Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Increasing the performance of mammalian perfusion culture system Drouin, Hans 2010

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

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
24-ubc_2010_fall_drouin_hans.pdf [ 10.15MB ]
Metadata
JSON: 24-1.0058878.json
JSON-LD: 24-1.0058878-ld.json
RDF/XML (Pretty): 24-1.0058878-rdf.xml
RDF/JSON: 24-1.0058878-rdf.json
Turtle: 24-1.0058878-turtle.txt
N-Triples: 24-1.0058878-rdf-ntriples.txt
Original Record: 24-1.0058878-source.json
Full Text
24-1.0058878-fulltext.txt
Citation
24-1.0058878.ris

Full Text

INCREASING THE PERFORMANCE OF MAMMALIAN PERFUSION CULTURE SYSTEM  by HANS DROUIN B.A.Sc., University of Sherbrooke, 2001 M.A.Sc., University of British Columbia, 2004   A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY  in THE FACULTY OF GRADUATE STUDIES (Chemical and Biological Engineering)     The University of British Columbia (Vancouver)   September 2010   Hans Drouin, 2010  ii Abstract Perfusion culture processes are potentially the most efficient way of producing large quantities of biopharmaceuticals.  However, these processes are not the most commonly used by industry in part due to a lack of simple solutions to perfusion challenges.  This thesis has investigated recombinant protein production instability, a strategy to improve low perfusion rate culture performances and the complications due to cell aggregate formation.  Human embryonic kidney 293 (HEK293) cells producing recombinant human interferon-alpha2b (IFN-α2b) were investigated as a model recombinant cell line.  These cells also were maintained for more than 4 weeks in batch cultures using three media with different osmolalities in order to evaluate production stability.  Exposure to high osmolality (~ 375 mOsm kg-1) gradually decreased the yield of IFN-α2b secreted by the viable cells.  Perfusion cultures validated the batch cultures with results showing that it was not possible to maintain stable production at elevated osmolality whereas at normal osmolality (~ 300 mOsm kg-1), the titer was maintained at 250 mg L-1.  A reduced perfusion rate strategy was explored to increase the product titer in a perfusion bioreactor using enriched media.  The HEK293 cell line was found to have a growth-associated production.  By increasing the bleed rate in order to increase growth, the perfusion process yielded an up to 35% increased IFN-α2b concentration.  Several modified medium conditions were investigated in batch cultures to help identify the main mechanism of aggregation observed in perfusion cultures.  The addition of dead cells in the batch case was found to yield cellular aggregation that was most similar to perfusion cultures.   The presence of aggregation did not affect the on-line monitoring of the viable cell concentration using a permittivity signal.  However, if cells in aggregates are  iii neglected, this can result in major cell specific-rate calculation errors.  The image analysis used to estimate the cellular content of aggregates was an efficient method of improving viable cell estimates and cell specific-rate analyses.  Overall, advances in the methods to more efficiently monitor and operate high performance perfusion-culture processes should expand their potential to fulfill the increasing demand for recombinant protein products from biotechnology.  iv Preface The work contained in this thesis was conducted within a collaboration between the University of British Columbia (UBC) and the Biotechnology Research Institute (BRI) in Montreal.  Hans Drouin performed all experiments, data analysis and writing at the BRI laboratories.  Hans Drouin was co-supervised by Dr. James M Piret (UBC) and Dr. Amine Kamen (BRI).  Dr. Yves Durocher (BRI) supplied the cell line and provided input into the development of analytical method and kinetic studies.  Stéphane Lanthier (BRI) provided assistance and guidance in setting up the perfusion bioreactor and provided valuable advice and discussions.  Sven Ansorge (BRI) provided assistance in setting up the perfusion bioreactor on-line determination of cell density by permittivity measurement and assisted with signal interpretation.  Since the work discussed in Chapters 3, 4 and 5 will be submitted as three separate manuscripts, these Chapters have been written up in an article format. Thus, there has been repetition of text, especially in the Materials and Methods sections.  v Table of Contents ABSTRACT .......................................................................................................... ii
 PREFACE............................................................................................................ iv
 TABLE OF CONTENTS ....................................................................................... v
 LIST OF TABLES................................................................................................ ix
 LIST OF FIGURES .............................................................................................. xi
 ACKNOWLEDGEMENTS................................................................................... xv
 1
 INTRODUCTION............................................................................................. 1
 1.1
 Thesis Objectives...................................................................................................... 3
 1.2
 References................................................................................................................. 5
 2
 BACKGROUND .............................................................................................. 6
 2.1
 Effects of Culture Environment on Mammalian Cells......................................... 6
 2.1.1
 Inhibitory Metabolites......................................................................................... 6
 2.1.2
 Osmolality........................................................................................................... 7
 2.2
 Cellular Aggregation ............................................................................................... 8
 2.3
 On-line Measurement of Cell Density .................................................................. 11
 2.4
 Perfusion Cultures ................................................................................................. 12
 2.4.1
 High Cell Concentration System ...................................................................... 12
 2.4.2
 Process Operation ............................................................................................. 14
 2.5
 Development Approaches for Perfusion Processes............................................. 16
 2.6
 Low Perfusion Rate Strategy ................................................................................ 18
 2.7
 Model Cell Line ...................................................................................................... 19
  vi 2.8
 References............................................................................................................... 20
 3
 INFLUENCE OF CULTURE MEDIUM COMPOSITION AND OSMOLALITY ON INTERFERON-ALPHA 2B PRODUCTION STABILITY IN HEK293 CELLS…............................................................................................................. 28
 3.1
 Introduction............................................................................................................ 28
 3.2
 Materials and Methods.......................................................................................... 31
 3.2.1
 Cell Line and Media.......................................................................................... 31
 3.2.2
 Cell Culture Experiments.................................................................................. 32
 3.2.3
 Perfusion Culture Operation ............................................................................. 33
 3.2.4
 Cell Specific Rate Equations ............................................................................ 36
 3.3
 Analytical Methods ................................................................................................ 37
 3.3.1
 Cell Concentration and Viability ...................................................................... 37
 3.3.2
 Culture Medium Analyses ................................................................................ 38
 3.4
 Results and Discussion........................................................................................... 38
 3.4.1
 Batch Culture Performance of 293-IFN Cells in Three Culture Media............ 38
 3.4.2
 Effect of Osmolality on Cell Production Stability............................................ 43
 3.4.2.1
 Freestyle293 Medium ................................................................................ 44
 3.4.2.2
 LC-SFM Medium ...................................................................................... 45
 3.4.2.3
 MLC-SFM Medium................................................................................... 46
 3.4.2.4
 Osmotic Stress ........................................................................................... 46
 3.4.3
 Effect of Selective Pressure .............................................................................. 47
 3.4.4
 Perfusion Culture Results ................................................................................. 48
 3.5
 Conclusions............................................................................................................. 50
 3.6
 References............................................................................................................... 52
 4
 THE INFLUENCE OF REDUCED PERFUSION RATES ON HEK CELL INTERFERON-ALPHA 2B PRODUCTION IN PERFUSION CULTURES ......... 55
 4.1
 Introduction............................................................................................................ 55
 4.2
 Materials and Methods.......................................................................................... 59
 4.2.1
 Cell Line and Media.......................................................................................... 59
 4.2.2
 Cell Culture Experiments.................................................................................. 60
 4.2.3
 Semi-Continuous Culture Experiments ............................................................ 61
 4.2.4
 Perfusion Culture Operation ............................................................................. 61
 4.2.4.1
 Perfusion Operation ................................................................................... 65
 4.2.4.2
 Model Development and Equations........................................................... 67   vii 4.2.5
 Analytical Methods........................................................................................... 68
 4.2.5.1
 Cell Concentration and Viability ............................................................... 68
 4.2.5.2
 Culture Medium Analyses ......................................................................... 69
 4.3
 Results and Discussion........................................................................................... 70
 4.3.1
 Low Perfusion Rate Culture Experiment – Low Bleed Rate Culture............... 70
 4.3.2
 Kinetic Profile of HEK293-IFN Cells .............................................................. 73
 4.3.3
 Identification of Possible Limitations at Low Perfusion Rate .......................... 75
 4.3.4
 Adapted Perfusion Process Strategy – Second Perfusion Culture .................... 77
 4.3.5
 Comparison of Perfusion Culture Strategies..................................................... 81
 4.4
 Conclusions............................................................................................................. 82
 4.5
 References............................................................................................................... 84
 5
 EVOLUTION OF CELLULAR AGGREGATION DURING HEK293 BATCH AND PERFUSION CULTURES.......................................................................... 90
 5.1
 Introduction............................................................................................................ 90
 5.2
 Materials and Methods.......................................................................................... 93
 5.2.1
 Cell Line and Media.......................................................................................... 93
 5.2.2
 Cell Culture Experiments.................................................................................. 95
 5.2.3
 Permittivity Measurements ............................................................................... 96
 5.2.4
 Perfusion Culture Operation ............................................................................. 97
 5.2.5
 Analytical Methods......................................................................................... 100
 5.2.5.1
 Cell Concentration and Viability ............................................................. 100
 5.2.5.2
 Culture Medium Analyses ....................................................................... 101
 5.2.5.3
 Cell Size Measurement Methods ............................................................. 101
 5.3
 Results and Discussion......................................................................................... 103
 5.3.1
 Aggregate Measurement Method Development and Validation .................... 103
 5.3.2
 Evolution of Cellular Aggregation in Perfusion Cultures............................... 106
 5.3.3
 Mechanisms of Cell Aggregation ................................................................... 109
 5.3.3.1
 Batch Experiments ................................................................................... 109
 5.3.3.2
 Comparison of Cell Aggregation Evolution in Perfusion and Batch Cultures 111
 5.3.4
 Effects of Cellular Aggregation on Permittivity Measurements..................... 113
 5.3.5
 Effects of Cellular Aggregation on Calculations and Production................... 117
 5.4
 Conclusions........................................................................................................... 120
 5.5
 References............................................................................................................. 121    viii 6
 CONCLUSIONS AND FUTURE WORK ..................................................... 124
 6.1
 References............................................................................................................. 129
 APPENDIX 1:  RAW DATA.............................................................................. 130
 APPENDIX 2:  MATLAB CODE....................................................................... 141
 APPENDIX 3:  PICTURE ANALYSIS PROTOCOL ......................................... 149
 APPENDIX 4:  EXAMPLE OF IMAGES........................................................... 151
 APPENDIX 5:  IMAGE J CODE ....................................................................... 154
   ix List of Tables TABLE 3-1:  PARTIAL COMPOSITION OF THE MEDIA USED BASED ON THE ADDITIVES AND AMINO ACID ANALYSIS. .................................................... 32
 TABLE 3-2: YIELD OF INTERFERON CONCENTRATION ON VIABLE CELL CONCENTRATION FOR DIFFERENT BATCHES STARTED AT 0, 2 AND 4 WEEKS IN FREESTYLE293, LC-SFM AND MLC-SFM MEDIA FOR DIFFERENT OSMOLALITY.................................................................................. 44
 TABLE 4-1: PARTIAL COMPOSITION OF THE MEDIA USED BASED ON THE ADDITIVES AND AMINO ACID ANALYSIS. .................................................... 60
 TABLE 4-2:  BIOREACTOR CONDITIONS INVESTIGATED IN THE FIRST PERFUSION EXPERIMENT. ................................................................................. 66
 TABLE 4-3:  BIOREACTOR CONDITIONS INVESTIGATED IN THE SECOND PERFUSION EXPERIMENT. ................................................................................. 66
 TABLE 4-4:  SUMMARY OF CALCULATED CELL SPECIFIC RATES FOR THE FIRST PERFUSION CULTURE. ............................................................................ 73
 TABLE 4-5 :  SUMMARY OF CALCULATED CELL SPECIFIC RATES FOR THE BATCH CULTURES PERFORMED WITH SPENT, ENRICHED SPENT AND FRESH MEDIA........................................................................................................ 76
 TABLE 4-6 :  SUMMARY OF CALCULATED CELL SPECIFIC RATES FOR THE ADAPTED STRATEGY PERFUSION CULTURE................................................ 81
 TABLE 5-1 :  PARTIAL COMPOSITION OF THE MEDIA USED BASED ON THE ADDITIVES AND AMINO ACID ANALYSIS. .................................................... 94
 TABLE 5-2.  COMPARISON OF PARTICLE DIAMETERS MEASURED WITH THE AUTOMATIC CEDEX AND WITH THE COMPUTER ASSISTED IMAGE ANALYSIS METHOD........................................................................................... 105
 TABLE 5-3. CORRELATION COEFFICIENTS OF PERMITTIVITY RESULTS VS. VIABLE CELL CONCENTRATIONS FOR DILUTION EXPERIMENTS IN BATCH AND PERFUSION CULTURES............................................................. 116
  x TABLE 5-4.  PRODUCTIVITY COMPARISON BETWEEN BATCH AND PERFUSION CULTURES. .................................................................................... 118
   xi List of Figures FIGURE 1-1 : SIMPLE SCHEMATICS OF A) BATCH PROCESS, B) FED-BATCH PROCESS AND C) PERFUSION PROCESS ........................................................... 2 FIGURE 2-1: SCHEMATIC OF A PERFUSION CULTURE SYSTEM ....................... 14 FIGURE 3-1.  SCHEMATIC OF THE PERFUSION SYSTEM. .................................... 34 FIGURE 3-2.  VIABLE CELL DENSITY AND VIABILITY (A), INTERFERON (B), GLUCOSE AND SERINE (C), AND LACTATE AND AMMONIA (D) CONCENTRATION PROFILES FOR HEK-293 CELLS IN BATCH CULTURE USING FREESTYLE MEDIUM. ............................................................................ 40 FIGURE 3-3.  VIABLE CELL DENSITY AND VIABILITY (A), INTERFERON (B), GLUCOSE AND SERINE (C), LACTATE AND AMMONIA (D) CONCENTRATION PROFILES FOR HEK-293 CELLS IN BATCH CULTURE USING LC-SFM MEDIUM. .................................................................................... 41 FIGURE 3-4.  VIABLE CELL DENSITY AND VIABILITY (A), INTERFERON (B), GLUCOSE AND SERINE (C), LACTATE AND AMMONIA (D) CONCENTRATION PROFILES FOR HEK-293 CELLS IN BATCH CULTURE USING MLC-SFM MEDIUM.................................................................................. 42 FIGURE 3-5.  APPARENT CELL SPECIFIC GROWTH RATES (A) AND CELL SPECIFIC PRODUCTION RATES (B) DURING HEK293 CELL BATCH CULTURE IN FREESTYLE MEDIUM, IN LC-SFM MEDIUM AND IN MLC- SFM MEDIUM......................................................................................................... 43 FIGURE 3-6.  APPARENT GROWTH RATES OF HEK-293 CELLS MAINTAINED IN FREESTYLE MEDIUM AT NORMAL (290 MOSM KG-1) AND HIGH (380 MOSM KG-1) OSMOLALITY................................................................................. 45 FIGURE 3-7.  YIELD OF INTERFERON CONCENTRATION ON VIABLE CELL CONCENTRATION FOR BATCHES STARTED FROM MAINTENANCE CULTURES AT DIFFERENT TIME AND CULTIVATED IN LC-SFM MEDIUM AT HIGH OSMOLALITY (375MOSM KG-1) (A) OR MLC-SFM MEDIUM AT NORMAL OSMOLALITY (300 MOSM KG-1) (B) USING OR NOT SELECTIVE PRESSURE............................................................................................................... 48  xii FIGURE 3-8.  VIABLE CELL AND INTERFERON CONCENTRATION PROFILES DURING PERFUSION CULTURE USING HIGH OSMOLALITY LC-SFM MEDIUM (375 MOSM KG-1). ................................................................................. 49 FIGURE 3-9.  VIABLE CELL AND INTERFERON CONCENTRATION PROFILES DURING PERFUSION CULTURE USING NORMAL OSMOLALITY MLC-SFM (300 MOSM KG-1).................................................................................................... 50 FIGURE 4-1:  SCHEMATIC OF THE PERFUSION SYSTEM. .................................... 63 FIGURE 4-2:  EVOLUTION OF VIABLE CELL AND IFN-Α2B CONCENTRATIONS DURING HEK293 CELL PERFUSION CULTURE COMPARED TO MODEL SIMULATED RESULTS. ........................................................................................ 71 FIGURE 4-3:  EVOLUTION OF GLUCOSE AND SERINE CONCENTRATIONS DURING HEK293 CELL PERFUSION CULTURE COMPARED TO MODEL SIMULATED RESULTS. ........................................................................................ 72 FIGURE 4-4: EVOLUTION OF LACTATE AND AMMONIA CONCENTRATIONS DURING HEK293 CELL PERFUSION CULTURE COMPARED TO MODEL SIMULATED RESULTS. ........................................................................................ 72 FIGURE 4-5 : RELATION BETWEEN SPECIFIC PRODUCTION RATE AND THE GROWTH RATE FOR SEMI CONTINUOUS CULTURES WITH AND WITHOUT CELL RETENTION. ............................................................................ 74 FIGURE 4-6 :  EVOLUTION OF VIABLE CELL AND IFN-Α2B CONCENTRATIONS FOR THE ADAPTED STRATEGY DURING SECOND HEK293 CELL PERFUSION CULTURE COMPARED TO MODEL SIMULATED RESULTS. . 78 FIGURE 4-7 :  EVOLUTION OF GLUCOSE AND SERINE CONCENTRATIONS FOR THE ADAPTED STRATEGY DURING SECOND HEK293 CELL PERFUSION CULTURE COMPARED TO MODEL SIMULATED RESULTS. . 79 FIGURE 4-8 : EVOLUTION OF LACTATE AND AMMONIA CONCENTRATIONS FOR THE ADAPTED STRATEGY DURING SECOND HEK293 CELL PERFUSION CULTURE COMPARED TO MODEL SIMULATED RESULTS. . 79 FIGURE 5-1 : SCHEMATIC OF THE PERFUSION SYSTEM. .................................... 98 FIGURE 5-2 : ANALYSIS OF PARTICLE DIAMETER SIZE DISTRIBUTION FOR A) 10 µM AND B) 19 µM DIAMETER CALIBRATION BEADS. .......................... 104  xiii FIGURE 5-3:  VIABLE CELL CONCENTRATION AND VIABILITY FOR HEK293 CELLS PERFUSION CULTURE PERFORMED AT A) LOW BLEED RATE (0.042 D-1) AND B) HIGH BLEED RATE (0.167 D-1). ........................................ 106 FIGURE 5-4 : PICTURES OF THE BIOREACTOR GLASS WALL DURING THE LOW BLEED RATE PERFUSION CULTURE (A) AND THE SECOND PERFUSION CULTURE USING HIGH BLEED RATE (B). .............................. 107 FIGURE 5-5 :  EVOLUTION OF AVERAGE PARTICLE DIAMETER DURING HEK293 CELL PERFUSION CULTURES AT LOW AND HIGH BLEED RATE. ................................................................................................................................ 108 FIGURE 5-6 : EVOLUTION OF  THE VOLUME FRACTION OF PARTICLES LARGER THAN 100 µM DURING HEK293 CELL PERFUSION CULTURES AT LOW AND HIGH BLEED RATE. ........................................................................ 108 FIGURE 5-7 : VIABLE CELL CONCENTRATION AND VIABILITY PROFILES FOR HEK293 CELLS IN BATCH CULTURE IN STANDARD CONDITIONS (A), WITH ADDITION OF DEAD CELLS (B), WITH ADDITION OF CALCIUM (C), AND AT LOW AGITATION RATE (D). ............................................................. 110 FIGURE 5-8 :  EVOLUTION OF AVERAGE PARTICLE DIAMETER DURING HEK293 CELL BATCH CULTURES IN STANDARD CONDITION (CONTROL), WITH ADDITION OF DEAD CELLS, WITH ADDITION OF CALCIUM AND AT LOW AGITATION RATE. ................................................ 111 FIGURE 5-9 :  PARTICLE SIZE DISTRIBUTION FOR BATCH CULTURES AROUND DAY 6 IN STANDARD CONDITIONS (A), WITH ADDITION OF DEAD CELLS (B), WITH ADDITION OF CALCIUM (C), AND AT LOW AGITATION RATE (D). ....................................................................................... 112 FIGURE 5-10 :  PARTICLE SIZE DISTRIBUTION FOR PERFUSION CULTURES AT DAY 6 FOR LOW BLEED RATE (A) AND HIGH BLEEB RATE (B) OPERATION.......................................................................................................... 113 FIGURE 5-11: PERMITTIVITY PROFILES FOR TWO PERFUSION CULTURES USING LOW BLEED RATE  (A) AND HIGH BLEED RATE (B). .................... 114 FIGURE 5-12 : EVOLUTION OF THE SPECIFIC PERMITTIVITY FOR TWO PERFUSION CULTURES USING HIGH BLEED RATE (SOLID CIRCLES, SOLID LINE) AND LOW BLEED RATE (CIRCLES, DASHED LINE)............ 115  xiv FIGURE 5-13 : EVOLUTION OF VIABLE CELL CONCENTRATION AND VIABILITY OF AGGREGATES FROM THE BATCH WITH CALCIUM MIXED WITH TRIPSIN. ..................................................................................................... 119  xv  Acknowledgements Trop souvent nous prenons la vie pour acquis et nous réalisons la chance que nous avons une fois confrontée à une situation difficile.  Je tiens à dédier ma thèse au professeur Bruce Bowen qui a été pour moi un modèle dans mon cheminement professionnel et personnel.  Il m’a permis d’appliquer concrètement des concepts théoriques pour simuler des phénomènes physiques et biologiques concrets.  Bruce m’a toujours impressionné par son grand professionnalisme mélangé à un côté humain authentique et son écoute franche.  Comme Bruce m’a récemment écrit: “I look on each new day as a gift that I should try to take full advantage of”.  La vie est l’ensemble de notre quotidien et il n’en tient qu’à nous d’en profiter, car sans s’en rendre compte, les journées disparaissent à un rythme effréné.  La rédaction de cette thèse s’est fait dans un contexte difficile pour moi, Bruce a été pour moi une source importante d’inspiration, de motivation et je suis très fier de pouvoir enfin terminer cette étape de vie.  Je voudrais remercier mes superviseurs Dr. James M Piret de l’Université de Colombie Britannique et Dr. Amine Kamen de l’Institue de Recherche en Biotechnologie (IRB) qui m’ont permis de vivre cette expérience à Montréal.  Tous les gens de l’IRB m’ont aussi beaucoup aidé à apprendre durant mon projet, ce fut pour moi une expérience riche professionnellement et aussi, personnellement.  J’aimerais remercier mon entourage proche, notamment Amélie qui m’a toujours supporté même à travers les moments difficiles et aussi Benoit et Jean-François pour leur confiance et leur encouragement.   1 1 Introduction Therapeutic molecules are being produced under industrial conditions to provide treatments for major human diseases including cancers, as well as viral and cardiovascular diseases.  Since the Canadian population is aging, the need for such treatments will certainly increase.  Mammalian cells are the most often used system for producing therapeutic molecules such as recombinant proteins (Wurm 2004), including monoclonal antibodies (MAbs) (Chu and Robinson 2001; Wood-MacKenzie 2003). These molecules are usually produced in suspension stirred tank bioreactors, up to 20,000 L (Lonza Biologics, Slough, UK).  The three main production methods used to culture the mammalian cells have been batch, fed-batch and perfusion processes.  Batch is the oldest form of production.  It involves inoculating a fixed volume of medium with cells and then culturing them until their viability decreases, usually within a week to ten days (Figure 1-1A).  To increase the production from these cultures, in repeated-batch cultures a large proportion (e.g. 90%) of the medium is periodically harvested and replaced with fresh medium.  The maximal cell concentration in these processes is primarily limited by the depletion of nutrients (Hu and Aunins 1997).  A more sophisticated process is the fed-batch culture whereby periodic feeding of nutrients allows the most productive phase of the culture to be prolonged (Figure 1-1B).  These cultures are limited as by-products build up, leading to decreased viability and protein productivity such that these processes are usually terminated after approximately 15 to 20 days of culture.  The perfusion culture is a continuous process that provides for additional nutrients to the cells and the removal of their waste products while retaining  2 the cells in the bioreactor (Figure 1-1C).  The use of an appropriate cell retention device is an essential part of the perfusion operation.  This device normally returns >90% of the cells to the bioreactor from the out-flowing product and spent medium (i.e., harvest) stream.  This serves to increase cell concentrations while maintaining a more constant level of nutrients and waste metabolites within the reactor.  A perfusion system can sustain a culture at steady state for over 6 months of operation (Leist et al., 1990).   Figure 1-1 : Simple schematics of A) batch process, B) fed-batch process and C) perfusion process Dashed lines represent discontinuous operations and solid lines represent continuous operations.  The red square represents the cell retention device.  Since the early 1990s, mainly fed-batch and perfusion processes have been used for the large-scale industrial production of biopharmaceuticals (Griffiths 1992; Hu and Aunins 1997; Lim et al. 2006).  The choice of the production method is influenced by many factors, such as the nature of the cells, the type of biopharmaceutical produced, the product stability, the cell sensitivity to environmental variation and the available equipment (Griffiths 1992; Kadouri and Spier 1997).  In addition, operator experience with the process plays a key role.   3 The FDA is encouraging biomanufacturers to use more sophisticated processes, with improved monitoring and process control (DePalma, 2004).  Perfusion process selection offers the opportunity of controlling the culture at a well-defined steady state, especially compared to the more widely used fed-batch process.  However, a number of challenges related to perfusion culture process are limiting the use of this method.  The development of a perfusion cell culture process usually takes more time (Angepat et al. 2005) and can be more difficult because of the numerous variables that need to be optimized.  The dilution effect created by continuous feeding decreases the product titer in the bioreactor compared to fed-batch.  Moreover, continuous processing can increase the formation of heterogeneous cell aggregates that can affect growth, productivity and product quality (Coppen et al. 1995; Renner et al. 1993).  The monitoring of aggregated cells and their effect during perfusion culture has not been studied extensively.  1.1 Thesis Objectives Based on the potential of perfusion culture processes to fulfill the increasing demand for biopharmaceuticals, their operational challenges and the lack of systematic studies, the overall goal of this thesis was to develop techniques to better understand and improve the performance of perfusion culture. Specifically, the main objectives were to: i. Investigate the effects of medium composition and osmolality on long-term productivity ii. Investigate the effects of reduced perfusion rates using enriched media on cellular productivity  4 iii. Investigate cell aggregation in perfusion cultures and explore the influence of aggregation on the use of an on-line method to monitor cell concentration. To carry out these studies, the human embryonic kidney cell line (HEK293) engineered to express human interferon-alpha 2b (IFN-α2b) was used as a model system.  In Chapter 2 the literature background to this work is reviewed.  Chapter 3 of the thesis presents the investigation into the effects of medium composition and osmolality on long- term productivity by testing various media in batch and perfusion cultures.  Based on the medium composition and osmolality selected in Chapter 3, a transient approach was then used to investigate the effect of perfusion rate on productivity as described in Chapter 4. In addition, a model of the perfusion process was applied to help evaluate the results from this study.  Finally, given the complication for the analysis of long-term perfusion cultures where cells tend to aggregate over time, Chapter 5 describes the investigation into how these aggregates form, their influence on productivity calculations and how to control aggregate formation.  In order to monitor the aggregate formation, a simple image analysis method was developed and compared to a commercially available cell size measurement technique.  The effects of aggregation on the permittivity signal and on the calculated yield and specific rates were also investigated.   5 1.2 References Angepat S, Gorenflo VM, Piret JM. 2005. Accelerating perfusion process optimization by scanning non-steady-state responses. Biotechnol Bioeng 92(4):472-8. Chu L, Robinson DK. 2001. Industrial choices for protein production by large-scale cell culture. Current Opinion in Biotechnology 12(2):180-187. Coppen SR, Newsam R, Bull AT, Baines AJ. 1995. Heterogeneity within Populations of Recombinant Chinese-Hamster Ovary Cells Expressing Human Interferon-Gamma. Biotechnology and Bioengineering 46(2):147-158. DePalma A. 2004. PAT: Taking process monitoring to next level. Genetic Engineering News 24(15):46-47. Griffiths JB. 1992. Animal-Cell Culture Processes - Batch or Continuous. Journal of Biotechnology 22(1-2):21-30. Hu WS, Aunins JG. 1997. Large-scale mammalian cell culture. Curr Opin Biotechnol 8(2):148-53. Kadouri A, Spier RE. 1997. Some myths and messages concerning the batch and continuous culture of animal cells. Cytotechnology 24(2):89-98. Leist CH, Meyer HP, Fiechter A. 1990. Potential and Problems of Animal-Cells in Suspension-Culture. Journal of Biotechnology 15(1-2):1-46. Lim AC, Washbrook J, Titchener-Hooker NJ, Farid SS. 2006. A computer-aided approach to compare the production economics of fed-batch and perfusion culture under uncertainty. Biotechnol Bioeng 93(4):687-97. Renner WA, Jordan M, Eppenberger HM, Leist C. 1993. Cell-cell adhesion and aggregation: Influence on the growth behavior of CHO cells. Biotechnol Bioeng 41(2):188-93. Wood-MacKenzie. 2003. Horizons. Wood MacKenzie Limited. Report nr Pharmaceutical Issue 6. Wurm FM. 2004. Production of recombinant protein therapeutics in cultivated mammalian cells. Nature Biotechnology 22(11):1393-1398.    6 2 Background 2.1 Effects of Culture Environment on Mammalian Cells 2.1.1 Inhibitory Metabolites During recombinant protein production, mammalian cells utilize nutrients to create energy, biomass and secreted metabolic byproducts (metabolites).  While nutrient concentrations decrease, metabolites may accumulate in the bioreactor and gradually become inhibitory and/or toxic for the cells.  In this case, the growth rate is diminished and cell death can increase as a function of the level of inhibiting metabolites.  In general, culture performance is calculated from product titer and maximal cell concentration and is expressed in terms of yield, cell specific production or growth rates. Lactate and ammonia are the most commonly analyzed metabolites that can affect culture performance (Banik and Heath 1994; Banik and Heath 1995; Burteau et al. 2003; Dong et al. 2005; Kempken et al. 1991; Mercille et al. 2000; Smith et al. 1991; Yang and Butler 2000; Zhang et al. 1993).  For mammalian cells, concentrations over 3 to 10 mM of ammonia and/or 2 to 5 g L-1 of lactate have been reported to be toxic (Bibila and Robinson 1995; Xing et al. 2008; Yang and Butler 2000).  In some cases, cultures have had a dramatic loss of viable cell mass, possibly due to other inhibitory metabolites (Konstantinov et al. 2006; Ronning et al. 1991; Yang et al. 2000; Zeng and Deckwer 1999).   7 2.1.2 Osmolality The osmolality of a solution is defined as the number of osmoles (Osm) of solute per kilogram or litre of solution.  In the human body, cells are exposed to an osmolality of ~300 mOsm kg-1.  Therefore, in standard mammalian cell culture the media osmolality is maintained between 260 – 320 mOsm kg-1 (Ozturk and Palsson 1990).  Numerous researchers have studied the impact of osmolality on cell culture.  A mechanistic study of hyperosmotic stress based on gene expression found that it caused many significant cellular responses related to ion transport, accumulation of osmolytes, cell cycle distribution, proliferation, cytoskeletal organization and metabolism (Shen and Sharfstein 2006).  Hyperosmotic conditions are known to induce cellular stresses and to reduce growth rates, but can also increase cell specific protein productivity (deZengotita et al. 2002; Ju et al. 2009; Kim et al. 2002; Lin et al. 1999; Ozturk and Palsson 1990; Wu et al. 2004).  Because the osmolality of culture media can be readily increased (e.g. by adding Na2+ ions with NaCl), this can be an attractive method to increase production yields.  However, the reduced growth rate can counter-balance gains in specific productivity thus lowering the final product concentration (Zhu et al. 2005).  Biphasic batch culture strategies have thus been designed to increase the osmolality level only after optimal or desired cell concentrations are reached (Ferreira et al. 2005; Kim et al. 2002).  Another strategy is to gradually increase osmolality so that the cells have time to adapt to the new conditions (Lin et al. 1999; Reddy and Miller 1994).  Most of these studies have been performed at batch time scales (up to 10 days) and the long-term effects of high osmolality have not been well documented.  8  2.2 Cellular Aggregation Cellular aggregation is a normal physiological phenomenon observed even in batch suspension cultures of non-anchorage-dependent mammalian cells.  Aggregation can result from tight intercellular junctions between cells, which are characterized by the presence of a proteinaceous layer at the contact point between the cell cytoplasmic membranes (Coppen et al. 1995; Peshwa et al. 1993).  There is no consensus on the minimal size of aggregates with some researchers considering particles larger than 25 µm as aggregates (Renner et al. 1993).  However, this criterion alone is not entirely appropriate since it could represent a single cell undergoing mitosis (the diameter of single mammalian cells ranges between 10 to 20 µm).  Aggregate formation during long-term cultivation and can be useful as a natural means of cell retention (Litwin 1992; Moreira et al. 1994a; Peshwa et al. 1993; Sen et al. 2001). However, if it is not controlled, cellular aggregation becomes a critical factor that can negatively influence cell growth and productivity (Renner et al. 1993).  Increasing cellular aggregate size heterogeneity can influence the culture’s performance (Coppen et al. 1995; Renner et al. 1993) and decrease culture reproducibility.  In particular, the viability of cells at the centre of aggregates can be reduced based on trypan blue dye staining (Coco-Martin et al. 1992) with the formation of necrotic cores in aggregates larger than 300 µm, due to oxygen mass transfer limitations (Jo et al. 1998; Sen et al. 2001).   9 Three different techniques are commonly used to measure aggregate size: (1) microscopic evaluation (Litwin 1992; Liu et al. 2009; Moreira et al. 1995; Moreira et al. 1994b; Sen et al. 2001), (2) the electrical sensing zone method (Boraston et al. 1992; Peshwa et al. 1993), and (3) the laser diffraction method (Illing and Harrison 1999; Renner et al. 1993).  Many commercially available techniques are not adequate to measure large aggregates.  For example, the Coulter Counter (Millipore, Billerica, MA) could not measure particles larger than 50 µm and would be rapidly clogged large cell aggregates.  Aggregate settling is another problem that causes measurement errors, especially using laser diffraction devices.  Several techniques have been reported to limit aggregate size and thereby prevent the formation of necrotic cores in large aggregates.  The most common techniques used are increased hydrodynamic forces (Illing and Harrison 1999; Moreira et al. 1995; Renner et al. 1993; Sen et al. 2001), mechanical dissociation (Moreira et al. 1994b) and the modification of medium composition (Boraston et al. 1992; Litwin 1992; Moreira et al. 1994a; Peshwa et al. 1993).  Compared to bacteria, mammalian cells are more shear- sensitive because of their size and lack of a cell wall.  Therefore, if hydrodynamic forces are to be used to control aggregation, the magnitude of the shear applied to control the size of aggregates should be carefully adjusted in order to limit cell damage.  In stirred suspension cultures, the agitation rate, expressed in revolution per minute (rpm), can be used to control cell aggregation (Ozturk 1996) (Moreira et al. 1995).  A high agitation rate can lyse cells or change cellular metabolism, such as increase the specific glucose consumption or lactate production rates (Mercille et al. 2000; Smith et al. 1991), as well  10 as reduce the cell growth rate (Werner et al. 1992).  Hence, a balance between disaggregation and cell damage needs to be maintained to keep culture performance high (Varley and Birch 1999).  Cell aggregation is a more important consideration in a continuously perfused process than it is in batch culture because of the long cultivation period and the possible aggregation increase that may occur when, for instance, using an acoustic separator as a cell retention device (Bierau et al. 1998).  Cellular aggregation can provide a natural means to enhance cell retention in perfusion cultures (Han et al. 2006; Liu et al. 2009; Liu et al. 2006).  In these studies, image analysis was used to calculate the average diameter of 20 to 60 HEK293 aggregates (>50 µm).  In the perfusion culture, the average aggregate diameter continued to increase without achieving a steady state.  After more than 20 days, the average aggregate diameter had increased to 352 µm and the aggregates were mainly spherical.  The culture performance (defined by growth rate, maximum cell density, glucose consumption rate and lactate production rate) was compared for suspended aggregates vs. monolayers and no differences were observed (Liu et al. 2006). The main advantage of aggregation in this case was to facilitate cell retention; the factors that influenced aggregation were not analyzed.  The formation of aggregates was also studied in batch and repeated-batch (i.e. repeated harvesting and cell dilution) cultures where aggregated cells were maintained for more than 25 days after selecting for aggregates as the batch inoculum (Moreira et al. 1994b).  They suggested that aggregate culture could be feasible in large-scale cultures even though growth rate and viability were reduced in the most aggregated cultures.  Also, no information was given regarding  11 the effect of cellular aggregation on cell productivity.  Overall, there is very limited information for perfusion culture processes at high cell concentrations under steady-state conditions compared to batch cultures.  Moreover, aggregate effects on culture performance such as protein production have not been documented.  2.3 On-line Measurement of Cell Density Improved on-line monitoring technologies have been advocated as a response to the initiative of the US Food and Drug Administration (FDA) on “Process Analytical Technology” (Hinz 2006).  This initiative supports innovation and improved efficiency in pharmaceutical development, manufacturing and quality assurance.  Dissolved oxygen and pH (base additions) are routinely monitored on-line to follow the evolution of cell cultures.  However, these parameters are indirect measurements of the viable cell concentration that can be greatly affected by changes in metabolic activity.  The use of radio-frequency impedance to measure the permittivity signal has been used to non-invasively measure cell density on-line (Ansorge et al. 2009; Carvell and Dowd 2006).  A probe, placed in the bioreactor, allows a continuous reading of the cell biovolume rather than the number of cells.  There are different commercially available devices that use the same principal such as the Biomass Monitor 220 (Aber Instruments, Aberystwyth, UK) or the Fogale Biomass System (Fogale Nanotech, Nîmes, France). The medium and cells are very complex mixtures of conducting and non-conducting molecules.  In the presence of an electric field, cells act as spherical capacitors where ions are pushed in the direction of the field (positive ions) or pushed in the opposite  12 direction (negative ions).  This displacement of ions, called charge separation or polarization at the pole of the cells, can be measured by their capacitance (C) in farads (F).  The use of different frequencies, i.e., the rate at which the electric field changes direction, allows the differentiation between viable and non-viable cells.  Thus, it is possible to estimate on-line the viable cell concentration by using the appropriate frequency to measure the capacitance of the culture.  2.4 Perfusion Cultures 2.4.1 High Cell Concentration System Compared to batch cultures, optimal perfusion culture should result in increased product titers, augmented cell specific productivities with maintained product quality and process consistency, as well as maximized viable cell concentrations.  Increasing the bioreactor cell concentration increases its volumetric productivity, assuming perfusion rates are increased sufficiently to maintain appropriate environmental conditions (Banik and Heath 1994; Banik and Heath 1995; Chu and Robinson 2001; Dalm et al. 2004; Hiller et al. 1993; Mullin 2004; Reuveny et al. 1986a; Spier 1991; Takazawa and Tokashiki 1989; Varley and Birch 1999; Yang et al. 2000).  In some cases, the cell specific productivity has been reported to increase at higher cell concentrations (Banik and Heath 1994; Banik and Heath 1995).  Most reported perfusion cultures have been operated around 107 cells mL-1 (Banik and Heath 1994; Hiller et al. 1993; Gray et al. 1996; Han et al. 2006; Kim et al. 2008; Leelavatcharamas et al. 1999; Liu et al. 2009; Liu et al. 2006; Zeng and Deckwer 1999) but there have been a few reports of cell concentrations greater  13 than 3 x 107 cells mL-1 (Dalm et al. 2004; Gramer and Britton 2002; Link et al. 2004). Considering that the theoretical natural cell concentration limit of solid tissue is approximately 109 cells mL-1, given the physical limitation of space, it appears that it is still possible to increase the viable biomass of cultivated cells.  However, as cell concentrations are increased, the capacity for O2/CO2 mass transfer and the capacity of the cell retention device can be exceeded (Chu and Robinson 2001; Ozturk 1996; Zhang et al. 1993).  The high perfusion rates needed to provide sufficient nutrients to high cell concentrations induces a diminution of product titer in the bioreactor because of dilution (Yang et al. 2000).  Compared to other culture processes, perfusion needs additional instrumentation (Kadouri and Spier 1997), more complex scale-up (Werner et al. 1992; Woodside et al. 1998) and longer runs for process optimization (Woodside et al. 1998) that require longer validation times (Kretzmer 2002).  Fed-batch is considered more robust relative to the higher risks of contamination or mechanical failures in perfusion processes (Lim et al. 2006).  However, perfusion cultures do offer significant advantages such as high cell concentration (Cortin et al. 2004; Dalm et al. 2004; Griffiths 1992; Voisard et al. 2003; Werner et al. 1992; Woodside et al. 1998), high volumetric productivity (Cortin et al. 2004; Deo et al. 1996; Griffiths 1992; Woodside et al. 1998), short residence time (Griffiths 1992; Ryll et al. 2000; Woodside et al. 1998; Wurm 2004) and steady-state operation (Furey 2000; Woodside et al. 1998).   14 2.4.2 Process Operation Perfusion cultures are fed continuously or periodically and harvested to maintain a constant bioreactor volume.  An increase in feed rate is usually associated with an increase in viable cell concentration (Bierau et al. 1998; Dalm et al. 2007; Dalm et al. 2004; Leelavatcharamas et al. 1999; Mercille et al. 2000; Vits and Hu 1992).  The control of the glucose concentration by the feed rate is very important and has to be kept in the appropriate range (Dowd et al. 2001; Konstantinov et al. 1996).  The bioreactor viable cell concentration (CB) is kept constant by removing or ‘bleeding’ non-viable cells (Figure 2-1).  By removing nonviable cells, the bleed maintains the viability of the culture (Banik and Heath 1995; Dalm et al. 2004; Zeng and Deckwer 1999).  Figure 2-1 presents the flows, with the associated cell concentrations, of a perfusion culture system.   Figure 2-1: Schematic of a perfusion culture system   15 At steady state, the feed rate can be expressed as a function of the harvest rate and bleed rate: € FF = FH + FB   (2-1) where FF is the feed flow rate (L day -1), FH is the harvest flow rate (L day -1) and FB is the bleed flow rate (L day-1).  The cell mass balance, over a constant volume, is given by € dCB dt = µCB − kdCB − FBCB V − FHCH V  (2-2) where t is the time (day), CH is the viable cell concentration of the harvest flow (cells L-1), V is the bioreactor volume (L),  is the true specific growth rate (day -1) and  is the specific death rate (day -1).  The cell retention device is a key component of a high cell concentration perfusion culture system.  Essentially, a cell retention device should provide a high separation efficiency (e.g. >90% of the viable cells retained) at high perfusion rates, for a long period of operation, without having substantial negative effects on the cell culture (Woodside et al., 1998).  The separation efficiency (SE) of this device is defined as  (2-3) High separation efficiency is needed for the operation of a high cell concentration perfusion culture, to yield an improved culture performance.   16 2.5 Development Approaches for Perfusion Processes When developing a cost-effective process, the factors for optimization include cell concentration, productivity, product quality or development time (Link et al. 2004). Examining every possible combination of operational settings like the pH, dissolved oxygen, nutrient concentrations and medium feed rates in a perfusion culture process to determine the optimum is not feasible.  Instead, experience from previous small-scale cultures is used and a limited number of potentially critical variables for the perfusion culture are then investigated.  Factorial design methods are used to minimize the number of experiments needed to screen for significant effects and possible interactions (May 2004).  Another approach is to develop a mathematical model of the system; however this approach is usually impractical due to a lack of sufficient understanding of the underlying cell biology (Versyck et al. 1997; Zeng and Deckwer 1999).  For perfusion cultures, one suggested optimization strategy is to aim for a high cell concentration and then optimize the other process variables (Banik and Heath 1995). Parallel experiments require multiple bioreactors and long startup times, while sequential experiments can be done within one bioreactor, reducing the cost and the complexity of the experiments.  Nonetheless, during a sequential experiment, the time between two steady-states is affected by the nature and the magnitude of the variation imposed (Miller et al. 1988; Miller et al. 1989).  In general it takes more than a week for steady state to be reached (Angepat et al. 2005; Reuveny et al. 1986b).   17 Due to the long period of time needed to test each different steady-state condition, the process development of continuous cultures can take much more time than that of batch or fed-batch systems (Paalme et al. 1995; Werner et al. 1992).  As a way to rapidly evaluate the physiological responses of microbial cultures, Paalme et al. (1995) developed a non-steady-state approach to analyzing continuous cultures.  Instead of waiting for the true steady state, they examined the non-steady-state response to a gradually changing dilution rate and they were able to obtain results for the entire growth rate range of E. coli 10 times faster than by standard methods.  A similar approach was used to evaluate the optimal culture temperature of Chinese hamster ovary (CHO) cells producing tissue plasminogen activator (t-PA) in a perfusion culture (Angepat et al. 2005).  The optimal temperature was found rapidly by measuring 3-day sequential transient responses at temperatures of 37, 35, 33 and 31 oC.  This approach was predicted to reduce the time required to find the optimal temperature by 59% as compared to the standard perfusion steady-state approach (Angepat et al. 2005).  This method, known as the transient scanning approach, appears to be an efficient method to greatly accelerate process development.  So far, transient scanning has only been applied to temperature optimization.  An attempt was made to optimize the pH of the same CHO cell culture (Reiter and Piret 2004) but the results were not conclusive because the CO2 used to reduce the pH inhibited the growth and biased the results.  Furthermore, the formation of cell aggregates complicated the data analysis.   18 2.6 Low Perfusion Rate Strategy Besides the dilution effect that causes a decrease in product titer, high perfusion rates can also induce process instabilities, such as poor cell retention and washout (Deo et al. 1996).  The negative effects of high dilution rate can be overcome by using low volumetric perfusion-rate process strategies.  This can also have the advantages of reducing media cost and volume handling (Nivitchanyong et al. 2007) as well as increasing titers and productivity.  A simple strategy to reduce the perfusion rate is done by reducing the bioreactor temperature to slow down the growth rate, then decreasing the feed rate to provide the smaller amount of nutrients needed (Gambhir et al. 1999; Gorenflo et al. 2002).  The success of this approach is dependent on the temperature response of the cell line.  For example, a decrease in perfusion rate can also lead to negative effects such as decreases in cell viability and growth rate (Nivitchanyong et al. 2007).  Konstantinov et al. (2006) described the «push-to-low» approach, where the feed rate was decreased gradually and the medium improved as needed.  The minimal dilution rate was established accordingly to the maximal residence time suitable for an unstable product. The maximal dilution rate was determined by the physical limitations of the system (maximum cell retention capacity).  The bleed rate was automatically adjusted to maintain a high viable cell concentration (~2 x 107 cells mL-1).  This culture strategy was developed for cell lines with non-growth-associated or inversely growth-associated production kinetics.  Goudar et al. (2006) increased the product titer and reduced the  19 medium cost using a similar approach for a non-growth-associated production CHO cell line.  2.7 Model Cell Line The human embryonic kidney 293 (HEK293) cell line (Graham et al. 1977) provides the potential to express recombinant proteins with fully human post-translational modifications.  These cells have been adapted to suspension culture in serum-free media, thus enabling their use for large-scale culture (Cote et al. 1998).  It has become an alternative cell line for industrial applications.  For example, HEK293 cells are used commercially to produce drotrecogin alfa (activated), a recombinant version of the endogenous activated Protein C, a drug named Xigris (Eli Lilly, Giessen, Germany).  The HEK293 cell line used in this work has been engineered for a high level expression of human interferon alpha 2b (IFN-α2b) by transfection (Loignon et al. 2008).  20 2.8 References Angepat S, Gorenflo VM, Piret JM. 2005. Accelerating perfusion process optimization by scanning non-steady-state responses. Biotechnol Bioeng 92(4):472-8. Ansorge S, Esteban G, Schmid G. 2009. Multi-Frequency Permittivity Measurements Allow For On-Line Monitoring of Changes In Intracellular Conductivity Due To Nutrient Limitations During Batch Cultivations Of CHO Cells. Biotechnol Progr in print. Banik GG, Heath CA. 1994. An Investigation of Cell-Density Effects on Hybridoma Metabolism in a Homogeneous Perfusion Reactor. Bioprocess Engineering 11(6):229- 237. Banik GG, Heath CA. 1995. Hybridoma Growth and Antibody-Production as a Function of Cell-Density and Specific Growth-Rate in Perfusion Culture. Biotechnology and Bioengineering 48(3):289-300. Berthold W, Kempken R. 1994. Interaction of cell culture with downstream purification: a case study. Cytotechnology 15(1-3):229-42. Bibila TA, Robinson DK. 1995. In pursuit of the optimal fed-batch process for monoclonal antibody production. Biotechnol Prog 11(1):1-13. Bierau H, Perani A, al-Rubeai M, Emery AN. 1998. A comparison of intensive cell culture bioreactors operating with hybridomas modified for inhibited apoptotic response. J Biotechnol 62(3):195-207. Boraston R, Marshall C, Norman P, Renner G, Warner J. Elimination of cell aggregation in suspension cultures of Chinese hamster ovary (CHO) cells. In: Spier RE, Griffiths JB, MacDonald C, editors; 1992. Burteau CC, Verhoeye FR, Mols JF, Ballez JS, Agathos SN, Schneider YJ. 2003. Fortification of a protein-free cell culture medium with plant peptones improves cultivation and productivity of an interferon-gamma-producing CHO cell line. In Vitro Cellular & Developmental Biology-Animal 39(7):291-296. Carvell J, Dowd J. 2006. On-line Measurements and Control of Viable Cell Density in Cell Culture Manufacturing Processes using Radio-frequency Impedance. Cytotechnology 50(1):35-48. Chu L, Robinson DK. 2001. Industrial choices for protein production by large-scale cell culture. Current Opinion in Biotechnology 12(2):180-187. Coco-Martin JM, Oberink JW, Vanderveldendegroot TAM, Beuvery EC. 1992. Viability Measurements of Hybridoma Cells in Suspension-Cultures. Cytotechnology 8(1):57-64.  21 Coppen SR, Newsam R, Bull AT, Baines AJ. 1995. Heterogeneity within Populations of Recombinant Chinese-Hamster Ovary Cells Expressing Human Interferon-Gamma. Biotechnology and Bioengineering 46(2):147-158. Cortin V, Thibault J, Jacob D, Garnier A. 2004. High-titer adenovirus vector production in 293S cell perfusion culture. Biotechnol Prog 20(3):858-63. Cote J, Garnier A, Massie B, Kamen A. 1998. Serum-free production of recombinant proteins and adenoviral vectors by 293SF-3F6 cells. Biotechnol Bioeng 59(5):567-75. Dalm MC, Lamers PP, Cuijten SM, Tjeerdsma AM, van Grunsven WM, Tramper J, Martens DE. 2007. Effect of feed and bleed rate on hybridoma cells in an acoustic perfusion bioreactor: metabolic analysis. Biotechnol Prog 23(3):560-9. Dalm MCF, Cuijten SMR, van Grunsven WMJ, Tramper J, Martens DE. 2004. Effect of feed and bleed rate on hybridoma cells in an acoustic perfusion bioreactor: Part I. Cell density, viability, and cell-cycle distribution. Biotechnology and Bioengineering 88(5):547-557. Deo YM, Mahadevan MD, Fuchs R. 1996. Practical considerations in operation and scale-up of spin-filter based bioreactors for monoclonal antibody production. Biotechnol Prog 12(1):57-64. deZengotita VM, Schmelzer AE, Miller WM. 2002. Characterization of hybridoma cell responses to elevated pCO(2) and osmolality: intracellular pH, cell size, apoptosis, and metabolism. Biotechnol Bioeng 77(4):369-80. Dong H, Tang YJ, Ohashi R, Hamel JF. 2005. A perfusion culture system using a stirred ceramic membrane reactor for hyperproduction of IgG2a monoclonal antibody by hybridoma cells. Biotechnol Prog 21(1):140-7. Dowd JE, Kwok KE, Piret JM. 2001. Glucose-based optimization of CHO-cell perfusion cultures. Biotechnology and Bioengineering 75(2):252-256. Ferreira TB, Ferreira AL, Carrondo MJ, Alves PM. 2005. Two different serum-free media and osmolality effect upon human 293 cell growth and adenovirus production. Biotechnol Lett 27(22):1809-13. Furey J. 2000. Continuous cell culture using the ATF system - A new way to grow suspension- or anchorage-dependent cells. Genetic Engineering News 20(10):52-52. Gambhir A, Zhang C, Europa A, Hu WS. 1999. Analysis of the use of fortified medium in continuous culture of mammalian cells. Cytotechnology 31(3):243-254. Gorenflo VM, Smith L, Dedinsky B, Persson B, Piret JM. 2002. Scale-up and optimization of an acoustic filter for 200 L/day perfusion of a CHO cell culture. Biotechnol Bioeng 80(4):438-44.  22 Goudar C, Biener R, Zhang C, Michaels J, Piret J, Konstantinov K. 2006. Towards industrial application of quasi real-time metabolic flux analysis for mammalian cell culture. Adv Biochem Eng Biotechnol 101:99-118. Graham FL, Smiley J, Russell WC, Nairn R. 1977. Characteristics of a human cell line transformed by DNA from human adenovirus type 5. J Gen Virol 36(1):59-74. Gramer MJ, Britton TL. 2002. Antibody production by a hybridoma cell line at high cell density is limited by two independent mechanisms. Biotechnology and Bioengineering 79(3):277-283. Gray DR, Chen S, Howarth W, Inlow D, Maiorella BL. 1996. CO2 in large-scale and high-density CHO cell perfusion culture. Cytotechnology 22(1-3):65-78. Griffiths JB. 1992. Animal-Cell Culture Processes - Batch or Continuous. Journal of Biotechnology 22(1-2):21-30. Han Y, Liu XM, Liu H, Li SC, Wu BC, Ye LL, Wang QW, Chen ZL. 2006. Cultivation of recombinant Chinese hamster ovary cells grown as suspended aggregates in stirred vessels. J Biosci Bioeng 102(5):430-5. Henry O, Kwok E, Piret JM. 2008. Simpler noninstrumented batch and semicontinuous cultures provide mammalian cell kinetic data comparable to continuous and perfusion cultures. Biotechnol Prog 24(4):921-31. Hiller GW, Clark DS, Blanch HW. 1993. Cell retention-chemostat studies of hybridoma cells-analysis of hybridoma growth and metabolism in continuous suspension culture in serum-free medium. Biotechnol Bioeng 42(2):185-95. Hinz DC. 2006. Process analytical technologies in the pharmaceutical industry: the FDA's PAT initiative. Anal Bioanal Chem 384(5):1036-42. Illing S, Harrison STL. 1999. The kinetics and mechanism of Corynebacterium glutamicum aggregate breakup in bioreactors. Chemical Engineering Science 54(4):441- 454. Jo EC, Yun JW, Jung KH, Chung SI, Kim JH. 1998. Performance study of perfusion cultures for the production of single-chain urokinase-type plasminogen activator (scu- PA) in a 2.5 l spin-filter bioreactor. Bioprocess Engineering 19(5):363-372. Ju HK, Hwang SJ, Jeon CJ, Lee GM, Yoon SK. 2009. Use of NaCl prevents aggregation of recombinant COMP-angiopoietin-1 in Chinese hamster ovary cells. J Biotechnol 143(2):145-50. Kadouri A, Spier RE. 1997. Some myths and messages concerning the batch and continuous culture of animal cells. Cytotechnology 24(2):89-98.  23 Kempken R, Buntemeyer H, Lehmann J. 1991. The medium cycle bioreactor (MCB): monoclonal antibody production in a new economic production system. Cytotechnology 7(2):63-74. Kim BJ, Oh DJ, Chang HN. 2008. Limited use of Centritech Lab II Centrifuge in perfusion culture of rCHO cells for the production of recombinant antibody. Biotechnol Prog 24(1):166-74. Kim MS, Kim NS, Sung YH, Lee GM. 2002. Biphasic culture strategy based on hyperosmotic pressure for improved humanized antibody production in Chinese hamster ovary cell culture. In Vitro Cell Dev Biol Anim 38(6):314-9. Konstantinov K, Goudar C, Ng M, Meneses R, Thrift J, Chuppa S, Matanguihan C, Michaels J, Naveh D. 2006. The "push-to-low" approach for optimization of high-density perfusion cultures of animal cells. Adv Biochem Eng Biotechnol 101:75-98. Konstantinov KB, Tsai YS, Moles D, Matanguihan R. 1996. Control of long-term perfusion Chinese hamster ovary cell culture by glucose auxostat. Biotechnology Progress 12(1):100-109. Kretzmer G. 2002. Industrial processes with animal cells. Applied Microbiology and Biotechnology 59(2-3):135-142. Leelavatcharamas V, Emery AN, Al-Rubeai M. 1999. Use of cell cycle analysis to characterise growth and interferon-gamma production in perfusion culture of CHO cells. Cytotechnology 30(1-3):59-69. Leist CH, Meyer HP, Fiechter A. 1990. Potential and Problems of Animal-Cells in Suspension-Culture. Journal of Biotechnology 15(1-2):1-46. Lim AC, Washbrook J, Titchener-Hooker NJ, Farid SS. 2006. A computer-aided approach to compare the production economics of fed-batch and perfusion culture under uncertainty. Biotechnol Bioeng 93(4):687-97. Lin J, Takagi M, Qu Y, Gao P, Yoshida T. 1999. Enhanced monoclonal antibody production by gradual increase of osmotic pressure. Cytotechnology 29:27-33. Link T, Backstrom M, Graham R, Essers R, Zorner K, Gatgens J, Burchell J, Taylor- Papadimitriou J, Hansson GC, Noll T. 2004. Bioprocess development for the production of a recombinant MUC1 fusion protein expressed by CHO-K1 cells in protein-free medium. Journal of Biotechnology 110(1):51-62. Litwin J. 1992. The growth of Vero cells in suspension as cell-aggregates in serum-free media. Cytotechnology 10(2):169-74. Liu H, Liu XM, Li SC, Wu BC, Ye LL, Wang QW, Chen ZL. 2009. A high-yield and scaleable adenovirus vector production process based on high density perfusion culture of HEK 293 cells as suspended aggregates. J Biosci Bioeng 107(5):524-9.  24 Liu XM, Liu H, Wu BC, Li SC, Ye LL, Wang QW, Huang PT, Chen ZL. 2006. Suspended aggregates as an immobilization mode for high-density perfusion culture of HEK 293 cells in a stirred tank bioreactor. Appl Microbiol Biotechnol 72(6):1144-51. Loignon M, Perret S, Kelly J, Boulais D, Cass B, Bisson L, Afkhamizarreh F, Durocher Y. 2008. Stable high volumetric production of glycosylated human recombinant IFNalpha2b in HEK293 cells. BMC Biotechnol 8:65. May RM. 2004. Uses and abuses of mathematics in biology. Science 303(5659):790-793. Mercille S, Johnson M, Lanthier S, Kamen AA, Massie B. 2000. Understanding factors that limit the productivity of suspension-based perfusion cultures operated at high medium renewal rates. Biotechnology and Bioengineering 67(4):435-450. Miller WM, Wilke CR, Blanch HW. 1989. Transient responses of hybridoma cells to nutrient additions in continuous culture: I. Glucose pulse and step changes. Biotechnol Bioeng 33(4):477-86. Miller WM, Wilke CR, Blanch HW. 1988. Transient Responses of Hybridoma Cells to Lactate and Ammonia Pulse and Step Changes in Continuous Culture. Bioprocess Engineering 3(3):113-122. Moreira JL, Alves PM, Aunins JG, Carrondo MJ. 1995. Hydrodynamic effects on BHK cells grown as suspended natural aggregates. Biotechnol Bioeng 46(4):351-60. Moreira JL, Alves PM, Rodrigues JM, Cruz PE, Aunins JG, Carrondo MJ. 1994a. Studies of baby hamster kidney natural cell aggregation in suspended batch cultures. Ann N Y Acad Sci 745:122-33. Moreira JL, Feliciano AS, Santana PC, Cruz PE, Aunins JG, Carrondo MJ. 1994b. Repeated-batch cultures of baby hamster kidney cell aggregates in stirred vessels. Cytotechnology 15(1-3):337-49. Mullin R. 2004. A new battlefield in biologics. Chemical & Engineering News 82(19):26-27. Nivitchanyong T, Martinez A, Ishaque A, Murphy JE, Konstantinov K, Betenbaugh MJ, Thrift J. 2007. Anti-apoptotic genes Aven and E1B-19K enhance performance of BHK cells engineered to express recombinant factor VIII in batch and low perfusion cell culture. Biotechnol Bioeng 98(4):825-41. Ozturk SS. 1996. Engineering challenges in high density cell culture systems. Cytotechnology 22(1-3):3-16. Ozturk SS, Palsson BO. 1990. Effect of Medium Osmolarity on Hybridoma Growth, Metabolism and Antibody Production. Biotechnology and Bioengineering 37:989-993.  25 Paalme T, Kahru A, Elken R, Vanatalu K, Tiisma K, Vilu R. 1995. The computer- controlled continuous culture of Escherichia coli with smooth change of dilution rate (A- stat). Journal of Microbiological Methods 24(2):145-153. Peshwa MV, Kyung YS, Mcclure DB, Hu WS. 1993. Cultivation of Mammalian-Cells as Aggregates in Bioreactors - Effect of Calcium-Concentration on Spatial-Distribution of Viability. Biotechnology and Bioengineering 41(2):179-187. Pissarra PD. 2004. Changes in the business of culture. Nature Biotechnology 22(11):1355-1356. Reddy S, Miller WM. 1994. Effects of abrupt and gradual osmotic stress on antibody production and content in hybridoma cells that differ in production kinetics. Biotechnol Prog 10(2):165-73. Reiter CJ, Piret JM. 2004. pH Effects on CHO Cultures and Transient Scanning. Vancouver: University of British Columbia. Renner WA, Jordan M, Eppenberger HM, Leist C. 1993. Cell-cell adhesion and aggregation: Influence on the growth behavior of CHO cells. Biotechnol Bioeng 41(2):188-93. Reuveny S, Velez D, Macmillan JD, Miller L. 1986a. Factors affecting cell growth and monoclonal antibody production in stirred reactors. J Immunol Methods 86(1):53-9. Reuveny S, Velez D, Miller L, Macmillan JD. 1986b. Comparison of Cell Propagation Methods for Their Effect on Monoclonal-Antibody Yield in Fermenters. Journal of Immunological Methods 86(1):61-69. Ronning OW, Schartum M, Winsnes A, Lindberg G. 1991. Growth limitation in hybridoma cell cultures: the role of inhibitory or toxic metabolites. Cytotechnology 7(1):15-24. Ryll T, Dutina G, Reyes A, Gunson J, Krummen L, Etcheverry T. 2000. Performance of small-scale CHO perfusion cultures using an acoustic cell filtration device for cell retention: Characterization of separation efficiency and impact of perfusion on product quality. Biotechnology and Bioengineering 69(4):440-449. Sen A, Kallos MS, Behie LA. 2001. Effects of hydrodynamics on cultures of mammalian neural stem cell aggregates in suspension bioreactors. Industrial & Engineering Chemistry Research 40(23):5350-5357. Shen D, Sharfstein ST. 2006. Genome-Wide Analysis of the Transcriptional Response of Murine Hybridomas to Osmotic Shock. Biotechnol Bioeng 93(1):132-145. Smith CG, Guillaume JM, Greenfield PF, Randerson DH. 1991. Experience in Scale-up of Homogeneous Perfusion Culture for Hybridomas. Bioprocess Engineering 6(5):213- 219.  26 Spier RE. 1991. Large-Scale Mammalian-Cell Culture - Methods, Applications and Products. Current Opinion in Biotechnology 2(3):375-379. Takazawa Y, Tokashiki M. 1989. High cell density perfusion culture of mouse-human hybridomas. Applied Microbiology and Biotechnology 32(1):280-284. Varley J, Birch J. 1999. Reactor design for large scale suspension animal cell culture. Cytotechnology 29(3):177-205. Versyck KJ, Claes JE, VanImpe JF. 1997. Practical identification of unstructured growth kinetics by application of optimal experimental design. Biotechnology Progress 13(5):524-531. Vits H, Hu WS. 1992. Fluctuations in Continuous Mammalian-Cell Bioreactors with Retention. Biotechnology Progress 8(5):397-403. Voisard D, Meuwly F, Ruffieux PA, Baer G, Kadouri A. 2003. Potential of cell retention techniques for large-scale high-density perfusion culture of suspended mammalian cells. Biotechnol Bioeng 82(7):751-65. Werner RG, Walz F, Noe W, Konrad A. 1992. Safety and Economic-Aspects of Continuous Mammalian-Cell Culture. Journal of Biotechnology 22(1-2):51-68. Woodside SM, Bowen BD, Piret JM. 1998. Mammalian cell retention devices for stirred perfusion bioreactors. Cytotechnology 28(1-3):163-175. Wu MH, Dimopoulos G, Mantalaris A, Varley J. 2004. The effect of hyperosmotic pressure on antibody production and gene expression in the GS-NS0 cell line. Biotechnol Appl Biochem 40(Pt 1):41-6. Wurm FM. 2004. Production of recombinant protein therapeutics in cultivated mammalian cells. Nature Biotechnology 22(11):1393-1398. Xing Z, Li Z, Chow V, Lee SS. 2008. Identifying inhibitory threshold values of repressing metabolites in CHO cell culture using multivariate analysis methods. Biotechnol Prog 24(3):675-83. Yang JD, Angelillo Y, Chaudhry M, Goldenberg C, Goldenberg DM. 2000. Achievement of high cell density and high antibody productivity by a controlled-fed perfusion bioreactor process. Biotechnology and Bioengineering 69(1):74-82. Yang M, Butler M. 2000. Effects of ammonia on CHO cell growth, erythropoietin production, and glycosylation. Biotechnology and Bioengineering 68(4):370-380. Zeng AP, Deckwer WD. 1999. Model simulation and analysis of perfusion culture of mammalian cells at high cell density. Biotechnology Progress 15(3):373-382.  27 Zhang S, Handacorrigan A, Spier RE. 1993. A Comparison of Oxygenation Methods for High-Density Perfusion Cultures of Animal-Cells. Biotechnology and Bioengineering 41(7):685-692. Zhu MM, Goyal A, Rank DL, Gupta SK, Vanden Boom T, Lee SS. 2005. Effects of elevated pCO2 and osmolality on growth of CHO cells and production of antibody-fusion protein B1: a case study. Biotechnol Prog 21(1):70-7.  28 3 Influence of Culture Medium Composition and Osmolality on Interferon-alpha 2b Production Stability in HEK293 Cells1 3.1 Introduction The stability of recombinant production cell lines is an important element in successful cell culture processes (Barnes et al. 2003), especially where scale-up requires many cell divisions.  Clonal drift or changes in the cell population over many cell divisions can affect the cell specific production rates, which may also have an impact on product quality (Guarna et al. 1995).  For example, the appearance of a cell subpopulation with a high growth rate and reduced recombinant protein expression may eventually outgrow the original production cell line affecting titers and productivity (Morrison et al. 1997). Cell line and recombinant protein expression stability is even more critical during perfusion where cells are maintained in culture for months.  Substantial efforts such as the evaluation of extracellular factors (medium composition, etc.) have been dedicated to understanding the cellular mechanisms of instability (Barnes and Dickson 2006), especially since cell line stability issues can be a critical challenge in securing clinical product approval.  The human embryonic kidney 293 (HEK293) cell line (Graham et al. 1977) provides the potential to express recombinant proteins with human post-translational modifications. These cells were adapted to suspension culture in serum-free media, thus enabling their  1 A version of this chapter has been submitted for publication.  Drouin H, Lanthier S, Piret JM, Kamen A and Durocher Y, (submitted), Influence of Culture Medium Osmolality on Interferon-alpha 2b Production Stability in HEK293 Cells.  29 use for large-scale culture (Cote et al. 1998).  Recently, these HEK293 cells have been engineered to produce high levels of human interferon-alpha 2b (IFN-α2b) (Loignon et al. 2008).  Interferon is a protein used in the treatment of viral diseases and some types of cancers (Bajetta et al. 2006; Dezube 2000; Kaluz et al. 1994; Mohanty et al. 2006; Vogel 1999).  Therapeutic interferon is mainly produced in E. coli (Babu et al. 2000; Pestka et al. 1987; Srivastava et al. 2005; Srivastava and Mukherjee 2005).  However, bacterially produced recombinant human IFNα2b is not properly folded and requires a refolding process that could reduce specific activity and overall recovery yield (Loignon et al. 2008).  A HEK293 cell line stably producing high levels of biologically active and glycosylated interferon may thus represent an efficient alternative to current expression platforms.  The impact of osmolality on cell culture has been studied extensively.  Standard mammalian culture media have osmolalities between 260 and 320 mOsm kg-1 similar to the osmolality of mammalian serum (290 to 300 mOsm kg-1) (Ozturk and Palsson 1990). A mechanistic study of hyperosmotic stress based on gene expression found that it caused many significant cellular responses related to ion transport: accumulation of osmolytes, cell cycle distribution, proliferation, cytoskeletal organization and cell metabolism (Shen and Sharfstein 2006).  Hyperosmotic conditions are known to increase cell specific protein productivity but, on the other hand, they also induce cellular stresses and reduce growth rates (Kim et al. 2002; Lin et al. 1999; Ozturk and Palsson 1990; Wu et al. 2004). Because the osmolality of a culture medium can be readily increased, it can be an attractive method of increasing production yields.  However, culture conditions have to  30 be adjusted carefully in order to avoid side effects that could reduce overall culture performance such as final viable cell and product concentrations.  Biphasic batch culture strategies have therefore been developed to increase the culture osmolality level only after an optimal cell concentration has been reached (Kim et al. 2002).  Other strategies have focused on adapting cells to a gradual increase in the culture osmolality (Lin et al. 1999; Reddy and Miller 1994).  Most of these studies have been performed on batch time scales (up to 10 days) and so the long-term effects of high osmolality are not well documented.  Moreover, many specialized media used in cell cultures have been developed with high osmolality.  In this chapter, the effects on perfusion culture cell growth and productivity from varying the medium composition and increased osmolality (such as could be experienced during fed-batch or perfusion cultures) was studied using HEK293 cells expressing recombinant IFN-α2b as a model system.  The cells were initially passaged in batch cultures for extended periods of time using three media at normal or high osmolality, with or without selective pressure, to assess production stability.  Subsequently, two culture medium conditions were selected and evaluated in perfusion cultures to confirm the effects of stable and unstable conditions.  Thus, it was possible to evaluate the potential benefit of these culture conditions for the stable production of IFN-α2b in perfusion culture.   31 3.2 Materials and Methods 3.2.1 Cell Line and Media HEK293 cells engineered to express recombinant human interferon alpha 2b (IFN-α2b) (Loignon et al. 2008) (HEK293-IFN) were maintained in suspension culture using three different media: i) Freestyle293 (GIBCO, Grand Island, NY); ii) low calcium hybridoma serum-free medium (LC-SFM); and iii) modified low calcium serum free medium (MLC- SFM).  Freestyle293 is a serum-free, chemically-defined commercial medium supplemented with GlutaMaxTM (a stable dipeptide form of glutamine).  LC-SFM is a transformed version of the commercial hybridoma serum-free medium (HSFM, GIBCO) that was custom-made with a low calcium content (Cote et al. 1998).  MLC-SFM was formulated with 12.5 g L-1 of LC-SFM basal powder preparation (~2 times the concentration normally used in LC-SFM).  In some experiments using LC-SFM and MLC-SFM, the pressure selective antibiotics G418 sulfate (Wisent, St-Bruno, QC, Canada) at 50 mg L-1 and blasticidin (InvivoGen, San Diego, CA, USA) at 5 mg L-1 were added.  The LC-SFM and MLC-SFM media were also supplemented with 0.1% (w/v) bovine serum albumin (BSA #A7030; Sigma, St. Louis, MO, USA), 1% (v/v) chemically defined lipids (GIBCO) and 0.1% Pluronic F68.  Table 3-1 summarizes the levels of additives and nutrients.  The osmolality was adjusted to 300 ± 10 mOsm kg-1 (Normal) or 375 ± 10 mOsm kg-1 (High) by adding sodium chloride (NaCl).  32   Table 3-1:  Partial composition of the media used based on the additives and amino acid analysis. These commercial media are proprietary formulations, and so they contain other unknown components.   Media   Freestyle293 LC-SFM MLC-SFM Osmolality (mOsm kg-1) 290 ± 10 300 ± 10 300 ± 10 Glucose (g L-1) 4.5  4.5 4.5 BSA 30%  (mL L-1) - 3.33 3.33 Lipids 1000x  (mL L-1) - 1 1 Pluronic F-68 (g L-1) 1 1 1 NaCl added (g L-1) - 4.58 2.05 Asp (mM) 0.86 0.06 0.13 Glu (mM) 0.06 0.09 0.16 Ser (mM) 3.51 0.50 0.96 Asn (mM) 0.28 0.38 0.63 Gly (mM) 0.00 0.23 0.46 Gln (mM) - 5.36 5.96 GlutaMax (mM) 4.10 - - His (mM) 0.26 0.32 0.63 Thr (mM) 0.95 1.16 2.15 Arg (mM) 1.92 1.10 2.19 Ala (mM) 0.00 0.07 0.12 Pro (mM) 0.17 2.15 3.67 Tyr (mM) 0.35 0.52 1.00 Cys (mM) 0.31 0.36 0.71 Val (mM) 1.51 1.20 2.27 Met (mM) 0.69 0.33 0.65 Ile (mM) 1.43 1.34 2.53 Leu (mM) 2.15 1.37 2.71 Lys (mM) 1.30 1.27 2.44 Phe (mM) 0.57 0.59 1.16  3.2.2 Cell Culture Experiments The HEK293-IFN cells were initially cultured in Freestyle293 medium and amplified to generate a Working Cell Bank (WCB).  For every experiment, cells were thawed from the WCB and transferred to a MLC-SFM static culture for 24 hours.  Then, a 125 mL shake flask (Corning, Corning, NY, USA) was inoculated at 0.30 x 106 cells mL-1 with the contents of the static culture and placed on a shaking platform at 115 rpm in a  33 humidified incubator controlled at 37 oC and 5% CO2.  Subsequent batch cultures were performed in 125 mL shake flasks with 30 mL working volumes and inoculated also at 0.30 x 106 cells mL-1.  Batches were cultivated for 8 days with daily sampling.  Long- term maintenance cultures were diluted with fresh medium every 2-3 days to keep the viable cell concentration between 0.3 x 106 and 1.0 x 106 cells mL-1.  These cultures were then maintained for several weeks and samples removed at various times to assess cell densities, titers and productivities.  3.2.3 Perfusion Culture Operation The perfusion experiments were performed in a 3.5 L Chemap bioreactor (type SG, Mannedorf, Switzerland) with a 2.73 ± 0.03 L working volume.  The schematic of the system is presented in Figure 3-1.  Agitation was provided by 2 pitched-blade impellers (3 blades, 45o) with a speed starting at 75 rpm and then raised to 100 rpm after 2.8 days. The ratio of impeller diameter to vessel diameter was 0.5.  The mixing was enhanced by 3 surface baffles placed at 120o.  A Chemap FZ-2000 control unit was used to control the agitation and the temperature (37oC).  The pH was monitored using a gel electrode (Mettler Toledo, Wilmington, MA, USA) while the dissolved oxygen (DO) was measured with a polarographic electrode (Ingold, Andover, MA, USA) both placed close to the impeller.  The pH and the DO were controlled via a SAFE8000 interface (Control Microsystems, Kanata, ON, Canada, operating with FIX Dmacs and MMI software Intellution, Norwood, MA, USA) at a pH of 7.1 and a DO of 30 – 40% air saturation as previously described (Kamen et al. 1996).  Oxygen was directly sparged into the bioreactor culture through a polypropylene sparger with a pore size of 250 µm (Porex  34 Technologies, Fairburn, GA, USA).  The sparging flow rate was constant and operated in a pulse mode.  CO2 was added to lower the pH while a solution of 7.5 % NaHCO3 was added to increase the pH.   Figure 3-1.  Schematic of the perfusion system.   35 The bioreactor was inoculated (at 0.40 x 106 cells mL-1) with HEK293-IFN cells from one 2L Erlenmeyer shake flask (500 mL working volume) and run in batch mode for 1 day.  Perfusion was then initiated at a constant feed rate with the media described in Table 3-1.  One or two 60 mL bioreactor samples were taken daily.  During the first 3 days, oxygen was provided through surface aeration and then by sparging after day 3.  A separate bleed line was used to periodically remove the culture (spent medium, viable and dead cells) at a constant rate after day 3 (see Figure 3-1).  The culture volume was kept at a constant level using an optical-based sensor (Electromatic S-System, Ville St-Laurent, QC, Canada) that triggered operation of the harvest pump for flow into a harvest bottle. The medium feed was manually adjusted while the cell bleed and separator operation were controlled via the SAFE8000 interface.  For the perfusion cultures, the bioreactor was equipped with an acoustic cell separator for cell retention (BioSep 1015, AppliSens, division of Applikon, Schiedam, Netherlands). The cell suspension was pumped into the 10L acoustic separator (2.1 MHz operating frequency) and the cells were retained in the chamber by the acoustic forces (Drouin et al. 2007).  The aggregated cells settled back to the bioreactor and the clarified liquid was pumped out of the system at a selected harvest flow rate.  The operation of the separator was controlled with a BioSep ADI 1015 electronic controller (AppliSens).  The acoustic separator was operated with air backflush every 30 minutes (Gorenflo et al. 2003).  The alternation between the stop time and the run time (called the duty cycle) was set to 55 s of run time with 5 s of stop time.  The perfusion cell separation efficiency (defined as the percentage of viable cells retained) was maintained at ~95% during the entire culture.  36 Peristaltic pumps (Masterflex L/S, Cole Parmer, Vernon Hills, IL, USA) were used for the feed, bleed, recirculation, harvest and backflush.  The flow rate in the external recirculation loop was constant.  Flow rates are presented as volume per bioreactor volume per day (vvd).  3.2.4 Cell Specific Rate Equations Cell culture productivities were evaluated by calculating the cell specific productivity and yield for each different batch.  The equations used to calculate the specific production rates were determined using material balances based on the viable cell concentration (Xv). The cell mass balance, over a constant volume, is given by € dXv dt = µappXv   (3-1) where t is the time (day) and  is the apparent specific growth rate (day -1).  The apparent specific growth rate is defined as: € µapp = µ − kd  (3-2) where  is the true specific growth rate (day -1) and  is the specific death rate (day -1). An interferon (IFN) balance around the bioreactor yields the following equation: € dIFN dt = qIFNXv   (3-3) where IFN is the product concentration (mg L-1) and qIFN is the cell specific production rate (pg cell-1 day-1).   37 The yield calculation was defined as the final IFN-α2b concentration per maximum viable cell concentration and expressed in pg cell-1.  This last calculation was used to compare batch productivities.  3.3 Analytical Methods 3.3.1 Cell Concentration and Viability The cell number and viability were counted in a hemocytometer (Hausser Scientific, Horsham, PA, USA) under a microscope. Because of cell aggregation in culture, a 500 µL culture aliquot was placed in a 6 mL polystyrene tube and agitated to disperse the cell clumps.  The use of this approach had the effect of decreasing the apparent cell viability; therefore, the viability percentage was estimated separately without dispersing the cells using erythrosin B exclusion (Garnier et al. 1994).  Large aggregates (diameter > ~100 µm) were excluded from the cell count since they greatly increased the cell count variability given the small sample volumes measured.  For batch culture sampling, one hemocytometer count was used for viability estimation while two were used to determine the total cell number.  For perfusion culture sampling, three hemocytometer counts were used for the viability measurement while six were used to assess total cell number. Multiple hemocytometer readings allowed the statistical analysis of the viable cell concentration measurements, presented in terms of standard deviations in the plots.   38 3.3.2 Culture Medium Analyses Culture samples were spun down for 3 min at 1000 rpm to remove the cells and the supernatants were frozen at -80oC for later analysis.  The glucose, lactate and ammonia concentrations were measured using the IBI Biolyser Rapid Analysis Systems (Kodak, New Haven, CT).  The osmolality was measured using an osmometer (Advanced Instruments Micro Osmometer Model 3300, Needham Heights, MA, USA).  Amino acids were analyzed by HPLC (WATERS Milford, MA, USA) (Kamen et al. 1991).  IFN-α2b was measured using a commercial ELISA kit (PBL InterferonSource, Piscataway, NJ, USA).  An internal standard was added to every ELISA plate to assess the variation of the assay as well as to provide a reference to standardize the results between assays.  3.4 Results and Discussion 3.4.1 Batch Culture Performance of 293-IFN Cells in Three Culture Media HEK293-IFN cells were batch cultivated in Freestyle293, LC-SFM or MLC-SFM (Table 3-1).  The HEK293-IFN cell line was initially cloned in Freestyle293 medium, but, because of cost considerations, this medium was not used for the perfusion process. Therefore, for the subsequent perfusion cultures, the cell line was tested only in LC-SFM and MLC-SFM media.  In the batch culture using the Freestyle293 medium, the viable cell concentration reached over 3 x 106 cells mL-1 and then the viability rapidly decreased (Figure 3-2A).  The IFN-α2b concentration increased steadily up to 168 mg L-1, with growth-associated  39 production kinetics (Figure 3-2B).  The cell growth limitation was most likely due to nutrient depletion.  Glucose was depleted after day 6 while the serine level was still over 1 mM after the day 8 (Figure 3-2C).  All amino acids were measured but only serine appeared to be limiting.  This trend has been repeatedly observed in independent batch cultures.  Previous HEK293 cell studies have shown that 20 mM lactate can reduce viability and 40 mM lactate can reduce protein production (Nadeau et al. 1996). Ammonia concentrations around 5 mM also have been reported to decrease cell specific productivity (Ferreira et al. 2005).  In this study, lactate reached 30 mM at day 4 (Figure 3-2D) and the cells continued to grow at a reduced rate.  The ammonia concentration reached a value over 3 mM late in the cell death phase (Figure 3-2D).  Thus, it is not likely that the ammonia accumulation limited cell growth.  40   Figure 3-2.  Viable cell density and viability (A), interferon (B), glucose and serine (C), and lactate and ammonia (D) concentration profiles for HEK293-IFN cells in batch culture using Freestyle293 medium. The osmolality level was ~290 ± 10 mOsm kg-1 for the entire culture.  Error bars represent the standard deviation of the repeat measurements.  In the LC-SFM batch culture, the viable cell concentration reached a plateau after 3 days at 1.70 x106 cells mL-1 (Figure 3-3A).  The viability was over 90 % for the first 7 days of the culture.  The interferon concentration increased during both the growth and stationary phases to reach 85 mg L-1 (Figure 3-3B).  The cell growth arrest corresponds to the depletion of the amino acid serine, while glucose was not limiting (Figure 3-3C). However, serine depletion is not clearly represented on Figure 3-3C because its concentration was not measured at day 3 or 4 and no sample was frozen for later amino acid measurement.  HEK293 cells were found to have a linear consumption of serine during the exponential phase.  This has been observed regularly in multiple batches using different media (data not shown).  Therefore, extrapolating the initial values on Figure  41 3-3C (dashed line) suggests that the serine level was depleted at day 3.  A high glucose concentration may have contributed to maintain the stationary phase culture viable and stable compared to the Freestyle293 batch results.  Since the metabolites did not reach inhibitory levels (lactate up to 12.8 mM and ammonia up to 3.18 mM, Figure 3-3D), nutrient depletion was believed to be the main limitation.   Figure 3-3.  Viable cell density and viability (A), interferon (B), glucose and serine (C), lactate and ammonia (D) concentration profiles for HEK293-IFN cells in batch culture using LC-SFM medium. Dashed line represent extrapolated concentration since no measurements were taken between day 2 and day 6 (C).  Osmolality was maintained at ~305 ± 10 mOsm kg-1 for the entire culture.  Error bars represent the standard deviation of the repeat measurements.  In the MLC-SFM batch culture, the viable cell concentration increased up to 2.69 x106 cells mL-1 (Figure 3-4A).  The viability started to decrease more rapidly shortly after the maximal viable cell concentration was reached.  The interferon concentration increased up to 147 mg L-1 with growth-associated production kinetics (Figure 3-4B). The cell growth arrest corresponds to the depletion of the amino acid serine; the glucose  42 concentration did not become limiting (Figure 3-4C).  The lactate concentration reached 16.0 mM and the ammonia concentration 3.34 mM (Figure 3-4D).  Thus, the lactate and ammonia levels were not considered inhibitory and nutrient depletion is likely to be the main factor limiting cell growth.   Figure 3-4.  Viable cell density and viability (A), interferon (B), glucose and serine (C), lactate and ammonia (D) concentration profiles for HEK293-IFN cells in batch culture using MLC-SFM medium. Osmolality was maintained at ~300 ± 9 mOsm kg-1 during culture.  Error bars represent the standard deviation of the repeat measurements.  Apparent cell growth rates and cell specific interferon production rates were calculated at different times during the batch cultures (Figure 3-5).  Overall, the reductions in growth rate correlated with a net decrease in the cell specific interferon production rates demonstrating predominantly growth-associated production kinetics.  Cell specific glucose consumption and cell specific lactate and ammonia production rates had similar trends, though lactate was consumed late in the cultures (Figure 3-4D).  43   Figure 3-5.  Apparent cell specific growth rates (A) and cell specific interferon production rates (B) during HEK293-IFN cell batch culture in Freestyle293, LC-SFM and MLC-SFM media. Negative values correspond to consumption rates and positive values to production rates.  Yield calculations for the Freestyle, MLC-SFM and LC-SFM cultures gave very similar results (55.2, 54.4 and 49.9 pg cell-1, respectively).  Interferon production by HEK293 is growth-associated (Figure 3-5) and Freestyle293, a medium specifically developed for HEK293, yielded the highest viable cell density resulting in the highest level of IFN-α2b. The level of glucose and glutamine (GlutaMaxTM in the Freestyle medium) were similar for all 3 media.  This suggests that other nutrients (e.g. amino acids) and/or vitamins/trace elements may have a significant impact on growth.  3.4.2 Effect of Osmolality on Cell Production Stability The stability of a cell line is defined here as its ability to produce a recombinant protein at a similar yield per cell in a culture process over a long period of time independent of the passage number.  The effect of osmolality on cell production stability was studied by comparing the 3 media at Normal (~300 mOsm kg-1) and High (NaCl supplemented to ~375 mOsm kg-1) osmolality.   44 3.4.2.1 Freestyle293 Medium In the Freestyle293 medium at 'Normal' osmolality, HEK293-IFN cells maintained, for more than 4 weeks of culture, a constant interferon yield per cell (final IFN-α2b per maximum viable cell number).  The initial batch was inoculated at week 0 with cells from the maintenance culture, had a yield per cell of 37.7 pg cell-1 while the batch inoculated at week 4, from the same maintenance culture, reached 40.1 pg cell-1 (Table 3-2).  The exponential growth rates were similar as well: 0.622 day-1 for the initial batch compared to 0.672 day-1 after 4 weeks.  However, increasing the medium osmolality to 378 mOsm kg-1 ('High' osmolality) decreased the maximum cell concentration to 1.83 x106 cells mL-1 (or by 40%) compared to ~ 3 x106 cells/mL for Freestyle293 'Normal' osmolality medium (Figure 3-2).  Since the production of IFN nonetheless reached 164 mg L-1, the yield of interferon was increased over 2-fold at High vs. Normal osmolality (Table 3-2).  However, the cells maintained in this High osmolality medium were not able to grow for more than 2 weeks (Figure 3-6) and the yield of interferon per cell also dropped to a very low level.  Thus, interferon production in Freestyle293 High osmolality medium was not possible for an extended period of time, as the culture was unstable.  Table 3-2: Yield of interferon concentration on viable cell concentration for different batches started at 0, 2 and 4 weeks in Freestyle293, LC-SFM and MLC-SFM media for different osmolality. Osmolality levels correspond to 300 ± 10 (normal), 330 ± 5  (intermediate) and 375 ± 10 (high) mOsm kg-1. ND: Not Determined.  Freestyle LC-SFM MLC-SFM Culture Time Normal High Normal High Normal Intermediate High week  pg cell-1  pg cell-1  pg cell-1  pg cell-1  pg cell-1  pg cell-1  pg cell-1 0 37.7 90.4 60.0 62.2 41.4 51.5 80.6 2 ND 0.5 51.4 ND 55.3 ND 53.5 4 40.1 - 51.9 28.0 53.0 58.3 49.3  45   Figure 3-6.  Apparent growth rates of HEK293 cells maintained in Freestyle medium at normal (290 mOsm kg-1) and high (380 mOsm kg-1) osmolality. Apparent growth rates were calculated between each dilution of long-term maintenance cultures.  3.4.2.2 LC-SFM Medium The LC-SFM medium culture at 'Normal' osmolality also maintained a constant interferon yield per cell for more than 4 weeks of culture in maintenance (Table 3-2) with an average yield of 54.4 pg cell-1 (standard deviation of 4.8 pg cell-1).  Exponential growth rates were likewise similar, with an initial rate of 0.722 day-1 compared to 0.696 day-1 after 4 weeks.  At 'High' osmolality and after 4 weeks, the batch yield was decreased by 55% compared with the initial batch (both inoculated with cells from the maintenance culture).  The yield continued decreasing so that after 8 weeks, the batch yield was 29% of the initial batch (part of the selective pressure experiment below). Exponential growth rates were around 0.653 day-1 with a standard deviation of 0.063 day-1 for batches at Normal and High osmolality.   46 3.4.2.3 MLC-SFM Medium In this study, the MLC-SFM medium was also tested at an intermediate osmolality, i.e., Normal (300 mOsm kg-1), Intermediate (330 mOsm kg-1) and High (371 mOsm kg-1) osmolalities.  At 'Normal' osmolality, batch yields were maintained constant for more than 4 weeks of culture (Table 3-2).  Prolonged cultures maintained in MLC-SFM were confirmed to be stable over 10 weeks in culture (part of the selective pressure experiment below).  Cells exposed to 'Intermediate' osmolality were stable for the 4 weeks tested. For the 'High' osmolality batch, the initial yield was significantly increased up to 80.6 pg cell-1, ~2-fold compared to ‘Normal’ osmolality (Table 3-2).  However, the yield decreased at 2 and 4 weeks, returning to the levels obtained at Normal and Intermediate osmolality.  Exponential growth rates were similar for all batches at different osmolality with an average of 0.647 day-1 (standard deviation of 0.051 day-1).  3.4.2.4 Osmotic Stress The different results obtained in the Normal and High osmolality media are consistent with the results reported when an osmotic stress is applied (Kim et al. 2002; Lin et al. 1999; Ozturk and Palsson 1990; Wu et al. 2004).  In Freestyle293 medium, the osmotic stress resulted in a doubling of the cell-specific yield compared to that obtained under normal conditions (Table 3-2).  However, the maximal viable cell concentration was reduced by almost the same factor.  In the MLC-SFM cultures, there was almost a 2-fold increase in yield (Table 3-2).  The intermediate osmolality, though more marginally increasing the yield compared to the normal osmolality level, had the same tendency.  47 Interestingly, the results for the LC-SFM at increased osmolality did not exhibit this trend.  3.4.3 Effect of Selective Pressure The cultures in the previous section were all performed using selection (blasticidin and G418 sulfate).  As the presence of selective agents may prevent some instability by killing cells that lose the expression cassette, we determined the production stability of cells cultivated with and without selection.  From previous results, LC-SFM at 'High' osmolality and MLC-SFM at 'Normal' osmolality were used as examples of media unable or able to sustain IFN-α2b productivity, respectively.  Cells exposed to the High osmolality LC-SFM medium experienced a sharper decrease in yield without selection than with selection (Figure 3-7A).  After 8 weeks in the High osmolality medium, the batch performed with cells maintained without selection had 13% of the initial yield compared to 29% with those maintained with selection (Figure 3-7A). The trend in the results of this experiment were replicated but for a shorter period of time due to a temperature control failure of the incubator (data not shown).  The mechanism responsible for this decrease in productivity is not known but it is possible that High osmolality may increase the genetic instability of the cell line or somehow favor promoter silencing. Batches performed using MLC-SFM at Normal osmolality experienced no drop in productivity whether they were maintained with or without selection (Figure 3-7B).  Therefore, this HEK293 cell line stably produces IFN-α2b  48 without selection.  This would be a marked advantage when cultivating cells in a perfusion culture where a large amount of selective agent addition would be needed.   Figure 3-7.  Yield of interferon concentration on viable cell concentration for batches started from maintenance cultures at different times and cultivated in LC-SFM medium at high osmolality (375mOsm kg-1) (A) or MLC-SFM medium at normal osmolality (300 mOsm kg-1) (B) using or not using selective pressure.  3.4.4 Perfusion Culture Results To further validate the trends observed in the batch cultures for the different medium compositions, 2 perfusion cultures were performed.  The first was operated at High osmolality to test for instability during perfusion culture while the second was operated at Normal osmolality to validate that the cell line could be cultivated in perfusion without production instability.  For the perfusion culture using LC-SFM medium at 'High' osmolality, the viable cell concentration increased over 9 days and then reached a pseudo-steady state (Figure 3-8). The glucose concentration was maintained at 4.9 mM (+/- 0.7 mM) and lactate at 16.6 mM (+/- 1.2 mM) (data not shown).  The interferon concentration gradually decreased after the cell accumulation phase.  49   Figure 3-8.  Viable cell and interferon concentration profiles during perfusion culture using High osmolality LC-SFM medium (375 mOsm kg-1). Feed rate was set to 0.510 vvd and bleed rate to 0.039 vvd.  Error bars represent standard deviations of measurements.  The perfusion culture using MLC-SFM at 'Normal' osmolality reached a pseudo-steady state after 8 days (Figure 3-9).  Viable cell and interferon concentrations were both relatively constant.  The culture was continued for more than 50 days without showing any signs of instability (data not shown).  This culture demonstrated the potential of HEK-293 cells for stable interferon production in a perfusion culture process.  50   Figure 3-9.  Viable cell and interferon concentration profiles during perfusion culture using Normal osmolality MLC-SFM (300 mOsm kg-1). Feed rate was set to 0.523 vvd and bleed rate to 0.050 vvd.  Error bars represent standard deviations of measurements.  Aggregated cells larger than 100 µm of diameter were not considered in the calculation of cell concentration.  3.5 Conclusions HEK293-IFN cells were readily adapted to suspension culture in 3 different media: Freestyle293, LC-SFM and MLC-SFM.  Whereas hyperosmotic stress (~ 375 mOsm kg-1) of the HEK293-IFN cells increased IFN-α2b production in initial batches of Feestyle293, LC-SFM and MLC-SFM media, the interferon production was not stably maintained over extended periods of culture.  Even in the case of the MLC- SFM medium, where the yields were increased initially, the production rates returned to the normal osmolality production levels within 4 weeks.  Therefore, although hyperosmotic conditions can increase the productivity in the short-term, these benefits were not sustained over longer-term cultures.  The presence of selection agents did not decrease stability at elevated osmolality.  Maintaining the osmolality at approximately 300 mOsm kg-1 (i.e. 'Normal') provided the most stable production pattern for all 3 of the  51 media investigated.  It is, therefore, recommended in long-term culture processes to monitor and avoid increases in the osmolality, such as by the use of reduced osmolality level medium.   52 3.6 References Babu KR, Swaminathan S, Marten S, Khanna N, Rinas U. 2000. Production of interferon-alpha in high cell density cultures of recombinant Escherichia coli and its single step purification from refolded inclusion body proteins. Appl Microbiol Biotechnol 53(6):655-60. Bajetta E, Del Vecchio M, Nova P, Fusi A, Daponte A, Sertoli MR, Queirolo P, Taveggia P, Bernengo MG, Legha SS and others. 2006. Multicenter phase III randomized trial of polychemotherapy (CVD regimen) versus the same chemotherapy (CT) plus subcutaneous interleukin-2 and interferon-alpha2b in metastatic melanoma. Ann Oncol 17(4):571-7. Barnes LM, Bentley CM, Dickson AJ. 2003. Stability of protein production from recombinant mammalian cells. Biotechnol Bioeng 81(6):631-9. Barnes LM, Dickson AJ. 2006. Mammalian cell factories for efficient and stable protein expression. Curr Opin Biotechnol 17(4):381-6. Cote J, Garnier A, Massie B, Kamen A. 1998. Serum-free production of recombinant proteins and adenoviral vectors by 293SF-3F6 cells. Biotechnol Bioeng 59(5):567-75. Dezube BJ. 2000. New therapies for the treatment of AIDS-related Kaposi sarcoma. Curr Opin Oncol 12(5):445-9. Drouin H, Ritter JB, Gorenflo VM, Bowen BD, Piret JM. 2007. Cell separator operation within temperature ranges to minimize effects on Chinese hamster ovary cell perfusion culture. Biotechnol Prog 23(6):1473-84. Ferreira TB, Ferreira AL, Carrondo MJ, Alves PM. 2005. Two different serum-free media and osmolality effect upon human 293 cell growth and adenovirus production. Biotechnol Lett 27(22):1809-13. Garnier A, Cote J, Nadeau I, Kamen A, Massie B. 1994. Scale-up of the adenovirus expression system for the production of recombinant protein in human 293S cells. Cytotechnology 15(1-3):145-55. Gorenflo VM, Angepat S, Bowen BD, Piret JM. 2003. Optimization of an acoustic cell filter with a novel air-backflush system. Biotechnol Prog 19(1):30-6. Graham FL, Smiley J, Russell WC, Nairn R. 1977. Characteristics of a human cell line transformed by DNA from human adenovirus type 5. J Gen Virol 36(1):59-74. Guarna MM, Fann CH, Busby SJ, Walker KM, Kilburn DG, Piret JM. 1995. Effect of cDNA copy number on secretion rate of activated protein C. Biotechnol Bioeng 46(1):22- 7.  53 Kaluz S, Kabat P, Gibadulinova A, Vojtassak J, Fuchsberger N, Kontsek P. 1994. Interferon alpha2b is the predominant subvariant detected in human genomic DNAs. Acta Virol 38(2):101-4. Kamen AA, Bédard C, Tom R, Perret S, Jardin B. 1996. On-line monitoring of respiration in recombinant-baculovirus-infected and uninfected insect cell bioreactor cultures. Biotechnology and Bioengineering 50:34-48. Kamen AA, Tom R, Caron AW, Chavarie C, Massie B, Archambault J. 1991. Culture of insect cells in a helical ribbon impeller bioreactor. Biotechnol Bioeng 38:619-628. Kim MS, Kim NS, Sung YH, Lee GM. 2002. Biphasic culture strategy based on hyperosmotic pressure for improved humanized antibody production in Chinese hamster ovary cell culture. In Vitro Cell Dev Biol Anim 38(6):314-9. Lin J, Takagi M, Qu Y, Gao P, Yoshida T. 1999. Enhanced monoclonal antibody production by gradual increase of osmotic pressure. Cytotechnology 29:27-33. Loignon M, Perret S, Kelly J, Boulais D, Cass B, Bisson L, Afkhamizarreh F, Durocher Y. 2008. Stable high volumetric production of glycosylated human recombinant IFNalpha2b in HEK293 cells. BMC Biotechnol 8:65. Mohanty SR, Kupfer SS, Khiani V. 2006. Treatment of chronic hepatitis B. Nat Clin Pract Gastroenterol Hepatol 3(8):446-58. Morrison CJ, McMaster WR, Piret JM. 1997. Differential Stability of Proteolytically Active and Inactive Recombinant Metalloproteinase in Chinese Hamster Ovary Cells. Biotechnol Bioeng 53(6):594 - 600. Nadeau I, Garnier A, Côté J, Massie B, Chavarie C, Kamen A. 1996. Improvement of Recombinant ProteinProduction with the Human Adenovirus/293S Expression System Using Fed-Batch Strategies. Biotechnol Bioeng 51:613-623. Ozturk SS, Palsson BO. 1990. Effect of Medium Osmolarity on Hybridoma Growth, Metabolism and Antibody Production. Biotechnology and Bioengineering 37:989-993. Pestka S, Langer JA, Zoon KC, Samuel CE. 1987. Interferons and their actions. Annu Rev Biochem 56:727-77. Reddy S, Miller WM. 1994. Effects of abrupt and gradual osmotic stress on antibody production and content in hybridoma cells that differ in production kinetics. Biotechnol Prog 10(2):165-73. Shen D, Sharfstein ST. 2006. Genome-Wide Analysis of the Transcriptional Response of Murine Hybridomas to Osmotic Shock. Biotechnol Bioeng 93(1):132-145.  54 Srivastava P, Bhattacharaya P, Pandey G, Mukherjee KJ. 2005. Overexpression and purification of recombinant human interferon alpha2b in Escherichia coli. Protein Expr Purif 41(2):313-22. Srivastava P, Mukherjee KJ. 2005. Kinetic studies of recombinant human interferon- alpha (rhIFN-α) expression in transient state continous cultures. Biochemical Engineering Journal 26:50-58. Vogel W. 1999. Treatment of acute hepatitis C virus infection. J Hepatol 31 Suppl 1:189- 92. Wu MH, Dimopoulos G, Mantalaris A, Varley J. 2004. The effect of hyperosmotic pressure on antibody production and gene expression in the GS-NS0 cell line. Biotechnol Appl Biochem 40(Pt 1):41-6.     55 4 The Influence of Reduced Perfusion Rates on HEK Cell Interferon-alpha 2b Production in Perfusion Cultures2  4.1 Introduction Many authors have compared the advantages of using perfusion vs. batch or fed-batch culture (Griffiths 1992; Kadouri and Spier 1997; Konstantinov et al. 2006; Smith et al. 1991; Werner et al. 1992).  Increasingly, industrial practice has tended to favor fed-batch cultures (Heath and Kiss 2007), especially for antibody production.  Perfusion cultures were usually chosen for their ability to provide high volumetric productivity and stable operation for long durations (Cortin et al. 2004; Dalm et al. 2004; Hu and Piret 1992; Woodside et al. 1998).  Another marked advantage of perfusion process is the short residence time that allows the production of unstable products (Griffiths 1992; Ryll et al. 2000; Woodside et al. 1998; Wurm 2004).  The perfusion feed and bleed rates are key operational parameters that influence the maintenance of a high cell concentration in perfusion cultures (Dalm et al. 2004; Zeng and Deckwer 1999).  Generally, an increased feed rate can positively affect the viable cell concentration (Dalm et al. 2004; Dong et al. 2005; Vits and Hu 1992; Zeng and Deckwer 1999) and, in some cases, the production as well (Mercille et al. 2000).  Alternatively, a high feed rate can also negatively impact the process in the form of incomplete medium utilization, system instability and most importantly, reduced product titer (Banik and  2 A version of this chapter will be submitted for publication.  Drouin H, Lanthier S, Durocher Y, Kamen A and Piret JM, The Influence of Reduced Perfusion Rates on HEK Cell Interferon-alpha 2b Production in Perfusion Cultures.  56 Heath 1995; Konstantinov et al. 1996).  The cell bleed rate is usually employed to avoid the build-up of dead cells, thereby increasing the percent viability of the culture (Banik and Heath 1994; Banik and Heath 1995; Dalm et al. 2004; Zeng and Deckwer 1999). The bleed rate also helps maintain steady state by controlling the viable cell concentration (Buntemeyer et al. 1991; Dalm et al. 2004; Drouin et al. 2007; Kempken et al. 1991).  Most perfusion cultures reported have used non-growth-associated production cell lines (Banik and Heath 1994; Banik and Heath 1995; Chen et al. 2004; Dalm et al. 2007; Dalm et al. 2004; Konstantinov et al. 2006; Smith et al. 1991; Takazawa and Tokashiki 1989; Vits and Hu 1992).  In this case, the process strategy aims to maximize the viable cell concentration to improve performance (Tang et al. 2007).  Zeng et al. (1999) modeled a perfusion culture at different bleed and perfusion rates and suggested that non-growth- associated production cell line bleed rates should be minimized.  Similarly, Banik et al. (1995) determined that the optimal perfusion strategy to maximize MAb production was to maximize viable cell concentration by using a low bleed rate.  Non-growth-associated production should be more energetically efficient at a low specific growth rate and a high cell density.  Leelavatcharamas et al. (1999) reported on perfusion culture of a growth- associated production cell line. They performed perfusion cultures without bleed and concluded that their system was not suitable for growth-associated production CHO cells.  In high cell concentration culture systems, the high feed rate required to provide sufficient nutrients and to remove inhibitory by-products can decrease the product titer as  57 well as induce process instabilities, such as cell retention failure due to the high perfusion rate (Konstantinov et al. 1996).  In order to overcome the negative effects of a high dilution rate, perfusion process strategies favor a low volumetric perfusion rate.  A simple strategy to reduce the perfusion rate is to reduce the bioreactor temperature, and then decrease the feed rate to provide the smaller amount of nutrients needed (Angepat et al. 2005; Chen et al. 2004).  The success of this approach is dependent on the temperature response of the cell line.  Konstantinov et al. (2006) described the «push-to-low» approach, where the feed rate was decreased gradually and the medium enriched as needed.  The minimal dilution rate was established accordingly to the maximal residence time suitable for an unstable product.  The maximal dilution rate was determined by the physical limitations of the system (maximum cell retention capacity).  The bleed rate was automatically adjusted to maintain a high viable cell concentration (~2 x 107 cells mL-1). This culture strategy was developed for cell lines with non-growth-associated or inversely growth-associated production kinetics.  Goudar et al. (2006) increased the product titer and reduced the medium cost using a similar approach for a non-growth- associated production CHO cell line.  At low dilution rates, waste products can build up in the bioreactor.  Lactate and ammonia are usually the most common waste products known to affect culture performance (Banik and Heath 1994; Banik and Heath 1995; Burteau et al. 2003; Dong et al. 2005; Kempken et al. 1991; Lehmann et al. 1990; Mercille et al. 2000; Smith et al. 1991; Yang et al. 2000; Zhang et al. 1993).  For mammalian cells, concentrations over 3 – 10 mM of ammonia or/and 1.98 – 4.95 g L-1 of lactate have been reported to be toxic  58 (Bibila and Robinson 1995; Yang et al. 2000).  In some cases, cultures have been seen to crash, possibly because of other unidentified metabolites (Konstantinov et al. 2006; Ronning et al. 1991; Yang et al. 2000; Zeng et al. 1998).  To determine the optimal culture conditions, most experimental strategies are based on reaching a steady state, i.e., cell, glucose and lactate concentrations with a standard deviation of less than 5% (Gambhir et al. 2003), before the operating conditions are changed.  A method using non- steady-state transient responses or a 'transient scanning approach' was developed to assist the process of optimizing perfusion cultures by not waiting for steady-state responses to be obtained (Angepat et al. 2005).  This method relied instead on the shorter-term trends obtained from the transient responses to significantly accelerate process optimization, with a reported ~60% reduction in development time.  In this study, the effect of using low perfusion rates on productivity was investigated. Building on the findings from Chapter 3, the perfusion rates of media feed at a 'Normal' osmolality of 300 mOsm kg-1 were varied in a transient fashion.  A first perfusion culture was performed using transient steps to evaluate different low feeding rate modes.  A simple mathematical model was used to predict results that were compared with the perfusion results.  Semi-continuous and batch cultures were carried out to better understand the low perfusion culture and provide information to develop an adapted perfusion process strategy.  This last strategy was applied to a second perfusion culture and the results were compared to the first culture.  The ultimate process objective was to increase product titer in the bioreactor.   59 4.2 Materials and Methods 4.2.1 Cell Line and Media Human embryonic kidney cells (HEK293) were first isolated in 1977 (Graham et al. 1977) and later adapted to grow in suspension culture (Cote et al. 1998).  More recently, 293-6E cells were engineered to express recombinant human interferon alpha 2b (IFN-α2b) (Loignon et al. 2008).  Two media were used for these experiments, which were derived from the low calcium serum-free medium (LC-SFM) used in Chapter 3.  LC-SFM is a modified version of the commercial hybridoma serum-free medium (H-SFM by GIBCO, Grand Island, NY) and custom-made with a low calcium content (Cote et al. 1998).  The first medium was a modified low calcium serum-free medium (MLC-SFM) and the second medium was an enriched low calcium serum free medium (ELC-SFM).  MLC-SFM and ELC-SFM were formulated using 12.5 g L-1 of LC-SFM basal powder (~2 times the amount used in LC- SFM).  Selective pressure additives were added to both media at 50 mg L-1 of G418 sulfate (Wisent, St-Bruno, QC, Canada) and 5 mg L-1 of blasticidin (InvivoGen, San Diego, CA, USA).  Both media were supplemented with 0.1% (w/v) bovine serum albumin (BSA; Intergen, Purchase, NY, USA), 1% (v/v) chemically defined lipids (GIBCO) and 0.1% Pluronic F68.  Table 4-1 summarizes the levels of additives and nutrients.  The essential and non-essential amino acids added to the ELC-SFM were from commercial preparations (Sigma).  Therefore, because the amino acids are present at different concentrations in these preparations, the amino acid ratios of the ELC-SFM and  60 the MLC-SFM were varied (Table 4-1).  The osmolality of both media was adjusted to 300 ± 10 mOsm kg-1 by adding sodium chloride.  Table 4-1: Partial composition of the media used based on the additives and amino acid analysis. These commercial media are proprietary formulations, and so they contain other unknown components.   Media   MLC-SFM ELC-SFM Ratio Osmolarity (mOsm/kg) 300 ± 10 300 ± 10 - Glucose (g/L) 4.5 4.5 1.00 BSA 30%  (mL/L) 3.33 3.33 1.00 Lipids 1000x  (mL/L) 1 1 1.00 Pluronic F-68 (g/L) 1 1 1.00 Asp (mM) 0.13 0.46 3.54 Glu (mM) 0.16 0.53 3.31 Ser (mM) 0.96 1.32 1.38 Asn (mM) 0.63 1.04 1.65 Gly (mM) 0.46 0.81 1.76 Gln (mM) 5.96 6.02 1.01 His (mM) 0.63 0.88 1.40 Thr (mM) 2.15 2.64 1.23 Arg (mM) 2.19 2.82 1.29 Ala (mM) 0.12 0.50 4.17 Pro (mM) 3.67 4.46 1.22 Tyr (mM) 1.00 1.25 1.25 Cys (mM) 0.71 0.85 1.20 Val (mM) 2.27 2.81 1.24 Met (mM) 0.65 0.78 1.20 Ile (mM) 2.53 3.03 1.20 Leu (mM) 2.71 3.21 1.18 Lys (mM) 2.44 2.91 1.19 Phe (mM) 1.16 1.40 1.21  4.2.2 Cell Culture Experiments The cells were grown initially in Freestyle293 medium (Invitrogen, Grand Island, NY, USA) and frozen to make a Working Cell Bank (WCB).  For every experiment, cells were thawed from this WCB and transferred into a MLC-SFM or ELC-SFM static culture for 24 hours.  Then, a 125 mL shake flask (Corning, Corning, NY, USA) was inoculated  61 with the contents of the static culture and placed on a rotary plate at 115 rpm in a humidified incubator controlled at 37 oC and 5% CO2.  Subsequent batch cultures were performed in a 125 mL shake flask with 30 mL working volume, inoculated at 0.30 x 106 cells mL-1.  Batches were maintained for 8 days with daily sampling.  4.2.3 Semi-Continuous Culture Experiments Semi-continuous cultures were performed in 125 mL shake flasks with a 20 mL working volume, inoculated at 0.30 x 106 cells mL-1.  The shake flasks were kept in a humidified incubator controlled at 37 oC and 5% CO2 on a rotary plate at 115 rpm.  Experiments were run with daily sampling, where a fraction of the cell suspension was removed and replaced by fresh medium every day, until a pseudo-steady state was reached (viable cell density stabilized for 4 days) normally at ~14 days.  Experiments were also carried out with cell retention where the removed fraction of the cell suspension was spun down for 5 min at 1000 rpm, the spent medium removed and the cells re-suspended in fresh medium before they were returned to the culture.  4.2.4 Perfusion Culture Operation The perfusion experiments were performed in a 3.5 L SG Chemap bioreactor (Mannedorf, Switzerland) with a 2.73 ± 0.03 L working volume.  The schematic of the system is presented in Figure 4-1.  Agitation was provided by 2 pitched-blade impellers (3 blades, 45o) with speeds starting at 75 rpm and then raised to 100 rpm after 2.8 days. The ratio of impeller to vessel diameter was 0.5.  The mixing was enhanced by 3 surface  62 baffles placed at 120o.  The Chemap FZ-2000 control unit was used to control the agitation and the temperature (37oC).  The pH was monitored using a gel electrode (Mettler Toledo, Wilmington, MA, USA) while the DO was measured with a polarographic electrode (Ingold, Andover, MA, USA).  The pH and the DO were controlled via a SAFE8000 interface (Control Microsystems, Kanata, ON, Canada, operating with FIX Dmacs and MMI software Intellution, Norwood, MA, USA) at a pH of 7.1 and at a DO of 30 – 40% of air saturation as previously described (Kamen et al. 1996).  Oxygen was directly sparged in the bioreactor culture through a polypropylene sparger with a pore size of 250 µm (Porex Technologies, Fairburn, GA, USA).  The sparging flow rate was constant and operated in a pulse mode.  CO2 was added to lower the pH, while a solution of 7.5 % NaHCO3 was added to increase the pH.   63  Figure 4-1:  Schematic of the perfusion system.  The bioreactor was inoculated (at 0.40 x 106 cells mL-1) with HEK293-IFN cells from one 2L Erlenmeyer shake flask (500 mL working volume) and run in batch mode for 1 day.  Perfusion was then initiated at a constant feed rate with the media described in Table 4-1.  Bioreactor samples were taken daily.  During the first 3 days, oxygen was provided through surface aeration and then by sparging after day 3.  A separate bleed line  64 was used to periodically remove the culture (spent medium, viable and dead cells) at a constant rate after day 3 (see Figure 4-1).  The culture volume was kept at a constant level using an optical-based sensor (Electromatic S-System, Ville St-Laurent, QC, Canada) that triggered operation of the harvest pump.  The medium feed was manually adjusted while the cell bleed and separator operation were controlled via the SAFE8000 interface.  For the perfusion cultures the bioreactor was equipped with an acoustic cell separator for cell retention (BioSep 1015, AppliSens, division of Applikon, Schiedam, Netherlands). The cell suspension was pumped into the 10L acoustic separator (2.1 MHz operating frequency) and the cells were retained in the chamber by the acoustic forces (Drouin et al. 2007).  The aggregated cells settled back to the bioreactor and the clarified liquid was pumped out of the system at a selected harvest flow rate.  The operation of the separator was controlled with a BioSep ADI 1015 electronic controller (AppliSens).  The acoustic separator was operated with air backflush every 30 minutes (Gorenflo et al. 2003).  The alternation between the stop time and the run time (called the duty cycle) was set to 55 s of run time with 5 s of stop time.  The perfusion cell separation efficiency (defined as the percentage of viable cells retained) was maintained at ~95% during the entire culture. Peristaltic pumps (Masterflex L/S, Cole Parmer, Vernon Hills, IL, USA) were used for the feed, bleed, recirculation, harvest and backflush.  The flow rate in the external recirculation loop was constant.  Flow rates are presented as volume per bioreactor volume per day (vvd).   65 Two perfusion cultures were performed: one using MLC-SFM and the second with ELC- SFM.  The bleed rate was increased from 0.042 day-1 for the first perfusion culture to 0.167 day-1 for the second experiment.  After perfusion was initiated, operational settings were kept constant in order to allow stabilization of the viable cell concentration for both cultures (i.e., pseudo-steady state was reached).  4.2.4.1 Perfusion Operation For the first perfusion culture (low perfusion rate culture), after a pseudo-steady state was reached with feeding mode A, the transient feed modes were then changed every 5 days (Table 4-2).  The reduction in dilution rate was targeted to be around 10%, matched by 10% increases in the nutrient content of the feed.  The overall content of the medium was increased 10%, except for BSA, lipids and pluronic that were kept constant.  The sodium chloride concentration was adjusted to maintain a fixed osmolality, ~300 mOsm kg-1. Nutrient feed rate increases are reported here as an average of all the other medium components relative to their feed rate during the pseudo-steady state (mode A, Table 4-2).  This feed rate was based on the mass of nutrients provided to the culture. Though there were up to ~10% variations, the objective was for the medium concentration and the feed rate to be adjusted so that the amount of nutrients provided remained approximately the same as during feeding mode A, thus a value of 1.00.  66  Table 4-2:  Bioreactor conditions investigated in the first perfusion experiment. Bleed rate was started at t = 4.0 day and kept constant at a rate of 0.042 day-1.  Experimental variations, from measurement variability and apparatus reading errors, are represented by “±” in the table. Transient Feeding Mode Medium Concentration Target Medium Concentration Measured (relative to 1.0x) Dilution Rate  (day-1) Medium Component Feed Rate  (relative to 1.0x) A 1.0x 1.00 ± 0.02 0.523 1.00 ± 0.07 B 1.1x 1.09 ± 0.03 0.493 1.02 ± 0.06 C 1.2x 1.18 ± 0.02 0.432 0.98 ± 0.06 D 1.3x 1.25 ± 0.03 0.382 0.91 ± 0.04 E 1.4x 1.36 ± 0.03 0.342 0.88 ± 0.05   The bleed rate was increased from 0.042 day-1 for the first perfusion culture to 0.167 day-1 for the second experiment.  For the second perfusion culture experiment (adapted perfusion culture), again after a pseudo-steady state had been reached with mode A, the first feed mode change lasted 5 days and the two subsequent modes were continued for 7 days (Table 4-3).  The medium used for the first perfusion was MLC- SFM and for the second perfusion culture, ELC-SFM (Table 4-1).  Table 4-3:  Bioreactor conditions investigated in the second perfusion experiment. Bleed rate was started at t = 4.2 day and kept constant at a rate of 0.167 day-1.  Experimental errors are represented by “±” in the table. Transient Feeding Mode Medium Concentration Target Medium Concentration Measured (relative to 1.0x) Dilution Rate  (day-1) Medium Component Feed Rate  (relative to 1.0x) A 1.00x 1.00 ± 0.02 0.569 1.00 ± 0.04 B 1.15x 1.10 ± 0.02 0.535 1.04 ± 0.03 C 1.30x 1.24 ± 0.03 0.402 0.88 ± 0.02 D 1.45x 1.36 ± 0.04 0.356 0.86 ± 0.02    67 4.2.4.2 Model Development and Equations A simple mathematical model based on the conservation of mass was developed to simulate the perfusion culture responses.  This model predicted the evolution of viable cell concentration, nutrient consumption and metabolite production assuming that the cell specific growth, consumption and production rates were constant, based on average rates calculated during the two initial phases of the culture (growth and equilibrium phases). In order to keep a simple model and obtain a clear idea of the effect of an operational change, the equations did not include a Monod-type dependence of substrate limited growth rate.  The differential equations that govern the mass balance were solved using a non-adaptative Runge-Kutta method of order 4 programmed in MatLab 6.1 (The MathWorks Inc, Natick, MA, USA).  The change of rates was induced by the depletion of the amino acid serine.  When the serine concentration reached a level lower than 0.05 mM, exponential rates (calculated based on growth phase data) were programmed to change to equilibrium rates (calculated based on pseudo-steady-state or equilibrium phase data).  The equations used to calculate the specific uptake and production rates were determined using material balances based on the viable cell concentration ( € Xv ).  First, the viable cell mass balance, over a constant volume, is given by € dXv dt = µappXv − B ⋅ Xv V − H 1− SE( )Xv V   (4-1) where t is the time (day),  the apparent specific growth rate (day -1), V the bioreactor volume (L), B the bleed flow rate (L day-1), H the harvest flow rate (L day-1) and SE the  68 separation efficiency of the cell retention device (Gorenflo et al. 2005).  The apparent specific growth rate is defined as: € µapp = µ − kd  (4-2) where  is the true specific growth rate (day -1) and  the specific death rate (day -1). The perfusion or dilution rate (D) is calculated from: € D = H + BV . (4-3) Finally, nutrient and metabolite balances around the bioreactor yield the following equations: € dS dt = −qsXv + D Sin − S( )   (4-4) € dP dt = qpXv + D Pin − P( )   (4-5) where S is the nutrient concentration (mM), P the metabolite concentration (mM) or product concentration (mg L-1), Sin is the nutrient concentration of the feed stream, Pin is the metabolite or product concentration of the feed stream, and qs and qp are the specific consumption and production rates (mmol cell-1 day-1).  4.2.5 Analytical Methods 4.2.5.1 Cell Concentration and Viability The total cell number and viability were counted in a hemocytometer (Hausser Scientific, Horsham, PA, USA) under a microscope. Because of cell aggregation in culture, a 500 µL culture aliquot was placed in a 6 mL polystyrene tube and agitated to disperse the  69 cell clumps.  The use of this approach had the effect of decreasing the apparent cell viability; therefore, the viability percentage was estimated separately without dispersing the cells using erythrosin B exclusion (Garnier et al. 1994).  However, large aggregates (diameter > ~100 µm) were excluded from the count since these could have created cell count errors by increasing falsely cell concentration considering the small volume measured.  For batch and semi-continuous culture sampling, one hemocytometer count was used for viability estimation while two were used to determine the total cell number. For perfusion culture sampling, three hemocytometer counts were used for the viability measurement while six were used to assess total cell number.  Multiple hemocytometer readings allowed the statistical analysis of the viable cell concentration measurements, presented in terms of standard deviations in the plots.  4.2.5.2 Culture Medium Analyses Culture samples were centrifuged for 3 min at 1000 rpm to remove the cells and supernatants were frozen at -80oC for later analysis.  The glucose, lactate and ammonium concentrations were measured using the IBI Biolyser Rapid Analysis System (Kodak, New Haven, CT, USA).  The osmolality was measured using an osmometer (Advanced Instruments Micro Osmometer Model 3300, Needham Heights, MA, USA).  Amino acids were analyzed by HPLC (Waters Milford, MA, USA) (Kamen et al. 1991).  IFN-α2b was measured using an ELISA kit (PBL Biomedical Laboratories, Piscataway, NJ, USA).  An internal standard was added to every ELISA plate to provide a reference to standardize the results between assays.   70 4.3 Results and Discussion 4.3.1 Low Perfusion Rate Culture Experiment – Low Bleed Rate Culture The perfusion culture experiment was designed to investigate reduced perfusion rate process performance using a 'transient scanning approach' (Angepat et al. 2005).  This approach enabled the evaluation of 4 different low feeding strategies compared to the pseudo-steady-state results reached with feeding mode A (Table 4-2).  The length of each transient feeding mode was selected to be 5 days based on literature and culture observations.  Steady state could be reached after more than 10 days in perfusion operation and the changes in the culture response during the last days is usually not pronounced compared to the changes during the first days following the culture modification.  Therefore, the transient approach was designed to study the major trends, e.g., the response in the cell concentration, viability and productivity under the new perfusion conditions.  It was then possible to improve process development performance without a need for long-term steady-state experiments.  However, it is suggested that a small number of steady-state production runs be performed in order to confirm the transient results before larger scale productions are carried out.  The cell growth profile was divided into two phases: the exponential growth and stationary phases.  This last phase was characterized by the limitation of one amino acid, serine, found to be the growth-limiting nutrient for HEK293-IFN using MLC-SFM (Drouin et al. Submitted-b).  The operating settings used during the experiment are described in Table 4-2.  The perfusion culture stabilized, reaching a pseudo-steady state that was maintained from t = 12.8 days until t = 17.7 days, while consuming all the serine  71 provided by the feed (mode A, Figure 4-2 and Figure 4-3).  The pseudo-steady state is defined as a stabilized viable cell concentration ( € Xv), less than 5% variation, for 5 consecutive days (mode A, Figure 4-2).  At pseudo-steady state, the € Xv  was around 5.3 x 106 cells mL-1 and the interferon-alpha 2b (IFN-α2b) stabilized around 260 mg L-1. The glucose concentration was maintained around 1.3 mM, by constant feeding and the serine concentration was close to 0 mM (mode A, Figure 4-3).  The lactate and ammonia concentrations were also at a pseudo-steady state, around 23.4 mM and 2.2 mM respectively (mode A, Figure 4-4).   Figure 4-2:  Evolution of viable cell and IFN-α2b concentrations during HEK293-IFN cell perfusion culture compared to model simulated results. Transient feeding modes A through E are defined in Table 4-2.  The error bars represent the standard deviations of the repeat measurements.   72  Figure 4-3:  Evolution of glucose and serine concentrations during HEK293-IFN cell perfusion culture compared to model simulated results. Transient feeding modes A through E are defined in Table 4-2.   Figure 4-4: Evolution of lactate and ammonia concentrations during HEK293 cell perfusion culture compared to model simulated results. Transient feeding modes A through E are defined in Table 4-2.  After the pseudo-steady state, the perfusion rate was decreased by 6% and feed nutrient concentration was increased by ~10% (mode B, Table 4-2).  The € Xv  started to decrease slowly while the product concentration did not change significantly.  A simple mathematical model based on constant cell specific rates was used to analyze the effects of the feeding changes.  A comparison between the expected results and actual measurements did not reveal any significant differences for the first transient that may be  73 partly due to the large variation in the IFN-α2b measurement compared to the small operational change.  However, it was clear during the second transient (mode C), that IFN-α2b was not accumulating as predicted by the model to higher levels in the bioreactor at the reduced dilution rate.  Subsequent transients yielded even lower concentrations, reaching 185 mg L-1 at the end of the last transient (mode E), showing the opposite trend predicted by the model.  Nonetheless, the lactate and ammonia ions did accumulate at values close to the model predictions.  For the last two modes (D and E), the medium component feed rates were decreased, relative to the pseudo-steady state (Table 4-2), probably causing aggravated nutrient limitations.  Table 4-4 shows clearly the decreased apparent growth rate and the decreased cell specific productivity.  Table 4-4:  Summary of calculated cell specific rates for the first perfusion culture. Transient feeding modes are described in Table 4-2. Mode µapp qIFN - day-1  pg cell-1 day-1 A 0.071 26.0 B 0.064 25.0 C 0.057 21.4 D 0.057 17.5 E 0.043 14.7  4.3.2 Kinetic Profile of HEK293-IFN Cells To better understand why the product titer was not concentrated in the first perfusion culture, semi-continuous experiments were performed to determine the kinetic profile of the HEK293 cells producing IFN-α2b using a similar feed strategy.  Small-scale semi- continuous cultures provide a practical means to approximate perfusion cultures (Angepat et al. 2005; Henry et al. 2008) and enable rapid exploration of multiple process  74 conditions.  One experiment was performed using 5 shake flasks at 5 different feed/bleed rates and media (MLC-SFM, Table 4-1 and Table 4-2).  Alternatively, semi-continuous cultures using cell retention were performed using the same bleed rate, but with 5 different feed rates and media (MLC-SFM, Table 4-1 and Table 4-2).  All semi- continuous cultures were continued until a pseudo-steady state was obtained.  This pseudo-steady state was characterized by at least 4 consecutive days with a similar viable cell concentration.  All cultures reached a pseudo-steady state and calculations using the average values from this pseudo-steady-state period (data not shown) were performed. Calculated specific INF-α2b production rates were plotted against growth rates (Figure 4-5).  At pseudo-steady state, the growth rate was equal to the bleed rate.   Figure 4-5 : Relation between specific production rate (qp) and the growth rate for semi-continuous cultures with and without cell retention. The dashed line represents linear trend line with the calculated R-squared value.  There was a clear correlation between growth and production, typical of a growth- associated production cell line (R2=0.9655).  The experiment with cell retention yielded similar results.  Thus, the IFN-α2b production of the HEK293-IFN cell line is very sensitive to changes in growth rates.  The decrease in productivity experienced in the first  75 perfusion culture could have resulted from the decrease in growth rate at reduced perfusion rates.  For a growth-associated producer, it is essential to maintain high growth rates.  4.3.3 Identification of Possible Limitations at Low Perfusion Rate Two additional factors were explored, metabolite inhibitions and nutrient limitations, to help further in guiding process optimization.  During the perfusion culture pseudo-steady state, harvested liquid from the culture was centrifuged to remove cells and other debris, and kept at 4oC (spent medium).  Batch cultures of HEK293-IFN cells were carried out using: i) The spent medium; ii) Enriched spent medium consisting of 50% spent medium and 50% fresh MLC-SFM (Table 4-1); and iii) Fresh MLC-SFM as a control.  Cells cultivated in fresh medium grew 30% faster than the cells cultivated in spent medium (Table 4-5).  The higher nutrient concentrations in the enriched spent medium did not influence the growth rate.  Cell specific IFN-α2b production was therefore significantly higher for the fresh medium culture, where growth rate was higher and maintained for a longer period.  The higher nutrient concentration in the enriched spent medium allowed cells to grow after the exponential phase at a slower rate but for a longer period of time compared to the spent medium culture (data not shown).  This situation could explain the slightly higher IFN-α2b concentration in the enriched spent medium culture. Nutrients did not affect the value of the cell specific growth rates during the exponential phase, but were necessary to maintain cell growth.  Considering that both  76 spent medium cultures did not grow the same as the control, metabolite inhibitions were believed to be the main growth limitation.  Table 4-5 :  Summary of calculated cell specific rates for the batch cultures performed with spent, enriched spent and fresh media. Spent media was harvested during the pseudo-steady state of the first perfusion culture (Mode A, Table 4-2).  Enriched spent medium was supplemented with half of the nutrients normally used with fresh media (MLC-SFM, Table 4-1).  Media  Spent Enriched Spent Fresh (1.0x) Maximum viable cell concentration (106 cells mL-1) 0.87  1.25 2.40 Exponential growth rate (day-1) 0.43 0.43 0.57  IFN-α2b production (pg cell-1 day-1) 10.0  14.0 32.2  Metabolite or by-product levels present in the spent media were found to be the principal cause of growth limitation in batch.  During the first perfusion culture, metabolite concentrations increased after each transient (Figure 4-4) due to the reduction of the dilution effect.  Therefore, metabolite inhibitions were believed to be the main factor that affected the apparent growth rate.  Moreover, medium component feed rates were lower for the last two transients compared the initial feed (Table 4-2).  This could explain the accentuated decrease of growth and production experienced during this period (modes D and E, Figure 4-2).  Perfusion culture success at low perfusion rate for a growth associated production cell line is related to the ability of maintaining a sufficient growth rate.  Low perfusion rate operation has then to be adjusted to allow adequate growth rate and prevent excessive metabolite build up.   77 4.3.4 Adapted Perfusion Process Strategy – Second Perfusion Culture In the case of growth-associated cell lines, it was hypothesized that process enhancement could be achieved by maximizing the growth rate (by increasing the bleed rate) and minimizing metabolite productions (by improving the nutrient balance in the medium feed).  A second perfusion culture was performed with this adapted process strategy.  The medium used was ELC-SFM (Table 4-1) where essential and non-essential amino acids were supplemented in larger quantities than for MLC-SFM.  Once a pseudo-steady state was reached after day 17, 3 transient feeding modes were tested to explore lower perfusion rate cultures (Table 4-3).  These transients were chosen to be longer because the medium concentration steps were greater, at 15% instead of 10% in the first perfusion culture.  Higher steps were expected to yield more significant changes.  Again, the growth profile was divided into two phases: exponential growth and stationary phases characterized by a pseudo-steady state where the viable cell concentration was stabilized (mode A, Figure 4-6).  Serine was the growth-limiting nutrient where all other amino acids were not depleted (data not shown).  This trend was observed in many independent cultures.  The perfusion culture stabilized reaching a pseudo-steady state where all serine provided was consumed (mode A, Figure 4-7).  Then the culture equilibrated from day 12 until day 17, where the viable cell concentration (Xv) stabilized for 5 consecutive days (mode A, Figure 4-6).  At pseudo-steady state, the Xv was around 5.9 x 106 cells mL-1 and the IFN-α2b stabilized around 229 mg L-1.  The glucose concentration was maintained around 3.8 mM with constant feeding and the serine concentration was steadily less than 0.10 mM (mode A, Figure 4-7).  Ammonia  78 concentrations were low at an average of 2.3 ± 0.2 mM (mode A, Figure 4-8).  The lactate concentration steadily increased after t = 10 days and its average value was 9.3 ± 0.6 mM (mode A, Figure 4-8).  The same simple model was also used to simulate the results assuming all specific rates were constant.  These results were then compared to the experimental data in order to understand the deviations from this model following each transient (Figure 4-6, Figure 4-7 and Figure 4-8).   Figure 4-6 :  Evolution of viable cell and IFN-α2b concentrations for the adapted strategy during second HEK293-IFN cell perfusion culture compared to model simulated results. Transient feeding modes A through D are defined in Table 4-3.  The error bars represent the standard deviations of the repeat measurements.  79   Figure 4-7 :  Evolution of glucose and serine concentrations for the adapted strategy during second HEK293-IFN cell perfusion culture compared to model simulated results. Transient feeding modes A through D are defined in Table 4-3.   Figure 4-8 : Evolution of lactate and ammonia concentrations for the adapted strategy during second HEK293-IFN cell perfusion culture compared to model simulated results. Transient feeding modes A through D are defined in Table 4-3.  The first transient was set to increase the medium concentration by 15% when compared to the pseudo-steady state modes (Table 4-3).  However, after analysis, the nutrient concentrations (serine as reference) were not as high as desired; only 10% more than the 1.00x medium.  This situation resulted from an analytical error that overestimated concentration results.  Two days before use, the enriched medium was prepared and the  80 feed concentration adjustment was done accordingly to the analytical results. Consequently, the feed rate was decreased by 6%.  No significant changes were observed in terms of the IFN-α2b concentration.  This could be caused by the small operational change compared to the experimental variation, but also because the IFN-α2b level might not have reached its true steady state during the first mode (mode A).  Most of the other curves followed similar trends compared to the model predictions.  The second transient did not show the expected results (mode C, Figure 4-6).  The feed rate was not adjusted according to the level of nutrients in the medium.  As a consequence, the medium component feed rate was 12% lower than the reference level and created a partial nutrient limitation (Table 4-3).  Also, IFN-α2b did not concentrate as predicted by the model.  For the last transient (mode D), the feed rate was adjusted to provide the same amount of nutrient as the previous transient.  The culture adapted to the new level of nutrient and the IFN-α2b concentration in the bioreactor reached 303.5 mg L-1, an increase of 35% when compared to the final pseudo-steady state value.  This last result confirms that it is possible to use a low perfusion rate strategy to concentrate IFN-α2b in the bioreactor for a growth-associated production cell line.  Calculations of the apparent growth rates during the different transient steps show similar results (Table 4-6).  No major growth inhibition was experienced by the culture.  81  Table 4-6 :  Summary of calculated cell specific rates for the adapted strategy perfusion culture. Transient feeding modes are described in the Table 4-3. Mode µapp qIFN - day-1  pg cell-1 day-1 A 0.182 21.4 B 0.179 20.7 C 0.179 15.3 D 0.178 17.7  4.3.5 Comparison of Perfusion Culture Strategies The increased bleed rate (from 0.042 day-1 adjusted to 0.167 day-1) in the second perfusion culture increased the growth rate and the use of a balanced medium i.e. ELC- SFM (Table 4-1) reduced metabolite productions.  In the adjusted strategy perfusion culture, the apparent growth rate was increased by 2.6-fold (Table 4-4 vs Table 4-6).  The lactate level was, for the entire culture, at least 50% lower when compared to the first perfusion culture, while the final ammonia concentration was reduced by 37%.  Another interesting difference was the viability during the culture.  In the first perfusion, the viability was around 79%.  A well-known effect of increasing the bleed rate is to increase the culture viability.  Therefore, the viability of the second perfusion was around 90%. The higher viability could reduce cell lysis (Heath and Kiss 2007) and thereby lower potential separator fouling problems (e.g. if a membrane was used).  The first perfusion yielded a lower apparent viable cell concentration, but higher IFN-α2b concentration than the second culture (5.3 x106 cells mL-1 and 260 mg L-1 vs. 6 x106 cells mL-1 and 229 mg L-1).  Moreover, the cell specific IFN-α2b production rate  82 was also higher despite a lower apparent growth rate (0.071 day-1 and 26.0 pg cell-1 day-1 vs. 0.182 day-1 and 21.4 pg cell-1 day-1).  Considering that HEK293 cells producing IFN- α2b are growth-associated producers, the cell specific productivity level should have been proportional to the growth rate.  However, cell aggregation was predominant in the first culture but almost absent in the second.  Large aggregates (>100 µm) were not considered for the cell concentration measurement; therefore true viable cell concentration was underestimated in the first perfusion culture (Drouin et al. Submitted-a).  At the beginning it was assumed that these large aggregates were not contributing significantly, but after analysis, the opposite was found to be true.  Lower aggregation in the second perfusion culture could be related to the adjusted strategy. Increasing the bleed rate continuously removed aggregates and prevented size growth. Also, the higher viability could have reduced aggregation because it was observed that dead cells could enhance the formation of clumped cells (Moreira et al. 1994).  Further work should be done to provide a better understanding of aggregation causes and effects in cell culture.  This is explored further in Chapter 5  4.4 Conclusions Most of the perfusion culture studies in the literature were performed using non-growth associated production cell lines.  In this study, HEK293-IFN cells were found to have growth-associated IFN-α2b production.  Therefore, an adapted culture strategy with increased bleed rates was developed in order to increase product titer.  Low perfusion rate operation was also explored for its potential to increase product titer while using a transient scanning approach to accelerate process development.  A simple mathematical  83 model based on constant cell specific rates was developed that reasonably well predicted the performance of perfusion cultures and helped compare the performance of the adapted feeding modes.  The development of an enriched medium, operating perfusion cultures with low perfusion rates and with an increased bleed rate increased productivity and reduced waste product accumulation as well as cell aggregation.  Moreover, the transient scanning technique provided an effective method for optimizing perfusion cultures within reduced experimental run time.   84 4.5 References Angepat S, Gorenflo VM, Piret JM. 2005. Accelerating perfusion process optimization by scanning non-steady-state responses. Biotechnol Bioeng 92(4):472-8. Babu KR, Swaminathan S, Marten S, Khanna N, Rinas U. 2000. Production of interferon- alpha in high cell density cultures of recombinant Escherichia coli and its single step purification from refolded inclusion body proteins. Appl Microbiol Biotechnol 53(6):655- 60. Bajetta E, Del Vecchio M, Nova P, Fusi A, Daponte A, Sertoli MR, Queirolo P, Taveggia P, Bernengo MG, Legha SS and others. 2006. Multicenter phase III randomized trial of polychemotherapy (CVD regimen) versus the same chemotherapy (CT) plus subcutaneous interleukin-2 and interferon-alpha2b in metastatic melanoma. Ann Oncol 17(4):571-7. Banik GG, Heath CA. 1994. An Investigation of Cell-Density Effects on Hybridoma Metabolism in a Homogeneous Perfusion Reactor. Bioprocess Engineering 11(6):229- 237. Banik GG, Heath CA. 1995. Hybridoma Growth and Antibody-Production as a Function of Cell-Density and Specific Growth-Rate in Perfusion Culture. Biotechnology and Bioengineering 48(3):289-300. Barnes LM, Bentley CM, Dickson AJ. 2003. Stability of protein production from recombinant mammalian cells. Biotechnol Bioeng 81(6):631-9. Barnes LM, Dickson AJ. 2006. Mammalian cell factories for efficient and stable protein expression. Curr Opin Biotechnol 17(4):381-6. Bibila TA, Robinson DK. 1995. In pursuit of the optimal fed-batch process for monoclonal antibody production. Biotechnol Prog 11(1):1-13. Buntemeyer H, Lutkemeyer D, Lehmann J. 1991. Optimization of serum-free fermentation processes for antibody production. Cytotechnology 5(1):57-67. Burteau CC, Verhoeye FR, Mols JF, Ballez JS, Agathos SN, Schneider YJ. 2003. Fortification of a protein-free cell culture medium with plant peptones improves cultivation and productivity of an interferon-gamma-producing CHO cell line. In Vitro Cellular & Developmental Biology-Animal 39(7):291-296. Chen Z-L, Wu B-C, Liu H, Liu X-M, Huang P-T. 2004. Temperature Shift as a Process Optimization Step for the Production of Pro-urokinase by a Recombinant Chinese Hamster Ovary Cell Line in High-Density Perfusion Culture. Journal of Bioscience and Bioengineering 97(4):239-243.  85 Cortin V, Thibault J, Jacob D, Garnier A. 2004. High-titer adenovirus vector production in 293S cell perfusion culture. Biotechnol Prog 20(3):858-63. Cote J, Garnier A, Massie B, Kamen A. 1998. Serum-free production of recombinant proteins and adenoviral vectors by 293SF-3F6 cells. Biotechnol Bioeng 59(5):567-75. Dalm MC, Lamers PP, Cuijten SM, Tjeerdsma AM, van Grunsven WM, Tramper J, Martens DE. 2007. Effect of feed and bleed rate on hybridoma cells in an acoustic perfusion bioreactor: metabolic analysis. Biotechnol Prog 23(3):560-9. Dalm MCF, Cuijten SMR, van Grunsven WMJ, Tramper J, Martens DE. 2004. Effect of feed and bleed rate on hybridoma cells in an acoustic perfusion bioreactor: Part I. Cell density, viability, and cell-cycle distribution. Biotechnology and Bioengineering 88(5):547-557. Dezube BJ. 2000. New therapies for the treatment of AIDS-related Kaposi sarcoma. Curr Opin Oncol 12(5):445-9. Dong H, Tang YJ, Ohashi R, Hamel JF. 2005. A perfusion culture system using a stirred ceramic membrane reactor for hyperproduction of IgG2a monoclonal antibody by hybridoma cells. Biotechnol Prog 21(1):140-7. Drouin H, Ansorge S, Lanthier S, Durocher Y, Kamen A, Piret JM. Submitted-a. Evolution of Cellular Aggregation during HEK293 Batch and Perfusion Cultures. Drouin H, Lanthier S, Piret JM, Kamen A, Durocher Y. Submitted-b. Influence of Culture Medium Osmolality on Interferon-alpha 2b Production Stability in HEK293 Cells. Process Biochemistry. Drouin H, Ritter JB, Gorenflo VM, Bowen BD, Piret JM. 2007. Cell separator operation within temperature ranges to minimize effects on Chinese hamster ovary cell perfusion culture. Biotechnol Prog 23(6):1473-84. Ferreira TB, Ferreira AL, Carrondo MJ, Alves PM. 2005. Two different serum-free media and osmolality effect upon human 293 cell growth and adenovirus production. Biotechnol Lett 27(22):1809-13. Gambhir A, Korke R, Lee J, Fu PC, Europa A, Hu WS. 2003. Analysis of cellular metabolism of hybridoma cells at distinct physiological states. J Biosci Bioeng 95(4):317-27. Garnier A, Cote J, Nadeau I, Kamen A, Massie B. 1994. Scale-up of the adenovirus expression system for the production of recombinant protein in human 293S cells. Cytotechnology 15(1-3):145-55. Gorenflo VM, Angepat S, Bowen BD, Piret JM. 2003. Optimization of an acoustic cell filter with a novel air-backflush system. Biotechnol Prog 19(1):30-6.  86 Gorenflo VM, Ritter JB, Aeschliman DS, Drouin H, Bowen BD, Piret JM. 2005. Characterization and optimization of acoustic filter performance by experimental design methodology. Biotechnol Bioeng 90(6):746-53. Goudar C, Biener R, Zhang C, Michaels J, Piret J, Konstantinov K. 2006. Towards industrial application of quasi real-time metabolic flux analysis for mammalian cell culture. Adv Biochem Eng Biotechnol 101:99-118. Graham FL, Smiley J, Russell WC, Nairn R. 1977. Characteristics of a human cell line transformed by DNA from human adenovirus type 5. J Gen Virol 36(1):59-74. Griffiths JB. 1992. Animal-Cell Culture Processes - Batch or Continuous. Journal of Biotechnology 22(1-2):21-30. Guarna MM, Fann CH, Busby SJ, Walker KM, Kilburn DG, Piret JM. 1995. Effect of cDNA copy number on secretion rate of activated protein C. Biotechnol Bioeng 46(1):22- 7. Heath C, Kiss R. 2007. Cell culture process development: advances in process engineering. Biotechnol Prog 23(1):46-51. Henry O, Kwok E, Piret JM. 2008. Simpler noninstrumented batch and semicontinuous cultures provide mammalian cell kinetic data comparable to continuous and perfusion cultures. Biotechnol Prog 24(4):921-31. Hu WS, Piret JM. 1992. Mammalian cell culture processes. Current Opinion in Biotechnology(3):110-114. Kadouri A, Spier RE. 1997. Some myths and messages concerning the batch and continuous culture of animal cells. Cytotechnology 24(2):89-98. Kaluz S, Kabat P, Gibadulinova A, Vojtassak J, Fuchsberger N, Kontsek P. 1994. Interferon alpha2b is the predominant subvariant detected in human genomic DNAs. Acta Virol 38(2):101-4. Kamen AA, Bédard C, Tom R, Perret S, Jardin B. 1996. On-line monitoring of respiration in recombinant-baculovirus-infected and uninfected insect cell bioreactor cultures. Biotechnology and Bioengineering 50:34-48. Kamen AA, Tom R, Caron AW, Chavarie C, Massie B, Archambault J. 1991. Culture of insect cells in a helical ribbon impeller bioreactor. Biotechnol Bioeng 38:619-628. Kempken R, Buntemeyer H, Lehmann J. 1991. The medium cycle bioreactor (MCB): monoclonal antibody production in a new economic production system. Cytotechnology 7(2):63-74.  87 Kim MS, Kim NS, Sung YH, Lee GM. 2002. Biphasic culture strategy based on hyperosmotic pressure for improved humanized antibody production in Chinese hamster ovary cell culture. In Vitro Cell Dev Biol Anim 38(6):314-9. Konstantinov K, Goudar C, Ng M, Meneses R, Thrift J, Chuppa S, Matanguihan C, Michaels J, Naveh D. 2006. The "push-to-low" approach for optimization of high-density perfusion cultures of animal cells. Adv Biochem Eng Biotechnol 101:75-98. Konstantinov KB, Tsai YS, Moles D, Matanguihan R. 1996. Control of long-term perfusion Chinese hamster ovary cell culture by glucose auxostat. Biotechnology Progress 12(1):100-109. Leelavatcharamas V, Emery AN, Al-Rubeai M. 1999. Use of cell cycle analysis to characterise growth and interferon-gamma production in perfusion culture of CHO cells. Cytotechnology 30(1-3):59-69. Lehmann J, Kempken R, Lutkemeyer D, Buntemeyer H. 1990. Economic Aspects of Medium Recycling. Trends in Animal Cell Culture Technology:55-59. Lin J, Takagi M, Qu Y, Gao P, Yoshida T. 1999. Enhanced monoclonal antibody production by gradual increase of osmotic pressure. Cytotechnology 29:27-33. Loignon M, Perret S, Kelly J, Boulais D, Cass B, Bisson L, Afkhamizarreh F, Durocher Y. 2008. Stable high volumetric production of glycosylated human recombinant IFNalpha2b in HEK293 cells. BMC Biotechnol 8:65. Mercille S, Johnson M, Lanthier S, Kamen AA, Massie B. 2000. Understanding factors that limit the productivity of suspension-based perfusion cultures operated at high medium renewal rates. Biotechnology and Bioengineering 67(4):435-450. Mohanty SR, Kupfer SS, Khiani V. 2006. Treatment of chronic hepatitis B. Nat Clin Pract Gastroenterol Hepatol 3(8):446-58. Moreira JL, Feliciano AS, Santana PC, Cruz PE, Aunins JG, Carrondo MJ. 1994b. Repeated-batch cultures of baby hamster kidney cell aggregates in stirred vessels. Cytotechnology 15(1-3):337-49. Morrison CJ, McMaster WR, Piret JM. 1997. Differential Stability of Proteolytically Active and Inactive Recombinant Metalloproteinase in Chinese Hamster Ovary Cells. Biotechnol Bioeng 53(6):594 - 600. Nadeau I, Garnier A, Côté J, Massie B, Chavarie C, Kamen A. 1996. Improvement of Recombinant ProteinProduction with the Human Adenovirus/293S Expression System Using Fed-Batch Strategies. Biotechnol Bioeng 51:613-623. Ozturk SS, Palsson BO. 1990. Effect of Medium Osmolarity on Hybridoma Growth, Metabolism and Antibody Production. Biotechnology and Bioengineering 37:989-993.  88 Pestka S, Langer JA, Zoon KC, Samuel CE. 1987. Interferons and their actions. Annu Rev Biochem 56:727-77. Ronning OW, Schartum M, Winsnes A, Lindberg G. 1991. Growth limitation in hybridoma cell cultures: the role of inhibitory or toxic metabolites. Cytotechnology 7(1):15-24. Ryll T, Dutina G, Reyes A, Gunson J, Krummen L, Etcheverry T. 2000. Performance of small-scale CHO perfusion cultures using an acoustic cell filtration device for cell retention: Characterization of separation efficiency and impact of perfusion on product quality. Biotechnology and Bioengineering 69(4):440-449. Shen D, Sharfstein ST. 2006. Genome-Wide Analysis of the Transcriptional Response of Murine Hybridomas to Osmotic Shock. Biotechnol Bioeng 93(1):132-145. Siegwart P, Cote J, Male K, Luong JH, Perrier M, Kamen A. 1999. Adaptive control at low glucose concentration of HEK-293 cell serum-free cultures. Biotechnol Prog 15(4):608-16. Smith CG, Guillaume JM, Greenfield PF, Randerson DH. 1991. Experience in Scale-up of Homogeneous Perfusion Culture for Hybridomas. Bioprocess Engineering 6(5):213- 219. Srivastava P, Bhattacharaya P, Pandey G, Mukherjee KJ. 2005. Overexpression and purification of recombinant human interferon alpha2b in Escherichia coli. Protein Expr Purif 41(2):313-22. Srivastava P, Mukherjee KJ. 2005. Kinetic studies of recombinant human interferon- alpha (rhIFN-α) expression in transient state continous cultures. Biochemical Engineering Journal 26:50-58. Takazawa Y, Tokashiki M. 1989. High cell density perfusion culture of mouse-human hybridomas. Applied Microbiology and Biotechnology 32(1):280-284. Tang YJ, Ohashi R, Hamel JF. 2007. Perfusion culture of hybridoma cells for hyperproduction of IgG(2a) monoclonal antibody in a wave bioreactor-perfusion culture system. Biotechnol Prog 23(1):255-64. Vits H, Hu WS. 1992. Fluctuations in Continuous Mammalian-Cell Bioreactors with Retention. Biotechnology Progress 8(5):397-403. Vogel W. 1999. Treatment of acute hepatitis C virus infection. J Hepatol 31 Suppl 1:189- 92. Werner RG, Walz F, Noe W, Konrad A. 1992. Safety and Economic-Aspects of Continuous Mammalian-Cell Culture. Journal of Biotechnology 22(1-2):51-68.  89 Woodside SM, Bowen BD, Piret JM. 1998. Mammalian cell retention devices for stirred perfusion bioreactors. Cytotechnology 28(1-3):163-175. Wu MH, Dimopoulos G, Mantalaris A, Varley J. 2004. The effect of hyperosmotic pressure on antibody production and gene expression in the GS-NS0 cell line. Biotechnol Appl Biochem 40(Pt 1):41-6. Wurm FM. 2004. Production of recombinant protein therapeutics in cultivated mammalian cells. Nature Biotechnology 22(11):1393-1398. Yang JD, Angelillo Y, Chaudhry M, Goldenberg C, Goldenberg DM. 2000. Achievement of high cell density and high antibody productivity by a controlled-fed perfusion bioreactor process. Biotechnology and Bioengineering 69(1):74-82. Zeng AP, Deckwer WD. 1999. Model simulation and analysis of perfusion culture of mammalian cells at high cell density. Biotechnology Progress 15(3):373-382. Zeng AP, Deckwer WD, Hu WS. 1998. Determinants and rate laws of growth and death of hybridoma cells in continuous culture. Biotechnol Bioeng 57(6):642-54. Zhang S, Handacorrigan A, Spier RE. 1993. A Comparison of Oxygenation Methods for High-Density Perfusion Cultures of Animal-Cells. Biotechnology and Bioengineering 41(7):685-692.    90 5 Evolution of Cellular Aggregation during HEK293 Batch and Perfusion Cultures3  5.1 Introduction It was noted in Chapter 4 that aggregation occurred in perfusion cultures especially at low perfusion rates, which resulted in problems in estimating the culture productivity. Ideally, bioreactor cell culture systems should be homogenous since heterogeneity can reduce overall culture performance (Coppen et al. 1995; Renner et al. 1993) and reproducibility.  Often aggregation has been ignored by researchers due to a lack of appropriate measurement techniques.  Cell aggregation is a normal physiological phenomenon and an important consideration in perfusion processes compared to batch cultures due to the long cultivation periods that can result in the generation of a wide range of aggregate sizes (20 to >1000 µm) depending on the length of the perfusion culture run.  In addition, the use of an acoustic separator as a cell retention device in perfusion cultures can actually enhance aggregation (Bierau et al. 1998). The aggregation process, which can also be observed in batch suspension cultures, results from cells forming tight intercellular junctions, which are characterized by the presence of a proteinaceous layer at the contact point between the cell cytoplasmic membranes (Coppen et al. 1995; Peshwa et al. 1993).  In the case of perfusion culture, aggregates increase in size throughout the cultivation period and can be a useful as a natural means of easing cell retention (Litwin 1992; Moreira et al. 1994a; Peshwa et al. 1993).  3 A version of this chapter will be submitted for publication.  Drouin H, Lanthier S, Durocher Y, Kamen A and Piret JM, Evolution of Cellular Aggregation during HEK293 Batch and Perfusion Cultures.  91 However, cell aggregation can easily become a critical factor that negatively influences cell productivity (Renner et al. 1993).  Cell viability, based on permeability assessed by the inclusion of trypan blue dye (Coco-Martin et al. 1992), decreases at the centre of the larger aggregates (>300 µm) due to the formation of necrotic core caused by oxygen mass transfer limitations (Jo et al. 1998; Sen et al. 2001).  In the literature, there is no consensus on the minimum size required in order for the mass of cells to be considered as an aggregate.  Some researchers consider particles larger than 25 µm as aggregates (Renner et al. 1993).  However, this criterion alone is not appropriate to distinguish between irreversibly aggregated cells from cells during mitosis.  While the evolution of aggregate sizes has been studied in batch and repeated-batch cultures (Moreira et al. 1994b), it has not been studied in a perfusion culture process at high cell concentrations and under steady-state conditions.  To study aggregate formation, three different techniques are commonly used to measure the aggregate diameter: i) microscopic evaluation (Litwin 1992; Moreira et al. 1994a; Moreira et al. 1995; Sen et al. 2001); ii) electrical sensing zone method (Boraston et al. 1992; Peshwa et al. 1993); and iii) laser diffraction method (Illing and Harrison 1999; Renner et al. 1993).  Each method has limitations in terms of the size of aggregates that can be measured corresponding to the apparatus dimension, such as the aperture diameter in the electrical sensing zone method.  There are also several different techniques presented in the literature whose purpose is to control or limit aggregate size and thereby prevent the formation of necrotic cores in large aggregates such as: i) application of increased hydrodynamic forces (Illing and Harrison 1999; Moreira et al. 1995; Renner et  92 al. 1993; Sen et al. 2001); ii) mechanical dissociation (Moreira et al. 1994a); and iii) modification of the medium composition (Boraston et al. 1992; Litwin 1992; Moreira et al. 1994a; Peshwa et al. 1993).  To further monitor a perfusion culture, it is essential to be able to determine the cell density. Usually, culture samples are removed periodically and analyzed. However, more useful on-line methods have been developed to monitor cell density based on permittivity measurements (Ansorge et al. 2007; Ansorge et al. 2009; Cannizzaro et al. 2003; Carvell and Dowd 2006; Ducommun et al. 2001; Ducommun et al. 2002; Zeiser et al. 1999; Zeiser et al. 2000). Permittivity-based biomass probes can be used in-situ, performing on-line measurements of the membrane enclosed volume fraction or biovolume of the cell suspension.  Theoretically, highly aggregated cell suspensions exhibit a different permittivity than non-aggregated cultures of similar biomass concentration (Davey 1993; Ron et al. 2009). This effect can be explained by a partial neutralization of membrane polarization that can be induced by cell-to-cell contact. Although extensive work on the aggregation of erythrocytes is available (Beving et al. 1994; Pribush et al. 1999), the effect of aggregation on determining cell density based on permittivity measurements has, to our knowledge, not been studied for mammalian cell cultures.  In this chapter, HEK293 cells producing recombinant IFN-α2b have been monitored in batch and perfusion cultures in order to investigate the evolution of cell aggregation using a simple analysis method.  Factors that could influence aggregation were initially studied in batch cultures and the results were compared to perfusion cultures.  Three techniques to induce cell aggregation were investigated in batch culture (addition of dead cells,  93 switch to a calcium-enriched medium and the use of a low agitated culture). The results were used to help determine the main aggregation mechanisms observed during perfusion culture.  To monitor cell density and investigate the evolution of cell aggregation in both batch and perfusion cultures, a simple image analysis method was developed to analyze samples and an on-line permittivity-based technique was evaluated to continuously monitor the perfusion cultures. The effects of aggregation on the permittivity signal and on the calculated cell specific productivity rate were then determined. Finally, the evolution of cell aggregation was monitored in two perfusion cultures, one using a low bleed rate (0.042 d-1) and the other using a high bleed rate (0.167 d-1).  5.2 Materials and Methods 5.2.1 Cell Line and Media Human embryonic kidney cells (HEK293) were first isolated in 1977 (Graham et al. 1977) and later adapted to grow in suspension culture (Cote et al. 1998).  Most recently, 293-6E cells were engineered to express recombinant human interferon alpha 2b (IFN- α2b) (Loignon et al. 2008).  The two media used for these experiments were derived from a low calcium serum-free medium (LC-SFM).  LC-SFM is a modified version of the commercial hybridoma serum- free medium (H-SFM by GIBCO, Grand Island, NY, USA) and custom-made with a low calcium content (Cote et al. 1998).  The first medium was a modified low calcium serum- free medium (MLC-SFM) and the second medium was an enriched low calcium serum-  94 free medium (ELC-SFM).  Both MLC-SFM and ELC-SFM were formulated with 12.5 g L-1 of LC-SFM basal powder (~twice the amount of powder used in LC-SFM). Selective pressure additives were added to both media at 50 mg L-1 of G418 sulfate (Wisent, St-Bruno, QC, Canada) and 5 mg L-1 of blasticidin (InvivoGen, San Diego, CA, USA).  Both media were supplemented with 0.1% (w/v) bovine serum albumin (BSA; Intergen, Purchase, NY, USA), 1% (v/v) chemically defined lipids (GIBCO) and 0.1% Pluronic F68.  Table 5-1 summarizes the levels of additives and nutrients.  The osmolality of both media was adjusted to 300 ± 10 mOsm kg-1 by adding sodium chloride.  Table 5-1 :  Partial composition of the media used based on the additives and amino acid analysis. These commercial media are proprietary formulations, and so they contain other unknown components.  Media   MLC-SFM ELC-SFM Osmolarity (mOsm kg-1) 300 ± 10 300 ± 10 Glucose (g L-1) 4.5 4.5 BSA 30%  (mL L-1) 3.33 3.33 Lipids 1000x  (mL L-1) 1 1 Pluronic F-68 (g L-1) 1 1 Asp (mM) 0.13 0.46 Glu (mM) 0.16 0.53 Ser (mM) 0.96 1.32 Asn (mM) 0.63 1.04 Gly (mM) 0.46 0.81 Gln (mM) 5.96 6.02 His (mM) 0.63 0.88 Thr (mM) 2.15 2.64 Arg (mM) 2.19 2.82 Ala (mM) 0.12 0.50 Pro (mM) 3.67 4.46 Tyr (mM) 1.00 1.25 Cys (mM) 0.71 0.85 Val (mM) 2.27 2.81 Met (mM) 0.65 0.78 Ile (mM) 2.53 3.03 Leu (mM) 2.71 3.21 Lys (mM) 2.44 2.91 Phe (mM) 1.16 1.40  95  5.2.2 Cell Culture Experiments The cells were grown initially in Freestyle293 medium (Invitrogen, Grand Island, NY, USA) and frozen to make a Working Cell Bank (WCB).  For every experiment, the cells were thawed from this WCB and transferred into a MLC-SFM or an ELC-SFM static culture for 24 hours.  Then, a 125 mL shake flask (Corning, Corning, NY, USA) was inoculated with the contents of the static culture and placed on a rotary plate at 115 rpm in a humidified incubator controlled at 37 oC and 5% CO2.  Subsequent batch cultures were performed in a 125 mL shake flask with a 30 mL working volume, inoculated at 0.30 x 106 cells mL-1.  The batches were cultivated for 8 days with daily sampling.  Three batches have been performed using different culture conditions.  One was inoculated with 0.30 x 106 cells mL-1 of viable cells and the same amount of dead cells. The dead cells were harvested from a batch cultivated until the viability was 0%.  These cells were then spun for 5 min at 1000 rpm to remove spent medium and re-suspended in fresh ELC-SFM medium.  CaCl2 was added to the second batch at a concentration of 1.0 mM.  The third batch was cultivated at a reduced agitation rate of 70 rpm.  After 5 days, these three batches and a control culture were diluted with PBS (Hyclone, Logan, UT, USA) to measure the permittivity signal at different cell concentrations but at the same aggregation level.   96 5.2.3 Permittivity Measurements The permittivity signal was measured by radio-frequency impedance as a non-invasive approach to measure the cell concentration on-line.  These online in-situ measurements during bioreactor cultures and ‘off-line’ measurement of samples diluted from shake flask cultures were performed with a Biomass Monitor 220 (Aber Instruments, Aberystwyth, UK). The system setup was carried out according to the manufacturer’s instructions.  A biomass probe (Aber Instruments) was mounted on the bioreactor described in section 5.2.4.  The same probe was also used to measure the permittivity signal in samples from the shake flasks.  The data were recorded on the monitor then transferred to a computer for analysis.  All values reported in this study refer to a dual frequency permittivity signal (i.e., the difference in permittivity at 0.6 and 19 MHz).  To correlate permittivity and off-line viable cell counts by linear regression, the values for bioreactor cultures during the exponential growth phase and dilution experiments were used to calculate the slope (viable cells/unit permittivity).  For aggregated cultures, sequential dilutions with PBS (Hyclone) of the cell suspension were performed after the end of each experiment to study the effect of aggregation on the permittivity signal. Measurements in shake flasks were done by installing the BM220 on top of an agitated water bath which was controlled at 37°C.  The culture permittivity was measured by placing the probe tip in the culture while agitation continued.  All reported values are the average of three subsequent measurements.   97 5.2.4 Perfusion Culture Operation The perfusion experiments were performed in a 3.5 L SG Chemap bioreactor (Mannedorf, Switzerland) with a 2.73 ± 0.03 L working volume.  The schematic of the system is presented in Figure 5-1.  Agitation was provided by 2 pitch-blade impellers (3 blades, 45o) with speeds starting at 75 rpm and then raised to 100 rpm after the 2nd day. The ratio of impeller to vessel diameter was 0.5.  The mixing was enhanced by 3 surface baffles placed at 120o.  A Chemap FZ-2000 control unit was used to control the agitation and the temperature (37oC).  The pH was monitored using a gel electrode (Mettler Toledo, Wilmington, MA, USA) while the DO was measured with a polarographic electrode (Ingold, Andover, MA, USA).  The pH and the DO were controlled via a SAFE8000 interface (Control Microsystems, Kanata, ON, Canada, operating with FIX Dmacs and MMI software Intellution, Norwood, MA, USA) at a pH of 7.1 and at a DO 30 – 40% of air saturation as previously described (Kamen et al. 1996).  Oxygen was directly sparged in the bioreactor culture through a polypropylene sparger with a pore size of 250 µm (Porex Technologies, Fairburn, GA, USA).  The sparging flow rate was constant and operated in a pulse mode.  CO2 was added to lower the pH, while a solution of 7.5 % NaHCO3 was added to increase the pH.   98  Figure 5-1 : Schematic of the perfusion system.  The bioreactor was inoculated (at 0.40 x 106 cells mL-1) with HEK293-IFN cells from one 2L Erlenmeyer shake flask (500 mL working volume) and run in batch mode for 1 day.  Perfusion was then initiated at a constant feed rate with the media described in Table 5-1.  Bioreactor samples were taken daily.  During the first 3 days, oxygen was provided through surface aeration and then by sparging after day 3.  A separate bleed line  99 was used to periodically remove the culture (spent medium, viable and dead cells) at a constant rate after day 3 (see Figure 5-1).  The culture volume was kept at a constant level using an optical-based sensor (Electromatic S-System, Ville St-Laurent, QC, Canada) that triggered operation of the harvest pump.  The medium feed was manually adjusted while the cell bleed and separator operation were controlled via the SAFE8000 interface.  For the perfusion cultures the bioreactor was equipped with an acoustic cell separator for cell retention (BioSep 1015, AppliSens, division of Applikon, Schiedam, Netherlands). The cell suspension was pumped into the 10L acoustic separator (2.1 MHz operating frequency) and the cells were retained in the chamber by the acoustic forces (Drouin et al. 2007).  The aggregated cells settled back to the bioreactor and the clarified liquid was pumped out of the system at a selected harvest flow rate.  The operation of the separator was controlled with a BioSep ADI 1015 electronic controller (AppliSens).  The acoustic separator was operated with air backflush every 30 minutes (Gorenflo et al. 2003).  The alternation between the stop time and the run time (called the duty cycle) was set to 55 s of run time with 5 s of stop time.  The perfusion cell separation efficiency (define as the percentage of viable cells retained) was maintained at ~95% during the entire culture. Peristaltic pumps (Masterflex L/S, Cole Parmer, Vernon Hills, IL, USA) were used for the feed, bleed, recirculation, harvest and backflush.  The flow rate in the external recirculation loop was constant.  Flow rates are presented as volume per bioreactor volume per day (vvd).   100 Two perfusion cultures were performed: one using MLC-SFM and the second with ELC- SFM.  The bleed rate was increased from 0.042 day-1 for the first perfusion culture to 0.167 day-1 for the second experiment.  After perfusion was initiated, operational settings were kept constant in order to allow stabilization of the viable cell concentration for both cultures (i.e., pseudo-steady state was reached).  5.2.5 Analytical Methods 5.2.5.1 Cell Concentration and Viability The total cell number and viability were counted in a hemocytometer (Hausser Scientific, Horsham, PA, USA) under a microscope. Because of cell aggregation in culture, a 500 µL culture aliquot was placed in a 6 mL polystyrene tube and agitated to disperse the cell clumps.  The use of this approach had the effect of decreasing the apparent cell viability; therefore, the viability percentage was estimated separately without dispersing the cells using erythrosin B exclusion (Garnier et al. 1994).  However, large aggregates (diameter > ~100 µm) were excluded from the count since these could have created cell count errors by increasing falsely cell concentration considering the small volume measured.  For batch culture sampling, one hemocytometer count was used for viability estimation while two were used to determine the total cell number.  For perfusion culture sampling, three hemocytometer counts were used for the viability measurement while six were used to assess total cell number.  Multiple hemocytometer readings allowed the statistical analysis of the viable cell concentration measurements, presented in terms of standard deviations in the plots.   101 5.2.5.2 Culture Medium Analyses Culture samples were spun down for 3 min at 1000 rpm to remove the cells and the supernatants were frozen at -80oC for later analysis.  The glucose, lactate and ammonium concentrations were measured using an IBI Biolyser Rapid Analysis System (Kodak, New Haven, CT, USA).  The osmolality was measured using an osmometer (Advanced Instruments Micro Osmometer Model 3300, Needham Heights, MA, USA).  Amino acids were analyzed by HPLC (WATERS Milford, MA, USA) (Kamen et al. 1991).  IFN-α2b was measured using an ELISA kit (PBL Biomedical Laboratories, Piscataway, NJ, USA). An internal standard was added to every ELISA plate to provide a reference to standardize the results between assays.  5.2.5.3 Cell Size Measurement Methods An automated CEDEX cell counter (Innovatis AG, Bielefeld, Germany) was used for direct cell size measurement based on automated image analysis.  The detectable cell diameter range was 4 - 40 µm because of the equipment dimensions.  The program was designed to measure the diameter of each individual cell present in a small aggregate but not to measure the diameter of a whole aggregate.  Thus, it was not practical to measure cell aggregates using this method.  CEDEX measurements were used to validate the image analysis measurement method developed by comparing the size distribution of particles within the detectable range of the CEDEX counter.  To measure aggregate diameters, a method was developed based on computer-assisted image analysis.  Images were taken with a camera (Model No. WVC54GC ver 1.1,  102 Linksys a Division of Cisco Systems Inc, Irvine, CA, USA) mounted on a microscope (Leica DMIL, type 090-131.001, Leica Mikroskopie & Systems GmbH, Wetzlar).  The microscope light was placed at 50 cm above the sample to allow uniform distribution of the light.  A 10X objective was used.  Cell samples were taken using cut pipette tips (opening diameter ~ 5 mm) in order to minimize the shear applied to the aggregates.  The cell culture sample was well suspended before 300 µL was taken and diluted in 5 mL of PBS (HyClone DPBS/modified w/o calcium and magnesium, HyClone Laboratories Inc, South Logan, Utah, USA).  This dilution was performed in a 10 mL circular culture dish (Thermo Fisher Scientific, Waltham, MA, USA) which was agitated without shearing the cells.  Random images were taken after the cells settled to a single layer at the bottom of the dish.  The images taken were then analyzed on a computer using the free image analysis software ImageJ (Abramoff et al. 2004).  An algorithm was written to convert pixels into a known dimension based on a calibration previously performed.  This calibration was performed by taking and analyzing images of micro particles based on polystyrene (Fluka Chemical Corp, Milwaukee, WI, USA) of diameter of 10 and 19 µm respectively.  The area calculated was transformed into a characteristic diameter, given by: € Dc = 4A π   (5-1)  where € Dc  is the characteristic diameter of the particle and A the calculated area.  Particles could then be compared independently of their shape and allow particle size distribution  103 plots to be constructed.  The characteristic diameter was used to calculate the particle total volume, Vtotal, given by € Vtotal = 4πDc,i3 3i∑   (5-2) where Dc,i represents the characteristic diameter of the ith particle.  5.3 Results and Discussion 5.3.1 Aggregate Measurement Method Development and Validation A method for automated image analysis was developed to allow easy analysis of aggregates without size limitations.  A camera mounted on a microscope and a computer were the basic components of this approach (see Materials and Methods section).  In order to validate the method, different tests were performed so that the results could be compared with known references.  The first test was to analyze beads with known diameters.  An analysis of images taken showed a very good correlation with the known results (Figure 5-2).  The size measurement algorithm was not developed to distinguish single particles touching each other but rather to measure the entire particle.  This fact explains why there is not only one peak at the known diameter but also peaks for larger particle diameters.  Those larger particles were multiple beads touching each other.   104  Figure 5-2 : Analysis of particle diameter size distribution for A) 10 µm and B) 19 µm diameter calibration beads. Dash line corresponds to the known diameter of the calibration beads.  As a second step of the validation process, the image analysis results of samples of HEK293-IFN cells were compared with those obtained using a commercially available CEDEX automatic cell counter.  This CEDEX method was designed to measure individual cell diameters and particles larger than 40 µm could not be analyzed because of apparatus dimension limitations.  Table 5-2 presents the comparison results.  Errors were calculated taking the CEDEX results as a reference.  Images of each sample (around 20 per sample) were processed by both methods.  Samples with a larger range of aggregate sizes gave larger errors.  This result was expected because the CEDEX algorithm was only able to measure a single cell diameter in an aggregate of multiple cells.  Sample 3 was formed from single cells only and the associated error was 0.6% (Table 5-2).  105  Table 5-2.  Comparison of particle diameters measured with the automatic CEDEX and with the computer assisted image analysis method. *: Average diameter measured by the CEDEX.  **: Average diameter measured by the image analysis method considering only particles smaller than 40 µm. Sample CEDEX* Image Analysis** Error - µm µm % 1 21.21 21.90 3.3 2 18.76 19.05 1.5 3 18.74 18.86 0.6 4 20.69 20.30 1.9 5 22.08 22.10 0.1 6 20.32 19.84 2.4 7 22.00 22.34 1.0 8 19.23 20.48 6.5 9 18.51 19.51 5.4 10 18.35 19.16 4.4  The success of this method depends on careful sample handling.  It is essential to keep the aggregates suspended during sampling in order to get a representative sample.  The volume of the sample is also a critical aspect in obtaining the right ratio between large and small particles.  It is useful to evaluate the number of very large particles per mL (or less) to determine whether pictures are truly representative of the culture.  This practice could become a useful criterion for rejecting pictures of a sample.  It is important to minimize shearing during sampling; cutting the pipette tip is the first step.  From our analysis, results obtained with 15 pictures or fewer gave a lot more variability.  The number of pictures taken needs to be sufficient to represent the culture distribution. Finally, tests performed with calibration beads at the beginning and at the end were used to confirm the size measurement results.   106 5.3.2 Evolution of Cellular Aggregation in Perfusion Cultures HEK293-IFN cells were cultivated in a continuous perfusion system using two different bleed rates: 0.042 and 0.167 d-1 using MLC-SFM and ELC-SFM, respectively.  In the perfusion culture at the higher bleed rate, media enriched with amino acids were used. Both cultures were brought to a pseudo-steady state and, for the first 30 days, the aggregated cells were analyzed. This pseudo-steady state was defined as one where the viable cell concentration ( € Xv) was stable for 5 consecutive days. The low bleed rate culture reached an average viable cell concentration of 5.3 x106 cells mL-1 and an average viability of 78.7% (Figure 5-3).   Figure 5-3:  Viable cell concentration and viability for HEK293-IFN cells perfusion culture performed at A) low bleed rate (0.042 d-1) and B) high bleed rate (0.167 d-1). Vertical dashed lines represent the start of the bleed operation at a steady rate.  The high bleed rate perfusion culture yielded a slightly higher viable cell concentration with a pseudo-steady-state average of 5.9 x106 cells mL-1 (Figure 5-3).  One of the main differences between the cultures was the viability.  The high bleed rate culture had an average viability of 90.0%, an increase of 11.3% compared to the low bleed rate culture. For both cultures, daily samples were analyzed and particle size distributions were determined.  The particle size distribution difference between the two cultures was  107 obvious even from visual observation.  The low bleed rate culture had many large aggregates visible through the bioreactor glass wall while it was uncommon to see large particles in the high bleed rate culture (Figure 5-4).   Figure 5-4 : Pictures of the bioreactor glass wall during the low bleed rate perfusion culture (A) and the second perfusion culture using a high bleed rate (B).  A quantitative analysis of the two cultures gave the anticipated results. The average particle diameter was significantly higher for the low bleed rate culture than for the high bleed rate culture (Figure 5-5). After the bleed was initiated, the average diameter for both cultures increased.  This increase was more significant for the low bleed rate and it continued to increase during the culture.  More experience with the image analysis technique helped to reduce the variability of the results obtained for the high bleed rate. At the end of the pseudo-steady state period, the average aggregate diameter was ∼5 times higher for the low bleed rate culture.  In this culture, the percentage of cell aggregates larger than 100 µm was around 64% toward the end versus only 15% for the high bleed rate culture (Figure 5-6).  Cell aggregates larger than 100 µm were excluded from the cell count because their inclusion could significantly affect the results.  The cell count analysis usually had a 15% variability, indicating that aggregation in the high bleed  108 rate culture should have a negligible effect on the cell specific calculations.  Furthermore, the results for the low bleed rate culture had a larger variability than those of the high bleed rate culture.  This large variation between data points is mainly due to the number of images taken versus the number of large aggregates analyzed.   Figure 5-5 :  Evolution of average particle diameter during HEK293-IFN cell perfusion cultures at low and high bleed rates. Vertical dashed line represents the start of the bleed operation at a steady rate.  Solid and dashed curves are trend lines from polynomial regressions.   Figure 5-6 : Evolution of the volume fraction of particles larger than 100 µm during HEK293 cell perfusion cultures at low and high bleed rates. Vertical dashed line represents the start of the bleed operation at a steady rate.  Solid and dashed curves are trend lines from polynomial regressions.   109 5.3.3 Mechanisms of Cell Aggregation 5.3.3.1 Batch Experiments HEK293-IFN cell aggregation mechanisms were studied in small-scale batch cultures. Cells in each of three different batches were exposed either to the presence of dead cells, or to the addition of calcium or to low hydrodynamic forces (low agitation rate).  These batches were then compared to a batch maintained using standard conditions (called the control culture), i.e., ELC-SFM media at 115 rpm.  The control culture reached a viable cell concentration of 2.5 x106 cells mL-1 and a viability of 90% (Figure 5-7).  The apparent growth rate was 0.54 day-1.  For these experiments, the batch cultures were stopped after 5 days to perform this test at high viability.  Thus, the final viable cell concentrations shown in Figure 5-7 are not the maximum values.  The batch culture started in the presence of dead cells grew similarly to the control.  The viability was lower at the beginning due to the addition of dead cells but reached 90% after 5 days of culture with a viable cell concentration of 3.1 x106 cells mL-1.  It is possible that the presence of dead cells induced a slight lag phase expressed by a lower growth rate of 0.45 day-1.  The calcium-enriched batch did not grow like the control batch.  The viable cell concentration did not reach 1 million cells mL-1 and the viability started to decrease drastically after 2 days of culture.  At 0.23 day-1, the growth rate was 43% lower than that of the control batch.  The last batch was exposed to a lower agitation rate and grew normally, reaching a viable cell concentration of 3.4 x106 cells mL-1 with a viability of 89%.  The apparent growth rate was similar to that of the control with a value of 0.50 day-1.   110  Figure 5-7 : Viable cell concentration and viability profiles for HEK293-IFN cells in batch culture at standard conditions (A), with addition of dead cells (B), with addition of calcium (C), and at low agitation rate (D).  Only the calcium-enriched batch presented a significant difference in growth (Figure 5-7).  Its final viable cell concentration was 70% lower than that of the control batch. However, large cell aggregates were not included in the calculation of the viable cell concentration.  The particle size analysis revealed interesting aspects.  Cells exposed to calcium rapidly formed large cell aggregates (Figure 5-8).  At day 6, the average particle diameter was 906 µm, 27 times higher than that of the batch control.  We consequently observed the direct effect of calcium additions to the medium.  It has been found previously that calcium increases the formation of tight junctions made of complex protein assemblies (Brackenbury et al. 1981; Peshwa et al. 1993).  The final average particle diameter was 167 µm for the culture exposed to a low agitation rate.  The agitation normally provides sufficient shear to limit the formation of aggregates by keeping the cells singly suspended (Moreira et al. 1995).  Thus, by lowering the agitation  111 rate, the cells are exposed to less shear (that help keep cells separated) and junctions are more easily created.  The break-up of aggregates becomes a function of the exposure to shear (Illing and Harrison 1999).  The aggregates were smaller for the culture exposed to dead cells, reaching, after 5 days, an average diameter of 49 µm but this was still 46% higher than the control average.  It has been observed that the release of nucleic acids from dead cells can cause increased cell aggregation (Renner et al. 1993).   Figure 5-8 :  Evolution of average particle diameter during HEK293-IFN cell batch cultures at standard condition (control), with addition of dead cells, with addition of calcium and at low agitation rate.  5.3.3.2 Comparison of Cell Aggregation Evolution in Perfusion and Batch Cultures The aggregate size distributions in batch and perfusion cultures were compared in order to reveal possible underlying mechanisms that cause aggregation in the latter.  All batches started with the same cell size distribution inoculum, with an average diameter of 25.8 µm, mostly single cells.  In Figure 5-9, cell size distributions are presented at day 6 for all 4 batches.  As described in the previous section, the average diameters and particle size distributions differed depending on the culture method.  The addition of calcium to  112 one of the batches caused a more rapid formation of aggregates.  In this case, after 6 days, cell aggregates of 1000 µm represented 88% of the volume fraction.  For cells exposed to low agitation, the average particle diameter increased significantly because of the presence of large aggregates of 522 µm.  The culture with the addition of dead cells showed some cell aggregates between 165 and 269 µm compared to the control that did not have any large aggregates.   Figure 5-9 :  Aggregate size distribution for batch cultures at day 6 at standard conditions (A), with addition of dead cells (B), with addition of calcium (C), and at low agitation rate (D). The vertical dashed line represents the average diameter for each culture.  In graph C, volume fraction of particles of 1000 µm reached 88%.  Perfusion culture particle size distributions were analyzed 6 days after inoculation (Figure 5-10).  The low bleed rate culture had larger cell aggregates with many falling in the range of 91 to 196 µm.  The high bleed rate culture did not have a significant number  113 of aggregates.  Interestingly, the high bleed rate particle size distribution was similar to the batch control distribution (Figure 5-9, A) while the low bleed rate distribution was similar to that of the batch with the addition of dead cells (Figure 5-9, B).  Another similarity was the fact that the low bleed rate had a viability lower than the high bleed rate culture.  Therefore, the main factor influencing cell aggregation in the low bleed rate perfusion culture appeared to be related to the higher concentration of dead cells.  It seems that dead cells enhance the aggregation process while increasing the bleed rate improves viability.  Apparently, cellular aggregation was reduced in higher viability cultures.   Figure 5-10 :  Aggregate size distribution for perfusion cultures at day 6 for low bleed rate (A) and high bleed rate (B) operation. The vertical dashed line represents the average diameter for each culture.  5.3.4 Effects of Cellular Aggregation on Permittivity Measurements On-line cell density was measured by permittivity signals during both perfusion cultures. The results demonstrated that permittivity measurements were a reliable indicator of the biomass content, i.e., of the viable cell count during all phases of the two cultures (Figure 5-11).  The measured permittivities correlated well with the offline cell counts during the  114 exponential growth phase when the bioreactor cultures were exhibiting a low aggregation rates and a high viability, with correlation coefficients (permittivity vs. viable cell concentration) of R2 = 0.98 and 0.94 for the low and high bleed rate cultures, respectively. In the later stages of the perfusion culture (days 7-30), the correlation coefficients for permittivity vs. viable cell concentration regressions were comparatively low.  However, the viable cell concentrations for both cultures during this period did not vary a lot and thus, linear regressions were not as representative.   Figure 5-11: Permittivity profiles for two perfusion cultures using low bleed rate  (A) and high bleed rate (B). The solid line represents the permittivity signal and open circles represent viable cell counts.  To evaluate if the on-line signal could yield a prediction of the viable cell count in the later culture phase, the evolution of the specific permittivity (corresponding to the permittivity per cell) was calculated for the two cultures (Figure 5-12). If the dielectric properties of the cell suspension are not altered, the specific permittivity is expected to be constant independent of the cell concentration. However, permittivity measurements also reflect changes in cell size, viability and physiological state (Ansorge et al. 2009; Cannizzaro et al. 2003; Ducommun et al. 2001; Noll and Biselli 1998; Zeiser et al. 2000). In this case, distinct differences between the growth and perfusion phases were observed  115 in the values of the specific permittivity. A small offset in the permittivity signal affected the early values for the high bleed rate perfusion culture but the subsequent values and their evolution were similar for both bleed rates. Whereas a higher specific permittivity was observed in the growth phase, the value declined to a lower but almost constant value during the perfusion phase. During that phase, the specific permittivity was not significantly different in the two cases. However, the particle size distributions were distinctly different for both cultures. The presence of large aggregates is expected to have an impact on the specific permittivity value (Davey 1993; Ron et al. 2009).  Considering that cell concentrations in the low bleed rate culture were underestimated, the specific permittivity results for this culture were most likely lower than for the culture at the high bleed rate. For both cultures, the medium conductivity was in a similar range of ~14-15 mS cm-1 (data not shown), so artifacts due to changes in conductivity were excluded as reasons for the changes in permittivity.   Figure 5-12 : Evolution of the specific permittivity for two perfusion cultures using high bleed rate (solid circles, solid line) and low bleed rate (circles, dashed line). To ensure a good signal to noise ratio, only data points with viable cell counts > 1 x 106 cells mL-1 were included in the graph.   116 Dilution experiments were performed for the different cultures to evaluate if the permittivity signal was affected significantly or not by the presence of aggregated cells. First, the high bleed rate bioreactor culture was diluted with PBS in several steps and the permittivity recorded (Table 5-3). The dilution resulted in a linear decrease in permittivity with a high correlation coefficient, indicating that the permittivity signal was a robust indicator of the viable cell count during the perfusion phase. In addition, the previously described batch experiments (section 5.3.3) were subjected to sequential dilutions with PBS (Table 5-3). The slope was similar for all experimental conditions with high R2 values. The only significant difference was observed for the batch culture at high calcium concentration, even though the addition of calcium did not significantly change the medium conductivity compared to that of the standard culture medium (data not shown).  The correlation coefficient was lower for this experimental condition as a result of a larger variability in the permittivity measurements for this highly aggregated cell suspension.  Table 5-3. Correlation coefficients of permittivity results vs. viable cell concentrations for dilution experiments in batch and perfusion cultures. Experimental Conditions Slope R2 Batch data Standard conditions 0.21 0.99 Low agitation 0.21 0.93 Addition of dead cells 0.20 0.96 Addition of calcium 0.40 0.74 Perfusion data High bleed rate 0.22 0.97  The overall results showed that, under typical operating conditions, with average particle diameters ranging from 30 to 166 µm, the differences in aggregation rates should not affect the permittivity and, therefore, should not lead to limitations when estimating the  117 viable cell count from permittivity measurements (Figure 5-11 and Figure 5-12). However, during the two independent perfusion cultivations, the correlation of permittivity with viable cell count from the exponential growth phase did not hold in the pseudo-steady-state perfusion phase. Permittivity measurements are a function of the biomass content of a cell suspension, but they also reflect changes in cell size, viability and physiological state as described previously. A large number of parameters could therefore be responsible for the differences observed in the two culture phases (growth and perfusion). The viability remained relatively high in both cultures (~80% in the low bleed rate and 90% in the high bleed rate culture) but decreased noticeably from the growth to the perfusion phase.  A more detailed discussion on this topic is beyond the scope of the current work.  Nevertheless, it is possible that the decrease in specific permittivity could be, at least in part, also related to increased cellular aggregation.  The small-scale batch experiments suggest that there may be an effect of a very high aggregation rate on the relationship between permittivity and viable cell count.  In summary, however, it can be concluded that on-line in situ permittivity measurements are a robust tool to measure viable biomass even when cellular aggregation occurs.  5.3.5 Effects of Cellular Aggregation on Calculations and Production The effect of aggregation on culture productivity is not well understood.  Most of the existing studies looked at cell viability inside aggregates in order to project a hypothesis on how it could affect the productivity rather than examining directly the product titer and culture productivity. IFN-α2b concentrations have been measured for both batch and perfusion cultures.  The results are presented in the Table 5-4.  Product levels for the two  118 cultures with the most aggregates showed similar concentrations to those obtained in the other corresponding cultures.  The batch with the addition of calcium reached an IFN-α2b concentration of 140.2 mg L-1, slightly lower than the control batch at 159.1 mg L-1.  For the low bleed rate perfusion culture, the pseudo-steady state product concentration was at 262.2 mg L-1, higher than for the high bleed rate culture at 225.0 mg L-1.  Surprisingly, the cell specific production rates for these two aggregated cultures were higher than in the other cultures.  However, it is essential to note that the calculations were performed using cell counts that did not include large cell aggregates.  The cell concentrations were not properly determined for these cultures.  Table 5-4.  Productivity comparison between batch and perfusion cultures. *Calculations did not include cells in aggregates.  **Calculations were done including an estimate of cells in aggregates. Experimental IFN Titer qIFN* qIFN** Conditions mg L-1 pg cell-1 day-1 pg cell-1 day-1 Batch Control 159.1 38.5 - + dead cells 181.5 31.0 - + calcium 140.2 52.8 37.0 Low RPM 147.6 31.0 - Perfusion Low Bleed 262.9 26.0 12.5 High Bleed 225.0 21.4 -  In order to evaluate the impact of aggregated cells on the calculation of cell culture metabolic rates, it is important to obtain a good estimate of the true cell concentration.  A cell sample of the batch with calcium was analyzed at day 6 using trypsin to break the aggregates.  A cell count was performed at different times in order to evaluate the action of trypsin on the aggregates (Figure 5-13).  Viable cell concentration leveled off at ~1.9 x106 cells mL-1 almost 9 times higher than for the first reading.  However, this value might still be underestimated because the aggregates did not break down completely after  119 coming in contact with trypsin.  Trypsin acts on both dead cells and the viable cells within the aggregate.  During this pseudo-equilibrium, viable cells were disaggregated while new viable cells were released from the aggregates.  The viability increased gradually possibly because dead cells were disaggregated faster than viable cells.   Figure 5-13 : Viable cell concentration and viability of aggregates from the batch with calcium exposed to trypsin. Error bars represent standard deviation of the cell count.  The trypsin experiment clearly showed that the cell concentration was significantly underestimated in the presence of large aggregates (>100 µm).  Viable cell concentrations were estimated for the low bleed rate perfusion culture by not including aggregates larger than 100 µm.  Therefore, new viable cell concentrations were approximated by adding cells from the fraction of particle larger than 100 µm and from the new cell concentration obtained during the trypsin experiment.  Based on new cell concentration estimates (that included cells in all of the aggregates) the cell specific production rates were recalculated (Table 5-4).  These corrected cell specific production rates were significantly lower for the two highly aggregated cultures.  This effect adds another cause for reproducibility problems during cell culture comparisons.  Thus, it is important to consider cell  120 aggregates in culture kinetic calculations.  The effect of aggregation on culture productivity seems to negatively influence cell specific productivity.  Only the batch with calcium gave a higher cell-specific production rate compared to those of the control batch, but as discussed, its cell concentrations were likely still underestimated.  However, cell specific productivity was mainly associated with the growth rate for this cell line (Drouin et al., submitted).  Therefore, cell aggregation could have had an impact on both the growth rate and the productivity.  5.4 Conclusions Cellular aggregation is a natural phenomenon that results in culture heterogeneity.  It is probable that many researchers have neglected analysis of aggregates due to a lack of convenient and effective measuring methods.  In this study, an automated image analysis technique was developed to measure aggregates of all sizes.  Results were comparable to those obtained from the more sophisticated commercial CEDEX analyzer in the range of small sized aggregates (< 40 µm).  The evolution of aggregates in batch and perfusion culture was efficiently monitored using this technique.  The on-line monitoring of the viable cell concentration by permittivity measurements was not influenced by aggregates with diameters up to 166 µm.  However, if larger aggregates are neglected, underestimation of cell density resulted in cell specific-rate estimation errors.  Based on batch and perfusion culture analyses, it appeared that the main cause of aggregation in perfusion cultures was due to the accumulation of dead cells.  The adapted perfusion process strategy developed in Chapter 4 demonstrated that aggregation could be reduced significantly by increasing the bleed rate.  121 5.5 References Abramoff MD, Magelhaes PJ, Ram SJ. 2004. Image Processing with ImageJ. Biophotonics International 11(7):36-42. Ansorge S, Esteban G, Schmid G. 2007. On-line monitoring of infected Sf-9 insect cell cultures by scanning permittivity measurements and comparison with off-line biovolume measurements. Cytotechnology 55:115-124. Ansorge S, Esteban G, Schmid G. 2009. Multi-Frequency Permittivity Measurements Allow For On-Line Monitoring of Changes In Intracellular Conductivity Due To Nutrient Limitations During Batch Cultivations Of CHO Cells. Biotechnol Progr in print. Beving H, Eriksson LEG, Davey CL, Kell DB. 1994. Dielectric properties of human blood and erythrocytes at radio frequencies (0.2–10 MHz); dependence on cell volume fraction and medium composition. European Biophysics Journal 23(3):207-215. Bierau H, Perani A, al-Rubeai M, Emery AN. 1998. A comparison of intensive cell culture bioreactors operating with hybridomas modified for inhibited apoptotic response. J Biotechnol 62(3):195-207. Boraston R, Marshall C, Norman P, Renner G, Warner J. Elimination of cell aggregation in suspension cultures of Chinese hamster ovary (CHO) cells. In: Spier RE, Griffiths JB, MacDonald C, editors; 1992. Brackenbury R, Rutishauser U, Edelman GM. 1981. Distinct calcium-independent and calcium-dependent adhesion systems of chicken embryo cells. Proc Natl Acad Sci U S A 78(1):387-391. Cannizzaro C, Gugerli R, Marison I, von Stockar U. 2003. On-line biomass monitoring of CHO perfusion culture with scanning dielectric spectroscopy. Biotechnol Bioeng 84(5):597-610. Carvell J, Dowd J. 2006. On-line Measurements and Control of Viable Cell Density in Cell Culture Manufacturing Processes using Radio-frequency Impedance. Cytotechnology 50(1):35-48. Coco-Martin JM, Oberink JW, Vanderveldendegroot TAM, Beuvery EC. 1992. Viability Measurements of Hybridoma Cells in Suspension-Cultures. Cytotechnology 8(1):57-64. Coppen SR, Newsam R, Bull AT, Baines AJ. 1995. Heterogeneity within Populations of Recombinant Chinese-Hamster Ovary Cells Expressing Human Interferon-Gamma. Biotechnology and Bioengineering 46(2):147-158. Cote J, Garnier A, Massie B, Kamen A. 1998. Serum-free production of recombinant proteins and adenoviral vectors by 293SF-3F6 cells. Biotechnol Bioeng 59(5):567-75.  122 Davey CL. 1993. The Biomass Monitor Source Book. Ltd AI, editor. Aberystwyth: Department of Biological Sciences, University of Wales. Drouin H, Lanthier S, Piret JM, Kamen A, Durocher Y. Submitted. Influence of Culture Medium Osmolality on Interferon-alpha 2b Production Stability in HEK293 Cells. Process Biochemistry. Drouin H, Ritter JB, Gorenflo VM, Bowen BD, Piret JM. 2007. Cell separator operation within temperature ranges to minimize effects on Chinese hamster ovary cell perfusion culture. Biotechnol Prog 23(6):1473-84. Ducommun P, Bolzonella I, Rhiel M, Pugeaud P, von Stockar U, Marison IW. 2001. On- line determination of animal cell concentration. Biotechnol Bioeng 72(5):515-22. Ducommun P, Kadouri A, von Stockar U, Marison IW. 2002. On-line determination of animal cell concentration in two industrial high-density culture processes by dielectric spectroscopy. Biotechnol Bioeng 77(3):316-23. Garnier A, Cote J, Nadeau I, Kamen A, Massie B. 1994. Scale-up of the adenovirus expression system for the production of recombinant protein in human 293S cells. Cytotechnology 15(1-3):145-55. Gorenflo VM, Angepat S, Bowen BD, Piret JM. 2003. Optimization of an acoustic cell filter with a novel air-backflush system. Biotechnol Prog 19(1):30-6. Graham FL, Smiley J, Russell WC, Nairn R. 1977. Characteristics of a human cell line transformed by DNA from human adenovirus type 5. J Gen Virol 36(1):59-74. Illing S, Harrison STL. 1999. The kinetics and mechanism of Corynebacterium glutamicum aggregate breakup in bioreactors. Chemical Engineering Science 54(4):441- 454. Jo EC, Yun JW, Jung KH, Chung SI, Kim JH. 1998. Performance study of perfusion cultures for the production of single-chain urokinase-type plasminogen activator (scu- PA) in a 2.5 l spin-filter bioreactor. Bioprocess Engineering 19(5):363-372. Kamen AA, Bédard C, Tom R, Perret S, Jardin B. 1996. On-line monitoring of respiration in recombinant-baculovirus-infected and uninfected insect cell bioreactor cultures. Biotechnology and Bioengineering 50:34-48. Kamen AA, Tom R, Caron AW, Chavarie C, Massie B, Archambault J. 1991. Culture of insect cells in a helical ribbon impeller bioreactor. Biotechnol Bioeng 38:619-628. Litwin J. 1992. The growth of Vero cells in suspension as cell-aggregates in serum-free media. Cytotechnology 10(2):169-74.  123 Loignon M, Perret S, Kelly J, Boulais D, Cass B, Bisson L, Afkhamizarreh F, Durocher Y. 2008. Stable high volumetric production of glycosylated human recombinant IFNalpha2b in HEK293 cells. BMC Biotechnol 8:65. Moreira JL, Alves PM, Aunins JG, Carrondo MJ. 1994a. Changes in animal cell natural aggregates in suspended batch cultures. Appl Microbiol Biotechnol 41(2):203-9. Moreira JL, Alves PM, Aunins JG, Carrondo MJ. 1995. Hydrodynamic effects on BHK cells grown as suspended natural aggregates. Biotechnol Bioeng 46(4):351-60. Moreira JL, Feliciano AS, Santana PC, Cruz PE, Aunins JG, Carrondo MJ. 1994b. Repeated-batch cultures of baby hamster kidney cell aggregates in stirred vessels. Cytotechnology 15(1-3):337-49. Noll T, Biselli M. 1998. Dielectric spectroscopy in the cultivation of suspended and immobilized hybridoma cells. J Biotechnol 63(3):187-98. Peshwa MV, Kyung YS, Mcclure DB, Hu WS. 1993. Cultivation of Mammalian-Cells as Aggregates in Bioreactors - Effect of Calcium-Concentration on Spatial-Distribution of Viability. Biotechnology and Bioengineering 41(2):179-187. Pribush A, Meiselman HJ, Meyerstein D, Meyerstein N. 1999. Dielectric approach to the investigation of erythrocyte aggregation: I. Experimental basis of the method. Biorheology 36(5):411-423. Renner WA, Jordan M, Eppenberger HM, Leist C. 1993. Cell-cell adhesion and aggregation: Influence on the growth behavior of CHO cells. Biotechnol Bioeng 41(2):188-93. Ron A, Fishelson N, Croitoriu N, Benayahu D, Shacham-Diamand Y. 2009. Theoretical examination of aggregation effect on the dielectric characteristics of spherical cellular suspension. Biophysical Chemistry 140(1-3):39-50. Sen A, Kallos MS, Behie LA. 2001. Effects of hydrodynamics on cultures of mammalian neural stem cell aggregates in suspension bioreactors. Industrial & Engineering Chemistry Research 40(23):5350-5357. Zeiser A, Bedard C, Voyer R, Jardin B, Tom R, Kamen AA. 1999. On-line monitoring of the progress of infection in Sf-9 insect cell cultures using relative permittivity measurements. Biotechnol Bioeng 63(1):122-6. Zeiser A, Elias CB, Voyer R, Jardin B, Kamen AA. 2000. On-line monitoring of physiological parameters of insect cell cultures during the growth and infection process. Biotechnol Prog 16(5):803-8.    124 6 Conclusions and Future Work Perfusion cultures have the potential to fulfill the increasing demand for biopharmaceuticals based on their high volumetric productivity.  However, many challenges limit industrial acceptance of this technology.  The longer-term duration (i.e. months) of perfusion cultures is a feature that makes them more dependent on the stability of recombinant protein productivity.  In this study, it was found that exposure to a high osmolality medium (~ 375 mOsm kg-1) resulted in decreased IFN-α2b secretion rates from HEK293 viable cells.  As a consequence, in order to maintain productivity over long perfusion runs, the medium osmolality needed to be maintained at approximately 300 mOsm kg-1.  Since an exposure period of hyperosmotic stress during culture has been reported to increase the cell specific productivity (Lin et al. 1999; Ozturk and Palsson 1990; Wu et al. 2004), it would be interesting to investigate transient exposures to hyperosmotic stress during perfusion cultures as an alternative to increase overall process productivity.  The flow through with cell retention inherent to perfusion culture would have the advantage that osmotic levels could be manipulated in this way. More broadly, investigations to determine if there are significant influences from other medium variables on production stability could identify other critical variables that should be more closely monitored and controlled in perfusion processes to maintain production stability.  The HEK293 cell line used in this study was found to have a growth-associated production of IFN and, therefore, it was important to maintain a high growth rate during culture to maintain high productivity.  Perfusion process operation has the advantage of  125 allowing control of the cell density and cell growth rate by varying the bleed rate.  The bleed rate is often minimized to obtain higher cell concentrations (Banik and Heath 1995; Leelavatcharamas et al. 1999).  However, it was shown that using an increased bleed rate during a perfusion culture for a growth-associated production cell line improved culture performance.  This process variable was found to be critical to developing a well-adapted culture strategy for this type of cell line.  For the perfusion culture of new cell lines the growth association of the production should be tested to evaluate how the bleed rate should be optimized.   Operation at low perfusion rates is a process option that can increase product titers to ease downstream recovery and also this can help to avoid exceeding the capacity of cell separators.  In this study, two approaches have been used to further improve perfusion performance.  Inspired by the push-to-low approach (Konstantinov et al. 2006), the feed rate was reduced while nutrient content was increased.  In addition, the use of a transient scanning approach (Angepat et al. 2005) to optimize reduced perfusion rates, allowed the rapid identification of a feeding strategy suitable for this HEK293 cell line.  Using this adapted strategy yielded IFN-α2b at concentrations up to 303.5 mg L-1, a 35% increase compared to the standard perfusion conditions.  A simple mathematical model was developed to help further analyze the results.  This model could be improved by accounting for inhibition due to metabolite accumulations, including any distinct influences on cell growth rates, survival and protein productivity.  Also, the influence by these variables on the crucial-to-maintain product quality (e.g. glycosylation) should be determined.  126  In perfusion culture, it is important to find a balance between nutrient addition to promote growth and productivity vs. the resulting metabolite accumulations that can inhibit the cell growth and reduce the culture productivity.  For the case of this HEK293 cell line, enriching the medium with amino acids at a low perfusion rate resulted in reduced metabolite levels and overall improved process performance.  Further reduction in perfusion rates should be explored with serine and other selectively-enriched media, based on nutrient depletion analysis.  Using selectively supplemented media should significantly reduce nutrient precipitation.  In this case, the media could be stored for longer periods of time.  Cellular aggregation is a commonly observed phenomenon in perfusion cell culture. However, if the cells in aggregates are neglected, major errors in the calculation of cell specific rates can result.  In this study, computer-aided image analysis was used to follow the evolution of cellular aggregation in batch and perfusion cultures.  Based on the batch results, the dead cells in perfusion were predicted to be a major cause of aggregation in perfusion cultures.  Thus, increasing the bleed rate was tested in perfusion in order to increase the culture viability and this did result in reduced aggregation.  To further confirm the influence on nonviable cells, longer-term tests could be performed by adding nonviable cells to perfusion cultures.  The developed image analysis method was a simple and efficient way to estimate the cellular content of aggregates and improve viable cell estimates and cell specific rate  127 analysis.  The image analysis technique could be further improved by increasing the number of images taken and by using a larger sample volume so that the analysis of the largest aggregates could be included without risk of increasing falsely the large particle volume fraction.  The results with the permittivity measurement could then be compared to explore how effective that method would be in analyzing the largest aggregates.  In addition, these results could be compared with packed cell volume measurements, a method that is used by industry to measure total biomass content.  The packed cell volumes should correlate well with total cell volume measurements based on a complete aggregate analysis.  In conclusion, perfusion process development could be facilitated and performance significantly improved by considering the following guidelines: • Maintain osmolality within a range that limits production instability during long-term cultivation; • Better understand cell line growth and production kinetics to define an adapted perfusion culture strategy, such as to maximize the bleed rate for growth-associated production cell lines; • Reduce perfusion rates using enriched medium to increase product titer; • Improve medium nutrient balance to reduce metabolite production by identifying key nutrients to reduce or supplement; • Use a transient scanning approach to accelerate process development; • Develop a model to predict and compare with culture results;  128 • Improve the monitoring of cellular aggregation to further enable optimization of the methods to reduce aggregation and to reduce cell concentration calculation errors.  129 6.1 References Angepat S, Gorenflo VM, Piret JM. 2005. Accelerating perfusion process optimization by scanning non-steady-state responses. Biotechnol Bioeng 92(4):472-8. Banik GG, Heath CA. 1995. Hybridoma Growth and Antibody-Production as a Function of Cell-Density and Specific Growth-Rate in Perfusion Culture. Biotechnology and Bioengineering 48(3):289-300. Konstantinov K, Goudar C, Ng M, Meneses R, Thrift J, Chuppa S, Matanguihan C, Michaels J, Naveh D. 2006. The "push-to-low" approach for optimization of high-density perfusion cultures of animal cells. Adv Biochem Eng Biotechnol 101:75-98. Leelavatcharamas V, Emery AN, Al-Rubeai M. 1999. Use of cell cycle analysis to characterise growth and interferon-gamma production in perfusion culture of CHO cells. Cytotechnology 30(1-3):59-69. Lin J, Takagi M, Qu Y, Gao P, Yoshida T. 1999. Enhanced monoclonal antibody production by gradual increase of osmotic pressure. Cytotechnology 29:27-33. Ozturk SS, Palsson BO. 1990. Effect of Medium Osmolarity on Hybridoma Growth, Metabolism and Antibody Production. Biotechnology and Bioengineering 37:989-993. Wu MH, Dimopoulos G, Mantalaris A, Varley J. 2004. The effect of hyperosmotic pressure on antibody production and gene expression in the GS-NS0 cell line. Biotechnol Appl Biochem 40(Pt 1):41-6.   130 Appendix 1:  Raw Data Chapter 3 data     Data for Figure 3-2 Time Xv Viability IFN Glucose Serine Lactate NH3 Osmolality day 106 cell ml-1 % mg L-1 mM mM mM mM mOsm kg-1 0,00 0,2 98,0 2,6 25,5 3,4 4,3 0,0 279 0,91 0,5 95,9 17,0 14,2 18,4 0,2 281 1,98 0,9 98,3 2,9 2,97 1,9 96,6 3,90 2,2 98,2 55,0 7,0 30,0 1,0 282 5,02 2,3 96,3 6,18 2,6 92,6 120,0 3,0 1,7 28,5 2,5 278 6,98 3,1 82,9 8,06 2,5 70,5 167,8 0,0 1,3 27,2 3,6 277 9,01 1,6 40,2 170,0 0,0 27,0 3,8 280 9,88 0,9 29,0 Data for Figure 3-3 Time Xv Viability IFN Glucose Serine Lactate NH3 Osmolality day 106 cell ml-1 % mg L-1 mM mM mM mM mOsm kg-1 0,00 0,3 97,3 11,3 19,5 0,4 2,9 0,8 308 0,95 0,5 98,2 1,93 0,8 97,2 38,3 15,3 0,1 12,2 1,6 308 2,99 1,4 97,0 4,20 1,6 96,0 60,4 12,2 12,8 2,3 306 6,11 1,7 93,0 83,7 10,0 0,0 10,8 3,0 304 7,00 1,7 92,0 7,99 1,6 86,0 84,9 8,0 0,0 10,8 3,2 302 Data for Figure 3-4 Time Xv Viability IFN Glucose Serine Lactate NH3 Osmolality day 106 cell ml-1 % mg L-1 mM mM mM mM mOsm kg-1 0,00 0,2 96,1 15,5 20,1 0,8 3,0 0,5 313 0,85 0,4 97,9 1,78 0,6 97,9 33,8 15,5 0,6 11,4 1,2 308 2,79 1,0 98,3 3,81 1,6 97,0 64,6 10,0 16,0 2,5 301 5,86 2,4 94,0 121,3 6,8 0,0 12,0 3,2 296 6,98 2,5 93,0 7,78 2,7 86,0 139,2 4,0 9,8 3,3 291 10,09 2,4 72,0 146,6 2,0 0,0 10,0 2,3 291 Data for Figure 3-5 Freestyle LC-SFM MLC-SFM Time !app qIFN Time !app qIFN Time !app qIFN day day-1 pg cell-1 day-1 day day-1 pg cell-1 day-1 day day-1 pg cell-1 day-1 0 - - 0 - - 0 - - 2 0,643 14,8 1,9 0,559 28,8 1,8 0,598 26,3 3,9 0,477 13,8 4,2 0,299 8,6 3,8 0,465 14,4 6,2 0,081 11,9 6,1 0,041 7,5 5,9 0,194 13,8 8,1 -0,022 9,8 8 -0,037 0,4 7,8 0,055 3,6 10,1 -0,055 1,3  131       Data for Figure 3-6 Time Estimated Growth Rate day (day-1) 4,00 0,624 5,95 0,443 8,79 0,360 11,79 0,143 12,83 0,132 14,68 -0,203 16,99 -0,101  132     133 Chapter 4 data     134   135      136       137    138 Chapter 5 data       139      140           141 Appendix 2:  MatLab Code Simple Perfusion Culture Model  % ************************************************ % MathLab program to simulate perfusion bioreactor operations % ************************************************  % To start the simulation write 'culturesim' in the command window  clear all, close all  % ************************* % Parameters definitions % *************************  duree=36*24;        % ‘durée’ is the simulation duration in hrs  % ******************** % Initial conditions % ********************  X0=4.5e5;    %cells/mL V0=2.75;      % Litres S0=1.081;     %mM Substrat L0=5.1;      %mM Metabolite P0=24.97;      %mg/L Produit  y_init=[V0 X0 S0 L0 P0];  % ************* % Time Vector % *************  dt=0.1; tspan=0:dt:duree;  % ****** % Solver % ******  y=ode4('eq_perf',tspan,y_init);  V=y(:,1); X=y(:,2); S=y(:,3); M=y(:,4); P=y(:,5);  Var_etat=[V X S M P];     142 % ************************************** % Obtention of flow rates and kinetics % **************************************  inc=1;  for time=tspan(1):dt:tspan(length(tspan))     Flow(inc,:)=operation(time);     Kin(inc,:)=kinetic(time,Var_etat(inc,:));     inc=inc+1; end  F0=Flow(:,1);            % Feeding F=Flow(:,2);             % Outlet B=Flow(:,3);             % Bleed mu=Kin(:,1);            %Growth rate kd=Kin(:,2);            %Dead rate qs=Kin(:,3);            %Consuption rate qm=Kin(:,4);            %Metabolite production rate qp=Kin(:,5);            %Product production rate   % *************** % Screen display % ***************  figure(1) subplot(211) plot(tspan,X) xlabel('Temps (h)') ylabel('Concentration Cellulaire') subplot(212) plot(tspan,S) xlabel('Temps (h)') ylabel('Substrat (mM)')  figure(2) subplot(211) plot(tspan,M) xlabel('Temps (h)') ylabel('Metabolite (mM)') subplot(212) plot(tspan,P) xlabel('Temps (h)') ylabel('Produit (mg/L)')  figure(3) subplot(311) plot(tspan,F0) xlabel('Temps (h)') ylabel('Alimentation (L/h)') subplot(312) plot(tspan,F) xlabel('Temps (h)') ylabel('Sortie (L/h)') subplot(313)  143 plot(tspan,B) xlabel('Temps (h)') ylabel('Bleed (L/h)')  figure(4) plot(tspan,V) xlabel('Temps (h)') ylabel('Volume (L)')  figure(5) plot(tspan,mu) xlabel('Temps (h)') ylabel('Taux de croissance')  figure(6) plot(tspan,qs) xlabel('Temps (h)') ylabel('Taux de consommation')  [d1,d2]=size(X); vec=zeros(d1,5); vec(:,1)=tspan(1,:)'; vec(:,2)=X(:,1); vec(:,3)=S(:,1); vec(:,4)=M(:,1); vec(:,5)=P(:,1);  fprintf('\n At t= %4.0f h',tspan(1,d1)) fprintf('\n Xv = %10.2e cell/ml',X(d1,1)) fprintf('\n S = %10.1f mM',S(d1,1)) fprintf('\n M = %10.1f mM',M(d1,1)) fprintf('\n P = %10.4f mg/L',P(d1,1))  save my_data.out vec –ASCII     function dy=eq_perf(t,y)  % ******************************************************* % Fonction for dynamic model for perfusion bioreactor procès % *******************************************************  t_feed1=1.132*24;                  % Time for beginning of feeding at 1.00x t_feed11=16.965*24;             % Time for beginning of feeding at 1.15x t_feed12=21.990*24;             % Time for beginning of feeding at 1.30x t_feed13=28.993*24;             % Time for beginning of feeding at 1.45x   % **************** % Fixe Parameters % ****************  eff=0.95;          % Separation efficiency of acoustic filter Sin=1.33;         % Substrat concentration in the feed (mmol/L)  144 Min=0;             % Metabolite concentration in the feed (mmol/L)  if t>t_feed11     Sin=1.48;      % Serine Concentration     Min=0; end  if t>t_feed12     Sin=1.67;     Min=0; end  if t>t_feed13     Sin=1.83;     Min=0; end   % ******************************* % Assignation of state variables % *******************************  V=y(1);         % Volume X=y(2);         % Cell Concentration S=y(3);         % Substrat M=y(4);        % Metabolite P=y(5);         % Product  Var_etat=[V X S M P];  Xh=X*(1-eff);   % Outlet cell concentration  % *********************************************************************** % Call Function « opération » that will define different flow rates % in function of time (feed, harvest, bleed) % ***********************************************************************  Flow=operation(t);  F0=Flow(1);      % Alimentation F=Flow(2);        % Sortie B=Flow(3);        % Bleed  % Remarque:   F0 = approximativement (B + F)  % ********************************************************************** % Call function "kinetic" that will return kinetic rates in function % of state variables and time % **********************************************************************  Kin=Kinetic(t,Var_etat);  mu=Kin(1);           %Growth rate kd=Kin(2);            %Dead rate qs=Kin(3);             %Consumption rate qm=Kin(4);           %Metabolite production rate  145 qp=Kin(5);            %Product production rate  % *************************** % General dynamic model % (Invariable ...) % ***************************  dV=F0-F-B; dX=mu*X-(F/V)*Xh-(B/V)*X-kd*X - X/V*dV; dS=-qs*X*1000+(F0/V)*(Sin)-(F+B)/V*S-S/V*dV;               % x1000 to get cell/L dM=qm*X*1000+(F0/V)*(Min)-(F+B)/V*M-M/V*dV;         % x1000 to get cell/L dP=qp*X*1000-(F+B)/V*P-P/V*dV;                                           % x1000 to get cell/L  % ************** % Vectorisation % **************  dy=[dV dX dS dM dP]';     function Flow=operation(t)  % *********************************************************** % Function used to define flow rate; feed, % outlet and bleed in function of time % ***********************************************************  t_feed1=1.132*24;                  % Beginning of feeding at 1.00x t_feed11=16.965*24;             % Beginning of feeding at 1.15x t_feed12=21.990*24;             % Beginning of feeding at 1.30x t_feed13=28.993*24;             % Beginning of feeding at 1.45x  t_bleed=4.080*24;                  % Beginning of bleeding operation  % **************************** % Definitino of feeding % ****************************  F0=0;       % L/h  if t>t_feed1     F0=0.569*2.75/24; end  if t>t_feed11     F0=0.533*2.75/24; end  if t>t_feed12     F0=0.400*2.75/24; end  if t>t_feed13     F0=0.354*2.75/24;  146 end   % ******************* % Definition of bleed % *******************  if t<=t_bleed     B=0.00623*2.75/24;           % L/h else     if t<10.87*24         B=(0.0241*t/24-0.0834)*2.75/24;    % L/h     else         B=0.1672*2.75/24;    % L/h     end end  % *********************** % Definition of outlet % ***********************   F=F0-B;  % ************** % Vecteurisation % **************  Flow(1)=F0; Flow(2)=F; Flow(3)=B;     function Kin=kinetic(t,var_etat)  % *********************************************** % Function that will return kinetics, growth rate, % consumption and production rates % in function of state variables and/or time % ***********************************************  V=var_etat(:,1); X=var_etat(:,2); S=var_etat(:,3); M=var_etat(:,4); P=var_etat(:,5);  % ******************************************* % Time related to operation % *******************************************  t_feed1=1.132*24;                  % Beginning of feeding at 1.00x t_feed11=16.965*24;             % Beginning of feeding at 1.15x t_feed12=21.990*24;             % Beginning of feeding at 1.30x  147 t_feed13=28.993*24;             % Beginning of feeding at 1.45x  t_sparging=3.0208*24;          % Beginning of air sparging  % ************************************** %    Growth rate % **************************************   % ****************** % Parametres to be fixed % ******************  mumax=0.01717;       % Growth rate maximal in bioreactor (h-1)     mu=mumax;             % Calculated during exponential phase      if S<0.07         mu=0.0078;  % Calculer apres aÈration, rÈgion limitante     end  % ************************************** %  Dead rate % **************************************  % ****************** % Parametres to be fixed % ******************  % Basis case  kd=0;                     % This rate has been included into growth rate (Apperent growth rate)    % ********************************************************* %     Consumption rate  (sernine, limiting nutrient) % *********************************************************  % qs  (mmol/cell/h)  qs=1.12e-11;        % Calculated from BP3 culture, batch mode      if S<0.06         qs=3.06e-13;     end  % *************************************************************** %       Metabolite production rate (Lactate or NH3) % ***************************************************************  % ****************** % Parametres a fixer % ******************  % qm  (mmol/cell/h)   148 %************************************************************************************ %LACTATE RATES   (modify initial Lac value in culturesim.m (LO) and feed in Eq_perf.m (Min)) %************************************************************************************ qm=1.89e-10;                  % Calculate during exponential phase      if S<0.07         qm=4.1906e-11;     % Calculate after air sparging in limiting region     end  %************************************************************************************ %NH3 RATES  (modify initial NH3 value in culturesim.m (LO) and feed in Eq_perf.m (Min)) %************************************************************************************ %qm=1.2725e-11;       % Calculated from days BP3 culture expo phase  %if S<0.07 %   qm=6.819e-12;  % Calculate after air sparging in limiting region %end  %************************************************************************************ % Glucose RATES  (modify initial Glc value in culturesim.m (LO) and feed in Eq_perf.m (Min)) %************************************************************************************ %qm=-1.525e-10;       % Calculer durant phase exponentielle  %if S<0.07  %  qm=-7.21e-11;  % Calculate after air sparging in limiting region  %  if M<=0  %      qm=0;  %  end  %end  % ********************************************* %           Product formation rate % *********************************************  %IFN production (mg/L),  %qp   (mg/cell/hrs)  qp=1.762e-9;              % Calculate during exponential phase before air sparging      if S<0.07         qp=8.9e-10;         % Calculate after air sparging, limiting region     end  % ************** % Vecteurisation % **************  Kin(1)=mu; Kin(2)=kd; Kin(3)=qs; Kin(4)=qm; Kin(5)=qp;  %fin   149 Appendix 3:  Picture Analysis Protocol How to use ImageJ macro MultipleAggAnalys  Description  The frame of this simple macro is a file opener that individually opens files from a text file list and processes them according to a sequence of commands before they are closed. This frame is useful for any application where several images (in one directory) are processed/analyzed in the same way. The main part of the macro is dedicated to the analysis of particles using the ImageJ Particle Analyser. Results are saved to an Excel file.  General Remarks  • To be able to run the macro the plugin “Excel Writer” should be installed. http://rsb.info.nih.gov/ij/plugins/excel-writer.html • The preparation of the images for the Particle Analyser (Contrast Enhancement, Thresholding) is largely dependent on the quality of the images and should be optimized for individual types of images. • The parameters within the sequence of commands that is responsible for setting the scale should be adjusted if different scales are used.   Also, the parameters for the particle analysis can be adjusted.  Running the macro  • The file opener requires that a text file with all the filenames of the image files that will be processed is generated. • The macro should be located in the ImageJ macro-folder • Open the macro in ImageJ with Plugins => Edit … • Carry out necessary modifications (change paths and filenames, adjust analysis algorithm, …) • Run Macro • First Dialog box asks to open  the textfile that contains the filenames • Second Dialog box asks to select the directory that contains the images • Scale is set • Images are processed sequentially and results are saved to an excel file. Create a worksheet for each image to allow simple identification of corresponding data.    150 VBA macros to sort/process/analyse data  In order to be able to process the large amount of data generated for each sample more easily, VBA macros have been developed.  Here are examples of these macros and their application.  1. Macro ProcessSortData  • This macro is especially useful if the results for each image are saved in a separate worksheet. • Before running this macro, the very first worksheet of the file (book) needs to be activated. • The macro goes through each worksheet, calculates the average diameter and copies it - together with other parameters - into a new worksheet, so that – in the end – all the essential data for one sample are in a single worksheet.  Before running each of the following macros you need to activate the newly created worksheet, which has to be named “Sheet1”.  2. Macro PerCentVolumeDistribution  • Generates a size distribution  3. Macro VolumeDistributionHistogram  • Another size distribution diagram. Particle sizes larger than 40 microns are pooled in ranges of 20.  4. Macro Histogram  • Real histogram (number/percent of particles over diameter) for particles larger than 50 microns   151 Appendix 4:  Example of images   Image of 10 µm calibration beads.    Image of cells in batch with normal conditions.   152  Image of cells in batch culture with the addition of calcium.    Image of cells in batch culture inoculated with dead cells.   153   Image of cells in perfusion culture at low bleed rate.    Image of cells in perfusion culture at high bleed rate.     154 Appendix 5:  Image J Code // This macro opens and processes a set of images from a text file list     macro "MultipleAggAnalys" {   //Select and open textfile that contains image file list       open("C:\\BP2\\BP2-31\\list1.txt");        contents = getInfo();       run("Close");       dir = getDirectory("Select the Directory that contains the images");       if (indexOf(contents, "Title: ")>=0)           exit("Text file containing list of files expected");       list = split(contents, "\n");   //sets scale, change path for scale image accordingly       open("C:\\BP3\\BP3-1\\Calibration964C.jpg");       setTool(4);   //different scales require different parameters       run("Set Scale...", "distance=14 known=19 pixel=1 unit=micrometer global");       run("Close");       run("Set Measurements...", "area mean min circularity redirect=None decimal=3");   //loop that opens and processes every image in the file list depending on the analysis algormithm that is applied       for (i=0; i<list.length; i++) {           file = dir+list[i];           open(file);           run("Enhance Contrast", "saturated=0.5");           run("8-bit");           run("Threshold");           run("Analyze Particles...", "size=80-Infinity circularity=0.00-1.00 show=Outlines display exclude include summarize");    155 //the following lines contain commands for particle analysis, results are saved to an excel file  //change path accordingly           run("Excel...", "select...=[C:\\BP2\\BP2-31\\Result.xls]");           run("Close");           run("Close");           selectWindow("Results");           run("Close");     }  }   

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            data-media="{[{embed.selectedMedia}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0058878/manifest

Comment

Related Items