay (Mollet et al.,2008) In Press) 60 nd THP1 2 assay (Mollet et al., In2008) Press) THP1 3rd assay (Mollet et al., 2008)In Press) 40 % Damaged cells % 20 0 102 103 104 105 106 107 108 109 Median maximum EDR [W.m-3] Figure 2.13. Experimental curves for the percentage of damage experienced by cells in a custom design microfluidic device for single abuse experiments. Adapted from Ma et al. (2002), Mollet et al. (2007) and Mollet et al., (2008). 114 C K J B F, H Cell Response A I G D E Wm 3 1 2 3 4 5 6 7 8 9 10 10 10 10 10 10 10 10 10 1 11 2 6 4 9 3, 5, 7, 10 8 Hydrodynamic Conditions Cell Response Mode of Symbol Cell Reference growth A CHO-K1, necrosis Anchoraged Gregoriades et al. (2000) Thomas et al. (1994); Zhang Hybridoma, necrosis Suspended B et al. (1993) C MCF-7, necrosis Suspended Ma et al. (2002) D Mouse myeloma, necrosis Suspended McQueen and Bailey (1989) E HeLa S3, mouse L929, necrosis Suspended Augenstein et al. (1971) F CHO-K1, SF-9, HB-24, necrosis Suspended Ma et al. (2002) Uninfected and viral infected Suspended Submitted G PERC6 cells, necrosis Enthomopathogenic nematodes, Suspended Fife et al. (2004) H necrosis I CHO-K1, apoptosis Anchorage Mollet et al. (2007) J THP-1, necrosis Anchorage Mollet et al. (In Press) K Algae, loss of flagella Suspension Hu et al. (2007) Continued… Figure 2.14. Summary of the reported energy dissipation rate at which cells are affected as well as the reported levels of energy dissipation rate in various bioprocess environments. Adapted from Ma et al. (2002) and Mollet et al. (2004). 115 Figure 2.14. (Continued). Hydrodynamic Conditions Symbol Process Description Reference Volume average in typical Agitation Varley and Birch (1999) 1 animal cell bioreactors. Volume average in a 10 L Agitation 2 mixing vessel (RT, 700 RPM) Maximum in the 10 L mixing Zhou and Kresta (1996) 3 Agitation vessel (RT, 700 RPM) Volume average in a 22,000 L Agitation 4 mixing vessel (RT, 240 RPM) Wernersson and Maximum in the 22,000 L mixing Tragardh (1999) Agitation 5 vessel Maximum in spinner vessel Agitation Venkat et al. (1996) 6 (200 RPM) Bubble Pure water, bubble diameter: Garcia-Briones et al. (1994) 7 rupture 6.32mm Boulton-Stone and Blake Bubble Pure water, bubble (1993); 8 rupture diameter:1.7mm Garcia-Briones et al. (1994) Flow through Pure water, 100 mL/min, 1 mm Mollet et al. (2004) 9 a pipe diameter Flow through a Flow through a 200 L Mollet et al. (2004) 10 micropipette micropipette tip in 0.2 sec tip Volume average in a highly Agitation Oh et al. (1992) 11 agitated animal cell bioreactor 116 5 10 104 3 10 (cells/bubble) 2 10 1 10 Number of Cellsof Number Associated Each with Bubble 108 107 10 Bul 6 0.1 1 k C 10 5 0.01 ell C 10 0.001 onc ion (g/L) (ce entr 8 Concentrat ll/m ation PF 6 L) Figure 2.15. A three dimensional plot of the number of cells associated with each bubble as a function of cell concentration (cell·mL 1) and Pluronic F 68 concentration. The dots indicated experimental data and the surface is the plot of a multiple variable regression. (Adapted from Ma et al., 2004). 117 CHAPTER 3 CELL DAMAGE IN A FLUORESCENT ACTIVATED CELL SORTER Cell Damage was evaluated in two different models of BectonDickenson flow cytometer: FACS Vantage and FACS Aria. The work related with FACS Vantage was published previously: Mollet, M., Godoy Silva, R., Berdugo, C, Chalmers, J.J. (2008). Computer Simulations of the Energy Dissipation Rate in a Fluorescence Activated Cell Sorter: Implications to Cells. Biotechnology and Bioengineering, Vol. 100, No. 2, June 1, 2008. The work reported in this published paper focused on the FACS Vantage. Chinese hamster ovary cells (CHO) and a human leukemia cell line (THP1) were used as model cells to evaluate integrity in cell sorting. Claudia Berdugo’s contribution to this work includes: 1. Experiments of cell sorting in FACS Vantage with THP1 cells. 2. Data and results included in Figure 4.7 as well as data from cell sorting included in Tables 4.3 and 4.4. 3. Editing of the document, bibliographical review, analysis and discussion. The work related with FACS Aria is described in this chapter. 118 3.1. ABSTRACT Cell damage in Fluorescent Activated Cell Sorter (FACS) device, model Aria was studied. Four human leukemia monocytic like cell lines were used in the study; this choice of cell types motivated by customer/user concerns that once separated in the FACS Aria, these cells would not grow. As was presented in our previous publications, cells can be exposed to very high (relative to what is known to damage cells) hydrodynamic forces when flowing through channels and nozzle in the sorting process. In previous studies, the level of hydrodynamic forces was quantified in FACS Vantage by using Computer Fluid Dynamic software. The results indicated that the levels of stress that cells experience in the instrument are in the range where they can cause lethal damage. A similar methodology was followed in this study to characterize the hydrodynamics in the FACS Aria; in addition studies were conducted to evaluate the growth behavior after stress as well as the effect of sorting on cell cycle. Given the variability of biological samples and in some cases the subjectivity in flow cytometry analysis, a number of assays were performed in order to developed statistical analysis to determine the actual damage or effect of sorting in cell populations. The results indicate that not only are cells damaged in a flow cytometer, but that this damage can vary from cell line to cell line as well as specific conditions/type of flow cytometer and flow conditions. In addition, the sensitivity of any specific cell line can be a function of the growth phase of the cell. 119 3.2. INTRODUCTION Damage due to hydrodynamic forces occurs in equipment where the cells are exposed to liquid flow, such as found in the flow cytometers and fluorescent activated cell sorting, FACS. Flow cytometry is a technique that combines optics, fluidics and electronics to analyze cellular populations in a variety of applications. In flow cytometry, cells are labeled with fluorescent molecules that bind specifically to surface receptors in the cell as well as intracellular components in fixed cells. It is also possible to label the cell with multiple fluorochromes simultaneously. Reports exist of over 10 different markers used simultaneously. Each cell produces a flash of fluorescence with intensity proportional to the amount of fluorochrome attached to the cell. The cells are suspended in a carrier fluid while the suspension flowing through the channels joins a large amount of buffer (sheath fluid). They come together in laminar regime. The sheath fluid surrounds the sample, spacing out the cells so that only one passes the laser beam at a time. Detectors measure the pulse of fluorescence emitted as the cell crosses the beam. Typically 2 3 detectors are used with different wavelength band pass filters which allow the detection of emissions from different fluorochromes in a single cell. In addition, two types of light scatter are measured: Forward scatter, which is proportional to the diameter of the cells, and Side scatter, which is related to the granularity of the cells. 120 The hydrodynamics properties of the flow cytometers can be analyzed by studying fluid flow through equivalent tubes or pipes. As the suspending medium must transport the particles along well defined paths through the sensing and measuring region, laminar flow is needed. Fluorescence activated cell sorting (FACS) is a widely used method to sort subpopulations of cells rapidly and at high purities. FACS is an integral part of many biological and medical research projects as well as in cell line development at biotechnology and pharmaceutical companies. There is very little information available, however, in the literature regarding cell death in FACS devices as a result of hydrodynamic forces, although Seidl et al. (1999) did discover a significant increase in necrosis and apoptosis in BT474 tumor cells after FACS. As operators at the Flow Core Lab at the Heart & Lung Institute at The Ohio State University, we have received concerns about THP1 cells being damaged inside the flow cytometer. The main concern regarded the difficulty of growing the cells after sorting. Previously in a related work we investigated possible cell damage inside the flow cytometer Vantage. The high level of Energy dissipation rate (EDR) calculated in the FACS Vantage nozzle was found to cause significant necrosis in various cell lines (Mollet, et al. 2008). The work in this chapter focuses on hydrodynamics studies and cell damage in a BD FACS Aria. This instrument is equipped with three lasers (488nm, 633nm and 405 nm), 13 fluorescent channels and 2 scatter channels. FACS Aria offers expanded detection of fluorescence with respect to the Vantage since the last one does not have 405 nm laser. 121 Another important difference between Vantage and Aria is the point of interrogation of cells by the laser; FACS Vantage interrogation point is after the nozzle while in the FACS Aria the interrogation point is in the cell stream line before the nozzle. Both instruments are widely used in biomedical research and it is important to evaluate the effect of the hydrodynamic stress that the FACS exerts on cells. In this study we have investigated the effect of sorting cells on the growth by comparing the cell cycle profile before and after sorting. The DNA content distribution of the populations evaluated after sorting show a decrease in the content of DNA in the G2 phase. G2 has been considered to be the phase of the cell cycle in which the status of DNA replication and repair is checked and the events required for the initiation of mitosis are staged. Cells growing exponentially show a typical cell cycle profile when they are analyzed in Flow cytometer, since the DNA content varies according to the growth phase of the cells. Thus, a population in phase G2 has double the content of DNA that populations in G1 phase. Other aspects studied include: growth after sorting, necrosis (LDH), effect of size of the nozzle on cell integrity, effect of serum on cell damage when cells are sorted. In addition, the stress that the cells experience in the FACS was quantified by simulating the hydrodynamics through the nozzle of 70 m. Simulations were performed using computational fluid dynamics packages. 122 3.3. MATERIALS AND METHODS 3.3.1. Cell culture Four cell lines were investigated, leukemia monocytic like cell lines THP 1, U937, K562 and HL60. Cells were grown in suspension in T 75 flasks in phenol red free RPMI 1640 media (Cellgro), containing 2mM L glutamine (Gibco) and supplemented with 10% fetal bovine serum (Hyclone, Logan, UT) . Cells were incubated at 37°C in a humidified 6 atmosphere of 5% CO2, until reaching a cellular concentration of approximately 1 x10 cell/mL. THP1 is a human acute monocytic leukemia cell line from ATCC routinely used in screening assays to assess cytotoxicity induced by potential agents. U937 is a lymphoma cell line used in studies of differentiation of monocytes. K562 is an erytrholeukemia cell line and HL60 is a myelocytic leukemia cell line, K562 and HL60 are used in cytotoxicity and differentiation studies (Murray, et al 1993). Growth kinetics after stress (sorting and single pass) were developed in TFlasks, cells were collected after stress and seeded at a density of 2x105 cell/mL in TFlasks of 25 cm2 with 7 mL working volume. Samples were taken every 12 hr for cell counts and glucose/lactate analysis. 123 3.3.2. Single shear stress studies In this type of experiments we reproduced the conditions by Ma et al. (2002) and Mollet et al. (2007). These experiments consisted of a single pass of the cells through the shear stress device (Figure 3.1). Single shear stress studies allow a comparison of the sensitivity of THP1, U937, K562 and HL60 cell lines to hydrodynamic stress with that of the other animal cell lines previously studied by our research group. Cell suspensions were pumped through the shear stress device at five different flow rates using a syringe pump, set up of this experiment is described in Figure 3.2. The corresponding values of EDR for every flow rate were previously calculated (Mollet et al, 2007), and are presented in Table 3.1. Samples are collected before and after flowing through the shear stress device. Samples are tested to measure sensitivity, based on LDH content. The experiments were done in triplicate. 3.3.3. Cell sorting Samples at a cell density of approximately 3×106 cells/mL, suspended in 1 mL of RPMI media were sorted using a BD FACS Aria with a 70 m nozzle, a sheath pressure of 70 psi and a sample differential pressure of 70.3 psi. The flow rate in the instrument was set at 3. At this flow rate, the number of events per second is in the range of 600 to 1000. 124 The sample for sorting was divided in three aliquots. The first aliquot was used as positive control; by freezing the sample at 80 °C the maximum amount of LDH released for the cell density sorted can be estimated. The second aliquot was sorted and the third sample was used as negative control; third aliquot was kept at room temperature during the sorting. The cell viability and concentration was determined before and after every assay by manual counting using a hemacytometer and trypan blue. After sorting, the cells were collected in polypropylene 12 mm x 75 mm tube. Next, cells were centrifuged at 250g for 5 min, the supernatant was used to evaluate cell death based on LDH (lactate dehydrogenase assay) released. Since during collection cells are diluted with sheath fluid from the flow cytometer, the dilution factor was considered in calculating cell damage. 3.3.4. Cell damage analysis Cell damage as a result of being sorted in the FACS Aria or being pumped through the shear stress device was analyzed by several methods including a total cell balance based on visual cell counting (cell concentration) and LDH (lactate dehydrogenase) analysis. LDH is an intracellular enzyme that is released into the culture medium when the cell membrane is compromised (necrosis). Samples were centrifuged at 250g to remove cells and the supernatant was dispensed into flat bottom 96 well plates. Measurements of LDH were made following the protocol of the CytoTox ONE assay kit (Promega, WI). Absorbance measurements at 490 nm were made using a SpectraMax 250 125 spectrophotometer and were compared to a standard curve for every cell line to calculate the number of necrotic cells. Fresh medium was used as blank; 690 nm was used as a reference wavelength. The percent cell damage was subsequently determined using the following relationship: FI − FI e c % cell damage = × 100 (Eq. 3.1) FI cl − FI c where FI refers to fluorescence intensity of the given reading, and the subscripts e, c, cl, refer to effluent, control, and completely lysed, respectively. 3.3.5. Cell cycle analysis In cell cycle studies, the cell suspension was divided in two before sorting. One sample was used as control and it was maintained at room temperature for the same time that the sample was sorted, the second sample was sorted. After sorting, cells were collected in polypropylene 12×75 mm tube with 1 mL of RPMI media and centrifuged at 350 × g for 7 minutes, sample control was also centrifuged at the same conditions. Next, the supernatant was discharged and the cell pellet was washed with PBS. Then, cells were fixed in 70% ethanol overnight at 4 ̊ C. Subsequently, the cells were washed twice with PBS and treated with a solution of propidium iodide at 50 g/mL (Sigma) and RNase at 5 g/mL (Roche ). Finally, cells were analyzed using a BD FACS Calibur. Fractions of cells in G0/G1, S, and G2 phase were analyzed using cell cycle analysis software, Modfit. 126 3.3.6. Cell arrest in G2 phase Paclitaxel (Taxol ) was purchased from MP Biomedicals, Inc. and dissolved in DMSO to a concentration of 586 nmol/mL (taxol solution). Different concentrations of taxol solution in media were tested to estimate the proper concentration which arrests the cells without causing damage. The concentration selected was 0.4 nmol/mL; this concentration does not exhibit effect on cell growth and proliferation, as was experimentally confirmed. Cells in a density of 2x105 cell/mL were treated during different times to find out the time of maximum percentage of cells in G2 phase. Additionally a protocol for arresting cells in G2 with hydroxyurea and nocodazole was developed as follow: Cells were seeded at a density of 2x105 cell/mL after 18 hr cells were treated with hydrodxyurea (0.5 mM) for 18 hr. Next, cells were centrifuged at 2000 rpm for 5 min, supernatant was discarded and pellet was washed three times with PBS. Then, cells were resuspended in fresh RPMI, 5% FBS, 1% penn/strep and incubated 3 hr at 37 C in %CO2 atmosphere. After 3 hr, cells were treated with Nocodazol (200 ng/mL) for 18 hr, after which cells were centrifuged at 2000 rpm for 5 min, supernatant was discarded and pellet was washed three times with PBS. Finally, cells were resuspended into media for cell sorting or fixed for cell cycle analysis. 3.3.7. Computational Fluid Dynamics (CFD) simulations The flow through the nozzle from a Becton Dickinson (BD) FACS Aria was simulated using the commercial, CFD program, FLUENT®, Ver. 6.2, 2d, dp. The dimensions of the 127 nozzle were determined based on diagrams obtained from BD as well as photos and dimensions determined using micro meter. Once the sketch of the nozzle was defined the drawings were entered into Gambit, the preprocessor included with FLUENT®, to create the corresponding geometry and mesh. An unstructured (pave) mesh consisting of quadrilateral and triangle elements was used to model the flow through the system. Due to the axial symmetry of the nozzle, only half of a two dimensional (2 D) nozzle was simulated. The nozzle used in the model was 70 m. Finally, the geometry and mesh were imported into FLUENT® to start the simulations The system was modeled with a mesh of different grid size in order to test grid independence. Table 3.2 indicates grid size evaluated and its correspondent node points. For comparison purposes, the diameter of a CHO cell is around 12 14 m For the simulations in FLUENT®, we used the segregated solver model, which performs iterations to sequentially solve the non linear equations for continuity and momentum. No energy balance was included (isothermal conditions assumed and negligible frictional losses). The implicit formulation was used to solve the equations for steady state operation assuming laminar flow, viscous fluid model with Pabsolute = 1 atmosphere at the center of the output face of the simulated channel (Figure 3.3. Point A) Boundary conditions involved non slip conditions for walls and constant velocity at the inlet of both the sheath fluid and the sample. No interaction with air was taken into account at the exit of the nozzle (i.e., no surface tension), as the air is outside of the limits of the simulated system. The schemes used for interpolation: for momentum, 1st order upwind; for 128 pressure, the standard FLUENT® interpolation; and for pressure velocity coupling, the SIMPLE scheme was used. All the simulations were performed using a Pentium® D CPU running at 3.2 GHz with 1.0 GB RAM and 8.85 GB hard disk. The sheath inlet velocities were determined for the 70 m nozzle by experimentally measuring, in duplicate, the volumetric flow rate, at three different sheath pressures (25, 45, 70 psi) and different sample flow rates. Sample flow rates in the FACS Aria are defined for a fixed scale that goes from 1 to 11, though it is not known the correspondent units for this scale. Sample was flowing out of the nozzle and volume was measured for a defined period of time. The sample inlet velocity was determined using the same methodology by measuring the sample uptake rate, at the same values of sheath pressures and sample flow rates. The results of sheath and sample inlet velocities are presented in Figure 3.4. These velocities were entered into FLUENT® and after convergence for each set of conditions; a user defined function was implemented to calculate the EDR at every node point by using Equation 3.2. For an incompressible, Newtonian fluid, the following equation can be used to determine EDR (ε): r r T r ε = [∇U + (∇U ) ]: ∇U (Eq. 3.2) r r Where is the dynamic viscosity, U is the velocity vector, ∇U is the velocity gradient r r tensor, and (∇U )T is the transpose of ∇U . Equation (3.2) is valid for any flow regime (i.e., laminar, transient or fully turbulent). 129 Different cells going into the channel in various initial positions will experience different values of EDR. The most important value of EDR that every cell experiences is the maximum because most likely it is the most damaging. In order to find the value of EDR that is representative of the particular flow rate for all the cells, a methodology was implemented previously (Mollet, et al 2007) was used in this study. Specifically, an option in FLUENT® allows the injection of virtual particles in a group or from a surface. Particles are released from the inlet of the sample tube and tracked through the entire geometry. Table 3.2 shows the type of injection defined in the simulations. Particle liquid interactions and particle particle interactions were not considered in the simulations. The EDR data obtained from the simulations is exported and processed in a program written in Perl®; which will report the maximum values of EDR for each particle. EDR data points were next imported into Excel® to create histograms of the maximum EDR each particle experienced for every flow rate. 3.4. RESULTS 3.4.1. Single shear stress studies We evaluated the cellular damage to THP1 cells when they are exposed to a range of EDR which are common in a FACS device. While we previously investigated possible cell damage inside the flow cytometer Vantage in CHO cells as well as THP1, cell 130 damage in FACS Aria have not previously been studied. These previous studies found that high levels of EDR calculated in the FACS Vantage nozzle cause significant necrosis in various cell lines (Mollet, 2008). Table 3.3, from Mollet et al. (2008) lists the maximum, simulated EDR for a range of operating conditions that are typically used in the BD FACS Vantage at the University Cell Analysis and Sorting Core Using the shear stress device, Ma et al. (2002) performed single pass experiments on CHO, HB 24, SF 9 and MCF 7 cell lines, and found that those cell lines can withstand high levels of energy dissipation ( in the range of approximately 107 108 W/m³). In contrast, THP1 cells are significantly more sensitive than these other cell lines. While these other cell lines can be exposed to a maximum EDR on the order of 1x107 W/m3, the maximum EDR that THP1 cells can tolerate is on the order of 3x106 W/m3. Figure 3.5 shows the comparison of the cell damage caused at different cell lines with THP1. Inspection of Figure 3.5 and Table 3.3 indicates that the potential of significant cell damage exist at all typically operating conditions in FACS instruments. In addition, the cell sensitivity of other monocytic like cell lines was investigated in order to elucidate if the high cell sensitivity was a characteristic of these types of cell lines. The cell lines studied (THP1, K562, U937, and HL60), are in vitro cell culture models used to study normal process of macrophage differentiation and cytotoxicity. The common characteristic between those cell lines is that they are progenitors of macrophages, though they are in different states of differentiation. As can be observed in 131 Figure 3.5, K562 and U937 present similar level of sensitivity to THP1 cells, while HL60 seem to be more resistant to the hydrodynamic stress conditions evaluated. It can be speculated that cells in different states of differentiation have a set of proteins that help them to cope with shear stress better, thus, HL60 being in an early stage of differentiation have different proteins and are more stress tolerant than THP1 or U937. These results must be observed as trend of cell sensitivity behavior, further investigation is necessary to establish the exact differences in cell sensitivity between HL60 and other monocytic like cell lines. However, there is a clear trend in the set of experiments developed which indicates that those cell lines are more sensitive than MCF7 and the industrial cell lines previously evaluated. 3.4.2. Growth after hydrodynamic stress exposure An interesting observation by a number of the customers that complained that FACS sorting damaged THP1 cells was that while the cells they received after sorting were viable based on trypan blue staining, the cell did not grow. Consequently, cells recovered after a single exposure in the shear stress device were collected in sterile conditions and incubated to evaluate their growth behavior. Growth studies were performed for every condition evaluated in the single pass experiment. No difference was observed by comparing the growth of cells exposed to 10, 30, 50 and 70 mL/min, corresponding to EDR values of 2.87x105, 2.27 x106, 6.45x106, 2.60x107, respectively. However, as can be seen in Figure 3.6 the lag phase for cells exposed to 90 mL/min (EDR = 1.09x108 W/m3) is longer than the control (cells that were not exposed to hydrodynamic stress). While at 132 first view, one might observe that the difference is not great, closer examination indicates that the cells do not begin to grow for approximately 60 hours after EDR exposure Although the lag phase is longer for the cells that were exposed to stress at 90 mL/min with respect to the control, after the cells begin to grow, there does not appear to be a significant difference in the growth rate between the stressed cells and the control. The growth rate was 0.0171 h 1 for the control and 0.0179 h 1 for the cells exposed to stress. In addition, the cultures reach about the same cell concentration after 150 hr and the growth behavior is similar until the end of the culture. A similar experiment was developed with cells sorted in a FACS Aria. Cells were recovered after sorting and seeded in the same conditions as control cells (cells from the same batch that were not sorted). Samples were taken at regular time intervals to evaluate growth behavior. Figure 3.7 shows a growth behavior that resembles that observed in cells exposed to stress in the shear device (Figure 3.6), the lag phase for the cells that were sorted is longer than the cells in the control. In a comparable way after cells overcome the lag phase after exposure in the flow contraction device, similar growth behavior is observed and moreover the maximum cell density reached is similar. In order to evaluate if an increase in the serum concentration help to improve the growth after sorting, cells were grown in media supplemented with 30% FBS. Negative and positive fractions were collected after sorting, then samples were centrifuged and the pellet was re suspended in media and incubated at the same time that control cells (cells 133 non sorted). Samples were taken at regular intervals of time for cell counting.. As can be observed in Figure 3.8 the lag phase is longer for sorted cells growing in media with 30%FBS than for the control cells. The last experiment developed to evaluate cell growth after sorting consisted of observing growth behavior of sorted cells in media supplemented with 0%, 10% and 30% FBS as well as conditioned media. Conditioned media was prepared from culture media of cells growing in exponential phase plus fresh media (50% each). Cells were harvested and resuspended into fresh media (0% or 10% FBS as indicated in Figure 3.9). After sorting the cell suspension was centrifuged at 350 g and cell suspensions were prepared in the correspondent media for evaluation. As can be observed in Figure 3.9 the worst case scenario was cells sorted in media at 0% FBS and growing in media at 0% FBS. Similar growth behavior was observed with cells sorted at 0% FBS and growing in media at 10% FBS. However, when cells sorted at 0% FBS were grown in media supplemented with 30% FBS plus conditioned media (CM), the lag phase was reduced. On the other hand, cells sorted in media at 10% FBS and grown in media supplemented with either 10%FBS or 30%FBS plus CM presented similar growth behavior. The previous results suggest that by sorting cells in media containing FBS the lag phase can be reduced. Using conditioned media might improve the growth behavior of cells sorted without FBS and cells sorted in media with 10% FBS reached slightly higher cell concentration when they were grown in media supplemented with 30%FBS plus conditioned media. Replicates of the experiments were performed and results consistently showed that cell growth behavior is improved when conditioned media is used. 134 The effect of sorting cells in media with FBS Vs sorting in media without FBS will be discussed later. 3.4.3. FACS sorting studies at different cell densities Sorting was evaluated in cell suspensions at different cell densities. Samples of approximately 3 to 94 ×106 cells/mL, suspended in RPMI media were sorted using a BD FACS Aria with a 70 m nozzle, a sheath pressure of 70 psi and a sample differential pressure of 70.3 psi Typically the cell concentration sorted in the FACS Aria is in the range of 1 to 5 x106 cell/mL, this number is established for every operator according to the specific objectives of the particular sort. In the case of the FACS Aria it has been observed that this range allows for efficient separation in a reasonable time. As can be observed in Figure 3.10 there is not a clear trend of cell damage with respect to the cell concentration that is sorted. Nevertheless, we observed damage in all cases in the range of 18.55% and 29.85%. 3.4.4. FBS protective effect in FACS sorting Cell damage was evaluated when cells were sorted in media containing 0%FBS versus media containing 10%FBS. Cells were counted in hematocytometer before and after 135 sorting, and visual counting was compared with Aria reports. Figure 3.11 show the results of sorting of THP1 cells in media with 0% FBS while Figure 3.12 shows the results of sorting THP1 cells in media with 10% FBS. The initial cell counts are in the range of samples routinely sorted in the flow lab facilities. Results indicate that the recovery of cells after sorting goes from 25% to 88 % for samples sorted in media with 0%FBS, while the recovery goes from 35% to 99% for samples sorted in media with 10%FBS (although 99% recovery was achieved only with two samples). Regarding the reports of events obtained from the instrument, the events seem to be more related with the initial counting, which might indicate that the counting of events is performed before the nozzle were the damage may occur. The measurements of LDH were used as indicative of cell damage, considering that LDH is more sensitive and accurate measurement than cell counting. Statistical software (JMP, release 6.0.0, SAS Institute, USA) was used for the analysis. A t test for unequal sample size was performed to evaluate the effect of using FBS during sorting. The null (Ho) and alternate (Ha) hypotheses were established as follow: Ho: There is no difference in the percentage of death observed between samples sorted with 0% FBS and 10% FBS. Ha: The damage in samples sorted with 0% FBS is greater than the damage in samples with 10% FBS. That is: 0% > 10% , which is equivalent with 0% 10% <0 . 136 Results from statistical analysis are presented in Figure 3.13. The mean of the samples is represented by diamonds in Figure 3.13 (a). In order to accept the alternate hypothesis p value should be greater than 0.05. From Figure 3.13 (b) it is possible to say with 95% of confidence that the damage in samples with 0% FBS is greater than the damage in samples with 10%FBS. 3.4.5. Effect of sorting on cell cycle In an attempt to understand what happen to a cell population when it is sorted, we conducted cell cycle analysis profiles before and after sorting. Figure 3.14 is a typical plot of cell cycle before and after sorting THP1 cells. As can be seen the G2 phase is diminished after the cell sorting in the order of 6 %. It is also noticeable the increasing percentage in sub G1 (apoptosis), which corresponds to gate M4. To verify if the population in G2 phase is mainly affected by hydrodynamic stress in the flow cytometer, we used taxol to arrest the population in G2 phase. Taxol is used in some cancer treatments to arrest the cells in G2 phase by interacting with the tubulin/microtubule system and acting as antimitotic agent. In this assay taxol was diluted in DMSO. In order to verify that the concentrations of DMSO used does not affect the cell cycle profile, we used an additional control which had the same concentration of DMSO used for the assay (no taxol present). As can be seen in Figure 3.15 cells resuspended in DMSO (panel B) have a similar cell cycle profile 137 to that of cells in the control without any reagent (Panel A). Panel C shows the cell cycle profile of cells arrested in G2 phase. The results obtained when analyzing cell cycle profiles of cells arrested in G2 phase where consistent with those previously obtained for THP1 where cells were non synchronized. G2 phase decreased in the order of 4.5, 8 and 11% for cells control, cells in DMSO and cells arrested with taxol respectively. Given the variability of the samples a question arose regarding the statistical significance of these assays. To answer this question a series of experiments were developed and cell cycle profile was determined for a number of samples as can be observed in Table 3.4. The majority of assays were developed with THP1 cells but, K562 cell line was also evaluated. Again consistent results were observed regarding the reduction in the fraction of cells in G2 phase after sorting. For statistical analysis only sorting with THP1 cells were considered. A test for paired samples was performed to evaluate the effect of sorting on cell cycle profile. The null (Ho) and alternate (Ha) hypotheses were established as follow: Ho: There is no difference in the G2 fraction between control samples and after sorting. Ha: G2 fraction before sorting is greater than G2 fraction after sorting. That is: Before > After, which is equivalent with Before After <0 . 138 Results from statistical analysis are presented in Figure 3.16. Since the parameter p value < t is less than 0.05, the alternate hypothesis can be accepted. And it might be suggested that cell cycle profile is affected after sorting by reducing the fraction of cells in G2 phase. Decreasing in G2 fraction could be due to the size of the cells, since in G2 phase cells are bigger because they have higher content of DNA. Flow cytometric analysis is based on selecting populations on the plots based on cell size, granularity and other parameters. Differences in samples and operators criteria give a subjective character to this technology. The analysis should be based on percentages of the selected population; hence performing material balances is a difficult task using this technology. 3.4.6. FACS Flow Rate Measurements Figure 3.4a presents the experimentally measured sample uptake flow rate as a function of the Aria flow rate scale for a 70 m nozzle and different sheath pressure. Figure 3.4b presents the experimentally measured sheath fluid flow rate as a function of the Aria flow rate scale for a 70 m nozzle and different sheath pressure. Experimental measurements allow determining the correspondent values of uptake sample flow rate with respect to the scale given by the software in the FACS Aria. The flow rate of sheath fluid is approximately constant along the flow rate Aria scale but as expected it is dependent on 139 the sheath fluid pressure. The estimated flow rates are part of the hydrodynamic characterization of the instrument FACS Aria and are required to determine the EDR values at given conditions in the CFD simulations. 3.4.7. CFD Simulations In order to quantify the hydrodynamic forces that the cells experience during cell sorting, a set of computer simulations was performed. Simulations were performed for the 70 m nozzle and a sheath fluid pressure of 70 psi since those are the conditions commonly used at the Ohio State University Cell Analysis and Sorting Core. The first challenge in developing the simulations was to define the hydrodynamic components of the flow through the FACS Aria. Limited information and drawings of some components were provided; however, the connections and transitions between fluidic elements it is not clearly demonstrated and detailed information of the interior components and dimensions of the nozzle was not provided. In consequence, the definition of the geometry to create the mesh was based on in house measurements and educated deductions, based on understanding the fluid dynamics of FACS instruments in combination with the drawings provided. Fluidic components can be observed in Figure 3.16. The system is notable for a lack of smooth transitions between the components, which causes challenges in meshing the geometries together. The rectangular cuvette and 140 circular nozzle are forced together with an o ring. An sketch of the geometry in 3D drawn in Gambit is presented in Figure 3.17a, however considering the symmetric characteristics of the channel a 2D channel was generated and half of the channel was actually simulated (Figure 3.17b). Figure 3.18 is an output of one of the simulations performed which are color coded to correspond to the EDR at the given location. As can be observed in Figure 3.18a the region of maximum EDR, as expected, is in the contraction where the nozzle is located. A detail of this region is shown in Figure 3.18 b. Results of the simulations performed are reported in Table 3.5; the values of EDR reported correspond to the maximum value a particle experience for a set of specific operating conditions. 3.5. DISCUSSION Results obtained in single shear stress studies with different cell lines corroborate the character cell line specific regarding cell sensitivity that has been reported previously (Mollet, et al. 2007, Seidl et al., 1999). Moreover, the screening methodology has potential of application for purposes of clone selection or as part of cell line characterization. An effort was made to evaluate cell sensitivity of cell lines of monocytic lineage after it was discovered the higher level of sensitivity of the monocytic like cell line THP1. A trend was observed that might imply these cell lines are more sensitive, however the cell 141 line HL60 presented lower levels of sensitivity than the other monocytic like cell lines evaluated (K562, THP1, U937). Possible reasons of the lower damage observed in HL60 include: earlier state of differentiation, size of the cells, presence of different set of proteins that help cells to survive better under hydrodynamic stress conditions. The consistent results of extended lag phase when growing cells after stress exposure (either in shear stress device or in flow cytometer) indicate that the cells are affected physiologically in response to hydrodynamic forces. Metabolic changes had been observed before in studies of physiological effects of flow sorting (Jochem, 2005, Haugen et al., 1987, Rivkin et al 1986). Extended growth lag phase was reported by Jochem et al. and Haugen et al 1987, when sorting marine phytoplankton cells; coincidentally with our study, sorted cells eventually resumed control growth rates after 48 hr of lag phase. Regarding changes in metabolism after sorting, Rivkin, et al. 1986 evaluated rates of photosynthetic carbon uptake by unsorted samples as well as positive and negative sorted cell fractions; they found that rates of carbon uptake were lower in sorted samples than for non sorted samples. From the information above reported and our own results it can be postulated that hydrodynamic stress might alter at least temporally metabolism on cell populations. Our approach to diminish the effect of shear stress on cell growth post stress by collecting and growing sorted cells in enriched media, shows a potential method to 142 overcome the problem of extended lag growth phase post sorting. A similar technique was proposed by Maxwell et al., 1997 in dealing with maintaining membrane integrity of spermatozoa after flow cytometric sorting. The authors proposed an optimized procedure to improve viability of sperm after sorting based on collecting fractions in seminal plasma. The protective effect of serum in hydrodynamic stress exposure has been addressed from physiological and physical point of view (Lopez, et al 2003, Martens et al., 1992, Ramirez et al. 1992, Kunas and Papoutsakis, 1990). It has been speculated that physical mechanisms of protection are attributed to absorption of certain compounds on or into the plasma membrane, modulation of the membrane fluidity, cells sheltered from hydrodynamic forces, by formation of a stable foam, between others. In this work recovery was measured based on visual count and cell death was estimated based on LDH released. The results indicated that cell damage is higher when no FBS is presented in the sorting media. The mean of damage for samples sorted with 0% FBS is 15.46 %, while the mean of damage for samples sorted in media with 10% FBS is 8.6%. Results obtained from cell cycle analysis indicate that the cell cycle population affected the most in sorting assays is G2 phase. Cells in G2 phase appear bigger than other populations since they contain enough amount of DNA for two daughter cells. It can be suggested that the bigger size can have a sort of spatial impediment when flowing 143 through channels. In addition, it has been also found that the size of cells is decreased after continuous hydrodynamic stress exposure (Godoy Silva, et al. 2009). Computer Fluid Dynamics has been used to characterized hydrodynamics in flow cytometer instruments (Yang et al, 2008, Lee et al., 2001, Mollet et al., 2007). In this work computer simulations were performed in order to estimate the level of hydrodynamic stress that the cells experience in the FACS Aria. The energy dissipation rate estimated from the simulations is in the order of 3.99 x105 to 5.14 x 105 W/m3, for the finest grid evaluated. These values are one order of magnitude lower than the minimum values estimated by Mollet et al., 2008 for FACS Vantage. Further analysis is suggested in order to confirm grid independence and verify the actual levels of stress in the FACS Aria. The applications of flow cytometry in biomedical research and pharmaceutical industry are constantly expanded. The latest advances and capabilities to detect simultaneously 13 colors represent savings in reagents and labor (Herzenberg et al, 2002). Still there is room for improvement in detection capabilities, speed of analysis and sorting, and of course optimizing design and operational conditions to diminish cell damage due to hydrodynamic stress. 144 3.6. REFERENCES Aslam, I.; Robins, A; Dowell, K, and Fishel, S. (1998). Isolation, purification and assessment of viability of spermatogenic cells from testicular biopsies of azoospermic men. Human Reproduction. 13 (3): 639 645. Garcia Briones, M. A. and Chalmers, J. J. (1994). Flow parameters associated with hydrodynamic cell injury. Biotechnology and Bioengineering. 44:1089 1098. Gregoriades, N.; Clay, J.; Ma, N.; Koelling, K. and Chalmers, J. J. (2000). Cell damage of microcarrier cultures as a function of local energy dissipation created by a rapid extensional flow. Biotechnology and Bioengineering. 69:171 182. Godoy silva, R., (2008). Lethal and sub lethal effects of hydrodynamic forces on animal cell culture. PhD Dissertation. Chemical and Biomolecular Engineering. The Ohio State University. Godoy silva, R., Chalmers, J.J., Casnocha, S.A., Bass, L.A., and Ma, N. (2009). Physiological Responses of CHO Cells to Repetitive Hydrodynamic Stress. Biotechnology and Bioengineering. 103(6):1103 1117. Godoy silva, R., Berdugo, C. and Chalmers, J. (Peer review). Literature Review: Mixing, Aeration and hydrodynamics in bioreactors. Wiley Encyclopedia of Industrial Biotechnology. Huh, D., Gu, W., Kamotani, Y, Grotberg, J. B. and Takayama, S. (2005). Microfluidics for flow cytometric analysis of cells and particles. Physiological measurement. 26: R73 R98. Haugen, E.M, Cucci, T.L, Yentsch, C.M., Grotberg, and Shapiro, L.P. (1987). Effects of Flow Cytometric Analysis on Morphology and Viability of Fragile Phytoplankton. Applied and Environmental Microbiology. 53 (11): 2677 2679. Herzenberg, L.A.; Parks, D.; Sahaf, B.; Perez, O.; Roederer, M., and Herzenberg, L.A. (2002). The History and Future of the Fluorescence Activated Cell Sorter and Flow Cytometry a View from Stanford. Clinical Chemistry. 48 (10): 1819 1827 145 Ho, C T; Lin, R Z.; Chang, H Y., and Liu, C H. (2005). Micromachined electrochemical T switches for cell sorting applications. Lab Chip. 5: 1248 1258. Jochem, F. (2005). Short Term Physiologic Effects of Mechanical Flow Sorting and the Becton Dickinson Cell Concentrator in Cultures of the Marine Phytoflagellata Emiliania buxleyi and Micromonas pusilla. Cytometry Part A. 65A:77 83 Kunas, K.T., and Papoutsakis, E.T. (1990). The protective effect of serum against hydrodynamic damage of hybridoma cells in agitated and surface aerated bioreactors. Journal of Biotechnology. 15(1 2):57 69 Lee, G B; Hung, C I.; Ke, B J; Huang, G R.; Hwei, B H, and Lai, H F. (2001). Hydrodynamic Focusing for a Micromachined Flow Cytometer. Journal of Fluids Engineering. 123:672 679 Lopez, E.; Figueroa, S.; Oset Gasque, M J. and Gonzalez, M.P. (2003). Apoptosis and necrosis: two distinct events induced by cadmium in cortical neurons in culture. British Journal of Pharmacology. 138:901 911 Ma, N.; Koelling, K. and Chalmers, J. J. (2002). The fabrication and use of a transient contractional flow device to quantify the sensitivity of mammalian and insect cells to hydrodynamic forces. Biotechnology and Bioengineering. 80:428 437. Martens, D.E.; Gooijer, C.D.; Beuvery, E.C. and Tramper, J. (1992). Effect of Serum Concentration on Hybridoma Viable Cell Density and Production of Monoclonal Antibodies in CSTRs and on Shear Sensitivity in Air Lift Loop Reactors. Biotechnology and Bioengineering. 39: 891 897 Maxwell, W.M.C.; Welch, G. R. and Johnson, L. A. (1997). Viability and membrane integrity of spermatozoa after dilution and flow cytometric sorting in the presence or absence of seminal plasma. Reproduction, Fertility and Development. 8:1165 1178 McQueen, A. and Bailey, J. E. (1989). Influence of serum level, cell line, flow type, and viscosity on flow induced lysis of suspended mammalian cells. Biotechnology Letters. 11:531 536. 146 McQueen A.; Melhoc, E. and Bailey, J. E. (1987). Flow effects on the viability and lysis of suspended mammalian cells. Biotechnology Letters. 9:831 836. Mollet, M.; Ma, N.; Zhao, Y.; Brodkey, R.; Taticek, R. and Chalmers, J. (2004). Bioprocess equipment: characterization of energy dissipation rate and its potential to damage cells. Biotechnology Progress 20:1437 1448. 2004. Mollet, M.; Godoy Silva, R.; Berdugo, C. and Chalmers, J. J. (2007). Acute Hydrodynamic Forces and Apoptosis: A Complex Question. Biotechnology and Bioengineering. 98 (4): 772 788. Mollet, M.; Godoy Silva, R.; Berdugo, C. and Chalmers, J. J. (2008). Computer Simulations of the Energy Dissipation Rate in a Fluorescence Activated Cell Sorter: Implications to Cells. Biotechnology and Bioengineering. 100 (2): 260 272. Murray, N.R., Baumgardner, G.P., Burns, D. J. and Fields, A.P. (1993). Protein Kinase C Isotypes in Human Erythroleukemia (K562) Cell Proliferation and Differentiation. The Journal of Biological Chemistry. 268 (21): 15847 15853. Pinkel, D. and Stovel, R. 1985. Flow Chambers and Sample Handling. In: Van Dilla, M. A.; Dean, P. N.; Laerum, O. D. and Melamed, M. R. (Eds). Flow Cytometry: Instrumentation and Data Analysis. Pp.3 128. London: Academic Press,. Ramirez, O.T. and Mutharasan, R. 1992. Effect of Serum on the Plasma Membrane Fluidity of Hybridomas: An Insight into Its Shear Protective Mechanism. Biotechnology Progress. 8: 40 50. Rivkin, R.B.; Phinney, D.A., and Yentsch, C.M. (1986). Effects of flow Cytrometric Analysis and Cell Sorting on Photosynthetic Carbon Uptake by Phytoplankton in Cultures and from Natural Populations. Applied and Environmental Microbiology. 52(4):935 938 Seidl, J.; Knuechel, R., and Kunz Schughart, L. A. (1999). evaluation of membrane physiology following fluorescence activated or magnetic cell separation. Cytometry. 36:102–111. 147 Shapiro, H. W. (2003). Practical Flow cytometry. 4th Ed. New Jersey: John Wiley & Sons. Taylor, G. I. (1934). The formation of emulsions in definable fields of flow. Proceedings of the Royal Society of London. Series A. 146:501 523. Yang, A S.; Hsieh, W H., and Tseng, L Y. (2008). Three dimensional unsteady calculations for a flow cytometer. Proc. IMechE. J. Mechanical Engineering Science. 222. Part C. Zanghi, J.A.; Fussenegger, M., and Bailey, J.E. (1999). Serum Protects Protein Free Competent Chinese Hamster Ovary Cells Against Apoptosis Induced by Nutrient Deprivation in Batch Culture. Biotechnology and Bioengineering. 64(1):108 119. Williams, L. and Peter, Y. (2007). Adult stem cells of the lung in organ dysfunction. Journal of Organ Dysfunction. 3:31 35. 148 Flow rate, Q Average velocity Median of maximum EDR range [W/m3] [mL /min] at the inlet [m/s] EDR[W/m3] 10 0.105 2.9x105 3 x104 1 x106 30 0.316 2.3x106 1 x106 1 x107 50 0.527 6.4x106 2 x106 2 x107 70 0.738 2.6x107 5 x106 8 x107 90 0.949 1.1x108 1 x107 4 x108 Table 3.1. Median and range of energy dissipation rates that cells are exposed to in the Torture Chambers at different volumetric flow rates (Data from Godoy Silva et al 2009). 149 Simulation Grid size Nodes Injection mode 1 0.05 18046 Surface 2 0.05 18046 Group 200 particles 3 0.01 432407 Surface 4 0.01 432407 Surface 5 0.01 432407 Surface 6 0.005 1717302 Group 200 particles 7 0.005 1717302 Surface 8 0.005 1717302 Surface Table 3.2. Grid size and type of injection used in the calculation of EDR 150 Nozzle Sheath sample Max. Nozzle Sheath Sample Max. diameter pressure differential EDR diameter pressure differential EDR ( m) (psi) (psi) (W·m 3) ( m) (psi) (psi) (W·m 3) 70 8 1.0 4.54×106 100 8 1.0 9.69×106 70 8 1.5 1.16×107 100 8 2.0 1.76×107 70 8 2.0 8.07×106 100 17 0.5 1.98×108 70 17 0.5 2.86×108 100 17 1.0 1.00×107 70 17 1.0 1.55×107 100 17 1.5 1.14×107 70 17 1.5 1.83×107 100 17 2.0 2.18×107 70 17 2.0 1.56×107 100 25 0.5 1.98×108 70 25 0.5 5.97×108 100 25 1.0 4.08×107 70 25 1.0 2.61×107 100 25 1.5 1.69×107 70 25 1.5 1.83×107 100 25 2.0 1.86×107 70 25 2.0 5.16×107 100 35 0.5 5.74×108 70 35 1.0 4.26×107 100 35 1.0 5.58×107 70 35 1.5 2.08×107 100 35 1.5 3.74×107 70 35 2.0 4.03×107 100 35 2.0 2.20E+07 100 45 0.5 7.16×108 100 45 1.0 1.34×108 100 45 1.5 2.58×107 100 45 2.0 4.68×107 Table 3.3. Highest level of EDR that a particle would experience in a BD FACSVantage for different operating conditions in a 70 and 100 m nozzle (Adapted from Mollet et al 2008). 151 G2 reduction, Cell line Control Sorted % THP1 25.92 19.28 6.64 THP1 26.51 17.23 9.28 THP1 25.02 18.56 6.46 THP1 24.01 13.7 10.31 THP1 24.68 18.06 6.62 THP1 25.02 16.85 8.17 THP1 15.94 8.6 7.34 THP1 21.43 16.88 4.55 THP1 28.76 15.66 13.1 THP1 19.14 15.6 3.54 K562 24.92 9.9 15.02 K562 36.78 22.41 14.37 K562 16.33 13.24 3.09 K562 22.25 11.3 10.95 K562 21.04 15.44 5.6 THP1 (In DMSO) 23.84 15.77 8.07 THP1 (In DMSO) 33.01 17.02 15.99 THP1 (Taxol) 37.14 26.13 11.01 THP1 (Taxol) 57.35 39.95 17.4 Table 3.4. Effect of sorting on cell cycle profile and fraction of cells in G2 phase. 152 Simulation Grid Flow rate Velocity Velocity sheath Max EDR, size (Aria scale) sample, m/s fluid, m/s W/m3 1 0.05 3 2.6*10-4 2.8*10-3 25779 2 0.05 3 2.6*10-4 2.8*10-3 25779 3 0.01 3 2.6*10-4 2.8*10-3 236853 4 0.01 11 1.3*10-3 2.8*10-3 290596 5 0.01 11 1.3*10-3 2.8*10-3 290596 6 0.005 3 2.6*10-4 2.8*10-3 411632 7 0.005 11 1.3*10-3 2.8*10-3 514680 8 0.005 1 5.2*10-4 2.8*10-3 399854 Table 3.5. Highest level of EDR that a particle would experience in a BD FACS Aria for different operating conditions in a 70 m nozzle. 153 (a) 76.4 6.4 Ø=16.8 r=19.987 10.0 5.0 17.5 r = 2.6 5.345 y (b) x 47.8 r = 2.6 16.8 0.227 6.4 30.0 21.2 6.4 21.2 21.2 y x 0.304 z A (c) Figure 3.1. Photograph (a), top view (b), and a perspective view (c) of the flow contraction device. All measurements are in millimeters. 154 Manual valve A Shear device (Torture chamber) Manual valve B NC Syringe pump Sampling tube LDH Cells reservoir Waste Dead cells Figure 3.2. Single Pass set up: Cells are centrifuged and resuspended into fresh media. They pass once through the Torture chamber at different flow rates (10, 30, 50, 70, 90 mL/min). 155 Sample tube Flow channel Light collecting lenses Sorting orifice Figure 3.3. Simplified sketch flow cytometer 156 0.18 0.16 0.14 0.12 0.1 0.08 0.06 70 psi 0.04 45 psi Sample flow rate rate flowmL/min , Sample 0.02 25 psi 0 0 2 4 6 8 10 12 Flow scale Aria 7 6 5 4 3 2 70 psi Sheath flow rate, rate, flowmL/min Sheath 1 45 psi 25 psi 0 0 2 4 6 8 10 12 Flow scale Aria Figure 3.4. Exit sheath flow rate as a function of flow rate scale and sheath pressure for the 70 m nozzle (a), and sample flow rate as a function of flow rate scale and sheath pressure for a 70 m nozzle (b). 157 95 CHO Sf9 85 MCF 7 75 Hybridoma 65 THP1 55 U937 K562 45 HL60 35 CHO6E6 25 Damage ratio ratio [%] Damage 15 5 5 1.00E+05 1.00E+06 1.00E+07 1.00E+08 1.00E+09 Maximum Local Energy Dissipation Rate [W.m 3] Figure 3.5. Cell damage estimated from single pass experiments in the shear stress device. 10 6 Control Stress TC 90 mL/min 1 0.1 Cellular concentration, cell/mL x 10 cell/mL concentration, Cellular 0.01 0 50 100 150 200 250 300 350 400 Time, h Figure 3.6. Growth kinetic of THP1 cells after stress exposure in shear stress device (TC), cells were exposed at 90 mL/min. 158 10 6 Control Sorted cells 1 0.1 0.01 ellular concentration, cell/mL 10 x cell/mL concentration, ellular C 0 50 100 150 200 250 300 350 400 Time, h Figure 3.7. Growth kinetic of THP1 cells after sorting in FACS Aria, cells were grown in media with 10% FBS. 10 6 Control Sorted cells positive Sorted cells negative 1 0.1 0.01 Cellular concentration, cell/mL x 10 cell/mL concentration, Cellular 0 50 100 150 200 250 300 350 400 Time, h Figure 3.8. Growth kinetic of THP1 cells after sorting in FACS Aria, cells were grown in media with 30% FBS. 159 3.5 Sort 10% Growth 30%+CM 3 Sort 10% Growth 10% 6 Sort 0% Growth 10% 2.5 Sort 0% Growth 0% 2 Sort 0% Growth 30% 1.5 1 ell concentration, cell/mL x10 cell/mL concentration, ell C 0.5 0 0 50 100 150 200 250 300 Time, hr Figure 3.9. Growth kinetic of THP1 cells after sorting in FACS Aria, cells were grown in media with 10% FBS, 30%FBS and conditioned media. 40 30 20 Cell death, % death, Cell 10 0 3 11 23 47 94 Cell concentration, cell/mL x106 Figure 3.10. Cells suspensions at different cell concentrations were sorted under the same conditions. Cell damage was calculated based on the amount of LDH in the supernatant after sorting. 160 1.E+07 Before sorting 9.E+06 After sorting 8.E+06 Aria Report 7.E+06 6.E+06 5.E+06 4.E+06 3.E+06 otal number of cells events cells of number or otal T 2.E+06 1.E+06 0.E+00 1 2 3 4 5 6 7 8 9 Sorting number Figure 3.11. THP1 cells sorted in media with 0% FBS. Cells were counted with hematocytometer before and after sorting. Aria reports number of events. 8.E+06 Before sorting 7.E+06 After Sorting Aria Report 6.E+06 5.E+06 4.E+06 3.E+06 2.E+06 Total number of cells events cells of or number Total 1.E+06 0.E+00 1 2 3 4 5 6 7 8 9 101112 Sorting number Figure 3.12. THP1 cells sorted in media with 10% FBS. Cells were counted with hematocytometer before and after sorting. Aria reports number of events. 161 % death (a) (b) Figure 3.13. Statistical analysis of THP1 cells sorted in media with 10% FBS Vs. cells sorted in media with 0%FBS. 162 ControlBefore before sorting sort.001 M1 M2 Marker Left, Right Events % Gated % Total M3 All 0, 1023 10000 100.00 75.24 G1 M1 154, 246 5028 50.28 37.83 ounts C M4 S M2 246, 339 2231 22.31 16.79 G2 M3 339, 438 2468 24.68 18.57 A M4 0, 154 86 0.86 0.65 0 20 40 60 80 100 120 140 100 12080604020 0 0 200 400 600 800 1000 FL2-A AfterNegative.003 sorting M1 Marker Left, Right Events % Gated % Total M2 M3 All 0, 1023 10000 100.00 76.39 G1 M1 154, 246 5858 58.58 44.75 S M2 246, 339 2045 20.45 15.62 Counts M4 G2 M3 339, 438 1806 18.06 13.80 A M4 0, 154 135 1.35 1.03 0 30 60 90 120 150120 90 60 30 0 0 200 400 600 800 1000 FL2-A Figure 3.14. Cell cycle profile before and after stress exposure in flow cytometer.. 163 Before sorting After sorting Control before sort.001 0 Control after sort.002 A A M1 G1 56.69% M1 G1 63.55% M2 M2 M3 S 15.38% M3 S 13.48% ounts G2 21.43% G2 16.88% C Counts M4 A 5.57 % M4 A 5.4 % 0 40 80 120 160 200160 120 80 40 0 0 20 40 60 80 100 12 80 60 40 20 0 0 200 400 600 800 1000 0 200 400 600 800 1000 FL2-A FL2-A DMSO before sort.003 DMSO after sort.004 B G1 55.28% B G1 61.45% M1 M1 M2 S 16.25% M2 S 14.57% M3 G2 23.84% M3 G2 15.77% Counts A 3.07 % Counts A 7.66 % M4 M4 0 20 40 60 80 100 80 60 40 20 0 100 80 60 40 20 0 0 200 400 600 800 1000 0 200 400 600 800 1000 FL2-A FL2-A TAXOL before sort.005 TAXOL after sort.006 C M1 G1 36.48 % C M1 G1 39.74 % M2 S 5.88 % M2 S 7.9 % M3 G2 37.14 % M3 G2 26.13% Counts Counts A 15.60 % A 22.54 % M4 M4 0 10 20 30 40 50 60 70 60 50 40 30 1020 0 70 60 50 40 30 1020 0 0 200 400 600 800 1000 0 200 400 600 800 1000 FL2-A FL2-A Figure 3.15. Cell cycle profile of cells arrested in G2 phase before and after stress exposure in flow cytometer. 164 20 After sorting G2 10 0 -10 -20 Control G2 10 15 20 25 30 Mean: (After sorting G2+Control G2)/2 Figure 3.16. Statistical analysis of fraction of THP1 cells in G2 phase before and after sorting. 165 Sample port Flow cell Cuvette Nozzle Figure 3.17. Photograph of BD FACS Aria Flow cell components. 166 Figure 3.18. View of the nozzle geometry and mesh used for the simulation. Geometry and mesh were built in Gambit. 167 Figure 3.19. Fluent output of the simulations of particles flowing through the nozzle The color coded figures correspond to the levels of EDR, in units of W·m 3. 168 CHAPTER 4 EFFECT OF IMPELLER-SPARGER CONFIGURATIONS ON MASS TRANSFER CAPABILITIES AND CELL CULTURE PERFORMANCE 4.1. ABSTRACT The performance of ten different configurations Impeller-Sparger was evaluated to select a configuration that supports growth and recombinant protein production at mild stress. A full factorial experimental design with two central points was used to characterize the mass transfer capabilities for every configuration. The best Impeller-Sparger configurations were chosen based on the kLa response surface model for testing in cell culture experiments. Cells Hamster Ovary (CHO cells) were grown to validate the performance of the configurations chosen in terms of cell growth, CO2 accumulation and productivity. The proposed configuration supported high cell density cultures through improved gas dispersion, acceptable shear rates and low foam formation. When sintered sparger was used CO2 accumulation was larger than when open tube sparger was used. The configurations evaluated did not affect product accumulation (titer). 169 4.2. INTRODUCTION Mammalian cell culture is a recognized technology for producing protein therapeutics. The interest in this technology had led to designing large-scale culture strategies. A critical stage in process development is technology transfer from laboratory scale to industrial scale. A successful scale up of the process guarantees that protein quality is equivalent across scales resulting in a product with consistently quality. The most attractive system for large scale cell culture processes is the stirred tank. Mixing is vital in order to make nutrients available for growing cells, keeping a homogeneous and controlled environment and re-suspending the cells. Due to fluid dynamics, however, suspended particles experience hydrodynamic forces along the flow. There is a common perception that mammalian cells are highly sensitive to hydrodynamic forces and that normal operation of bioprocess equipment can create forces strong enough to damage cells. An additional common perception is that cell damage can range from an effect on metabolism, to reduction in growth rate and/or specific productivity, to outright destruction of the cells, with the subsequent complications of cell debris in the culture. The constant interest in the area of shear sensitivity and strategies to minimize impact of shear stress and improving mass transfer is noticeable in the state of art which shows recent publications in the area (Frahm, et al 2009, Matsunaga, et al., 2009, Amano, et al., 2008, Koynov, et al., 2007, Yang, et al., 2007, Jain and Kumar, 2008). 170 There is a number of parameters reported as key elements in designing bioreactors, among the most significant we can find: volumetric mass transfer coefficient (kLa), low volumetric power input, low shear stress, ( Jain and Kumar, 2008, Frahm et al., 2009). In general, a balance should be accomplished to obtain the best mass transfer capabilities and low shear stress during bioreactor design. Since, mass transfer capabilities are determined by the bioreactor’s agitation and aeration configuration, the design of impeller and sparger regulates the hydrodynamic conditions in the bioreactor. Different approaches have been considered to evaluate oxygen supply and effects of aeration and agitation in cell cultures. Most of these approaches involved numerical calculations and simulations which represent an advantage in the sense that many conditions can be explored without actual experiments. Nonetheless, if time and resources are available experimental assays are the most reliable way in solving problems and finding answers. Furthermore, a combination of computational and experimental approaches is desired in bioreactor design. Although studies on impeller, spargers and combination of them have been performed for long time (Aunins, et al., 1989, Martinez, et al 1989, Kawase, et. Al., 1992, Moreira, et al., 1995, Sardeing, et al., 2004, Puthli, 2005), results reported might be no consistent. A possible reason for contradictory results is that different attributes results in different mass transfer, shear dynamics and mixing profiles in the bioreactor, which affects cell culture performance and the subsequent protein production. Moreover there are 171 biological aspects that are system specific such as shear sensitivity, sensitivity to O2 and CO2, nutrient requirements between others (Nienow, 2006). Hence development of studies system specific should be considered in designing or optimizing cell cultures. In this work, the performance of ten different Impeller-Sparger configurations was evaluated. A full factorial experimental design with two center points was used to characterize the mass transfer capabilities for every configuration. The best Impeller- O2 sparger configuration was chosen based on the K L a response surface model for testing in cell culture experiments. The evaluation included different types of sparger and impeller as well as relative location of impeller along the shaft and clearance between impellers. Results indicated that the volumetric oxygen mass transfer coefficient is affected by the relative sparger location is less affected by clearance between impellers. Once the best impeller sparger configuration was identified a cell culture experiment was carried out to assess its effects on culture performance. The proposed configuration supported high cell density cultures through improved gas dispersion, acceptable shear rates and low foam formation. 172 4.3. MATERIALS AND METHODS O2 4.3.1. Methodology to evaluate K L a O2 Measurement of K L a was carried out in glass bioreactors at 2L working volume. In order to aid in finding suitable operating conditions, a chemically defined media was used O2 to determine K L a . Prior to any measurement dissolved oxygen probes were calibrated at 0-100% dissolved oxygen values with air. Probes were placed in bioreactors with Phosphate Buffered Saline (PBS) solution and autoclaved. Probes were recalibrated in the media to 0% dO2 by stripping oxygen using pure nitrogen injected into the bioreactor through the sintered sparger. 100% dO2 calibration point was achieved by injecting air (21% O2) into the bioreactor until no variation was seen in the transmitter box. O2 For all K L a measurements air was used and, to strip dissolved oxygen out of the bioreactor between measurements, pure nitrogen was injected through the sparger. All measurements started at 0% dO2 value up to 80-90% saturation. Air flow rate was measured through mass flow controllers. Combinations of three different types of impellers and 3 types of spargers were evaluated. Figure 4.1 shows two of the impellers evaluated. Rushton impeller and Pitch 173 Blade Turbine (PBT) impeller; the third impeller is a modification of PBT impeller which consists in larger blades. Spargers evaluated include: open tube sparger, sintered sparger of 50 m pore size and sintered sparger of 100 m pore size. Sparger position was evaluated in three different locations: below the lower impeller center, edge of impeller and side of the bioreactor, the experimental design included also distance between sparger and lower impeller (h) as well as interimpeller clearance (C). Figure 4.2 shows schematically the variables involved in the evaluation of the effect of sparger location on O2 O2 K L a . To evaluate the effect of Antifoam on K L a Antifoam at a concentration of100 ppm was used. 4.3.2. Experimental design The experimental design for the sparger location according with Figure 4.2 is in Table 4.1. The separation between impellers is highly variable, going from separation of 1 diameter impeller (D) until 2D where the flow patterns generated by each impeller may interfere. The total power drawn into the system can be estimated by the sum of the power drawn by each impeller as long as the patterns do not significantly interfere with each other (Godoy-silva, et al 2010) In this work we chose to evaluate a separation between impellers of 1D (4.3 cm) and 1.5D (6.5 cm). The off bottom clearance (h), also varies from 0.1-0.3 diameter of vessel (T) in this work 2 positions were evaluated correspondent to 0.3T (4.5 cm) and 0.2 T (2 cm). 174 Table 4.2 shows the different configurations impeller-sparger evaluated. A Central O2 Composite Design (CCD) was created to map the K L a capabilities of the 2L working volume bioreactors (JMP, release 6.0.0, SAS Institute, USA). Two factors were included, flow rate of air through the sparger and agitation speed. Each experimental condition was run in duplicate with 2 center points for a total of 12 observations. Since there were 2 bioreactors available, measurements were assigned randomly to each one of the bioreactors for a total of 6 measurements per bioreactor. Tables 4.3 to 4.5 show the experimental conditions for the configurations including perforated tube, sintered sparger 100 m and sintered sparger 50 m respectively. O2 4.3.3. Volumetric oxygen mass transfer coefficient calculation K L a This parameter was used to evaluate the mass transfer capabilities in different impeller- O2 sparger configurations in bioreactors. Volumetric oxygen mass transfer coefficient K L a , is a parameter commonly used as criteria for scaling up in aerobic fermentations since oxygen uptake could become a limiting factor in the culture. A critical step in aerobic systems is oxygen transport from the gas/liquid interface to the bulk of the liquid, hence it is required to guarantee an efficient design that considers the effect of mass transfer. O2 Using the dynamic method, K L a was calculated using: dC L = K O2 a(C* − C)− Q X (1) dt L O2 O2 175 Where: O2 K L a = Volumetric oxygen mass transfer C * = Saturation concentration of oxygen in water O2 C = Concentration of oxygen in water at time t Q X = Oxygen Uptake Rate (OUR) O2 The term Q X in (1) becomes zero since there are no cells in the liquid, then (1) O2 reduces to: dC L = K O2 a()C * − C (2) dt L O2 L Integrating (2) between the conditions 1 @ t1 = 0, C= CL 2 @ t2 = t, C= C L Eq. (2) yields: ln(C * − C 2 ) = −K O2 a(t − t )+ ln(C * − C1 ) (3) O2 L L 2 1 O2 L Then, the slope that results from by plotting ln(C * − C 2 )/(C * − C1 ) vs. time will O2 L O2 L O2 O2 yield K L a . Units of K L a will be reported in {1/hr} 176 O2 Once all K L a values were calculated, they were used for analysis using JMP (SAS Institute) to estimate the prediction expression and identify the significance of all terms involved. 4.3.4. Power number and Energy Dissipation rate calculation The power number (power transferred to the liquid during agitation) can be calculated O2 from the K L a relationships: α O2 P β K L a = k ()vs (4) V where: P = Power transferred to the liquid [=] W V = Volume of liquid in the bioreactor [=] m3 vs = Superficial velocity of the gas [=] m/s k, α, β are constants that depend on the system vs can be calculated as Q v = g (5) s Π T 2 4 where: 3 Qg = Gas flow rate [=] m /s T = Bioreactor diameter [=] m The power (P) transferred to the liquid during agitation can be calculated as: 177 3 5 P = P0 ρN D (6) where: P = Power [=] W P0 = Power number, impeller dependent [=] -- ρ = Density of the liquid [=] kg/m3 N = Agitation speed [=] s-1 D = Impeller diameter [=] m P0 and the constants k, α, β were fitted to the model by minimizing the sum of the O2 square error (SSE) between the measured and calculated K L a values using an Excel spreadsheet. P0 is used to calculate P/V that can be considered as the average EDR in the reactor. Energy Dissipation Rate, EDR can be used to represent hydrodynamic force in bioprocesses. 4.3.5. Cell Culture Experiments A clonal CHO cell line was used to evaluate cell culture experiments in order to assess the configuration’s effects on cell culture performance. A set of 3 L bioreactors (nominal volume) with 2 L working volume (Applikon, Inc., Foster city, CA), were used in the production step. Bioreactors were controlled at 37° C using a heating blanket. The dissolved oxygen concentration was controlled at 30% of air saturation by sparging air, pH was controlled at 7.0 by bicarbonate- CO2 system. Bioreactors are sampled daily for analysis (cell density, viability, DO, pH, glucose and lactate). A chemically defined 178 media was used in cell culture experiments. The composition of the media is a proprietary formulation. 4.4. RESULTS AND DISCUSSION O2 4.4.1. Estimation of K L a at 2L scale O2 Calculation of K L a was carried out using the dynamic method, which consists of: - De-oxygenation using Nitrogen - Aeration and monitoring dissolved oxygen concentration as a function of time - Calculate oxygen transfer using the slope of the aeration curve To illustrate this procedure, Table 4.6 shows the data collected from the analysis of configuration C2 (Perforated tube and Dual impeller PBT-Rushton). Data was collected and retrieve from the DCS (Distributed Control System) historian. Figure 4.3 shows the oxygen saturation curve, from which the saturation point and the range of highest rate of change can be estimated. Punctual dissolved oxygen concentrations (CL) were used to calculate ln(C * − C 2 )/(C * − C1 ) using equation (3). Notice that the points used to O2 L O2 L calculate the slope are in the range of O% dissolved oxygen up to 98% saturation. Then the slope that results from plotting ln(C * − C 2 )/(C * − C1 ) vs. time will yield K O2 a (See O2 L O2 L L O2 Figure 4.4). Units of K L a will be reported in {1/hr}. 179 4.4.2. Data analysis for different configurations Data collected from every configuration was analyzed using statistical software JMP ® 8.0 (SAS Institute Inc.), to estimate prediction expression and identify the significance of all terms involved. Prediction expressions allow determining mass transfer capabilities at different conditions of the parameters evaluated (agitation- RPM, air flow rate- Qg). Based on statistical parameters it is possible to determine if the prediction expression is O2 acceptable to evaluate K L a in a particular configuration. Figure 4.5 and Tables 4.6 and 4.7 summarize the procedure to calculate the prediction expression for configuration C2 (Perforated tube and Dual impeller PBT-Rushton). O2 Figure 4.5 shows the Actual vs. Predicted K L a after fitting a response surface to data O2 obtained from the K L a assay. The value RSquare in Table 6, is a measure of the proportion of variation around the mean explained by the linear or polynomial model. The value goes from 0 to 1. If RSquare value is 1, it would mean that the model fits perfectly. A larger Rsquare value and smaller Root Mean Square indicates a better fitted model. RSquare Adj is a value of Rsquare adjusted to make it more comparable over models with different numbers of parameters. Parameters in Table 4.6 for the configuration C2 indicate that the prediction equation obtained accounts for 86% of the variability. Table 4.7 lists the parameter estimates of the linear model. The prediction formula is the linear combination of these estimates with the values of their corresponding variables (Ref: JMP Manual). The significance of a parameter or 180 interaction of parameters is determined by the p-value. A value below 0.05 indicates that the parameter is significantly different from zero. Thus, from Table 4.7 it can be suggested that the parameters Qg (p-value 0.0009), RPM (p-value 0.0004) and the interaction Qg*RPM (p-value 0.0228) are all significant. Then the formula to calculate O2 K L a for the configuration C2 is given by: Qg −125 RPM − 300 RPM − 300 Qg −125 O2 K L a = .6 32 + .2 68* + .3 02* + .1 48* * 75 100 100 75 O2 For every configuration the K L a prediction equation was estimated following the procedure explained above. Table 4.8 shows the prediction equation for the configurations evaluated. First order interactions are statistically significant for all the configurations except for configurations C6 (Sparger: Perforated tube, Impeller: PBT Modified) and C7 (Sparger: Perforated tube, Impeller: Dual Impeller PBT downpumping- Rushton). Comments on the prediction equations for the configurations evaluated: O2 Configuration 2: In the original model the residuals increase as K L a increase, after applying a BoxCoxY transformation the residual by prediction plot shows a better distribution and the interaction RPM*Qg is not significative anymore. However, the elimination of the term RPM*Qg from the prediction equation, results in a less accurate fit. Therefore, the original model was chosen. 181 Configuration 3: There was not a significative improvement in the residuals distribution after transformation. Main effects and interaction RPM*Qg are significative. O2 In this case we choose K L a model (original model) with all effects. Configuration 4: The original model without transformation is good enough and O2 there is not a significant improvement after transformation, therefore the K L a original model was chosen. Configuration 5: The original model is similar to that of the transformation. O2 However the normal distribution for the transformation fits better, therefore the K L aX model (BoxCoxY transformation) was chosen. Configuration 6: The original model and the transformation are similar. The normal distribution for the transformation fits better, and the effect Qg*RPM is not O2 significant after transformation, therefore the K L aX model (BoxCoxY transformation) with the main effects was chosen. Configuration 7: The residuals for the original model look similar to that of the transformation. The normal distribution seems to fit better for the transformation. O2 Interaction between RPM*Qg is not significant according with K L a model (original) O2 and K L aX model (transformed), RSquare for kLa model is 0.94 and RSquare for O2 O2 K L aX model is 0.96 therefore the K L aX model was chosen in this case. Configuration 10: The residuals for the original model look similar to that of the transformation. The normal distribution seems to fit better for the transformation O2 O2 ( K L aX ). RSquare for the transformed model is 0.97, the K L aX model was chosen for this configuration. 182 4.4.3. Surface Response Analysis 4.4.3.1. Effect of sparger position on mass transfer coefficient Different dispersion patterns are observed depending on the location of the sparger with respect to the lower impeller as well as the distance between impellers and the distance between the lower impeller and the sparger, this different patterns affect the mass transfer in the system therefore we evaluated those variables to choose the location that result in the higher mass transfer coefficients. O2 Figure 4.6 shows the K L a surface profiles for the experimental design to evaluate location of the sparger according to Figure 4.2 and experimental design in Table 4.1. The volumetric oxygen mass transfer coefficient is dramatically affected by sparger location with respect to the lower impeller (Side of the tank, edge of the impeller, center of impeller) and is less affected by high of the sparger with respect to the lower impeller and clearance between impellers. It seems that a distance of 0.33*diameter (6.5 cm) of the tank is better than 0.25*diameter (4.5 cm). In other words, the distance between impellers for the 3 L bioreactor should be about 6.5 cm in order to increase the mass transfer O2 coefficient. On the other hand the maximum K L a when the sparger was located at the center below the lower impeller was 9.41 h-1 while it increases up to 12.24 h-1 when is located at the side or at the edge of the impeller. 183 4.4.3.2. Effect of antifoam on mass transfer coefficient The mass transfer coefficient was reduced in about 40% when the media has 100 ppm of antifoam. (See Figure 4.7). Although there is a significant decrease in the mass transfer capabilities the concentration of antifoam used in this assay is likely to be superior to the O2 actual concentration used in the cultures. Moreover, the K L a reached is estimated to be sufficient to support the cell culture. 4.4.3.3. Analysis of reactor configurations that include perforated tube The best configuration according with the surface plot on Figure 4.8 is that with the combination PBT- Rushton impeller. As can be seen in Table 4.10 this configuration O2 presented the higher mass transfer coefficient in this group (C2- Sparger at center: K L a : 2.03 – 13.43 h-1). The first row in Table 4.10 corresponds to the same configuration but in this assay the sparger was mistakenly located at the side of the impeller, the results corroborate the observations in section 4.3.3.1, where it was determined that this location has lower efficiency in terms of mass transfer. 4.4.3.4. Analysis of reactor configurations that include sintered sparger The best configuration according with the surface plot on Figure 4.9 is the one that includes sintered 50 m sparger and PBT impeller. As can be seen in Table 4.11 this 184 O2 configuration presented the higher mass transfer coefficient in this group ( K L a : 0.68 – 17.81 h-1). Figure 4.9 shows clearly better mass transfer capabilities for configurations with 50 m sparger than from those with 100 m sparger. The impeller PBT Mod (larger blades), did not show any improvement with respect to the regular PBT, and the effect on mass transfer coefficient seems to be more affected by the pore size of the sparger rather than the type of impeller in this set of experiments. 4.4.3.5. Analysis of reactor configurations that include PBT down-pumping and rushton impeller in the bottom. Figure 4.10 shows the surface plot for these configurations. In these set of experiments other combinations were tested including PBT down pumping, Dual impeller (PBT – Rushton) with Rushton impeller in the top, and PBT Mod – Rushton, and their performance was compared with one of the best configurations previously evaluated (C2). As can be seen in Figure 4.10 and Table 4.12 none of these configurations is superior in terms of mass transfer coefficient with respect to the configuration C2. 4.4.4. EDR calculation in the configurations evaluated. Energy Dissipation Rate (EDR) is a parameter used to estimate the level of hydrodynamic forces in a bioprocess unit. The power input for the best configurations was calculated in order to determine what EDR the cells are exposed in every configuration and estimate the damage by comparison with previous works. Gas flow 185 O2 rates, agitation speeds and measured K L a were tabulated and formulas (4) – (6) used to O2 calculate K L a . The values of power number (P0) and the constants k, α, β were fitted by minimizing the sum of the square error (SSE). Then the impeller power number (P0) obtained is used for calculations of power per volume or EDR. Tables 4.13 and 4.14 summarize the assumptions and results of the fitted model for configurations C2 and C5 respectively. The EDR was calculated from equations 4, 5 and 6. The values of EDR for every configuration are listed next: Configuration C2 (Perforated tube, dual impeller): 0.019 W/m3 Configuration C5 (Sintered sparger, PBT impeller): 0.021 W/m3 The levels of EDR reported in various bioprocess environments have been reported and they can be observed in Figure 4.11. The EDR estimated for configurations C2 and C5 are about 6 orders of magnitude lower than the EDR that has been reported causes cell damage. 4.4.5. Cell Culture. Cell culture experiments were developed to validate the performance of the configurations chosen in terms of cell growth, CO2 and productivity. Bioreactors were sampled daily to determine TCC, VCC, viability, metabolites, dO2, dCO2 accumulation, and product accumulation (titer). 186 Figure 4.12, 4.13 and 4.14 show the kinetic behavior of cell culture in the bioreactors with best configurations in terms of mass transfer capabilities. As can be seen in Figure 4.12 growth does not seem to be affected by the different hydrodynamic conditions evaluated. On the other hand, as can be seen in Figure 4.13 significant CO2 accumulation was observed when sintered sparger was used, however the titer reached in the configurations evaluated seems not to be affected by the different hydrodynamic conditions evaluated in spite of the accumulation of CO2 (See Figure 4.14). 4.5. CONCLUSIONS The effect of different hydrodynamic conditions on mass transfer capabilities and culture performance was evaluated in this work. Different hydrodynamic conditions were assessed for various impeller/sparger configurations. O2 After evaluating the effect of impeller and sparger location on K L a it was determined that volumetric mass transfer coefficient is affected by relative sparger location and less affected by clearance between impellers. Results indicate that the greatest values of overall mass transfer coefficient were obtained with configuration that includes dual impeller (PBT in the top and Rushton in the bottom) 187 and open tube sparger. The second best configuration includes PBT impeller and sintered sparger of 50 m. Configurations proposed support high cell density cultures through improved gas dispersion. Significant CO2 accumulation was observed when sintered sparger was used. Different configurations evaluated did not affect product accumulation (titer) design and the sample volumetric flow rate in that increasing A correlation was defined for each configuration evaluated that allows the calculation of kLa in a range of conditions. Those correlations represent a model that can be evaluated by simulating the hydrodynamics in the bioreactor. Ongoing studies in our laboratory are focused on further investigating numerical simulations using CFD to estimate the mass transfer coefficient for the configurations evaluated. The goal is to find an agreement between experimental analysis and numerical simulations for calculating mass transfer coefficients. 188 4.6. REFERENCES Amano, K., Haga, R. and Murakami, S. 2008. Expressions of mass transfer coefficient of bubbles and free surface of culture tanks using the k-e turbulence model. J Ind Microbiol Biotechnol. 35: 525-531. Aunins, J.G., Woodson, B.A., Hale, T. and Wang, D. 1989. Effects of Paddle Impeller Geometry on Power Input and mass transfer in small-scale animal cell culture vessels. Biotechnology and Bioengineering. 34: 1127-1132. Bird, R. B.; Stewart, W. E.; Lightfoot, E. N. (2001). Transport Phenomena,2nd Ed. New York: John Wiley and Sons. Brodkey, R. S. (1995). The Phenomena of Fluid Motions. P. 129. New York: Dover Publishing. Frahm, B., Brod, H., and Langer, U. 2009. Improving bioreactor cultivation conditions for sensitive cell lines by dynamic membrane aeration. Cytotechnology. 59: 17 – 30. Garcia-Briones, M. A. and Chalmers, J. J. (1994). Flow parameters associated with hydrodynamic cell injury. Biotechnology and Bioengineering. 44:1089-1098. Graf, R., Apenberg, S., Freyberg, M. and Friedl, P. (2003). A common mechanism for the mechanosensitive regulation of apoptosis in different cell types and for different mechanical stimuli. Apoptosis. 8: 531–538. Gregoriades, N.; Clay, J.; Ma, N.; Koelling, K. and Chalmers, J. J. (2000). Cell damage of microcarrier cultures as a function of local energy dissipation created by a rapid extensional flow. Biotechnology and Bioengineering. 69:171-182. Godoy-silva, R, Berdugo, C, Chalmers, J. Mixing, aeration and hydrodynamics in bioreactors. In review. Jain, E., and Kumar, A. 2008. Upstream processes in antibody production: Evaluation of critical parameters. 26: 46-72. 189 Kawase, Y., Halard, B. and Moo-Young, M. 1991. Liquid-Phase mass transfer coefficients in bioreactors. Biotechnology and Bioengineering. 39: 1133-1140. Koynov, A., Tryggvason, G., Khinast, J.G. 2007. Characterization of the localized hydrodynamic shear forces and dissolved oxygen distribution in sparged bioreactors. 97 (2): 317 – 331 Ma, N.; Koelling, K. and Chalmers, J. J. (2002). The fabrication and use of a transient contractional flow device to quantify the sensitivity of mammalian and insect cells to hydrodynamic forces. Biotechnology and Bioengineering. 80:428-437. Mollet, M.; Ma, N.; Zhao, Y.; Brodkey, R.; Taticek, R. and Chalmers, J. (2004). Bioprocess equipment: characterization of energy dissipation rate and its potential to damage cells. Biotechnology Progress 20:1437-1448. 2004. Mollet, M.; Godoy-Silva, R.; Berdugo, C. and Chalmers, J. J. (2007). Acute Hydrodynamic Forces and Apoptosis: A Complex Question. Biotechnology and Bioengineering. 98 (4): 772-788. Matsunaga, N., Kano, K., Maki, Y., and Dobashi, T. 2009. Culture scale-up studies as seen from the viewpoint of oxygen supply and dissolved carbon dioxide stripping. Journal of Bioscience and Bioengineering. 107 (4): 412 – 418. Martinez, A., Galindo, E. and Salvador, M. 1989. Sparger position effect over kLa in bench and pilot stirred tank fermentors. Journal of fermentation and bioengineering. 68 (1): 71-73. Marks, D. 2003. Equipment design considerations for large scale cell culture. Cytotechnology 42: 21-33. Mollet, M., Ma, N., Zhao, Y., Brodkey, R., Taticek and Chalmers, J. 2004. Biotechnol. Prog. 20(5):1437-144 190 Moreira, J. L., Cruz, P.E., Santana, P.C. and Feliciano, A.S. 1995. Influence of power input and aeration method on mass transfer in a laboratory animal cell culture vessel. J. Chem. Tech. Biotechnol. 62: 118-131 Nienow, A.W. 2006. Reactor engineering in large scale animal cell culture. Cytotechnology 50: 9-33 Puthli, M.S., Rathod, V.K., and Pandit, A.B. 2005. Gas-liquid mass transfer studies with triple impeller system on a laboratory scale bioreactor. Biochemical Engineering Journal. 23: 25-30 Sardeing, R., Aubin, J., Poux, M and Xuereb, C. 2004. Gas-liquid mass transfer. Influence of Sparger location. Trans IChemE, Part A. September 2004. Chemical Engineering Research and Design. 82(A9):1161-1168 Yang, J-D., Lu, C., Stasny, B., Henley, J., Guinto, W., Gonzalez, C. Gleason, J., Fung, M., Collopy, B., Benjamino, M., Gangi, J., Hanson, M., Ille, E. 2007. Fed-Batch bioreactor process scale-up from 3L to 2500L scale for monoclonal antibody production from cell culture. Biotechnology and Bioengineering. 98 (1): 141-154. 191 Position sparger High (h), cm Clearance (C), cm Side -2 4.3 Side -2 4.3 Side -2 6.5 Side -2 6.5 Side -4.5 4.3 Side -4.5 4.3 Side -4.5 6.5 Side -4.5 6.5 Edge -2 4.3 Edge -2 4.3 Edge -2 6.5 Edge -2 6.5 Edge -4.5 4.3 Edge -4.5 4.3 Edge -4.5 6.5 Edge -4.5 6.5 Centered -2 4.3 Centered -2 4.3 Centered -2 6.5 Centered -2 6.5 Centered -4.5 4.3 Centered -4.5 4.3 Centered -4.5 6.5 Centered -4.5 6.5 Table 4.1. Experimental design to evaluate effect of sparger position. Side: Sparger is located at side of the impeller near to the wall of the tank. Edge: Sparger is located below the edge of the impeller. Centered: Sparger is located under the impeller in the center. High (h): Distance of the sparger with respect to the lower impeller, negative (-) indicates negative value in the axes with respect to the impeller located in the center (0, 0, 0). Clearance (C): distance between impellers. 192 Configuration Sparger Impeller 1 SIN – 5 m PBT 2 Perforated tube Dual Impeller (PBT – Rushton) 2A Perforated tube Dual Impeller (PBT – Rushton) 3 SIN – 100 m PBT 3A SIN – 100 m (Centered) PBT 4 Perforated tube PBT 5 SIN – 50 m Dual Impeller (PBT_Modified –Rushton) 6 Perforated tube PBT_Modified 7 Perforated tube Dual Impeller (PBT-Downpumping – Rushton) 8 SIN – 100 m PBT_Modified 9 SIN – 50 m PBT_Modified 10 Perforated tube Dual Impeller (PBT_bottom –Rushton top) 11 Perforated tube Dual impeller (PBT – Rushton) Table 4.2. Configurations impeller-sparger evaluated. Flow rate Agitator speed Pattern (sccm) (RPM) +- 200 200 +- 200 200 -- 50 200 -- 50 200 00 125 300 00 125 300 ++ 200 400 ++ 200 400 -+ 50 400 -+ 50 400 00 125 300 00 125 300 O2 Table 4.3. Experimental table to test K L a capabilities of the 2L bioreactors for configurations that include perforated tube. (00) indicates center points, (-+ or similar pattern ) indicate corner points. 193 Flow rate Agitator speed Pattern (sccm) (RPM) +- 100 200 +- 100 200 -- 10 200 -- 10 200 00 55 300 00 55 300 ++ 100 400 ++ 100 400 -+ 10 400 -+ 10 400 00 55 300 00 55 300 O2 Table 4.4. Experimental table to test K L a capabilities of the 2L bioreactors for configurations that include sintered sparger. (00) indicates center points, (-+, or similar pattern) indicate corner points. Flow rate Agitator speed Pattern (sccm) (RPM) +- 100 200 +- 100 200 -- 5 200 -- 5 200 00 52.5 300 00 52.5 300 ++ 100 400 ++ 100 400 -+ 5 400 -+ 5 400 00 52.5 300 00 52.5 300 O2 Table 4.5. Experimental table to test K L a capabilities of the 2L bioreactors for configurations that include sintered sparger of 50 m size pore. (00) indicates center points, (-+, or similar pattern) indicate corner points. 194 Configuration C2 Dual Impeller: PBT – Rushton Sparger: Perforated tube N= 400 rpm, Qg=200 sccm, T = 37 °C * 2 * 1 Time (min) dO2 (%) ln(C − C )/(C − C ) O2 L O2 L 0 0.4051 1 0.4051 2 0.2893 3 0.2893 4 0.7523 5 17.9398 -0.1902 6 34.8380 -0.4207 7 48.6690 -0.6593 8 59.8669 -0.9054 9 69.4734 -1.1790 10 76.6782 -1.4482 11 82.9572 -1.7619 12 87.8183 -2.0977 13 91.9849 -2.5163 14 94.9363 -2.9755 15 97.3380 -3.6185 16 99.3634 -5.0493 17 99.7974 -6.1944 18 99.7974 -6.1944 19 99.7974 -6.1944 20 99.7974 -6.1944 O2 Table 4.6. Data treatment to calculate K L a , Configuration C2 . 195 RSquare 0.893009 RSquare Adj 0.852888 Root Mean Square Error 1.487576 Mean of Response 6.323167 Observations (or Sum Wgts) 12 Table 4.7. Summary of Fit. Term Estimate Std Error t Ratio Prob>|t| Intercept 6.3231667 0.429426 14.72 <.0001 Flow rate Qg 2.67975 0.525938 5.10 0.0009 RPM (200,400) 3.01725 0.525938 5.74 0.0004 Qg*RPM 1.47825 0.525938 2.81 0.0228 Table 4.8. Parameter Estimates. 196 Conf. Empirical formula Qg −125 RPM − 300 RPM −300 Qg −125 O2 C2 K L a = .6 32 + .2 68* + .3 02* + .1 48* * 75 100 100 75 Qg − 55 RPM − 300 RPM − 300 Qg − 55 O2 C3 K L a = .5 22 + .3 87 * + .1 79 * + .1 34 * * 45 100 100 45 Qg −125 RPM − 300 RPM − 300 Qg −125 O2 C4 K L a = .3 85 + .1 63* + .1 71* + .0 79* * 75 100 100 75 Qg − 52 5. RPM −300 RPM − 300 Qg − 52 5. O2 C5 K L a = .6 03+ .5 95* + .1 79* + .1 56* * 47 5. 100 100 47 5. 197 C6 Qg −125 RPM − 300 O2 K L a = .1 11+ .0 84* + .0 71* 75 100 Qg −125 RPM − 300 O2 C7 K L a = .5 54 + .1 61* + .2 81* 75 100 Q − 55 Q −55 O2 g RPM − 300 RPM −300 g C8 K L a = .2 53+ .2 06* + .1 45* + .0 66* * 45 100 100 45 Q − 52 5. Q − 52 5. O2 g RPM − 300 RPM − 300 g C9 K L a = .7 89 + .6 29* + .1 84* + .1 62* * 47 5. 100 100 47 5. Q −125 Q −125 O2 g RPM − 300 RPM − 300 g C10 K L a = .3 70 + .1 36* + .1 90* + .0 39* * 75 100 100 75 O2 Table 4.9. Prediction correlations to calculate K L a for the configurations evaluated. 1 97 Configuration Impeller kLa range 2 PBT - Rushton 1.55 - 6.51 2A PBT - Rushton 2.03 - 13.43 4 PBT Mod - Rushton 1.34 - 8.03 6 PBT Modified 0.66 - 3.8 O2 Table 4.10. K L a range reached with configurations including perforated tube and a combination of impellers. Configuration Impeller O2 K L a range 3A (100 m) PBT 0.86 - 12.2 8 (100 m PBT Modified 0.61 - 7.98 5 (50 m) PBT 0.68 - 17.81 9 (50 m) PBT Modified 0.85 - 17.12 O2 Table 4.11. K L a range reached with configurations including sintered sparger and a combination of impellers. Configuration Impeller O2 K L a range 2A PBT - Rushton 2.03 - 13.43 7 PBT Downpum 1.58 – 10.42 10 Rushton top- PBT bottom 1.46 - 8.35 4 PBT Mod. - Rushton 1.34 - 8.03 O2 Table 4.12. K L a range reached with configurations including sintered sparger and a combination of impellers. 198 Data for calculations Results Tank diameter T= 0.129 m Liquid density ρ= 1050 kg/m3 Po 0.937 (assumed) Agitation speed N= 370 1/min k 0.204 Impeller diameter Di= 0.043 m α 0.508 Liquid volume in V= 0.0025 m3 β 0.672 the bioreactor Table 4.13. Input data and power number and constants calculations using measured O2 K L a for configuration C2 (Dual impeller and perforated tube). Data for calculations Results Tank diameter T= 0.129 m Liquid density ρ= 1050 kg/m3 Po 0.839 (assumed) Agitation speed N= 370 1/min k 10.165 Impeller diameter Di= 0.043 m α 0.309 Liquid volume in V= 0.0025 m3 β 0.952 the bioreactor Table 4.14. Input data and power number and constants calculations using measured O2 K L a for configuration C5 (PBT impeller and sintered sparger). 199 (a). Rushton (b). PBT Figure 4.1. Type of impellers evaluated: Rushton (a), Pitch Balde Turbine (b). C h Sparger location Figure 4.2. Geometrical configuration to evaluate effect of impeller and sparger O2 location on K L a . h: distance between lower impeller and sparger, C: distance between impellers, sparger location: side, center, edge. 200 Dual Imp PBT Rushton. Perforated tube N= 400 rpm, Qg = 200 sccm 120 100 80 , % 2 60 dO 40 20 0 0 5 10 15 20 25 Time, min O2 Figure 4.3. Sample calculation of K L a from collected data. DO profile. Time, min 0 5 10 15 0 1 2 y = 0.2613x + 1.1282 3 R2 = 0.9942 ln(C* CL2)/(C* CL1) 4 O2 Figure 4.4. Sample calculation of K L a from collected data. Obtaining slope. 201 15 10 kLa Actual kLa 5 0 0 5 10 15 kLa Predicted P=0.0003 RSq=0.89 RMSE=1.4876 O2 Figure 4.5. Actual vs. Predicted K L a after fitting a response surface to data O2 collected during the K L a assay for configuration C2. O2 Figure 4.6. Effect of location of sparger on K L a Sparger was located in three different positions: At side of the impeller, near to the wall of the tank, edge of the impeller and centered under the impeller. 202 O2 Figure 4.7. Surface response results on effect of antifoam on K L a . Legend C2_PBT-Rushton C2A_PBT-Rush.Centered C4_PBTMOD-RUSH. C6_PBTMOD O2 Figure 4.8. Effect of configuration impeller/sparger on K L a . Perforated tube and combination of impellers. 203 Legend C3A SIN100 - PBT C8 SIN100 - PBT Mod C5 SIN50 - PBT C9 SIN50 - PBT Mod O2 Figure 4.9. Effect of configuration impeller/sparger on K L a . Sintered sparger and combination of impellers. O2 Figure 4.10. Effect of configuration impeller/sparger on K L a . Perforated tube and combination of impellers including dual impeller (PBT down – pumping and rushton impeller). 204 C K J B F, H Cell Response A I G D E Wm 3 1 2 3 4 5 6 7 8 9 10 10 10 10 10 10 10 10 10 1 11 2 6 4 9 3, 5, 7, 10 8 Hydrodynamic Conditions Cell Response Mode of Symbol Cell Reference growth A CHO-K1, necrosis Anchoraged Gregoriades et al. (2000) Thomas et al. (1994); Zhang Hybridoma, necrosis Suspended B et al. (1993) C MCF-7, necrosis Suspended Ma et al. (2002) D Mouse myeloma, necrosis Suspended McQueen and Bailey (1989) E HeLa S3, mouse L929, necrosis Suspended Augenstein et al. (1971) F CHO-K1, SF-9, HB-24, necrosis Suspended Ma et al. (2002) Uninfected and viral infected Suspended Submitted G PERC6 cells, necrosis Enthomopathogenic nematodes, Suspended Fife et al. (2004) H necrosis I CHO-K1, apoptosis Anchorage Mollet et al. (2007) J THP-1, necrosis Anchorage Mollet et al. (In Press) K Algae, loss of flagella Suspension Hu et al. (2007) Continued… Figure 4.11. Summary of the reported energy dissipation rate at which cells are affected as well as the reported levels of energy dissipation rate in various bioprocess environments. Adapted from Ma et al. (2002) and Mollet et al. (2004). 205 Figure 4.11. (Continued). Hydrodynamic Conditions Symbol Process Description Reference Volume average in typical Agitation Varley and Birch (1999) 1 animal cell bioreactors. Volume average in a 10 L Agitation 2 mixing vessel (RT, 700 RPM) Maximum in the 10 L mixing Zhou and Kresta (1996) 3 Agitation vessel (RT, 700 RPM) Volume average in a 22,000 L Agitation 4 mixing vessel (RT, 240 RPM) Wernersson and Maximum in the 22,000 L mixing Tragardh (1999) Agitation 5 vessel Maximum in spinner vessel Agitation Venkat et al. (1996) 6 (200 RPM) Bubble Pure water, bubble diameter: Garcia-Briones et al. (1994) 7 rupture 6.32mm Boulton-Stone and Blake Bubble Pure water, bubble (1993); 8 rupture diameter:1.7mm Garcia-Briones et al. (1994) Flow through Pure water, 100 mL/min, 1 mm Mollet et al. (2004) 9 a pipe diameter Flow through a Flow through a 200 L 10 Mollet et al. (2004) micropipette micropipette tip in 0.2 sec tip Volume average in a highly Agitation Oh et al. (1992) 11 agitated animal cell bioreactor 206 1.20 1.00 0.80 0.60 0.40 Rxt 515 Perforated tube A Rxt 523 Perforated tube B Normalized Cell Count Cell Normalized Rxt 619 Sintered Sparger A 0.20 Rxt 623 Sintered Sparger B Rxt 517 Perforated tube A Rxt 520 Perforated tube B 0.00 0 2 4 6 8 10 12 14 16 18 20 22 Elapsed Time Figure 4.12. Cell growth in bioreactors with best configurations. 1.20 1.00 2 0.80 0.60 Normalized CO Normalized 0.40 Rxt 515 Perforated tube Rxt 523 Perforated tube 0.20 Rxt 619 Sintered Sparger Rxt 623 Sintered Sparger Rxt 617 Perforated tube 0.00 0 2 4 6 8 10 12 14 16 18 20 22 Elapsed Time Figure 4.13. CO2 profile in cell culture evaluation. 207 1.2 1.0 0.8 0.6 Normalized Titer Normalized 0.4 Reactor 515 Perforated tube Reactor 523 Perforated tube Reactor 619 Sintered Sparger 0.2 Reactor 623 Sintered Sparger Reactor 517 Perforated tube Reactor 520 Perforated tube 0.0 0 2 4 6 8 10 12 14 16 18 20 22 Elapsed Time Figure 4.14. Recombinant protein profile in cell culture evaluation. 208 CHAPTER 5 EFFECT OF HYDRODYNAMIC CONDITIONS ON CELL CYCLE STRESS PROTEINS AND RECOMBINANT PROTEIN PRODUCTIVITY 5.1. ABSTRACT Stress proteins are a set of proteins expressed when cells are exposed to unfavorable conditions. They are also known as heat shock proteins and have been studied in the context of heat and nutritional stress, but the expression of stress proteins in response to hydrodynamic stress has not been determined, or at least not generally reported. In this work, we explored the relationship between stress proteins, cell productivity and cell cycle. Stress proteins’ profiles were determined by following production of a recombinant protein expressed by the CHO cells in a 2L bioreactor using flow cytometry. In addition, cell cycle profiles were evaluated under different shear stress conditions. Although, it has been hypothesized that the production of recombinant proteins by mammalian cells is associated, or modulated, with a particular cell cycle growth phase, there is not a consensus in the literature regarding this link. 209 Understanding the relationship between cell cycle phase and productivity could help improve process design by developing mathematical models to predict productivity as well as devising culture strategies to control the preferential cell cycle phase. Results obtained in this work, suggest that HSP70 in CHO cell culture is constitutively expressed at very low levels along the culture. On the other hand, HSP90 seems to be strongly expressed at the beginning of the culture and its expression tends to decrease along the culture. The expression of stress proteins in bioreactor show a different profile from those on TFlasks and the expression of HSP90 under stress was corroborated with a highly sensitive cell line. In regarding with cell cycle studies, specific productivity seems to be associated with G1 phase and not significant differences were observed in cell cycle profile due to different hydrodynamic conditions. 5.2. INTRODUCTION 5.2.1. Cell Damage Mechanisms The effects of hydrodynamic stress on the cells can be classified as lethal or sub lethal. The sub lethal effects can be considered as part of a hypothetic mechanism based on the response of the cells under stress (hydrodynamic stress or other types of stress such as temperature, nutrients, amino acids analogs, oxygen deprivation, pH). Most studies have 210 been oriented toward explaining the lethal effects; however, sub lethal effects can become very important if they affect the quality of the proteins and productivity. If sub lethal effects are presented, there are response mechanisms that involve signal reception and transduction including: altering membrane integrity, passive transport, ions canals, and changes in membrane fluidity (Al Rubeai, et al., 1993, Al Rubeai, et al., 1994, Martens, et al., 1991). These effects on the membrane might induce reaction chains that affect protein synthesis. Specific receptors in the plasma membrane indicate that cells respond at a molecular level under mechanical forces such as agitation. There is evidence that cells respond under different types of stress such as nutritional and oxygen deprivation by expressing a set of proteins known as Heat shock proteins or Stress Proteins (Subjeck et al, 1986), 1986). It is, therefore, likely that mechanical stress induces the expression of this set of proteins or a sub set of them. If this is the case, we could establish models that help to improve the understanding of cell damage at molecular level. That improved understanding would help us to design control mechanisms that contribute to the enhancement of cell survival at optimal conditions and improve protein quality and productivity. Likewise, the rate of DNA synthesis and the relative proportions of cells in the various stages of the cell cycle are influenced by agitation and gas sparging. For that reason, the evaluation of cell cycle profiles can help to elucidate the mechanism of cell damage if the cell cycle profile is affected under hydrodynamic stress. 211 5.2.2. Mammalian Stress Response Cells respond against adverse conditions by different mechanisms. A well known defense mechanism is the expression of a set of proteins commonly known as Heat Shock Proteins (HSP). The name was applied after discovering the expression of certain proteins in response to an abrupt increase in temperature. From a physiological standpoint, the expression of these proteins is a protective “modus operandi”. Oehler et al. (2001) report on the expression of the Heat Shock Protein HSP70 as a result of fever in mammalians. Fever has been considered a response system against infections. It is suggested that the expression of HSP70 is fever induced and is cell type dependent. Heat Shock proteins are expressed under different types of insults besides heat. Stress proteins expression has been reported in different biological phenomena such as embryogenesis and differentiation, viral infection, growth state and metabolism, protection from phenocopy (variation caused by unusual environment conditions), thermo tolerance, UV radiation and osmotic challenge. (Welch, 1992; Oelher et al., 2001). Stress proteins are very important in the synthesis of a number of proteins since they act as chaperons and help with the proper folding of the protein (Pelicano et al., 2006). These proteins are also involved in defense mechanisms preventing the cell from overcoming apoptosis (Lanneau, et al. 2008). It has also been postulated that stress proteins are involved in a phenomenon known as acquired thermotolerance, which consists of a type of resistance that cells acquire. Cells that survive a heat shock can easily survive a second and stronger heat shock. 212 The importance of this set of proteins has been verified in various studies. It has been observed that after deletion of HSP genes, cells are unable to survive if they are exposed to extreme environments. Stress proteins have been linked to some diseases and therefore most of the studies in stress proteins have been done with human cells and primary cultures. 5.2.2.1. Structure and function of Stress Proteins Classification of stress proteins is based on characteristics such as size, function, mode of regulation, cellular localization and homology. Since originally there was incomplete knowledge of the function of these proteins, they were identified and classified according to their size in an electrophoresis gel. For a number of cell lines, stress proteins have been detected that show molecular sizes of approx. 8, 22, 23, 25, 28 46, 56, 58, 72, 73, 76, 90, 97 and 110 kDa (Welch, 1992). Based on their function or mode of induction, there are two well known groups: heat shock proteins (HSP), and glucose regulated proteins (GRP). The set of HSP includes stress proteins that are strongly expressed after heat shock. The heat shock response was the first extreme condition which revealed the expression of stress proteins. A subset known as major stress proteins induced by heat shock exhibit molecular size of 68, 70, 89 and 110 kDa. A different subset of stress proteins expressed under heat shock exhibit low molecular size on the order of 8, 28 and 58. Glucose regulated proteins GRP are expressed under nutritional deficiency, mainly glucose deprivation. The identified GRP 213 showed molecular size of 78, 94 and 170 (Welch, 1992) or 76, 97 (Subjeck and Shyy, 1986). Despite the above classification, stress proteins can be expressed under nutritional deficiency and heat shock as well as other kinds of stresses. Curiously, the literature does not indicate a specific molecular size for some proteins. The reason for this could be that the molecular sizes are very close, making it hard to distinguish between these proteins using such a technique. In any case, the literature on the subject is not conclusive in defining the molecular size and classification of certain stress proteins. Some classifications are unique and they can also be combined. For the purposes of this work we will consider the classifications based on size and function. Stress proteins have various functions. For example they are integral in the proper folding of protein involved in apoptosis. Stress proteins are attached to certain amino acids before folding, and they help them to move to a different place where they are phosphorylated. Once phosphorylated they can act in chain reactions and cascades to avoid or promote apoptosis. Janig et al. (2005) explain how misfolded and aggregated proteins could be related to stress proteins in diseases like Alzheimer, Parkinson, Huntington and other diseases characterized by the occurrence of abnormal proteins accumulated inside the cells or in between them. Janig, et al (2005) formulated possible mechanisms of the stress response. For instance, they postulate that under stress conditions proteins form misfolded 214 intermediates. The stress protein HSP27 and a chaperone named Clusterin work together to stabilize these intermediate states in soluble complexes in order to avoid irreversible aggregation. Furthermore, the stress protein HSP70 has the capability to refold misfolded intermediates in the cytoplasm. Young et al (2003) elaborate on the interaction between HSP70 and HSP90. These proteins are involved in the transport of polypeptides across membranes, and also a relationship between these proteins and cytoskeleton components is established in Young’s work. To summarize, stress proteins can play an important role in the quality of the proteins by monitoring their proper folding, and hence, affect the characteristics and activity of the proteins. 5.2.2.2. Induction of response Expression of heat shock proteins has been reported under different stress conditions and phenomena such as: heat shock, starvation and differentiation. Also, some of these proteins can be expressed constitutively but there is a significant increase of expression under stress conditions. Stress proteins are also induced by chemical compounds, including: amino acid analogs, alcohols, heavy metals, sulfhydryl reagents, oxidizing agents, gene expression inhibitors, steroid hormones, suboptimal temperatures, anoxia, nutrient starvation and viral infection 215 (Huhtala et al., 2005; Welch, 1992). Table 5.1 shows the type of stress that has been evaluated in the context of stress protein expression. 5.2.2.3. Stress proteins and hydrodynamic stress Since the cells express stress proteins under insults such as heat shock and nutritional stress (among others), it seems likely that in vivo circulating cells as well as in vitro cultured cells experiencing hydrodynamic stress insults would produce stress proteins. Huhtala et al. (2005) studied the response of insect cells to bioreactor environmental stresses including: temperature, oxygen deprivation, salinity, ethanol, pH changes and shear stress. For shearing experiments, they used a spinner culture and silicon capillary tubing. Huhtala, et al reported a shear stress of 0.2 to 1.6 N/m2 in the experiments with capillary tubing and for the spinner experiments they evaluated shear stress correspondent to agitation up to 300 rpm. Their results indicate that the conditions evaluated do not induce expression of stress proteins, however some observations are worth noting. First, the levels of stress exposure are in the range were no cell damage has been observed. Figure 1 shows reported levels of energy dissipation rate in various bioprocess environments. The values reported by Huhtala, et al correspond to 2.6x103 W/m3 for the tubing experiments and approx. 8.6x103 W/m3 for the spinner. The second point to note is the time of exposure and the time of sampling after exposure. Cells were exposed to stress for short periods of time (30 min or 1 hr), in addition, it is not clear the times when the expression of stress protein is evaluated. Changes in expression of stress proteins can be observed at different times after exposure depending on type of stress and 216 times of exposure. For example, Bachelet, et al. (1998) founded an increasing expression of stress proteins during seven hours after heat stress exposure followed by a decrease in the expression. CHO cells are probably the cell type most frequently employed by industry for the production of monoclonal antibodies and therapeutic recombinant proteins (Godoy Silva, 2008). Stress proteins in CHO cells have been studied in the context of heat and nutritional stress but, to the knowledge of the author, the expression of stress proteins in response to hydrodynamic conditions in CHO cell cultures has not been reported. As a result, we are proposing to study the effect of hydrodynamic stress on the production of HSP. Cell damage caused by a variety of devices due to hydrodynamic forces is always a concern in biotechnological research and development. Although numerous studies have explored the consequences of this damage, there is a lack of understanding of the mechanism of such damage at the cellular level. Pharmaceutical industries require strategies to overcome challenges in animal cell cultures in order to improve productivity. One of these challenges is cell damage and the resultant loss of productivity and/or decreased quality of recombinant proteins. The project is relevant in that it addresses a specific industrial problem. Cell damage can also influence biomedical applications. Cells manipulated in instruments such as diagnosis and analysis devices can experience hydrodynamic forces. A 217 contribution in this area is required to improve the understanding of cell damage phenomena at the cellular level and to define strategies to overcome these challenges. Although considerable research has been devoted to the family of stress proteins expressed due to heat shock and nutritional stress, less attention has been paid to stress protein expression due to hydrodynamic stress. In this work we want to explore the expression of stress proteins under mechanical stress. Since these proteins can be related with proper folding of proteins it is important to increase the knowledge of the behavior of these proteins in cell culture under different hydrodynamic conditions and to investigate the effect of the stress proteins on the quality of recombinant proteins. In a similar way we want to know the effect of different hydrodynamic conditions on cell cycle profile along the culture and to evaluate the cell cycle phase where the cells are the most productive. 5.3. MATERIALS AND METHODS 5.3.1. Cell line Two CHO cell lines were used in this study. CHO6E6 (ATCC CRL 11398) was used to establish the base line of expression of stress proteins HSP25, HSP70 and HSP90. This cell line was used for studies in T Flasks, and continuous hydrodynamic stress exposure in bioreactors. Cells were maintained by subculturing in spinner flasks in an incubator at 218 37°C and an atmosphere of 5% CO2. For the experiments with CHO6E6 the commercial medium CD OptiCHO (Gibco Invitrogen) was used. A second clonal CHO cell line was used for studies in bioreactors at different hydrodynamic conditions. A chemically defined media was used with this cell line. THP1 is a human acute monocytic leukemia cell line from ATCC routinely used in screening assays to assess cytotoxicity induced by potential agents. THP1 cells were grown in suspension in 75 mL TFlasks in phenol red free RPMI 1640 media containing 2mM l glutamine (Gibco) and supplemented with 10% FBS (Hyclone, Logan, UT). THP1 was chosen as a model to study chronic shear stress exposure since it is a highly sensitive cell line to hydrodynamic stress. 5.3.2. Static Cell Cultures Since some stress proteins are expressed constitutively, it is important to know the expression profile of the stress proteins evaluated under non hydrodynamic stress conditions in order to compare them with the expression profile of cells exposed to different hydrodynamic conditions. Cells were seeded in 25cm2 Tflasks with 7 mL culture media at cell concentration of 0.1x105 cell/mL, CD OptiCHO medium was used for cells growing in suspension. 219 Samples were taken at regular intervals of time to evaluate growth kinetics, glucose lactate kinetics and expression of stress proteins. 5.3.3. Bioreactors In a related study (Chapter 4 in this document) a set of bioreactors with different impeller sparger configurations (different hydrodynamic conditions) was evaluated in order to choose the best configuration in terms of mass transfer capabilities. Two configurations were selected which provide good mass transfer capabilities and hold different hydrodynamic conditions. The configurations chosen were used to evaluate the expression of stress proteins under moderated hydrodynamic stress in bioreactors as well as cell cycle profile and its relationship with recombinant protein production. Glass 3 L bioreactors with 2 L working volume (Applikon, Inc., Foster City, CA) were used as production step bioreactors. Production step refers to the production of recombinant protein. Bioreactors were controlled at 37°C using a heating blanket, dissolved oxygen concentration was controlled at 30% of air saturation by sparging air, pH was controlled at 7.0 by bicarbonate CO2 system. Bioreactors were sampled daily for analysis: cell density, viability, metabolites, cell cycle profiles and expression of stress proteins. 220 5.3.4. Continuous stress In addition, a methodology proposed by Godoy et al. 2009 was performed to evaluate the expression of stress proteins under continuous hydrodynamic stress. The methodology is detailed elsewhere (Godoy Silva et al, 2009) briefly: Cells were grown in a 3 L total volume bioreactor (Applikon, Inc., Foster City, CA)), a recirculation system was installed that allows the re circulation of cells taken from the bioreactor and passed them through a flow contraction device, and subsequently sending them back to the bioreactor. A dual syringe pump operated in continuous mode and a set of solenoid valves (Harvard Apparatus) allow the permanent uptake and return of the cells to the bioreactor. When cells are passing through the contraction device they are exposed to a defined shear stress that is related to the flow rate at which the pump is operated. Specifically the cells were recirculated at 30 mL/min, which corresponds to a range of energy dissipation rate (EDR) of 1x106 to 1x107 W/m3 and the median of maximum EDR is 2.3x106 W/m3. Recirculation started on day 4 of culture and it was stopped when glucose was depleted (day 7 of culture). Figure 5.2, shows the set up for the continuous stress exposure in bioreactor. Continuous stress experiments were also performed with the cell line THP1, a recirculation loop was installed in a mini bioreactor, 250 mL total volume (working volume of 150 mL). THP1cells were recirculated at two different stress conditions: The first condition was 50 mL/min (Median of maximum EDR 6.4x106 W/m3) during 30 min 221 (10 recirculation cycles). The second condition was 90 mL/min (Median of maximum EDR 1.1x108 W/m3) during 17 min (10 recirculation cycles). 5.3.5. Stress Proteins analysis 5.3.5.1. Cell fixation Cells were washed twice with PBS and re suspended at a concentration of 1x107 cells/mL. Fixation was performed by re suspending the cell pellet in 1 mL of a 4% paraformaldehyde solution. Cell suspension was kept at 4 ºC for 20 min. Cells were then washed twice in staining buffer which contains PBS plus 1% FBS. Finally, the cell pellet was re suspended in 90% FBS/10% DMSO solution and stored at 80 ºC for intracellular staining at a later time. 5.3.5.2. Staining and analysis After samples were thawed at 37 ºC, cells were washed two times in staining buffer. A cell suspension of 1x106 cells/mL was then re suspended in 1x BD Perm/Wash buffer and incubated for 15 minutes. Next, aliquots of sample were taken for controls and stress proteins analysis. Cells were permeabilized with Perm/Wash buffer (BD 554714) and then stained with the proper antibody. Controls were performed with unstained samples and isotypes controls IgG PE (Assay Designs SAB 600PE), IgG DyLight (Assay Designs SAB 600 488). Stress proteins antibodies used in different analysis: HSP90 PE (Assay 222 Designs SPA 830PE), HSP90 DyLight (Assay Designs SPA 830 488), HSP70 PE (Assay Designs SPA 810PE), HSP70 DyLight (Assay Designs SPA 810 488). The cell suspension was incubated at 4 ºC for 40 min in the dark. Finally, cells were washed with 1xBD perm/wash buffer (1 mL/wash) and re suspended in Staining buffer for flow cytometric analysis. Stained cells were examined using a flow cytometer BD Biosciences FACS Calibur, equipped with a single argon ion laser with the excitation wavelength at 488 nm and emission detected at 585/42. Samples from studies with cell line CHO6E6 were analyzed with flow cytometer BD LSRII, equipped with four lasers (488nm, 633nm, 405 nm and 355), 12 acquisition channels and 2 scatter channels. Cells were analyzed using Cell Quest Pro software. 5.3.6. Analytical methods Production bioreactors were sampled daily to determine TCC, VCC, viability, metabolites (glucose, lactate), dissolved oxygen (dO2) and dissolved carbon dioxide (dCO2) accumulation. Cell free samples for future reference were also prepared. Protein titer was determined by Biacore Concentration. Cell counts and viability were determined using an automated cell counter based on the trypan blue exclusion method (Vicell, Beckman Coulter), when viability in the culture dropped to less than 90% as indicated by the automated cell counter, manual viability 223 determination was implemented. Metabolites were measured using YSI (YSI 2700) or Nova Bioprofile 400; both instruments are enzyme based electrochemical reactions. 5.3.7. Cell cycle analysis Cells were centrifuged at 350 × g for 7 minutes. Then, cells were washed with PBS and fixed in 70% ethanol overnight at 4 ̊ C. After that, cells were washed with PBS and treated with a solution that contains RNase solution at 5 g/mL (Roche) and propidium iodide at 50 g/mL(Sigma). The cells were then analyzed using a BD FACS Calibur. The fractions of the cells in G0/G1, S, and G2 phase were analyzed using cell cycle analysis software, Modfit. 5.3.8. Parameter calculation 5.3.8.1. Growth rate For production bioreactors, the maximum growth rate was calculated for each one of the cultures to have a meaningful comparison basis. The maximum growth rate was estimated as described by Goudar et al. (2005). The suggested logistic equation takes the form: A X = (2) exp()k t + C ×exp()− t dmax max where 224 6 X, viable cell density [=] 10 cells/ml max, maximum growth rate [=] 1/day k , maximum death rate [=] 1/day dmax t, time [=] day 5.3.8.2. Integral of viable cell concentration (IVC) Integral of Viable Cells was calculated using VCC data and applying the trapezoid rule following the formula, i (VCCn +VCCn−1) IVC = ∑[ (tn − tn−1) ] (3) n=1 2 Where: IVC, Integral of Viable Cell density [=] (106 cells/ml).day t, culture time [=]days VCC, Viable cell density [=] 106 cells/ml ni number of samples taken throughout the culture, for calculation purposes i=4 5.3.8.3. Specific Productivity Specific Productivity was calculated as the slope of the best fit line correlating product titer (mg/L) to Integral of Viable Cell density (IVC). 225 5.4. RESULTS AND DISCUSSION Cell cultures were performed in bioreactors with two different configurations in order to evaluate the effect of different hydrodynamic conditions on cell cycle, expression of stress proteins and productivity of a recombinant protein. In addition, a set of experiments was developed to evaluate the difference between expression of stress proteins in TFlasks and bioreactors. Also, the expression of stress proteins under critical conditions was performed with a highly sensitive cell line. 5.4.1. Effect of configuration on cell cycle profile The purpose of this part of the work was to identify cell cycle profiles in cultures of a recombinant Chinese Hamster Ovary cell line in 2L bioreactor using flow cytometry analysis. Comparison was established between cell cycle profiles for cell cultures under different shear stress and environmental conditions. Four bioreactors were followed up in this part of the study. Two of them were operated with perforated tube sparger and dual impeller (PBT top and Rushton bottom), the other two bioreactors were operated with sintered sparger and PBT impeller. Figure 5.3 (a), 5.3 (b), 5.4 (a) and 5.4 (b), show growth kinetics, viability, glucose and lactate kinetics respectively. No significant differences were observed in terms of growth and metabolites consumption/production for the four bioreactors evaluated. 226 Shear stress is reported to be related with cell cycle (Dutton, et al 2006)). In addition, it has been hypothesized that the production of recombinant proteins by mammalian cells is associated with a particular cell cycle phase (Table 2). However, as can be seen in Table 2, there is not a consensus in the literature regarding to the cell cycle phase where cells are the most productive and even for the same product, reports of different productivity with different cycles have been observed. Figure 5.5, panel A, shows a diagram of the cell cycle. During a growth phase, cells are constantly produced by division; they follow a cycle that involves 4 phases. The first step is G1 phase where the cell senses that internal and external conditions are proper for a new round; in S phase, DNA is synthesized; and in G2 the material is arranged for every cell. G2 is a check point which means if the cell senses that there is something wrong, the cell goes to apoptosis. Conversely, if everything is “right” the cell goes through mitosis. The amount of DNA content is increasing during S phase and its maximum in G2 phase. Since the amount of DNA in every phase is different, flow cytometry is routinely used to evaluate the cell cycle for cells in culture. Figure 5.5, panel B, shows a typical histogram obtained from flow cytometer analysis of growing cells. In the x axis the content of DNA is plotted. The first peak with less content of DNA corresponds to cells in G1 phase, the next peak toward the right indicated with blue lines and with greater content of DNA corresponds to cells in S phase and the last peak with maximum amount of DNA corresponds to cells in G2 phase. 227 Daily cell cycle analysis allows determining the cell cycle profile along the culture. It is possible to observe how the average population is cycling (See Figure 5.6 panel A). In addition it is possible to observe how the cell cycle is affected by external changes as well as physicochemical conditions. The question that arises is how this cell cycle profile is related to product productivity, and if so, how could we manipulate the culture to potentially maintains the cell population in the more productive cell cycle phase. To figure out the relationship between product production and cell cycle, product concentration as a function of the individual subpopulations was plotted. The fraction of individual sub. As can be observed in Figure 5.6, panel B, the protein production was directly proportional to the population in phase G1, which strongly suggests that secretion occurs primarily during this cell cycle phase. It is possible to develop correlations to describe the relationship between product concentration and the other populations but they will be variable and certainly non linear relationships, as can be judged from the plots on Figure 5.6 panels C and D. Figure 5.7 shows a comparison of cell cycle profile between two cultures performed in reactors with different impeller/sparger configurations and therefore different hydrodynamic conditions. As can be observed there is not a significant difference between the cell cycle profiles for these reactors configurations. The relationship between subpopulations and titer shows a linear relationship with respect to cells in G1 for both configurations, and moreover the linear relationship shows a similar slope, which indicates a qp in the same order (Figure 5.8). 228 To summarize this section we can make the following observations: Based on the non conclusive literature and even contradictory findings regarding the cell cycle phase where the cells are more productive, it is valid to speculate that the relationship cell cycle/ productivity is cell line specific. For the clonal cell line evaluated in this study, protein production was mainly directly proportional to the G1. There were no differences in the cell cycle profile regarding to different impeller/sparger configurations evaluated. No lethal damage was observed at the EDR calculated for each configuration. If sub lethal damage was caused, it did not seem to affect the cell cycle profile. 5.4.2. Effect of configuration on expression of stress proteins In addition to product production, the expression of stress proteins was followed to investigate how the cells respond to environmental changes as well as different hydrodynamic conditions. Following the protocol explained in section 5.3.5 the analysis was developed based on single parameter histograms. Figure 5.9 (b) shows a typical histogram of a single parameter, the horizontal axis represents the fluorescent parameter and the vertical axis represents the number of cells. First cells are identified and selected by adjusting voltages for the forward scatter channel (FSC) and side scatter channel (SSC) as can be observed in Figure 5.9 (a). The histogram is linked to the selected region so the analysis is done for 229 the viable cell population. Negative controls (unstained or isotype sample) are used to set the voltage in the region of low mean fluorescence intensity. Once the sample is tested, the positive fraction (cells expressing the stress protein), is estimated by comparison with respect to the negative control. Since populations vary along the culture and hence the auto fluorescence fluctuates; a proper control is required every day to estimate the fraction of cells actually expressing the stress proteins. A variety of statistical information can be obtained from the histograms to estimate the fraction of cells expressing the stress protein. Sample size was set as 10000 in order to have a target population large enough for acceptable statistics. Geometric mean (GeoMean) indicates the central tendency or typical value of the fluorescence for a selected region. Thus, the level of expression in each sample can be calculated based on the fluorescence in order to determine the expression profile along the culture. The relative expression can be calculated based on the difference of the GeoMean of the sample minus the control or based on the ratio as follow: GeoMeanSample RelativeExpressionHSP = GeoMeanUnstained Or RelativeExpressionHSP = GeoMeanSample − GeoMeanControl 230 The difference was used to observe the trend of expression along the culture, the ratio was used to estimate the relationship of fraction of cells expressing the stress protein with respect to the level of product obtained. A comparison was established between the expression of HSP70 and HSP90 associated to the integral of viable cell concentration (IVC) for the two configurations (impeller/sparger) evaluated. Figure 5.10 presents the expression of HSP70_IVC versus product concentration and Figure 5.11 presents the expression of HSP90_IVC versus product concentration. As can be observed in Figure 5.10, the expression of HSP70 seems to be directly proportional to the product production; a trend also appears to indicate that the expression of HSP70 is also dependent on the configuration of the bioreactor. On the other hand, from Figure 5.11, it can be suggested that there is a nonlinear relationship between expression of HSP90 with respect to product concentration and it did not seem to be affected by the configuration evaluated. The main purpose of analyzing the expression of stress proteins in industrial cell lines is to determine if these proteins are somehow related with the expression of recombinant protein. If a positive correlation was obtained, it could be advised to propose strategies to manipulate the expression of these proteins in order to increase productivity. From the analysis obtained regarding the relationship between expression of stress proteins and the recombinant protein, it is plausible that such relationship exists, however further research is required in order to specify the existent relationship as well as the directions required 231 to manipulate the expression of stress proteins with the purpose of improve quantity or quality of the final product. Further analysis indicates the profile of expression of stress proteins along the culture. As can be observed in Figures 5.12 and 5.13, the expression of HSP90 tends to decrease along the culture in all cases, while HSP70 is expressed at very low levels and with a slight increase with time of culture. The level of expression seems to be cell strain specific, thus the maximum mean fluorescent intensity of HSP90 expression strain A is 400 and goes to less than 100 ( Figure 5.12 (a) and Figure 5.13 (a)), while maximum mean fluorescent intensity of HSP90 expression in strain B is 250 and goes to less than 50 at the end of the culture. The question that arose from the previous analysis was if the expression profile of stress proteins was exclusive of bioreactors or if a similar pattern would be observed in a static culture under non stress hydrodynamic conditions. To answer this question the expression profile of stress proteins was evaluated in a cell culture for a bioreactor and a set of static culture as controls. 5.4.3. Expression of Stress Proteins in Reactor Vs. TFlask The experiment was developed with the ATCC cell line CHO 6E6 in a commercial media (Opti CHO Gibco). To eliminate error sources a sample from a cell bank was thawed and seeded in a TFlask containing the same type of media used in the reactor. 232 Consecutive passes were performed in spinner until the proper cell concentration was reached to start simultaneously the bioreactor as well as the experiments in TFlasks. A set of 40 T Flasks containing 7 mL of media was seeded. Reactor and TFlasks were inoculated with approximately 0.2x106 cell/mL. Samples were taken from the bioreactor approximately every twelve hours for analysis (cell count, glucose, lactate), at the same times two Tflasks were removed from the incubator and used for growth kinetic, metabolite and stress proteins analysis of control samples. Figure 5.14 shows the growth kinetics and viability of the cell cultures. As can be observed in Figure 5.14 (a), similar level of cell concentration was reached in bioreactor as well as TFlasks. Viability of TFlasks follow the same profile as for bioreactor except for the last two points of culture. Figure 5.15 (a) shows glucose consumption profile. Glucose is completely depleted in the bioreactor while in TFlasks glucose still remains by the end of the culture. The slope of lactate production in the bioreactor indicates a lower rate of production than in TFlasks (Figure 5.15 b). Figure 5.16 shows the comparison on expression of stress proteins between reactor and TFlasks. The stress proteins evaluated in this assay are HSP70, HSP90 and HSP25. The relative expression was normalized to facilitate comparison. As can be observed in Figure 5.16 panels a and b, the expression of HSP90 falls from 1 to around 0.2 in the bioreactor while in the TFlasks, the relative expression of HSP90 oscillates between the maximum value and 0.4,which averages to a relatively constant expression of HSP90. Likewise, the 233 pattern of expression for HSP25 indicates a decrease in the expression of HSP25 in the bioreactors (Figure 5.16 c). Conversely, the pattern in the TFlasks indicates a fast decrease during the first hours of culture followed by a constant expression period and an increase at the end of the culture. It is important to highlight that the maximum mean fluorescent intensity of HSP90 expression for this cell line is around 250, which is larger than the maximum mean fluorescent intensity for HSP25 which is around 125. To increase the level of shear stress in the bioreactor a recirculation loop was installed and the cells were forced to pass through a flow shear device as described in section 5.3.4. Cells were re circulated at 30 mL/min which correspond to a level of stress of 2.3x106 W/m3. Re circulation was started when the cells were in exponential phase (81 hr of culture) and it run continuously until glucose was depleted (140 hr of culture). Since HSP90 can be expressed under nutritional stress, re circulation was stopped in order to separate the possible stress inductors, i.e. shear stress from recirculation from nutritional stress due to absent of glucose. From the pattern of expression of proteins in the reactor during the continuous stress interval (Figure 5.16, re circulation period is indicated with dashed lines), some observations can be drawn. Expression of HSP90 seems to be un affected in the immediate hours after starting the continuous stress, an increase on expression is observed after 120 hours of culture, it is possible that at this point the cells started sensing limited glucose availability, other than that no noticeable event was detected in the culture. After continuous stress stops, there is a sharp decrease in the expression of 234 HSP90, followed by an increase after 160 hr of culture and a decreased trend at the end of the culture. A similar set of observations can be enunciated for HSP25, although the pattern of expression of HSP25 during the continuous stress interval shows a sharp decrease after starting re circulation followed by a slight increase in expression until the point where re circulation is stopped when a sharp decrease is observed that resembles what is observed for HSP90. Regarding HSP70, in spite of the low level of expression it is noticeable the increase in the level of expression when the continuous stress starts, after about 24 hr of continuous stress the expression of HSP70 decreases and it is trigger again when the re circulation stops. Ongoing experiments are being performed in order to confirm the observations regarding the expression of stress proteins during the continuous stress interval. Furthermore, the results obtained in this experiment confirm the pattern of expression of HSP90 and HSP70 obtained with the clonal CHO cell line from GSK. In addition, the patterns of expression observed in control experiments developed in TFlasks under non hydrodynamic stress conditions are different from those obtained in the bioreactor. 5.4.4. Expression of Stress proteins in a highly sensitive cell line THP1 is a cell line with a level of sensitivity orders of magnitude higher than CHO and other cell lines previously evaluated in Chalmers’ group (Ma et al., 2002, Mollet et al., 235 2007, Godoy et al., 2009). Based on the high sensitivity of this cell line, THP1 has been used in Chalmer’s group as model to study shear stress cell sensitivity. A cell suspension of THP1 was exposed to continuous stress for a short period of time to avoid 100% lethal cell damage. Then, cells were harvested and incubated at 37 °C and 5% CO2 atmosphere in order to let them recover. Samples were taken at different times to evaluate stress proteins expression. Figure 5.17 and 5.18 show the histograms obtained from samples of cells that were exposed to 50 mL/min and 90 mL/min respectively. As can be observed in Figure 5.17 the mean of fluorescence intensity for the sample that was exposed to shear stress at 50 mL/min, after one hour of recovery after stress, is similar to that from the control for both HSP90 and HSP70, which suggest no additional expression of the stress proteins. In the histogram for the sample taken 6 hr into the recovery, the mean of fluorescence intensity for HSP90 is shifted to the right, implying an increase in the expression of the stress protein, however, sample at 8 hr presents the same mean intensity fluorescence and sample at 10 hr is slightly shifted. Presumably a transient increase of expression is presented around 6 hr after stress. Further experiments need to be developed in order to eliminate possible error in the test or experiment. Regarding with HSP70 expression, from the histograms we can infer that no expression is stimulated by exposing the cells to shear stress. 236 Conversely, cells that were exposed to 90 mL/min seem to increase the expression of HSP90 as can be seen in Figure 5.18 for samples taken 2 hr and 7 hr in recovery after stress. The histograms of HSP70 for the samples that were exposed to shear stress are slightly shifted from the histograms of the control samples implying that HSP70 is not trigger by the hydrodynamic stress at which the cells were exposed. A different set of cells were exposed to 90 mL/min to corroborate the previous results. The histograms obtained for samples taken at similar time, intervals, yielding the same trend and are consistent with those presented here, based on the results we can speculate that the expression of stress protein HSP90 is stimulated by mechanical forces. 5.5. CONCLUSIONS We investigated the relationship between cell cycle and product expression as well as the effect of different hydrodynamic conditions. The results suggest that for the clonal cell line evaluated G1 phase of the cell cycle may be more conducive to produce the recombinant protein. Accordingly, specific culture conditions that promote extension of the cells in G1 phase stage could have a positive impact on productivity. Further systematic analysis oriented to arrest cell population in G1 can help to elucidate the productivity pattern of this recombinant protein. It has been stated that DNA synthesis and the relative proportions of the cells in different stages of the cell cycle are influenced by hydrodynamic forces. Nevertheless, the 237 comparison between two different hydrodynamic conditions evaluated in this work did not yield any significant difference in terms of the fractions of cells in every cell cycle stage. Presumably, the conditions evaluated are mild enough to not cause any perturbation in the cell cycle resulting in similar cell cycle profiles. The conditions evaluated are in the range of typical operational conditions that provide good mass transfer capabilities, and yet allow cells to grow without being exposed to rough hydrodynamic conditions. The study of stress protein expression due to different hydrodynamic conditions, surely shows a potential area of research where much information can be obtained to improve knowledge about the hydrodynamic stress response in cell culture. Cell culture stress response studies have been focused on stress conditions such as nutritional stress, heat stress, osmotic stress rather than hydrodynamic stress. Nonetheless, literature is rich on studies of hydrodynamic stress effects on animal cells dealing with cellular stress and health diseases. The foundation in those studies is the hydrodynamic stress that the cells experience in flowing stream of blood. It has been established that the shear stress in veins is in the order of 0.2 N/m2 (90 W/m3) while in arteries is in the order of 2.5 N/m2 (6.25*103 W/m3) (Rosenson Schloos, 1999). Hydrodynamic stress in the human body has been associated with cardiovascular diseases like atherosclerosis and hypertension. Lessons can be learned from the human body and studies of those diseases in order to understand hydrodynamic stress response and cell damage in animal cell cultures. 238 In fact, studies indicate that expression of stress proteins increases due to shear stress. Rosensen Schools et al, 1999, reported increased expression of HSP70 after exposing leukocytes to shear stress in a flow chamber. Likewise, Hochleitner et al 2000, exposed umbilical vein endothelial cells to shear stress in a rotational viscometer which stimulated the expression of HSP60. Ongoing research in the area is attempting to elucidates the molecular mechanism that link the expression of these proteins with the mechanical stress. In a similar way additional research is required in order to elucidate the mechanism that results in the profile of expression of stress proteins that was observed in cultures of CHO and clonal CHO cell lines in bioreactors, as well as THP1 cell line exposed to chronic shear stress. Certainly, the results obtained in this work suggest a relationship between hydrodynamic stress and expression of stress proteins. The type of stress protein and the level of expression seem to be dependent on cell type and even differences can be observed between clones of the same cell line. Studies on expression of other stress proteins as well as chronic stress at higher flow rate (higher EDR) can help to elucidate the mechanism and threshold of shear stress that trigger the expression of stress proteins in cell cultures. Although not considered in this study other relevant effects of stress proteins can be further investigated. For example, apoptosis has been directly linked with expression of stress proteins in vivo as well as in vitro. HSP27, HSP70 and HSP90 are considered anti 239 apoptotic stress proteins while HSP60 is considered pro apoptotic. Indeed, cell culture may be optimized by manipulating stress protein expression in order to control apoptosis. 240 5.6. REFERENCES Al Rubeai, M. and Emery, A. N. (1990). Mechanisms and kinetics of monoclonal antibody synthesis and secretion in synchronous and asynchronous hybridoma cell cultures. Journal of Biotechnology. 16: 67 76. Al Rubeai, M.; Emery, A. N.; Chalder, S. and Golodman, M.H. (1993). A flow cytometric study of hydrodynamic damage to mammalian cells. Journal of Biotechnology. 31: 161 177. Al Rubeai, M.; Singh, R.P.; Goldman, M.H. and Emery, A. N. (1995). Death Mechanisms of Animal Cells in Conditions of Intensive Agitation. Biotechnology and Bioengineering. 45: 463 472. Bachelet, M.; Marlethoz, E.; Banzet, N.; Souil, E.; Pinot, F.; Polla, C.Z.; Durand, P.; Bouchaert, I. and Polla B. (1998). Flow cytometry is a rapid and reliable method for evaluating heat shock protein 70 expression in human monocytes. Cell Stress and Chaperones. 3 (3): 168 – 176. Bi, J. X.; Shuttleworth, J. and Al Rubeai, M. (2004). Uncoupling of cell growth and proliferation results in enhancement of productivity in p21CIP1 arrested CHO cells. Biotechnology and Bioengineering. 85: 721 749. Bluestein, M. and Mockros, L. F. (1969). Hemolysis effects of energy dissipation in flowing blood. Medical and Biological Engineering and Computing. 7(1):1 6. Cleary, S.; Cao, G.; Liu, L.; Egle, P.M. and Shelton, K. (1997). Stress Proteins are not induced in mammalian cells exposed to radiofrequency or microwave radiation. Bioelectromagnetism. 18 (7): 499 505. Deneke, S.; Harford, P.; Lee, K.; Deneke, C.; Wright, S. and Jenkinson, S. (1997). Induction of Cystine Transport and other stress proteins by disulfiram: Effects on Gluthathione levels in cultured cells. Am. J. Respir. Cell Mol. Biol. 17: 227 234. 241 Gething, M.J. and Sambrook, J. (1992). Protein folding in the cell. Nature. 355 (2), 33 45. Gregoriades, N.; Clay, J.; Ma, N.; Koelling, K. and Chalmers, J. J. (2000). Cell damage of microcarrier cultures as a function of local energy dissipation created by a rapid extensional flow. Biotechnology and Bioengineering. 69:171 182. Haug, G.; Leemhuis, J.; Tiemann, D.; Meyer, D. K., Aktories, K. and Barth, H. (2003). The Host Cell Chaperone HSP90 is essential for translocation of the binary clostridium botulinum C2 toxin into the cytosol. The Journal of Biological Chemistry. 278 (34); 32266 – 32274. Huhtala, A.; Linko, P. and Mutharasan, R. (2005). Protein response of insect cells to bioreactor environmental stresses. Journal of Biotechnology. 118: 278 289. Janig, E.; Stumptner, C.; Fuchsbichler, A.; Denk, H. and Zatloukal, K. (2005). Interaction of stress proteins with misfolded keratins. European Journal of Cell Biology. 84; 329 – 339. Kromenaker, S. J. and Srienc, F. (1994). Cell cycle kinetics of accumulation of heavy and light chain immunoglobulin proteins in a mouse hybridoma cell line. Cytotechnology. 14: 205 218. Kunas, K. T. and Papoutsakis, E. T. (1990a). Damage mechanisms of suspended animal cells in agitated bioreactors with and without bubble entrainment. Biotechnology and Bioengineering. 36:476 483. Lanneu, D.; Brunet, M.; Frisna, E.; Solary, E.; Fontenay, M. and Garrido, C. (2007). Journal of Cellular and Molecular Medicine Postprint 10.1111/j.1582 4934.2008 . Ma, N.; Koelling, K. and Chalmers, J. J. (2002). The fabrication and use of a transient contraction flow device to quantify the sensitivity of mammalian and insect cells to hydrodynamic forces. Biotechnology and Bioengineering. 80:428 437. 242 Martens, D.E.; de Gooijer, C.D.; Beuvery, E.C.; and Tramper, J. (1992). Effect of serum concentration on hybridoma viable cell density and production of monoclonal antibodies in CSTR and on shear sensitivity in air lift loop reactors. Biotechnology and Bioengineering. 39:891 897 Mollet, M.; Ma, N.; Zhao, Y.; Brodkey, R.; Taticek, R. and Chalmers, J. J. (2004). Bioprocess Equipment: Characterization of Energy Dissipation Rate and its potential to Damage Cells. Biotechnology Progress. 20:1437 1448. Mollet, M.; Godoy Silva, R.; Berdugo, C. and Chalmers, J. J. (2007). Acute Hydrodynamic Forces and Apoptosis: A Complex Question. Biotechnology and Bioengineering. 98 (4): 772 788. Mollet, M.; Godoy Silva, R.; Berdugo, C. and Chalmers, J. J. (in Press). Computer simulations of the energy dissipation rate in a fluorescence activated cell sorter: implications to cells. Biotechnology and Bioengineering. Oehler, R.; Pusch, E.; Zellner, M.; Dungel, P.; Hergovics, N.; Homoncik, M., Eliasen, M., Brabec, M. and Roth, E. (2001). Cell type specific variations in the induction of hsp70 in human leukocytes by feverlike whole body hyperthermia. Cell Stress & Chaperones. 6 (4), 306 315. Pelicano, H.; Carew, JS.; McQueen, TJ.; Andreeff, M.; Plukett, W.; Keating, MJ and Huang, P. (2006). Targeting Hsp90 by 17 AAG in leukemia cells: mechanism for synergistic and antagonistic drug combinations with arsenic trioxide and Ara C. Leukemia. 20: 610 619. Stolzing, A.; Sethe, S., and Scutt, A. (2006). Stressed Stem Cells: Temperature Response in Aged Mesenchymal Stem Cells. Stem Cells and Develompent. 15: 478 – 487. Subjeck, J.R.; Shyy, T. T., (1986). Stress Protein systems of mammalian cells. Am. J. Physiol. 250, C1 C17. Thomas, X.; Campos, L.; Mounier, C.; Cornillon, J.; Flandrin, P.; Le, Q., Piselli, S., and Guyota, D. (2005). Expression of heat shock proteins is associated with major adverse pronostic factors in acute myeloid leukemia. Leukemia Research. 29: 1049 1058 . 243 Thomas, C. R.; Al Rubeai, M. and Zhang, Z. (1994). Prediction of mechanical damage to animal cells in turbulence. Cytotechnology. 15: 329 335. Varley, J. and Birch, J. (1999). Reactor design for large scale suspension animal cell culture. Cytotechnology. 29:177 205. Venkat, R.; Stock, R. and Chalmers, J. J. (1996). Study of hydrodynamics in microcarrier culture spinner vessels: A particle tracking velocimetry approach. Biotechnology and Bioengineering. 49:456 466. Welch, W., (1992). Mammalian Stress Response: Cell Physiology, Structure/Function of Stress Proteins, and Implications for Medicine and Disease. Physiological Reviews. 72 (4): 1063 – 1081. Young, J.; Barral, J.M. and Hartl. F.U. (2003). More than folding: localized functions of cytosolic chaperones. TRENDS in Biochemical Sciences. 28 (10): 541 547. Zhang, Z.; Al Rubeai, M. and Thomas, C. R. (1993). Estimation of disruption of animal cells by turbulent capillary flow. Biotechnology and Bioengineering. 42: 987–993. Zhou, G. and Kresta, S. M. (1996b). Impact of Tank Geometry on the Maximum Turbulence Energy Dissipation Rate for Impellers. AIChE Journal. 42:2476 2490. 244 Stress Agent Cell line HSP Reference expressed C. botulinum C2 toxin Vero HSP90 Haug et al., C. perfringens iota toxin HeLa 2003 C2IN C3 fusion toxin J774.A1 Anticancer agent (17 AAG) HL 60 HSP90 Pelicano et Jurkat al., 2006 Heat shock Whole blood simple: HSP70 Oehler et al., different types 2001 Radiofrequency/microwave CHO N.D. Cleary et al., radiation HeLa 1996 Chemotherapy drugs Blood samples HSP27 Thomas et al., HSP60 2005 HSP70 HSP90 HSP110 Heat shock Human monocytes HSP70 Bachelet et al., 1998 Disulfiram (Drug for Hyperbaric CHO Cystine Deneke et al., hyperoxia) transport 1997 Heat shock Mesenchymal Stem HSP 27 Stolzing et Cold shock Cells HSP 70 al., 2006 HSP 90 Table 5.1. Stress agents reported that induce stress proteins expression. 245 Cell line Product Phase cell cycle Reference higher productivity CHO β galactosidase S phase Kompala et al 1993 CHO t PA G2 Aggeler et al 1982 Epithelial cells Acid phosphatase Late G1 Herz et al 1991 CHO Dhfr S Mariani et al 1981 CHO t PA G1 Dutton et al 2006 Table 5.2. State of art on relationship between protein expression and cell cycle phases. 246 C K J B F, H Cell Response A I G D E Wm 3 1 2 3 4 5 6 7 8 9 10 10 10 10 10 10 10 10 10 1 11 2 6 4 9 13 12, 14 3, 5, 7, 10, 8 Hydrodynamic Conditions Cell Response Mode of Symbol Cell Reference growth A CHO-K1, necrosis Anchoraged Gregoriades et al. (2000) Thomas et al. (1994); Zhang Hybridoma, necrosis Suspended B et al. (1993) C MCF-7, necrosis Suspended Ma et al. (2002) D Mouse myeloma, necrosis Suspended McQueen and Bailey (1989) E HeLa S3, mouse L929, necrosis Suspended Augenstein et al. (1971) F CHO-K1, SF-9, HB-24, necrosis Suspended Ma et al. (2002) Uninfected and viral infected Suspended Submitted G PERC6 cells, necrosis Enthomopathogenic nematodes, Suspended Fife et al. (2004) H necrosis I CHO-K1, apoptosis Anchorage Mollet et al. (2007) J THP-1, necrosis Anchorage Mollet et al. (In Press) K Algae, loss of flagella Suspension Hu et al. (2007) Continued… Figure 5.1. Summary of the reported energy dissipation rate at which cells are affected as well as the reported levels of energy dissipation rate in various bioprocess environments. Adapted and improved from Ma et al. (2002) and Mollet et al. (2004). 247 Figure 5.1. (Continued). Hydrodynamic Conditions Symbol Process Description Reference Volume average in typical Agitation Varley and Birch (1999) 1 animal cell bioreactors. Volume average in a 10 L Agitation 2 mixing vessel (RT, 700 RPM) Maximum in the 10 L mixing Zhou and Kresta (1996) 3 Agitation vessel (RT, 700 RPM) Volume average in a 22,000 L Agitation 4 mixing vessel (RT, 240 RPM) Wernersson and Maximum in the 22,000 L mixing Tragardh (1999) Agitation 5 vessel Maximum in spinner vessel Agitation Venkat et al. (1996) 6 (200 RPM) Pure water, bubble diameter: Bubble rupture Garcia-Briones et al. (1994) 7 6.32mm Boulton-Stone and Blake Pure water, bubble Bubble rupture (1993); 8 diameter:1.7mm Garcia-Briones et al. (1994) Flow through Pure water, 100 mL/min, 1 mm Mollet et al. (2004) 9 a pipe diameter Flow through Flow through a 200 L 10 a micropipette Mollet et al. (2004) micropipette tip in 0.2 sec tip Volume average in a highly Agitation Oh et al. (1992) 11 agitated animal cell bioreactor Laboratory scale centrifuge plus Centrifugation Hutchinson et al. (2005) 12 rotational device 13 Centrifugation Disc stack centrifuge Neal, et al. (2002) Pilot multichamber bowl Centrifugation Boychin, et al (2001) 14 centrifuge operated at 167 rps 248 chronic exposure of suspended Adapted with permission from Godoy silva et Diagram of the experimental setup for continuous, 5.2. igure animal cells to high levels of hydrodynamic forces. al (2009). F 249 1.20 1.00 0.80 0.60 (a) 0.40 Normalized Cell count Cell Normalized Perforated tube, strain A 0.20 Perforated tube, Strain B Sintered Sparger, strain A Sintered Sparger, strain B 0.00 0 2 4 6 8 10 12 14 16 18 20 22 Elapsed Time , day 1.20 1.00 0.80 (b) 0.60 Normalized VCCNormalized 0.40 0.20 Perforated tube, strain A Perforated tube, strain B Sintered sparger, strain A Sintered sparger, strain B 0.00 0 2 4 6 8 10 12 14 16 18 20 22 Elapsed Time Figure 5.3. Growth kinetic (a) and viability (b) for four cell cultures performed in bioreactor with two different impeller/sparger configurations. One configuration operated with perforated tube sparger and dual impeller (PBT in the top and Rushton bottom), the second configuration operated with sintered sparger (50 m) and PBT impeller. Two strains of the CHO clonal cell line were used strain A and strain B were seeded in each bioreactor as is indicated in the label. 250 1.20 Perforated tube, strain A Perforated tube, strain B 1.00 Sintered sparger, strain A Sintered sparger, strain B 0.80 0.60 (a) 0.40 Normalized Glucose Glucose Normalized 0.20 0.00 0 2 4 6 8 10 12 14 16 18 20 22 Elapsed time 1.20 Perforated tube, strain A Perforated tube, strain B 1.00 Sintered sparger, strain A Sintered sparger, strain B 0.80 0.60 (b) 0.40 Normalized Lactate Normalized 0.20 0.00 0 2 4 6 8 10 12 14 16 18 20 22 Elapsed Time Figure 5.4. Concentration of glucose (a) and lactate (b) for four cell cultures performed in bioreactor with two different impeller/sparger configurations. One configuration operated with perforated tube sparger and dual impeller (PBT in the top and Rushton bottom), the second configuration operated with sintered sparger (50 m) and PBT impeller. Two strains of the CHO clonal cell line were used strain A and strain B were seeded in each bioreactor as is indicated in the label. 251 A B Debris Dip G1 Dip G2 Dip S 0 30 60 90 120 150 Channels (FL2-A) Figure 5.5. Cell cycle diagram (A) Typical cell cycle histogram obtained in flow cytometer (B). Content of DNA increases along the x axis. First peak correspond to cells in G1 phase, small area in the center with lines correspond to S phase, peak further right correspond to cells in G2 phase. 252 1 A B 1.2 0.8 1 0.8 0.6 G2 0.6 S 0.4 G1 0.4 A 0.2 0.2 Normalized Titer Normalized 0 Fraction of population of Fraction 0 0 10 20 30 40 50 60 70 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Day G1_IVC C 1.2 D 1.2 1 1 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 Normalized Titer Normalized Normalized Titer Normalized 0 0 0 5 10 15 20 25 30 0 1.5 3 4.5 6 7.5 9 S_IVC G2_IVC Figure 5.6. Cell cycle profile (A), Relationship between cell cycle and titer: subpopulation in G1 phase (B), subpopulation in G2 phase (C), and subpopulation in S phase (D). 253 A Open tube Double impeller G2 S G1 A Fraction of population of Fraction 0 1 2 3 4 5 6 7 8 10 11 12 13 14 15 16 17 18 19 20 Time B Sintered sparger PBT impeller G2 S G1 A Fraction of population of Fraction 0 1 2 3 4 5 6 7 8 1011121314151617181920 Time Figure 5.7. Cell cycle profile comparison between reactors with different hydrodynamic conditions. Two impeller/sparger configurations were evaluated: A. Open tube sparger and dual impeller Rushton bottom – PBT top. B. Sintered sparger and PBT impeller. 254 1.2 1.0 0.8 0.6 0.4 Normalized Titer 0.2 R523 Open tube_Dual imp R623 Sintered_PBT 0.0 0 20 40 60 80 100 120 G1_IVC, % Figure 5.8. Relationship between cell cycle and titer subpopulation in G1 phase for two different reactor configurations. 1. Open tube sparger and dual impeller Rushton bottom – PBT top. 2. Sintered sparger and PBT impeller. 255 Unstained Cells TFla#174A8F.fcs Aggregates Dead cells (a) SSC-A Live cells Debris 0 200 400 400 200 0 600 1000 800 0 200 400 600 800 1000 FSC-A R5: Live cells R6: Debris R7: Dead cells R8: Agregates IgG PE Unstained HSP90 - PE (b) Counts 0 0 40 80 120 160 200 100 101 102 103 104 PE-A Figure 5.9. Typical plots for stress proteins analysis in flow cytometry. Dot plot (a) indicates gated populations for live cells, dead cells, debri and aggregates. Histogram to estimate mean fluorescence intensity of samples as indicative of level of expression of stress proteins (b). 256 6 R619 Sintered_PBT Strain A R623 Sintered_PBT Strain B 5 R515 Open tube_Dual imp Strain A R523 Open tube_Dual imp Strain B 4 3 2 HSP70_IVC 1 0 0 0.2 0.4 0.6 0.8 1 1.2 Normalized Titer Figure 5.10. Analysis of relationship between population expressing HSP70 and specific productivity. 257 30 25 20 15 HSP90_IVC 10 R619 Sintered_PBT Strain A 5 R623 Sintered_PBT Strain B R515 Open tube_Dual imp Strain A R523 Open tube_Dual imp Strain B 0 0 0.2 0.4 0.6 0.8 1 1.2 Normalized Titer Figure 5.11. Analysis of relationship between population expressing HSP90 and specific productivity. 258 600 HSP90 500 HSP70 400 (a) 300 200 Relative expressionHSPRelative 100 0 0 100 200 300 400 500 600 Time, hr 300 HSP90 250 HSP70 200 (b) 150 100 Relative expressionHSPRelative 50 0 0 100 200 300 400 500 600 Time, hr Figure 5.12. Profile of stress proteins expression. Two different strains of a clonal CHO cell line where seeded in two bioreactors with the configuration sintered sparger and PBT Impeller. Strain A (a), and Strain B (b). 259 600 HSP90 500 HSP70 400 300 (a) 200 100 Relative expressionHSP90Relative 0 0 100 200 300 400 500 600 Time, hr 300 HSP90 250 HSP70 200 150 (b) 100 50 Relative expressionHSP90Relative 0 0 100 200 300 400 500 600 Time, hr Figure 5.13. Profile of stress proteins expression. Two different strains of a clonal CHO cell line where seeded in two bioreactors with the configuration open tube sparger and dual impeller (PBT top, Rushton bottom). Strain A (a), and Strain B (b) 260 4 ) 3.5 6 3 2.5 2 (a) 1.5 1 Density (cell 10 (cell * / mL Density Reactor 0.5 Tflask 1 Tflask 2 0 0 25 50 75 100 125 150 175 200 Time , hr 1.2 ) 6 1.0 0.8 (b) 0.6 0.4 Reactor Density (cell 10 (cell * / mL Density 0.2 Tflask 1 Tflask 2 0.0 0 25 50 75 100 125 150 175 200 Time , hr Figure 5.14. Growth kinetic of cell line CHO 6E6. (a), and viability, (b), for cell culture in agitated bioreactor (2 L working volume) and TFlasks, to compare profile of stress protein expression. 261 5 4.5 4 3.5 3 Reactor (a) 2.5 Tflask 1 2 Tflask 2 Glucose, g/L Glucose, 1.5 1 0.5 0 0 25 50 75 100 125 150 175 200 225 Time, hr 4.5 4 3.5 3 2.5 (b) 2 1.5 Reactor Lactate, g/L Lactate, 1 Tflask 1 0.5 Tflask 2 0 0 25 50 75 100 125 150 175 200 225 Time , hr Figure 5.15. Concentration of glucose, (a), and lactate, (b), as a function of time of two batchs of cells growing in TFlasks Vs. cells growing in agitated bioreactor (2L working volume). Cell line CHO 6E6. 262 1.2 1.2 Tflask 1 Reactor Tflask 2 1 1 SP90 H 0.8 0.8 0.6 0.6 expression, 0.4 0.4 elative Relative expression, expression, HSP90 Relative R 0.2 0.2 0 0 0 20 40 60 80 100 120 140 160 180 200 220 0 20 40 60 80 100 120 140 160 180 200 220 Time, hr Time, hr 1.2 1.2 Tflask 1 Reactor Tflask 2 1 1 0.8 0.8 0.6 0.6 0.4 0.4 Relative expression, expression, HSP25 Relative Relative expression expression HSP25 Relative 0.2 0.2 0 0 0 20 40 60 80 100 120 140 160 180 200 220 0 20 40 60 80 100 120 140 160 180 200 220 Time, hr Time, hr 4 4 Reactor Tflask 1 3 3 Tflask 2 2 2 1 1 0 0 0 20 40 60 80 100 120 140 160 180 200 220 0 20 40 60 80 100 120 140 160 180 200 220 1 1 2 2 3 3 Relative expression expression HSP70 Relative Relative expression expression HSP70 Relative 4 4 5 5 Time, hr Time, hr Figure 5.16. Comparison on expression of stress proteins between reactor and TFlasks. Cell line CHO 6E6. 263 HSP90 HSP70 t= 1 hr t=1 hr Counts Counts 0 1 2 3 4 0 40 0 80 120 160 200 1 2 3 4 0 0 10 40 80 120 160 200 10 10 10 10 10 10 10 10 10 FITC PE FL2 HSP90 HSP70 t=6 hr t=6 hr Counts Counts 100 101 102 103 104 0 40 0 80 120 160 200 1 2 3 4 0 0 40 80 120 160 200 10 10 10 10 10 FITC PE FL2 HSP90 HSP70 t=8 hr t=8 hr Counts Counts 0 1 2 3 4 0 0 10 40 80 120 160 200 10 10 10 10 0 1 2 3 4 FITC 0 10 40 80 120 160 200 10 10 10 10 PE FL2 HSP90 HSP70 t=10 hr t=10 hr Counts Counts 0 1 2 3 4 0 1 2 3 4 0 0 10 40 80 120 160 200 10 10 10 10 0 10 40 80 120 160 200 10 10 10 10 FITC PE FL2 Figure 5.17. THP1 cell suspension was subjected to recycle through the TC at 50 mL·min 1, corresponding to 6.4×106 W·m 3, cells passed 10 times through TC. After stress, cells were incubated at 37 °C in a 5% CO2 atmosphere. Samples were taken 1, 6, 8 10 hr in recovery to evaluate HSP70 and HSP90 expression. Purple line corresponds to unstained control, green line corresponds to control (non stressed cells), red line correspond to sample cells stressed at 50 mL/min. 264 HSP90 HSP70 t=0 hr t=0 hr Counts Counts 0 0 40 80 120 160 200 0 40 80 120 160 200 100 101 102 103 104 100 101 102 103 104 FITC PE FL2 HSP90 HSP70 t=2 hr t=2 hr Counts Counts 0 0 40 80 120 160 200 0 40 80 120 160 200 100 101 102 103 104 0 1 2 3 4 FITC 10 10 10 10 10 PE FL2 HSP90 HSP70 t=7 hr t=7 hr Counts Counts 0 0 40 80 120 160 200 0 40 80 120 160 200 0 1 2 3 4 100 101 102 103 104 10 10 10 10 10 FITC PE FL2 Figure 5.18. THP1 cell suspension was subjected to recycle through the TC at 90 mL·min 1, corresponding to 1.1×108 W·m 3, cells passed 10 times through TC. After stress, cells were incubated at 37 °C in a 5% CO2 atmosphere. Samples were taken 2 and 7 hr in recovery to evaluate HSP70 and HSP90 expression. Purple line corresponds to unstained control, green line corresponds to control (non stressed cells), red line correspond to sample cells stressed at 90 mL/min. 265 CHAPTER 6 CONCLUSIONS AND RECOMMENDATIONS This study has been oriented to understand the effect of hydrodynamic forces on mammalian cell cultures. My approach to accomplish that goal includes: hydrodynamic studies in bioprocess and biomedical equipment, cell sensitivity screening for a number of human and industrial cell lines and physiological characterization of cells after stress exposure. Cell sensitivity screening was performed with the shear stress device developed earlier in our group. An improvement was made based on suggestions stated in Dr. Godoy-Silva’s dissertation (2008), regarding the use of ½ inches polycarbonate instead of 1 inch, this improvement avoids deformation during autoclave step. By using the methodology of single pass, it was possible to establish breakthrough curves for the cell lines evaluated. It is worth to point out that we found out in this work that THP1 is a very sensitive cell line, which could answer some questions regarding delays in growth after passing this cell through flow cytometer for sorting purposes. This result can be considered remarkable from two perspectives. First, it allow us to consider THP1 as candidate for studying cell damage mechanism, given that effects due to hydrodynamic stress might be observed 266 easier and faster using this cell line. Secondly, the methodology results attractive for clones selection in the industry, even though the differences in sensitivity between industrial cell lines are not as big as the observed between THP1 and other cell lines, small differences in sensitivity might represent drastic differences in cell damage at large scale. In fact, the technology has been already used in some pharmaceutical industries with that purpose. As part of my PhD training I was assigned to give instruction in the use of the device to two pharmaceutical companies. The results are not presented in this document because they are protected under confidentiality agreement. Hydrodynamic studies were performed in bioreactors bench scale used in the pharmaceutical industry. This part of the work was performed as part of my internship at GlaxoSmithKline (GSK). Although the hydrodynamic characterization in bioreactors seems to be a mature area, there is still interest in developing strategies that allow scaling up processes with the confidence of controlling shear stress impact and improving mass transfer capabilities. Hydrodynamic conditions in a bioreactor are regulated by impeller/sparger configuration. In this part of my study I evaluated different impeller/sparger configurations and a number of empirical correlations were obtained to describe their hydrodynamic behavior. A complement to this experimental work could be the hydrodynamic characterization of the same configurations using computer simulations. Based on the correlations and mass transfer capabilities two designs were chosen to evaluate cell culture performance. In general the configurations selected provide proper 267 conditions to obtain high cell density and similar yield of product. The main difference observed was the accumulation of CO2 when the sintered sparger was used. Although, this concentration of CO2 did not affect growth behavior, it could be speculated that other important aspects in the cell culture might be affected such as the quality of the recombinant protein. Also as part of my internship in GSK, the reactor configurations discussed above, were also used to evaluate the effect of different hydrodynamic conditions on cell cycle, stress proteins expression and its relationship with the production of the recombinant protein. The research reported here would seem to indicate that there is a direct relationship between the population in G1 phase of the cell cycle and the expression of recombinant protein. The analysis of expression of the stress proteins HSP70 and HSP90 suggested a particular pattern of expression. Population expressing HSP90 does not show a clear correlation with titer of recombinant protein while population expressing HSP70 seem to be directly related with titer. Incidentally, the levels of expression of HSP70 are very low and a slight increase along the culture was observed as opposed to HSP90, which showed a larger level of expression at the beginning of the culture with a decreasing tendency. Although exploratory, this study offered some insight into the mechanism and possible relationship between cell damage and expression of stress proteins. The study of stress protein expression was complemented at OSU with an ATCC CHO cell line. Consistent results were obtained in terms of expression of stress proteins profile along the culture, moreover control experiments were included in TFlasks where no 268 hydrodynamic forces were involved. A different expression profile of stress proteins was observed in the control experiments, which was characterized by a tendency to oscillate around the higher values of expression as opposed to expression of stress proteins in reactors where a decreasing tendency was observed. Results also suggested that the level of stress protein and the type of stress protein expressed is cell type and even cell clone dependent. The findings in this work regarding expression of stress proteins in different hydrodynamic conditions may imply that stress proteins are involved in the chain of response mechanism of cells against hydrodynamic stress. Further research is required in order to define the link between the expression of stress proteins and other physiological responses. In addition, study of stress proteins in the context of apoptosis and hydrodynamic stress may contribute not only to understand cell damage mechanism at molecular level but also may give directions on ways of controlling apoptosis in cell cultures. Cell damage was also studied in Fluorescent Activated Cell Sorter Aria. Analysis performed include: Lethal cell damage based on released of LDH, cell growth behavior after sorting, protective effect of fetal bovine serum (FBS), and hydrodynamic characterization. Overall, results indicated that cell damage as result of sorting is in the range of 8.6% to 15.5% for cells sorted in media with 0% and 10% FBS respectively. 269 Growth behavior of cells after stress was evaluated. A delayed in the growth of sorted cells was observed was well as cells stressed on shear stress device. Enriched media helped cells to resume growth after stress. It is recommended to use enriched media in collecting and growing steps after sorting. Overall, the results obtained in this study may represent a contribution toward understanding cell damage mechanism as a response to hydrodynamic stress. 270 BIBLIOGRAPHY Aloi, L. E. and Cherry, R. S. (1996). Cellular response to agitation characterized by energy dissipation at the impeller tip. Chemical Engineering Science . 51 : 1523-1529. Al-Rubeai, M.; Emery, A. N.; Chalder, S. and Goldman, M. H. (1993). A flow cytometric study of hydrodynamic damage to mammalian cells. Journal of Biotechnology . 31 (2): 161-177. Al-Rubeai, M.; Oh, S. K. W.; Musaheb, R. and Emery, A. N. (1990). Modified cellular metabolism in hybridomas subjected to hydrodynamic and other stresses. Biotechnology Letters . 12 : 323-328. Al-Rubeai, M.; Singh, R. P.; Goldman, M. H. and Emery, A. N. (1995). Death mechanisms of animal cells in conditions of intense agitation. Biotechnology and Bioengineering . 45 : 463-472. Amano, K., Haga, R. and Murakami, S. 2008. Expressions of mass transfer coefficient of bubbles and free surface of culture tanks using the k-e turbulence model. J Ind Microbiol Biotechnol. 35: 525-531. Amanullah, A.; Serrano-Carreon, L.; Castro, B.; Galindo, E. and Nienow, A. W. (1998). The influence of impeller type in pilot scale xanthan fermentations. Biotechnology and Bioengineering . 57 (1):95-108. Asay, T. and Kono, T. (1982). Estimation of oxygen absorption coefficient and power consumption in a stirred-tank fermentor. Journal of Fermentation Technology . 60 (3):265- 268. 271 Aslam, I.; Robins, A; Dowell, K, and Fishel, S. (1998). Isolation, purification and assessment of viability of spermatogenic cells from testicular biopsies of azoospermic men. Human Reproduction . 13 (3): 639-645. Augenstein, D. C. ; Sinskey, A J. and Wang, D. I. C. (1971). Effect of shear on the death of two strains of mammalian tissue cells. Biotechnology and Bioengineering . 13 :409-418. Aunins, J. G.; Woodson, B. A. Jr.; Hale, T. K. and Wang, D. I. C. (1989). Effects of paddle impeller geometry on power input and mass transfer in small-scale animal cell culture vessels. Biotechnology and Bioengineering . 34 (9):1127-1132. Backer, M. P.; Metzger, L. S.; Slaber, P. L.; Nevitt, K. L. and Boder, G. B. (1988). Large scale production of monoclonal antibodies in suspension culture. Biotechnology and Bioengineering . 32 :993-1000. Bachelet, M.; Marlethoz, E.; Banzet, N.; Souil, E.; Pinot, F.; Polla, C.Z.; Durand, P.; Bouchaert, I. and Polla B. (1998). Flow cytometry is a rapid and reliable method for evaluating heat shock protein 70 expression in human monocytes. Cell Stress and Chaperones . 3 (3): 168 – 176. Barnabé, N. and Butler, M. (2000). The effect of glucose and glutamine on the intracellular nucleotide pool and oxygen uptake rate of a murine hybridoma. Cytotechnology. 34 : 47–57. Bavarian, F.; Fan, L. S. and Chalmers, J. J. (1991). Microscopic visualization of insect cell-bubble interactions. I: Rising bubbles, air-medium interface and the foam layer. Biotechnology Progress . 7:140-150. Bi, J. X.; Shuttleworth, J. and Al-Rubeai, M. (2004). Uncoupling of cell growth and proliferation results in enhancement of productivity in p21 CIP1 -arrested CHO cells. Biotechnology and Bioengineering . 85 : 721-749. Birch, J.R.; Boraston, R. and Wood, L. (1985). Bulk production of monoclonal antibodies in fermenters. Trends in Biotechnology . 3:162–166. 272 Birch, J.R.; Thompson, P. W; Boraston, R.; Oliver, S. and Lambert, K. (1987). The large scale production of monoclonal antibodies in airlift fermenters. In: Plant and Animal Cells: Process Possibilities . Webb, C. and Mavituna, F. (Eds). Ellis Horwood. Chichester, UK. Bird, R. B.; Stewart, W. E. and Lightfoot, E. N. (2006). Transport Phenomena. 2 nd Ed. John Wiley & Sons, New York. Bluestein, M. and Mockros, L. F. (1969). Hemolysis effects of energy dissipation in flowing blood. Medical and Biological Engineering and Computing. 7(1):1-6. Boraston, R.; Thompson, P. W.; Garland, S. and Birch, J. R. (1983). Growth and oxygen requirement of antibody-producing mouse hybridoma cells in suspension culture. Developments in Biological Standardization . 55 :103-111. Born, C.; Zhang, Z.; Al-Rubeai, M. and Thomas, C. R. (1992). Estimation of disruption of animal cells by laminar shear stress. Biotechnology and Bioengineering . 40 (9): 1004- 1010. Boulton-Stone, J. M. and Blake, J. R. (1993). Gas bubbles bursting at a free surface. Journal of Fluid Mechanics . 254 :103-111. Brodkey, R. S. (1995). The Phenomena of Fluid Motions . Dover Publishing, New York. Bryant, J. C.; Schilling, E. L.; Earle, W. R. and Evans, V. J. (1960). Massive fluid suspension cultures of certain mammalian tissue cells. III. Exploratory supplementation experiments with chemically defined medium NCTC 109. Journal of the National Cancer Institute . 24 :859-871. Butler, M. (2005). Animal cell cultures: recent achievements and perspectives in the production of biopharmaceuticals. Applied Microbiology and Biotechnology. 68 : 283– 291. Carrel, A. (1912). On the permanent life of tissues outside the organism. Journal of Experimental Medicine. 15 (5):516-528. 273 Carvalhal, A. V.; Marcelino, I. and Carrondo, M. J. T. (2003). Metabolic changes during cell growth inhibition by p27 overexpression. Applied Microbiology and Biotechnology . 63 :164–173. Cartwright, 1994. Animal Cells as bioreactors. Cambridge University Press. Chalmers, J. J. and Bavarian, F. (1991). Microscopic visualization of insect cell-bubble interactions II: The bubble film and bubble rupture. Biotechnology Progress . 7:151-158. Chattopadhyay, D.; Rathman, J. and Chalmers, J. J. (1995a). The protective effect of specific medium additives with respect to bubble rupture. Biotechnology and Bioengineering . 45 :473-480 Chattopadhyay, D.; Rathman, J. and Chalmers, J. J. (1995b). Thermodynamic approach to explain cell adhesion to gas-liquid interfaces. Biotechnology and Bioengineering . 48 :649-658. Chen, A. K.; Latz, M. I. and Frangos, J. A. (2003). The use of dinoflagellate bioluminescence to characterize cell stimulation in bioreactors. Biotechnology and Bioengineering . 83 (1):93-103. Cherry, R. S. and Papoutsakis, E. T. (1989). Hydrodynamic effects on cells in agitated tissue culture reactors. Bioprocess Engineering . 1:29-41. Cherry, R. S. and Papoutsakis, E. T. (1988). Physical mechanisms of cell damage in microcarrier bioreactors. Biotechnology and Bioengineering . 32 :1001-1014. Chien, S. (2007). Mechanotransduction and endothelial cell homeostasis: the wisdom of the cell. American Journal of Physiology - Heart and Circulation Physiology . 292 :1209- 1224. Chisti, Y. (1993). Animal cell culture in stirred bioreactors: observations in scale-up. Bioprocess Engineering . 9:191-196. 274 Chisti, Y. (1999). Shear sensitivity. In: M.C. Flickinger and S.W. Drew, (eds). Encyclopedia of Bioprocess Technology: Fermentation, Biocatalysis, and Bioseparations . Vol. 5. John Wiley & Sons. New York. Pp. 2379-2406. Chisti, Y. (2001). Hydrodynamic damage to animal cells. Critical reviews in biotechnology . 21 (2): 67-110. Cho, M. H. and Wang, S. S. (1988). Enhancement of oxygen-transfer in hybridoma cell- culture by using a perfluorocarbon as an oxygen carrier. Biotechnology Letters . 10 :855- 860. Chu, L. and Robinson D. K. (2001). Industrial choices for protein production by large- scale cell culture. Current Opinion in Biotechnology . 12 :180–187. Clay, J. D. (1997). Molecular degradation of polymer solutions in fast transient extensional flow. Ph.D. Thesis. Chemical Engineering Department. The Ohio State University. Cleary, S.; Cao, G.; Liu, L.; Egle, P.M. and Shelton, K. (1997). Stress Proteins are not induced in mammalian cells exposed to radiofrequency or microwave radiation. Bioelectromagnetism . 18 (7): 499-505. Costes, J. and Couderc, J.P. (1988). Study by laser Doppler anemometry of the turbulent flow induced by a Rushton turbine in a stirred tank: Influence of the size of the units - II. Spectral analysis and scales of turbulence. Chemical Engineering Science . 43 (10):2765- 2772. Croughan, M. S.; Hamel, J. F. and Wang, D. I. C. (1987). Hydrodynamic effect on animal cells grown in microcarrier cultures. Biotechnology and Bioengineering . 29 : 130-141. Croughan, M. S.; Sayre, E. S. and Wang, D. I. C. (1989). Viscous reduction of turbulent damage in animal cell cultures. Biotechnology and Bioengineering . 33 : 862-872. Cui, Y. Q. ; van der Lans, R. G. J. M. and Luyben, K. C. A. M. (1996). Local power uptakes in gas liquid systems with single and multiple Rushton turbines. Chemical Engineering Science . 51 (11): 2631-2636. 275 Cutter, L. A. (1966). Flow and Turbulence in a Stirred Tank. AIChE Journal . 12 :35-45. Deneke, S.; Harford, P.; Lee, K.; Deneke, C.; Wright, S. and Jenkinson, S. (1997). Induction of Cystine Transport and other stress proteins by disulfiram: Effects on Gluthathione levels in cultured cells. Am. J. Respir. Cell Mol. Biol . 17 : 227-234. Deshpande, R. R. and Heinzle, E. (2004). On-line oxygen uptake rate and culture viability measurement of animal cell culture using microplates with integrated oxygen sensors. Biotechnology Letters . 26 : 763–767. deZengotita, V. M.; Kimura, R. and Miller, W. M. (1998). Effects of CO 2 and osmolality on hybridoma cells: growth, metabolism and monoclonal antibody production. Cytotechnology . 28 :213-227. deZengotita, V. M.; Miller, W. M., Aunins, J. G. and Zhou, W. (2000). Phosphate Feeding Improves High-Cell-Concentration NS0 Myeloma Culture Performance for Monoclonal Antibody Production. Biotechnology and Bioengineering . 69 (5):566-576. Discher, D. E.; Mohandas, N. and Evans, E. A. (1994). Molecular maps of red cell deformation: hidden elasticity and in situ connectivity. Science . 266 (5187):1032-1035. Dorresteijn, R. C.; Degooijer, C. D.; Tramper, J. and Beuvery, E. C. (1994). A method for simultaneous determination of solubility and transfer coefficient of oxygen in aqueous media using off-gas mass spectrometry. Biotechnology and Bioengineering . 43 :149-154. Drapeau, D.; Luan, Y. T.; Whiteford, J. C.; Lavin, D. P. and Adamson, S. R. (1990). Cell culture scale-up in stirred tank reactors. Presented at the Annual Meeting of Society of Industrial Microbiology, Orlando, FL. USA. Ducommun, P.; Bolzonella, I.; Rhiel, M.; Pugeaud, P.; von Stockar, U. and Marison, I. W. (2001). On-Line Determination of Animal Cell Concentration. Biotechnology and Bioengineering . 72 (5): 515-522 Eyer K.; Oeggerli, A. and Heinzle, E. (1995). On-line gas analysis in animal cell cultivation: II. Methods for oxygen uptake rate estimation and its application to controlled feeding of glutamine. Biotechnology and Bioengineering . 45 : 54-62. 276 Fife, J. P.; Derksen, R. C.; Ozkan, H. E.; Grewal, P. S.; Chalmers, J. J. and Krause, C. R. (2004). Evaluation of a contraction flow field on hydrodynamic damage to enthomopathogenic nematodes – A biological pest control agent. Biotechnology and Bioengineering . 86 (1):96-107. Frahm, B., Brod, H., and Langer, U. 2009. Improving bioreactor cultivation conditions for sensitive cell lines by dynamic membrane aeration. Cytotechnology. 59: 17 – 30. Fong S. S.; Joyce A. R. and Palsson, B. Ø. (2005). Parallel adaptive evolution cultures of Escherichia coli lead to convergent growth phenotypes with different gene expression states. Genome Research . 15 :1365-1372. Galindo, E. and Nienow, A. W. (1993). Performance of the Scaba 6SRGT agitator in mixing of simulated xanthan gum broths. Chemical Engineering and Technology . 16 (2):102-108. Garcia, M. A. and Chalmers, J. J. (1994). Flow parameters associated with hydrodynamic cell injury. Biotechnology and Bioengineering. 44 :1089-1098. García-Briones, M. A. and Chalmers, J. J. (1992). Cell-bubble interactions: Mechanisms of suspended cell damage. Annals of the New York Academy of Sciences . 665 :219-229. Garcia-Briones, M. A.; Brodkey, R. S. and Chalmers, J. J. (1994). Computer simulations of the rupture of a gas bubble at a gas-liquid interface and its implications in animal cell damage. Chemical Engineering Science . 49 (14):2301-2320. Gething, M.J. and Sambrook, J. (1992). Protein folding in the cell. Nature . 355 (2), 33- 45. Gimenes, M. A. P.; Carlos, L. C. S.; Faria L. F. F. and Pereira, N. Jr. (2002). Oxygen uptake rate in production of xylitol by Candida guilliermondii with different aeration rates and initial xylose concentrations. Applied Biochemistry and Biotechnology . 98- 100 (1-9):1049-1059. Glasgow, L. A.; Hua, J.; Yiin, T.-Y. and Erickson, L. E. (1992). Experiment studies of interfacial phenomena in sparged reactors. In: Tatterson, G B. and Calabrese, R. V. 277 (Eds.). Process Mixing: chemical and biological applications . AIChE Symposium Series. Vol. 88, Issue 286. Pp.103-108. American Institute of Chemical Engineers. New York. Gódia, F. and Cairó, J. J. (2006). Cell metabolism. In: S.S. Ozturk and W-S. Hu (Eds). Cell culture technology for pharmaceutical and cell-based therapies . Pp. 81-112. CRC Press. Boca Raton, FL., USA. Godoy-silva, R., (2008). Lethal and sub-lethal effects of hydrodynamic forces on animal cell culture. PhD Dissertation. Chemical and Biomolecular Engineering. The Ohio State University. Godoy-silva, R., Chalmers, J.J., Casnocha, S.A., Bass, L.A., and Ma, N. (2009). Physiological Responses of CHO Cells to Repetitive Hydrodynamic Stress. Biotechnology and Bioengineering . 103 (6):1103-1117. Godoy-silva, R., Berdugo, C. and Chalmers, J. (Peer review). Literature Review: Mixing, Aeration and hydrodynamics in bioreactors. Wiley Encyclopedia of Industrial Biotechnology. Goldblum, S.; Bae, Y. K.; Hink, W. F. and Chalmers, J. J. (1990). Protective effect of methylcellulose and other polymers on insect cells subjected to laminar shear stress. Biotechnology Progress . 6:383-390. Goldman, W. H. (2000). Mechanical manipulation of animal cells: cell indentation. Biotechnology letters . 22 : 431-435. Goto, T.; Mochizuki, G. and Kikuchi, K.-I. (2001a). Perfluorocarbon-mediated aeration applied to recombinant protein production by virus-infected insect cells. Biochemical Engineering Journal . 7 (1):69-78. Goto, T.; Mochizuki, G. and Kikuchi, K.-I. (2001b). A novel column fermentor having a wetted-wall of perfluorocarbon as an oxygen carrier. Biochemical Engineering Journal . 8(2):165-169. 278 Graf, R.; Apenberg, S.; Freyberg, M. and Friedl, P. (2003). A common mechanism for the mechanosensitive regulation of apoptosis in different cell types and for different mechanical stimuli. Apoptosis . 8: 531-538. Gray, D. R.; Chen, S.; Howarth, W.; Inlow, D. and Maiorella, L. (1996). CO 2 in large- scale and high-density CHO cell perfusion culture. Cytotechnology . 22 :65-78. Gregoriades, N.; Clay, J.; Ma, N.; Koelling, K. and Chalmers, J. J. (2000). Cell damage of microcarrier cultures as a function of local energy dissipation created by a rapid extensional flow. Biotechnology and Bioengineering . 69 :171-182. Griffiths, J. B. (2000). Animal cell products, overview. In: Spier RE (ed). Encyclopedia of cell technology , Vol. 1. Pp 71–76. Wiley, New York, USA. Guck, J.; Ananthakrishnan, R.; Mahmood, H.; Moon, T. J.; Cunningham, C. C. and Käs, J. (2001). The optical stretcher: a novel laser tool to micromanipulate cells. Biophysical Journal . 81 : 767-784. Haga, J. H.; Li, Y.-S. J. and Chien, S. (2007). Molecular basis of the effects of mechanical stretch on vascular smooth muscle cells. Journal of biomechanics . 40 : 947- 960. Handa, A.; Emery, A. N. and Spier, P. E. (1987). On the evaluation of gas-liquid interfacial effects on hybridoma viability in bubble column bioreactors. Developments in Biological Standardization . 66 : 241-253. Handa-Corrigan, A.; Emery, A. N. and Spier, R. E. (1989). Effects of gas-liquid interfaces on the growth of suspended mammalian cells: mechanisms of cell damage by bubbles. Enzyme and Microbial Technology . 11 : 230-235. Harris, R.; Ng, K. W. and Wraith, A. E. (2005). Spargers for controlled bubble size by means of the multiple slot disperser (MSD). Chemical Engineering Science . 60 : 3111 – 3115. Harrison, R.G. (1907). Observations on the living developing nerve fiber. Proceedings Proceedings of the Society for Experimental Biology and Medicine. 4:140–3. 279 Haug, G.; Leemhuis, J.; Tiemann, D.; Meyer, D. K., Aktories, K. and Barth, H. (2003). The Host Cell Chaperone HSP90 is essential for translocation of the binary clostridium botulinum C2 toxin into the cytosol. The Journal of Biological Chemistry . 278 (34); 32266 – 32274. Haugen, E.M, Cucci, T.L, Yentsch, C.M., Grotberg, and Shapiro, L.P. (1987). Effects of Flow Cytometric Analysis on Morphology and Viability of Fragile Phytoplankton. Applied and Environmental Microbiology . 53 (11): 2677-2679. Herzenberg, L.A.; Parks, D.; Sahaf, B.; Perez, O.; Roederer, M., and Herzenberg, L.A. (2002). The History and Future of the Fluorescence Activated Cell Sorter and Flow Cytometry a View from Stanford. Clinical Chemistry . 48 (10): 1819-1827 Hempel, D. C. (1988). Fundamentals of scale-up for biotechnological processes in stirred fermentors. In: Finn, R. K.; Präve, P.; Schlingmann, M.; Crueger, W.; Esser, K.; Thaver, R. and Wagner, F. (eds.). Biotechnology Focus 1: Fundamentals, applications, information . Hanser Publishers. Munich. Pp. 51-94. Hensler, W.T. and Agathos, S.N. (1994). Evaluation of monitoring approaches and effects of culture conditions on recombinant protein production in baculovirus-infected insect cells. Cytotechnology . 15 :177–186. Hesse, F.; Ebel, M.; Konisch, N.; Sterlinski, R.; Kessler, W. and Wagner, R. (2003) Comparison of a Production Process in a Membrane-Aerated Stirred Tank and up to 1000-L Airlift Bioreactors Using BHK-21 Cells and Chemically Defined Protein-Free Medium. Biotechnology Progress . 19 : 833-843. Hiller, G. W.; Aeschlimann, A. D.; Clark, D. S. and Blanch, H. W.(1991). A kinetic analysis of hybridoma growth and metabolism in continuous suspension culture on serum-free medium. Biotechnology and Bioengineering . 38 :733-741. Hirtenstein, M. D. and Clark, J. M. (1980). Critical parameters in the microcarrier culture of animal cells. In: Richards, R. and Rajan, K. (eds). Tissue Culture in medical research . Vol. 2. Pergamon Press, Oxford. Ho, C-T; Lin, R-Z.; Chang, H-Y., and Liu, C-H. (2005). Micromachined electrochemical T-switches for cell sorting applications. Lab Chip . 5: 1248-1258. 280 Hoh, J. H. and Schoenenberger, C. A. (1994). Surface morphology and mechanical properties of MDCK monolayers by atomic force microscopy. Journal of Cell Science . 107 (5): 1105-1114. Hu, W.S.; Meier, J. and Wang, D.I.C. (1986). Use of Surface Aerator to Improve Oxygen Transfer in Cell Culture. Biotechnology and Bioengineering . 28 : 122-125. Hu, W.; Gladue, R.; Hansen, J.; Wojnar, C. and Chalmers, J. J. (2007). The Sensitivity of the Dinoflagellate Crypthecodinium cohnii to Transient Hydrodynamic Forces and Cell- Bubble Interactions. Biotechnology Progress . 23 :1355-1362. Huh, D., Gu, W., Kamotani, Y, Grotberg, J. B. and Takayama, S. (2005). Microfluidics for flow cytometric analysis of cells and particles. Physiological measurement . 26 : R73- R98. Huhtala, A.; Linko, P. and Mutharasan, R. (2005). Protein response of insect cells to bioreactor environmental stresses. Journal of Biotechnology . 118 : 278-289. Hudcova, V.; Machon, V. and Nienow, A. W. (1989). Gas liquid dispersion with dual Rushton turbine impellers. Biotechnology and Bioengineering . 34 :617-628. Jan, D. C. H.; Petch, D. A.; Huzel, N. and Butler, M. (1997). The effect of dissolved oxygen on the metabolic profile of a murine hybridoma grown in serum-free medium in continuous culture. Biotechnology and Bioengineering . 54 : 153–164. Jain, E., and Kumar, A. 2008. Upstream processes in antibody production: Evaluation of critical parameters. 26: 46-72 Janig, E.; Stumptner, C.; Fuchsbichler, A.; Denk, H. and Zatloukal, K. (2005). Interaction of stress proteins with misfolded keratins. European Journal of Cell Biology . 84 ; 329 – 339. Jöbses, I.; Martens, D. and Tramper, J. J. (1991). Lethal events during gas sparging in animal cell culture. Biotechnology and Bioengineering . 37 : 484-490. 281 Jones, W. R.; Ting-Beall, H. P.; Lee, G. M.; Kelley, S. S.; Hochmuth, R. M. and Guilak, F. (1999). Alterations in the Young’s modulus and volumetric properties of chondrocytes isolated from normal and osteoarthritic human cartilage. Journal of Biomechanics . 32: 119-127. Jordan, M.; Sucker, H.; Einsele, A.; Widmer, F. and Eppenberger, H. M. (1994). Interactions between animal cells and gas bubbles: the influence of serum and Pluronic F- 68 on the physical properties of the bubble surface. Biotechnology and Bioengineering . 43(6):446-454. Joshi, J. B. ; Elias, C. B. and Patole, M. S. (1996). Role of hydrodynamic shear in the cultivation of animal, plant and microbial cells. Chemical Engineering Journal . 62 :121- 141. Jochem, F. (2005). Short-Term Physiologic Effects of Mechanical Flow Sorting and the Becton-Dickinson Cell Concentrator in Cultures of the Marine Phytoflagellata Emiliania buxleyi and Micromonas pusilla. Cytometry Part A . 65A :77-83 Ju, L.-K. and Armiger, W. B. (1992). Use of perfluorocarbon emulsion in cell-culture. BioTechniques . 12 :258-263. Jüsten, P; Paul, G. C., Nienow, A. W. and Thomas, C. R. (1998) Dependence of Penicillium chrysogenum growth, morphology, vacuolation and productivity in fed-batch fermentations on impeller type and agitation intensity. Biotechnology and Bioengineering . 59 (6): 762-775. Kamen, A.A.; Chavarie, C.; Andre, G. and Archambault, J. (1992). Design parameters and performance of a surface baffled helical ribbon impeller bioreactor for the culture of shear sensitive cells. Chemical Engineering Science . 27 : 2375-2380. Kawase, Y., Halard, B. and Moo-Young, M. 1991. Liquid-Phase mass transfer coefficients in bioreactors. Biotechnology and Bioengineering. 39: 1133-1140. Keane, J. T.; Ryan, D. and Gray, P. P. (2003). Effect of shear stress on expression of a recombinant protein by Chinese hamster ovary cells. Biotechnology and Bioengineering . 81 (2):211-220. 282 Kioukia, N.; Nienow, A. W.; Al-Rubeai, M. and Emery, A. N. (1992). The impact of fluid dynamics on the biological performance of free suspension animal cell culture: further studies. Food and Bioproducts Procesing (Transactions of the Institution of Chemical Engineers. Part C.). 70 :143-148. Kioukia, N.; Nienow, A. W.; Al-Rubeai, M. and Emery, A. N. (1996). The influence of agitation and sparging on the growth rate and infection of insect cells in bioreactors and a comparison with hybridoma culture. Biotechnology Progress . 12 : 779-785. Kompala, D. S. and Ozturk, S. S. (2006). Optimization of high cell density perfusion reactors. In: S.S. Ozturk and W-S. Hu (Eds). Cell culture technology for pharmaceutical and cell-based therapies . Pp. 387-416. CRC Press. Boca Raton, FL., USA. Kossen, N. W. F. (1994). Scale-up. In: Galindo, E. and Ramirez, O. T. (eds.). Advances in Bioprocess Engineering . Kluwer Academic Publishers. Dordrecht. Pp. 53-65. Koynov, A., Tryggvason, G., Khinast, J.G. 2007. Characterization of the localized hydrodynamic shear forces and dissolved oxygen distribution in sparged bioreactors. 97 (2): 317 – 331 Kresta, S. (1998). Turbulence in stirred tanks: anisotropic, approximate and applied. Canadian Journal of Chemical Engineering . 76 :563-575. Kresta, S. M., Mao, D. and Roussinova, V. (2006). Batch blend time in square stirred tanks. Chemical Engineering Science . 61 (9):2823-2825. Kretzmer, G. (2002). Industrial processes with animal cells. Applied Microbiology and Biotechnology . 59 :135–142. Kromenaker, S. J. and Srienc, F. (1994). Cell cycle kinetics of accumulation of heavy and light chain immunoglobulin proteins in a mouse hybridoma cell line. Cytotechnology . 14 : 205-218. Kulkarni, A. A. and Joshi, J. B. (2005). Bubble Formation and Bubble Rise Velocity in Gas-Liquid Systems: A Review. Industrial & Engineering Chemistry Research . 44 : 5873-5931. 283 Kunas, K. T. and Papoutsakis, E. T. (1990a). Damage mechanisms of suspended animal cells in agitated bioreactors with and without bubble entrainment. Biotechnology and Bioengineering . 36 :476-483. Kunas, K. T. and Papoutsakis, E. T. (1990b). The protective effect of serum against hydrodynamic damage of hybridoma cells in agitated and surface aerated bioreactors. Journal of Biotechnology . 15 :57-70. Kunkel, J. P.; Jan, D. C. H.; Jamieson, J. C. and Butler, M. (1998). Dissolved oxygen concentration in serum-free continuous culture affects N-linked glycosylation of a monoclonal antibody. Journal of Biotechnology . 62 :55–71. Lakhotia, S. K. ; Bauer, K. D. and Papoutsakis, E. T. (1992). Damaging agitation intensities increase DNA synthesis rate and alter cell-cycle phase distributions of CHO cells. Biotechnology and Bioengineering . 40 : 978-990. Lanneu, D.; Brunet, M.; Frisna, E.; Solary, E.; Fontenay, M. and Garrido, C. (2007). Journal of Cellular and Molecular Medicine Postprint 10.1111/j.1582-4934.2008 . Langheinrich, C. and Nienow, A. W. (1999). Control of pH in large-scale, free suspension animal cell bioreactors: alkali addition and pH excursions. Biotechnology and Bioengineering. 66: 171–179. Langheinrich, C.; Nienow, A. W.; Eddleston, T.; Stevenson, N. C.; Emery, A. N.; Clayton, T. M.; Slater, N. K. H. (2002). Oxygen transfer in stirred bioreactors under animal cell culture conditions. Food and bioproducts processing . 80 (1): 39-44. Lavery, M. and Nienow, A.W. (1987). Oxygen Transfer in Animal Cell Culture Medium. Biotechnology and Bioengineering . 30 (3): 368 – 373. Lehmann, J.; Vorlop, J. and Buntemayer, H. (1988). Bubble-free reactors and their development for continuous culture with cell recycle. In: Spier R.E. and Griffiths J.B. (eds), Animal Cell Biotechnology. Vol. 3. Academic Press Ltd., London, UK, pp. 221– 237. 284 Lehoux, S.; Castier, Y. and Tedgui, A. (2006). Molecular mechanisms of the vascular responses to haemodynamic forces. Journal of Internal Medicine . 259 : 381-392. Lee, G-B; Hung, C-I.; Ke, B-J; Huang, G-R.; Hwei, B-H, and Lai, H-F. (2001). Hydrodynamic Focusing for a Micromachined Flow Cytometer. Journal of Fluids Engineering . 123 :672-679 Lopez, E.; Figueroa, S.; Oset-Gasque, M-J. and Gonzalez, M.P. (2003). Apoptosis and necrosis: two distinct events induced by cadmium in cortical neurons in culture. British Journal of Pharmacology . 138 :901-911 Li, C. and Xu, Q. (2007). Mechanical stress-initiated signal transduction in vascular smooth muscle cells in vitro and in vivo. Cellular Signaling . 19 :881-891. Li, Y.-S. J.; Haga, J. H. and Chien, S. (2005). Molecular basis of the effects of shear stress on vascular endothelial cells. Journal of biomechanics . 38 : 1949-1971. Lin, A. A. and Miller, W. M. (1992). Modulation of glutathione level in CHO cells. Effects of oxygen concentration and previous exposure to hypoxia. Annals of the New York Academy of Sciences . 665 :117-126. Linek, V.; Moucha, T. and Sinkule, J. (1996). Gas–liquid mass transfer in vessel stirred with multiple-impellers I: Gas–liquid mass transfer characteristics in individual stages. Chemical Engineering Science . 51 (12): 3203-3212. Lowe, K. C.; Davey, M. R. and Power, J. B. (1998). Perfluorochemicals: their applications and benefits to cell culture. Trends in Biotechnology . 16 :272-277. Lütz, S.; Rao, N. N. and Wandrey, C. (2006). Membranes in biotechnology. Chemical Engineering and Technology . 29 (12):1404–1415. Ma, N.; Chalmers, J. J.; Aunins, J. G.; Zhou, W. and Xie, L. (2004). Quantitative Studies of Cell Damage in Sparged Bioreactors Using A Bubble Collector to Segregate Bubble Rupture Events. Biotechnology Progress. 20 (4):1183 – 1191. 285 Ma, N.; Koelling, K. and Chalmers, J. J. (2002). The fabrication and use of a transient contraction flow device to quantify the sensitivity of mammalian and insect cells to hydrodynamic forces. Biotechnology and Bioengineering . 80 :428-437. Ma, N.; Mollet, M. and Chalmers, J. J. (2006). Aeration, Mixing and Hydrodynamics in Bioreactors. In: S.S. Ozturk and W-S. Hu (Eds). Cell culture technology for pharmaceutical and cell-based therapies . Pp. 225-248. CRC Press. Boca Raton, FL., USA. Machon, V.; Vlcek, J.; Skrivanek, J. (1985). Dual Impeller Systems for Aeration of Liquids: An Experimental Study. In: Proc. 5 th European Conference on Mixing . Würzburg, Germany. Pp. 155-169. MacIntyre, F. (1972). Flow patterns in breaking bubbles. Journal of Fluid Mechanics. 77 :5211-5228. Mahaffy, R. E.; Shih, C. K.; MacKintosh, F. C and Kas, J. (2000). Scanning probe-based frequency-dependent microrheology of polymer gels and biological cells. Physical Review Letters . 85 :880–883. Malek, A. M. and Izumo, S. (1994). Molecular aspects of signal transduction of shear stress in the endothelial cell. Journal of Hypertension . 12 (9):989-999. Matsunaga, N., Kano, K., Maki, Y., and Dobashi, T. 2009. Culture scale-up studies as seen from the viewpoint of oxygen supply and dissolved carbon dioxide stripping. Journal of Bioscience and Bioengineering. 107 (4): 412 – 418. Martens, D.E.; de Gooijer, C.D.; Beuvery, E.C.; and Tramper, J. (1992). Effect of serum concentration on hybridoma viable cell density and production of monoclonal antibodies in CSTR and on shear sensitivity in air-lift loop reactors. Biotechnology and Bioengineering. 39 :891-897 Martinez, A., Galindo, E. and Salvador, M. 1989. Sparger position effect over kLa in bench and pilot stirred tank fermentors. Journal of fermentation and bioengineering. 68 (1): 71-73. 286 Mardikar, S. H. and Niranjan, K. (2000). Observations of the shear damage to different animal cells in a concentric cylinder viscometer. Biotechnology and Bioengineering . 68 (6): 697-704. Marks, D. M. (2003). Equipment design considerations for large scale cell culture. Cytotechnology . 42 : 21–33. Martens, D.E.; Gooijer, C.D.; Beuvery, E.C. and Tramper, J. (1992). Effect of Serum Concentration on Hybridoma Viable Cell Density and Production of Monoclonal Antibodies in CSTRs and on Shear Sensitivity in Air-Lift Loop Reactors. Biotechnology and Bioengineering . 39 : 891-897 Maxwell, W.M.C.; Welch, G. R. and Johnson, L. A. (1997). Viability and membrane integrity of spermatozoa after dilution and flow cytometric sorting in the presence or absence of seminal plasma. Reproduction, Fertility and Development . 8:1165-1178 McQueen, A. and Bailey, J. E. (1989). Influence of serum level, cell line, flow type, and viscosity on flow-induced lysis of suspended mammalian cells. Biotechnology Letters . 11 :531-536. McQueen A.; Melhoc, E. and Bailey, J. E. (1987). Flow effects on the viability and lysis of suspended mammalian cells. Biotechnology Letters. 9:831-836. McIntire, L. V.; Frangos, J. A.; Rhee, B. G.; Eskin, S. G. and Hall, E. R. (1987). The effect of fluid mechanical stress on cellular arachidonic acid metabolism. Annals of the New York Academy of Sciences . 516 :513-524. McQueen, A. and Bailey, J. E. (1989). Influence of serum level, cell line, flow type and viscosity on flow-induced lysis of suspended mammalian cells. Biotechnology Letters . 11 :531-536. McQueen, A.; Meilhoc, E. and Bailey, J. E. (1987). Flow effects on the viability and lysis of suspended mammalian cells. Biotechnology Letters . 9(12):831-836. Meier, S. J. (2005). Cell culture scale-up: mixing, mass transfer, and use of appropriate scale-down models. Biochemical engineering XIV. Harrison Hot Springs, Canada. 287 Menisher, T.; Metghalchi, M. and Gutoff, E. B. (2000). Mixing studies in bioreactors. Bioprocess engineering . 22 (2):115-120. Metz, B.; Kossen, N. W. P. and van Suijdam, J. C. (1979). The rheology of mold suspensions. Advances in Biochemical Engineering . 11 :103-156. Michaels, J. D.; Nowak, J. E. ; Mallik, A. K.; Koczo, K.; Wasan, D. T. and Papoutsakis, E. T. (1995). Interfacial properties of cell culture media with cell-protecting additives. Biotechnology and Bioengineering . 47 (4): 420-430. Midler, J. Jr. and Finn, R. K. (1966). A model system for evaluating shear in the design of stirred fermentors. Biotechnology and Bioengineering . 8:71-84. Miller, W. M.; Wilke, C. R. and Blanch, H. W. (1987). Effect of dissolved oxygen concentration on hybridoma growth and metabolism in continuous culture. Journal of Cell Physiology . 132 :524-530. Miller, W. M.; Wilke, C. R. and Blanch, H. W. (1988). Transient responses of hybridoma metabolism to changes in the oxygen supply rate in continuous culture. Bioprocess Engineering . 3:103-113. Mollet, M.; Ma, N.; Zhao, Y.; Brodkey, R.; Taticek, R. and Chalmers, J. J. (2004). Bioprocess Equipment: Characterization of Energy Dissipation Rate and its potential to Damage Cells. Biotechnology Progress . 20 :1437-1448. Mollet, M.; Godoy-Silva, R.; Berdugo, C. and Chalmers, J. J. (2007). Acute Hydrodynamic Forces and Apoptosis: A Complex Question. Biotechnology and Bioengineering . 98 (4): 772-788 . Mollet, M., Godoy-Silva, R., Berdugo, C, Chalmers, J.J. (2008). Computer simulations of the energy dissipation rate in a fluorescence activated cell sorter: implications to cells. Biotechnology and Bioengineering . 100(2): 260-272. Molowa, D. T. and Mazanet, R. (2003). The state of biopharmaceutical manufacturing. Biotechnology Annual Review . 9:285–302. 288 Morão, A.; Maia, C. I.; Fonseca, M. M. R.; Vasconcelos, J. M. T. and Alves, S. S. (1999). Effect of antifoam addition on gas-liquid mass transfer in stirred fermenters. Bioprocess and Biosystems Engineering , 20 (2): 165 – 172. Moreira, J. L.; Cruz, P. E.; Santana, P. C. and Feliciano, A. S. (1995). Influence of power input and aeration method on mass transfer in a laboratory animal cell culture vessel. Journal of Chemical Technology and Biotechnology . 62 :118-131. Mostafa, S. S. and Gu, X. (2003). Strategies for improved dCO 2 removal in large-scale fed-batch cultures. Biotechnology Progress . 19 (1):45-51. Motobu, M.; Wang, P. C. and Matsumura, M. (1998). Effect of shear stress on recombinant Chinese hamster ovary cells. Journal of Fermentation and Bioengineering . 85 (2): 190-195. Mufti, N. A. and Shuler, M. L. (1995). Induction of cytochrome P-450IA1 activity in response to sublethal stresses in microcarrier-attached Hep G2 cells. Biotechnology Progress . 11 (6): 659-663. Murhammer D.W. and Goochee C. F. (1990). Sparged animal cell bioreactors: mechanism of cell damage and Pluronic F-68 protection. Biotechnology Progress . 6: 391–397. Murhammer, D. W. and Goochee, C. F. (1990). Sparged animal cell bioreactors: mechanism of cell damage and Pluronic F-68 protection. Biotechnology Progress . 6: 391–397. Murray, N.R., Baumgardner, G.P., Burns, D. J. and Fields, A.P. (1993). Protein Kinase C Isotypes in Human Erythroleukemia (K562) Cell Proliferation and Differentiation. The Journal of Biological Chemistry . 268 (21): 15847-15853. Nehring, D.; Czermak, P.; Vorlop, J. and Lübben, H. (2004). Experimental Study of a Ceramic Microsparging Aeration System in a Pilot-Scale Animal Cell Culture. Biotechnology Progress . 20 :1710-1717. 289 Nienow, A. W. (1997). On impeller circulation and mixing effectiveness in the turbulent flow regime. Chemical Engineering Science . 52 (15):2557-2565. Nienow, A. W. (2006). Reactor engineering in large scale animal cell culture. Cytotechnology . 50 :9-33. Nienow, A. W.; Langheinrich, C.; Stevenson, N. C.; Emery, A. N.; Clayton, T. M. and Slater, N. K. H. (1996). Homogenization and oxygen transfer rates in large agitated and sparged animal cell bioreactors: Some implications for growth and production. Cytotechnology. 22 : 87-94. Oehler, R.; Pusch, E.; Zellner, M.; Dungel, P.; Hergovics, N.; Homoncik, M., Eliasen, M., Brabec, M. and Roth, E. (2001). Cell type-specific variations in the induction of hsp70 in human leukocytes by feverlike whole body hyperthermia. Cell Stress & Chaperones . 6 (4), 306 -315. Oh, S. K. W.; Nienow, A. W.; Al-Rubeai, M. and Emery, A. N. (1989). The effect of agitation intensity with and without continuous sparging on the growth and antibody production of hybridoma cells. Journal of Biotechnology . 12 :45-62. Oh, S. K. W.; Nienow, A. W.; Al-Rubeai, M. and Emery, A. N. (1992). Further studies of the culture of mouse hybridoma in an agitated bioreactor with and without continuous sparging. Journal of Biotechnology . 22 :245-270. Oldshue, J. Y. (1983). Fluid Mixing Technology . McGraw-Hill Publications Co. New York. USA. Orton, D. and Wang, D.I.C. (1991). Fluorescent visualization of cell death in bubble aerated bioreactors. Cell Culture Engineering III, United Engineering Foundation. February 2-7, 1991. Ozturk, S. S. (1996) Engineering challenges in high cell density cell culture systems. Cytotechnology. 22: 3–16. 290 Ozturk, S. S. (2006). Cell culture technology – An Overview. In: S.S. Ozturk and W.-S. Hu (eds). Cell culture technology for pharmaceutical and cell-based therapies . Pp. 1-13. CRC Press. Boca Raton, FL., USA. Ozturk, S. S. and Palsson, B. O. (1990). Effects of dissolved oxygen on hybridoma cell growth, metabolism and antibody production kinetics in continuous culture. Biotechnology Progress . 6:437-446. Packer, L. and Fuehr, K. (1977). Low oxygen concentration extends the lifespan of cultured human diploid cells. Nature . 267 :423-425. Palomares, L. A. ; Lopez, S. ; Ramirez, O. T. (2004). Utilization of oxygen uptake rate to assess the role of glucose and glutamine in the metabolism of infected insect cell cultures. Biochemical Engineering Journal . 19 (1): 87-93. Papadaki, M. and Eskin, S.G. (1997). Effects of fluid shear stress on gene regulation of vascular cells. Biotechnology Progress . 13 (3):209-221. Parker, B. and Barsom, G. (1970). Biological and chemical significance of surface microlayers in aquatic ecosystems. BioScience . 20 :87-93. Passini, C. A. and Goochee, C. F. (1989). Response of a mouse hybridoma cell line to heat shock, agitation and sparging. Biotechnology Progress . 5: 175-188. Pelicano, H.; Carew, JS.; McQueen, TJ.; Andreeff, M.; Plukett, W.; Keating, MJ and Huang, P. (2006). Targeting Hsp90 by 17-AAG in leukemia cells: mechanism for synergistic and antagonistic drug combinations with arsenic trioxide and Ara-C. Leukemia . 20 : 610-619. Pinkel, D. and Stovel, R. 1985. Flow Chambers and Sample Handling. In: Van Dilla, M. A.; Dean, P. N.; Laerum, O. D. and Melamed, M. R. (Eds). Flow Cytometry: Instrumentation and Data Analysis. Pp.3-128. London: Academic Press,. Puthli, M.S., Rathod, V.K., and Pandit, A.B. 2005. Gas-liquid mass transfer studies with triple impeller system on a laboratory scale bioreactor. Biochemical Engineering Journal. 23: 25-30 291 Ramirez, O.T. and Mutharasan, R. 1992. Effect of Serum on the Plasma Membrane Fluidity of Hybridomas: An Insight into Its Shear Protective Mechanism. Biotechnology Progress . 8: 40-50. Ramirez, O. T. and Mutharasan, R. (1990a). Cell cycle- and growth phase-dependent variations in size distribution, antibody productivity and oxygen demand in hybridoma cultures. Biotechnology and Bioengineering . 36 (8): 839-848. Ramirez, O. T. and Mutharasan, R. (1990b). The role of the plasma membrane fluidity on the shear sensitivity of hybridomas grown under hydrodynamic stress. Biotechnology and Bioengineering . 36 (9):911-920. Research and Market (2007). Monoclonal Antibodies 2007: Competitive Landscape & Pipeline Insight. http://www.researchandmarkets.com/reportinfo.asp?report_id=367599 Research and markets, 2008. Biologics Pipelines: Who has the next promising recombinant proteins? http://www.researchandmarkets.com/reports/603343 Rhiel, M.; Mitchell-Logean, C. M. and Murhammer, D. W. (1997). Comparison of Trichoplusia ni BTI-Tn-5B1-4 (High Five TM ) and Spodoptera frugiperda Sf-9 insect cell line metabolism in suspension cultures. Biotechnology and Bioengineering. 55 :909–920. Rivkin, R.B.; Phinney, D.A., and Yentsch, C.M. (1986). Effects of flow Cytrometric Analysis and Cell Sorting on Photosynthetic Carbon Uptake by Phytoplankton in Cultures and from Natural Populations. Applied and Environmental Microbiology . 52 (4):935-938 Rocha-Valadez, J. A.; Hassan, M.; Corkidi, G.; Flores, C.; Galindo, E. and Serrano- Carreon, L. (2005) 6-pentyl-α-pyrone production by Trichoderma harzianum : the influence of energy dissipation rate and its implications on fungal physiology. Biotechnology and Bioengineering . 91 (1):54-61. Sauer, P. W.; Burky, J. E.; Wesson, M. C.; Sternard, H. D. and Qu, L. (2000). A High- Yielding, Generic Fed-Batch Cell Culture Process for Production of Recombinant Antibodies. Biotechnology and Bioengineering . 67 (5): 585-597. 292 Sardeing, R., Aubin, J., Poux, M and Xuereb, C. 2004. Gas-liquid mass transfer. Influence of Sparger location. Trans IChemE, Part A. September 2004. Chemical Engineering Research and Design. 82(A9):1161-1168 Schürch, U.; Kramer, H.; Einsele, A.; Widmer, F. and Eppenberger, H. M. (1988). Experimental evaluation of laminar shear stress on the behaviour of hybridoma mass cell cultures, producing monoclonal antibodies against mitochondrial creatine kinase. Journal of Biotechnology . 7(3): 179-184. Sen, A.; Kallos, M. S. and Behie, L. A. (2001). Effects of hydrodynamics on cultures of mammalian neural stem cell aggregates in suspension bioreactors : Novel reactor engineering for the new millennium. Industrial and Engineering Chemistry Research . 40 (23): 5350-5357. Seidl, J.; Knuechel, R., and Kunz-Schughart, L. A. (1999). evaluation of membrane physiology following fluorescence activated or magnetic cell separation. Cytometry. 36 :102–111. Shapiro, H. W. (2003). Practical Flow cytometry. 4 th Ed. New Jersey: John Wiley & Sons. Shiloach, J. and Fass, R. (2005). Growing E. coli to high cell density—A historical perspective on method development. Biotechnology Advances . 23 :345–357. Singh, V. (1999). Disposable bioreactor for cell culture using wave-induced agitation. Cytotechnology . 30 : 149–158. Sinskey, A. J.; Fleischaker, R. J.; Tyo, M. A.; Giard, D. J. and Wang D. I. C. (1981). Production of cell-derived products: virus and interferon. Annals of the New York Academy of Sciences. 369 :47-59. Smith, C. M.; Marks, A. D. and Lieberman M. A. (2005). Marks' Basic Medical Biochemistry: A Clinical Approach . 2 nd Ed. Lippincott Williams & Wilkins. Philadelphia. USA. 293 Stolzing, A.; Sethe, S., and Scutt, A. (2006). Stressed Stem Cells: Temperature Response in Aged Mesenchymal Stem Cells. Stem Cells and Develompent . 15 : 478 – 487. Subjeck, J.R.; Shyy, T.-T., (1986). Stress Protein systems of mammalian cells. Am. J. Physiol . 250, C1-C17. Sundararajan, A. and Ju, L.-K. (2006). Use of cyanobacterial gas vesicles as oxygen carriers in cell culture. Cytotechnology . 52 :139–149. Tatterson, G. B. (1991). Fluid mixing and gas dispersion in agitated tanks . New York. Mc Graw Hill. Taylor, G. I. (1934). The formation of emulsions in definable fields of flow. Proceedings of the Royal Society of London. Series A . 146 :501-523. Telling, R.C. and Elsworth, R. (1965). Submerged culture of hamster kidney cells in a stainless steel vessel. Biotechnology and Bioengineering . 7(3):417-434. Thomas, X.; Campos, L.; Mounier, C.; Cornillon, J.; Flandrin, P.; Le, Q., Piselli, S., and Guyota, D. (2005). Expression of heat-shock proteins is associated with major adverse pronostic factors in acute myeloid leukemia. Leukemia Research . 29 : 1049-1058 . Thomas, C. R.; Al-Rubeai, M. and Zhang, Z. (1994). Prediction of mechanical damage to animal cells in turbulence. Cytotechnology . 15 : 329-335. Thomson Centerwatch (2007). Drugs approved by the FDA. http://www.centerwatch.com/patient/drugs/druglist.html Tomos, D. (2000). The plant cell pressure probe. Biotechnology Letters . 22 :437-442. Tramper, J. J.; Joustra, D. and Vlak, J. M. (1987). Bioreactor design for growth of shear sensitive insect cells. In: Webb, C. and Mavituna, F. (Eds). Plant and animal cell cultures: process possibilities . Ellis Horwood. Chichester, U. K. Pp 125-136. 294 Trinh, K.; García-Briones, M. A.; Hink, F. H. and Chalmers, J. J. (1994). Quantification of damage to suspend insect cells as a result of bubble rupture. Biotechnology and Bioengineering . 43 :37-45. Tzima, E. (2006). Role of Small GTPases in Endothelial Cytoskeletal Dynamics and the Shear Stress Response. Circulation Research . 98 :176-185. Uhl, V. W. and Gray, J. B. (1967). Mixing: theory and practice . Vol. II. Academic Press. San Diego, USA. p. 254. Ujcová, E.; Fencl, Z.; Musílková, M. and Seichert, L. (1980). Dependence of release of nucleotides from fungi on fermentor turbine speed. Biotechnology and Bioengineering . 22 : 237-241. Van’t Riet, K. (1979). Review of measuring methods and results in non-viscous gas liquid mass transfer in stirred vessels. Industrial and Engineering Chemistry Process Design and Development . 18 (3):357-363. Varley, J. and Birch, J. (1999). Reactor design for large scale suspension animal cell culture. Cytotechnology . 29 :177-205. Venkat, R.; Stock, R. and Chalmers, J. J. (1996). Study of hydrodynamics in microcarrier culture spinner vessels: A particle tracking velocimetry approach. Biotechnology and Bioengineering . 49 :456-466. Vorlop, J. and Lehmann, J. (1989). Oxygen transfer and carrier mixing in large scale membrane stirred cell culture reactors. In: Spier R.E., Griffiths J.B., Stephenne J. and Crooy P.J. (Eds). Advances in Animal Cell Biology and Technology for Bioprocesses . Butterworths, Svenoaks, UK, pp. 366–369. Wang, N; Butler, J. P. and Ingber, D. E. (1993). Mechanotransduction across the cell surface and through the cytoskeleton. Science . 260 (5111): 1124-1127. Wernersson, E. S. and Trägårdh, C. (1999). Scale-up of Rushton turbine-agitated tanks.. Chemical Engineering Science . 54 (19): 4245-4256. 295 Welch, W., (1992). Mammalian Stress Response: Cell Physiology, Structure/Function of Stress Proteins, and Implications for Medicine and Disease. Physiological Reviews . 72 (4): 1063 – 1081. Williams, L. and Peter, Y. (2007). Adult stem cells of the lung in organ dysfunction. Journal of Organ Dysfunction . 3:31-35. Wong, T. K. K.; Nielsen, L. K.; Greenfield, P. F. and Reid, S. (1994). Relationship between oxygen uptake rate and time of infection of Sf9 insect cells infected with a recombinant baculovirus. Cytotechnology . 15 :157–167. Wu, J. (1996). Insights into protective effects of medium additives on animal cells under fluid stresses: the hydrophobic interactions. Cytotechnology . 22 (1-3):103-109. Wu, J. ; Ruan, Q. and Lam, H. Y. P. (1997). Effect of surface-active medium additives on insect cell surface hydrophobicity relating to cell protection against bubble damage. Enzyme and Microbial Technology . 21 (5):341-348. Wurm, F. M. (2004). Production of recombinant protein therapeutics in cultivated mammalian cells. Nature Biotechnology . 22 (11): 1393-1398. Yang, J-D., Lu, C., Stasny, B., Henley, J., Guinto, W., Gonzalez, C. Gleason, J., Fung, M., Collopy, B., Benjamino, M., Gangi, J., Hanson, M., Ille, E. 2007. Fed-Batch bioreactor process scale-up from 3L to 2500L scale for monoclonal antibody production from cell culture. Biotechnology and Bioengineering. 98 (1): 141-154. Yang, A-S.; Hsieh, W-H., and Tseng, L-Y. (2008). Three-dimensional unsteady calculations for a flow cytometer. Proc. IMechE. J. Mechanical Engineering Science . 222 . Part C. Yoon, S. and Konstantinov, K. B. (1994). Continuous, real-time monitoring of the oxygen-uptake rate (OUR) in animal-cell bioreactors. Biotechnology and Bioengineering. 44 :983-990. Young, J.; Barral, J.M. and Hartl. F.U. (2003). More than folding: localized functions of cytosolic chaperones. TRENDS in Biochemical Sciences . 28 (10): 541-547. 296 Zeng, A.-P. and Bi, J.-X. (2006). Cell culture kinetics and modeling. In: S.S. Ozturk and W-S. Hu (eds). Cell culture technology for pharmaceutical and cell-based therapies . Pp. 299-348. CRC Press. Boca Raton, FL., USA. Zhang, Z. and Thomas, C. R. (1993). Modeling of animal cell damage in turbulent flows. In: Nienow A. W. (Ed). Proceedings of the 3 rd . International conference on bioreactor and bioprocess fluid dynamics . Mechanical Engineering Publications Ltd. London. Pp. 475-482. Zhang, Z.; Al-Rubeai, M. and Thomas, C. R. (1993). Estimation of disruption of animal cells by turbulent capillary flow. Biotechnology and Bioengineering . 42 : 987–993. Zhang, S.; Handa-Corrigan, A. and Spier, R. E. (1992). Oxygen transfer properties of bubbles in animal cell culture media. Biotechnology and Bioengineering . 40 (2):252-259. Zhang, Z.; Al-Rubeai, M. and Thomas, C. R. (1993). Estimation of disruption of animal cells by turbulent capillary flow. Biotechnology and Bioengineering . 42 : 987–993. Zhang, Z.; Ferenczi, M. A. and Thomas, C. R. (1992). A micromanipulation technique with a theoretical cell model for determining mechanical properties of single mammalian cells. Chemical Engineering Science . 47 (6): 1347-1354. Zhou, G. and Kresta, S. M. (1996a). Distribution of Energy Dissipation Between Convective and Turbulent Flow For Three Frequently Used Impellers. Chemical Engineering Research and Design . (Transactions of the Institution of Chemical Engineers, part A). 74 :379-389. Zhou, G. and Kresta, S. M. (1996b). Impact of Tank Geometry on the Maximum Turbulence Energy Dissipation Rate for Impellers. AIChE Journal . 42 :2476-2490. Zanghi, J.A.; Fussenegger, M., and Bailey, J.E. (1999). Serum Protects Protein-Free Competent Chinese Hamster Ovary Cells Against Apoptosis Induced by Nutrient Deprivation in Batch Culture. Biotechnology and Bioengineering . 64 (1):108-119. Zhou, W. and Hu, W.-S. (1994). On line characterization of a hybridoma cell culture process. Biotechnology and Bioengineering . 44 : 170-177. 297 Zhu, M. M.; Goyal, A.; Rank, D. L.; Gupta, S. K.; VandenBoom, T. and Lee, S. S. (2005), Effects of Elevated pCO2 and Osmolality on Growth of CHO Cells and Production of Antibody-Fusion Protein B1: A Case Study. Biotechnology Progress . 21 (1): 70-77. 298