UNIVERSITY OF CALGARY. Serum-free Co-expansion of Mesenchymal Stem Cells and Chondrocytes as Aggregates. in Suspension Bioreactors.

Transcription

1 UNIVERSITY OF CALGARY Serum-free Co-expansion of Mesenchymal Stem Cells and Chondrocytes as Aggregates in Suspension Bioreactors by Madiha Khurshid A THESIS SUBMITTED TO THE FACULTY OF GRADUATE STUDIES IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE DEPARTMENT OF CHEMICAL AND PETROLEUM ENGINEERING CALGARY, ALBERTA DECEMBER, 2014 Madiha Khurshid 2014

2 Abstract Defects in articular cartilage can develop into osteoarthritis, a condition that affects 10% of Canadians. A clinical method of cartilage repair involves the implantation of articular chondrocytes (ACs) into the defect. Mesenchymal stem cells (MSCs) can also be used in this application, after differentiation into chondrocytes. It is difficult to expand either of these cell populations without decreasing the quality of cartilage tissues produced. However, the co-culture of ACs and MSCs can produce enhanced cartilage tissues in tissue culture flasks (T-flasks). The scale-up of this co-expansion under serum-free conditions is necessary for clinical translation. A serum-free medium (SFM) called PPRF-msc6 was shown to support cell coexpansion. Additionally, these cells were successfully co-expanded in scalable suspension bioreactors as three-dimensional aggregates, achieving cell densities of 95,000 cells/ml and producing more matrix than in T-flasks. Suspension bioreactors can co-expand ACs and MSCs as aggregates using SFM for potential utility in cartilage repair strategies. ii

3 Acknowledgements I would like to thank all the individuals who supported me throughout my thesis project. First, I would like to acknowledge my supervisor, Dr. Arin Sen, without whom this thesis would not be possible. He encouraged me to pursue my master s degree in the first place and taught me a great deal throughout this degree program, everything from technical knowledge to communication skills. I am also grateful to him for supporting me in the difficult times during this degree program. In addition, I want to thank my co-supervisor, Dr. Adetola Adesida, for his encouragement, expertise and guidance. His contribution was invaluable to my project and to my understanding of the biology behind it. I would like to thank my lab mates for their assistance as well as comradery over the years. They made the lab a great place to work. I also want to acknowledge Ms. Aillette Mulet- Sierra for carrying out several assays and coordinating the shipments with me. It was a lot of work for her and I really appreciated it. Lastly, I would like to thank my family and my husband for their support. Also, I really appreciate all the rides they gave me to and from the lab at odd times and their patience at all the times I couldn t be there with them. It has been a long road and I m thankful that they were with me on this journey. iii

4 Table of Contents Abstract... ii Acknowledgements... iii Table of Contents... iv List of Tables... vii List of Figures and Illustrations... viii List of Symbols and Abbreviations... xiv CHAPTER 1: SCOPE OF THESIS Motivation Thesis Overview...3 CHAPTER 2: BACKGROUND Introduction Articular Cartilage and Osteoarthritis Current Treatments for Cartilage Repair Cell Culture for Cartilage Repair Co-culture of Chondrocytes and Mesenchymal Stem Cells Cell Culture in Hypoxia Serum-Free Media Cell Culture Scale-up Suspension Bioreactors Scale-up as Aggregates Objectives...23 CHAPTER 3: MATERIALS AND METHODS Introduction Cell Lines Cell Culture Media Serum-Containing Media Serum-Free Media Static Culture Procedures Gelatin Coating Cell Culture Cryopreservation of Single Cells Cell Resuscitation Bioreactor Culture Procedures Bioreactor Preparation Bioreactor Inoculation Bioreactor Feeding Serial Passaging of Cells in Bioreactors Analytical Procedures Aggregate Dissociation Manual Cell Density and Viability Measurements Automated Cell Density and Viability Measurements Aggregate Characterization...41 iv

5 3.6.5 Cell Culture Media Analysis Histology Gene Expression GAG and DNA Content Analysis Design of Experiments Passage Level Donors and Replicates Factorial Experiments Statistics and Error Analysis...52 CHAPTER 4: BIOPROCESS OVERVIEW Statement of Contribution Title Page Abstract Introduction Materials and Methods Cell harvest and expansion Media screening Bioreactor and static co-culture Nutrient and metabolite analysis Aggregate characterization Biochemical assays Histology Gene expression Statistics Results Serum-Free Media Screening in Static Co-culture Extension of Viable Bioreactor Co-culture Period Serial Passaging in Bioreactor Co-Culture Feeding in Bioreactor Co-Culture Comparison of Bioreactor and Static Co-culture Protocols Discussion Conclusions Competing Interests Author s Contributions Acknowledgements Ethical considerations List of Abbreviations CHAPTER 5: EFFECT OF AGITATION RATE ON CO-CULTURED AGGREGATES IN BIOREACTORS Introduction Growth Phase Agitation Rate Cell Quantity Aggregate Characteristics Cell Quality Inoculation Phase Agitation Rate v

6 5.3.1 Cell Quantity Aggregate Characteristics Cell Quality Conclusions CHAPTER 6: EFFECT OF OXYGEN TENSION AND DONORS IN BIOREACTOR AND STATIC CO-CULTURE Introduction Role of Oxygen Tension in Static Co-culture Role of Oxygen Tension in Bioreactor Co-culture Comparison of Bioreactor Co-culture to Static Co-culture Effect of Different Donors on Bioreactor Co-culture Conclusions CHAPTER 7: CONCLUSIONS AND RECOMMENDATIONS Conclusions Recommendations CHAPTER 8: REFERENCES APPENDIX A: PREPARATION OF PPRF-MSC APPENDIX B: FACTORIAL EXPERIMENT RESULTS APPENDIX C: AGGREGATE ECCENTRICITY APPENDIX D: COPYRIGHT PERMISSION vi

7 List of Tables Table 2.1: Culture of mammalian cells as aggregates in smooth-walled spinner flask bioreactors as reported in the literature Table 3.1: Donor Information...26 Table 3.2: List of primer sequences used in RT-PCR...46 Table 4.1: Primer sequences used in quantitative real-time polymerase chain reaction. All primers were purchased from Life Technologies Table 4.2: Kinetic parameters for hmsc and hac bioreactor co-culture for batch and fed-batch operation modes Table 4.3: Kinetic parameters for hmsc and hac co-cultures in static and bioreactor conditions. Literature values for bovine ACs (92), hmscs (80,93,94) and murine MSCs (95) in static culture are included for comparison. Note that literature values for hac/hmsc co-culture in static conditions were not available, and literature values for bioreactor culture of hacs or hmscs were also not available Table 6.1: Kinetic parameters for hmsc and hac bioreactor co-cultures in normoxic and hypoxic conditions. Literature values are shown for bovine ACs (92), hmscs (80,93,94) and murine MSCs (95) in static culture only under normoxia, because values for static co-culture, bioreactor culture and hypoxic conditions are not available Table A.1: Formulation of PPRF-msc6 (-) Table B.1: Results of Factorial Experiment vii

8 List of Figures and Illustrations Figure 2.1: Healthy and osteoarthritic knee joints. Modified from (27)....5 Figure 2.2: Stages in the formation of hypertrophic cartilage in vivo and in vitro. The cell types (highlighted in pink), factors (green) and matrix proteins and enzymes (blue) are also shown during the progression of this process. Adapted from (1,6,13)....7 Figure 2.3: Safranin O staining of pellets that were isolated and expanded in hypoxia or normoxia and subsequently, differentiated under both oxygen tension levels. Modified from (52) Figure 3.1: Factorial experiment design to test effect of oxygen tension and donors and compare static co-culture to bioreactor co-culture. There were eight cell pools of different donors, cell types and oxygen tensions. All eight conditions were tested in duplicates Figure 4.1: Serum-free media screening. Three media (SCM, TheraPEAK and PPRFmsc6) were screened for their ability to support the proliferation of hacs and hmscs in static co-culture. All experiments were conducted in duplicate. Cells were passaged three times for 6 days each without medium changes. The first passage was inoculated at a 1 hac to 3 hmsc ratio and each passage was inoculated at 20,000 cells/ml in 4 ml of medium. A) Cell densities at day 6 in the three media are shown over three passages. Error bars show range of data for duplicate cell counts from duplicate conditions. Photomicrographs of cells in static co-culture at the third passage on day 6 are shown at 5x magnification for B) SCM, C) TheraPEAK and D) PPRF-msc Figure 4.2: Photomicrographs of co-cultured aggregates. hmscs were stained with PKH67 (green) and hacs were stained with PKH26 (red) prior to inoculation in bioreactors in a ratio of 1 hac to 3 MSCs. Bioreactors were inoculated with 60 ml of SCM and agitated at 60 rpm. Representative samples of aggregates were imaged A) 4, B) 7 and C) 10 days after inoculation Figure 4.3: Serial passaging of cells in bioreactor co-culture. Bioreactors were operated at 60 rpm for the first two days followed by 80 rpm. Cells were cultured for 10 days without medium changes for each of two passages. Aggregates harvested from the first bioreactor were dissociated with trypsin-edta and inoculated (20,000 cells/ml) into new bioreactor vessels as single cells in fresh medium. A) Cell densities in bioreactor co-culture in two consecutive passages are shown. Error bars show range of data for duplicate cell counts from duplicate conditions. B) Safranin O staining of cells co-cultured in the first passage is shown. C-D) Photomicrographs of bioreactor co-cultures for each of two passages are shown at 10x magnification after 10 days in culture viii

9 Figure 4.4: Feeding cells in bioreactor co-culture cell density and cell morphology. Effect of feeding was tested in bioreactor co-culture with regards to cell density and morphology. A) Cell densities are shown in bioreactor co-culture in the batch and fed-batch conditions. Error bars show range of data. Green arrows indicate time points for 50% medium change for the fed-batch condition. Photomicrographs of hmsc and hac aggregates at B-C) 10 days and D-E) 16 days in culture are shown at 10x magnification Figure 4.5: Feeding cells in bioreactor co-culture GAG levels and aggregate morphology. A) GAG, B) DNA and C) GAG/DNA of the aggregates are shown in the batch and fed-batch conditions after 19 days in culture. Error bars show standard error of the mean. Safranin O staining of cells co-cultured in the D) batch and E) fed-batch conditions are shown. F) Average aggregate diameter is shown over the culture period. Error bars show standard deviation. Green arrows indicate time points for 50% medium change for the fed-batch condition. G) Aggregate diameter distribution after 16 days in culture is shown Figure 4.6: Feeding cells in bioreactor co-culture nutrient consumption and waste production. The cumulative A) glucose consumption, B) lactic acid production, C) glutamine consumption and D) ammonia production are shown in both conditions. Error bars show range of data for duplicate samples from duplicate cultures. Green arrows indicate time points for 50% medium change in the fedbatch condition Figure 4.7: Comparison of bioreactor and static culture protocols cell density and cell morphology. Cell densities in A) static co-culture and B) bioreactor co-culture are shown. Arrows indicate time points for 50% medium change. Photomicrographs of co-cultures of hmscs and hacs are shown at 10x magnification after 10 days in culture in C) static culture flasks and D) bioreactors. White ovals highlight areas of stratified monolayers Figure 4.8: Comparison of bioreactor and static culture protocols GAG levels and aggregate morphology. A) Average aggregate diameter and B) average aggregate eccentricity are shown. Error bars show standard deviation. Green arrows indicate time points for 50% medium change. C) Aggregate diameter distribution after 16 days in culture is also shown. D) GAG, E) DNA and F) GAG/DNA in the bioreactor and static conditions are shown after 10 days in culture. Error bars show standard error of the mean Figure 4.9: Comparison of bioreactor and static culture protocols gene expression levels and GAG deposition. The gene expression, relative to β-actin, for the static and bioreactor conditions is shown after 10 and 16 days in culture respectively for C) collagen I, collagen X, aggrecan, COMP, and D) collagen II. Error bars show range of data. Safranin O staining of cells co-cultured in the bioreactor condition is shown after C) 10 and D) 16 days in culture ix

10 Figure 4.10: Comparison of bioreactor and static culture protocols nutrient consumption and waste production. The cumulative A) glucose consumption, B) lactic acid production, C) glutamine consumption and D) ammonia production are shown in both conditions. Error bars show range of data for duplicate samples from duplicate cultures. Arrows indicate time points for 50% medium change..91 Figure 5.1: Growth phase agitation rate was tested in bioreactor co-culture with respect to cell densities and viabilities. Bioreactors were inoculated in a 1 hac to 3 hmsc ratio with 20,000 cells/ml in 125 ml medium and operated at 60, 80 and 100 rpm. Cell densities in each agitation rate were different from cell densities at other agitation rates (p<0.05). Error bars show range of data of duplicate cell counts from duplicate cultures Figure 5.2: Aggregate morphology is shown for three agitation rates. Photomicrographs of co-cultured aggregates A-C) 1 day and D-F) 10 days in culture are shown at 10x magnification Figure 5.3: Aggregate diameters are shown at several time points during the bioreactor co-culture for three agitation rates. A) Average aggregate diameters and B) maximum aggregate diameters in the three agitation rates is shown. Error bars on average diameters show standard deviation of 20 samples from duplicate cultures. Error bars on maximum diameters show range of data from duplicate cultures Figure 5.4: Aggregate diameter distribution is shown for three agitation rates after A) 10 and B) 14 days in culture Figure 5.5: A) Aggregate densities and B) eccentricities are shown at three agitation rates. Error bars on aggregate densities show range of data from duplicate cultures. Error bars on eccentricities show standard deviation of 20 samples from duplicate cultures Figure 5.6: Growth phase agitation rate was tested with regards to GAG deposition. A) GAG/DNA of the aggregates at different agitation rates is shown after 16 days in culture. Error bars show standard error of the mean. B) Safranin O staining of aggregates cultured at 80 rpm is shown after 16 days in culture. Scale bar represents 100 µm. Data was collected by Aillette Mulet-Sierra Figure 5.7: Growth phase agitation rate was tested with regards to gene expression. Gene expression of co-cultured cells relative to β-actin at different agitation rates is shown for A) collagen I, B) collagen X, COMP, C) collagen II and aggrecan. Error bars show range of data from duplicate cultures. An asterisk (*) indicates a statistically significant difference from the 60 rpm condition. Data was collected by Aillette Mulet-Sierra Figure 5.8: Inoculation phase agitation rate was tested in bioreactor co-culture with respect to cell densities and viabilities. Bioreactors were operated at either 60 rpm or 80 rpm for the first two days, followed by 80 rpm for the remainder of the x

11 culture. Error bars show range of data of duplicate cell counts from duplicate cultures Figure 5.9: Aggregate morphology is shown for two inoculation agitation rates. Photomicrographs of aggregates cultured at A-D) 60/80 rpm and E-H) 80/80 rpm are shown at 10x magnification after 1, 6, 10 and 12 days in culture. Scale bars represent 200 µm Figure 5.10: A) Aggregate densities and B) eccentricities are shown for two inoculation agitation rates. Error bars on aggregate densities show range of data from duplicate cultures. Error bars on eccentricities show standard deviation of 20 samples from duplicate cultures Figure 5.11: A) Average and B) maximum aggregate diameters are shown at several time points during the bioreactor co-culture for two inoculation agitation rates. Error bars on average diameters show standard deviation of 20 samples from duplicate cultures. Error bars on maximum diameters show range of data from duplicate cultures Figure 5.12: Aggregate characteristics are shown at two inoculation agitation rates. A) Aggregate diameter distribution is shown after 10 days in culture for both conditions. B) Packing density is shown at days 2, 6, 8 and 10 in the two conditions. Error bars show range of data from duplicate cultures Figure 5.13: Inoculation phase agitation rate was tested with regards to GAG deposition. A) GAG/DNA is shown in both conditions after 13 days in culture. Error bars show standard error of the mean. Safranin O staining of aggregates cultured at inoculation agitation rates of B) 60 rpm and C) 80 rpm are shown after 13 days in culture. Scale bars represent 100 µm. Data was collected by Aillette Mulet-Sierra Figure 5.14: Inoculation phase agitation rate was tested with regards to gene expression. Expression of A) collagen I, B) collagen II and X, aggrecan and COMP, relative to β-actin, is shown in both conditions after 13 days. Error bars show range of data from duplicate cultures. Data was collected by Aillette Mulet- Sierra Figure 6.1: Oxygen tension was tested in hmsc and hac static co-culture in duplicates. T-flasks were inoculated in a 1 hac to 3 hmsc ratio with 20,000 cells/ml in 4 ml PPRF-msc6. A) Cell densities and viabilities in normoxic (21% O2) and hypoxic conditions (3% O2) are shown. Error bars show range of data of duplicate cell counts from duplicate cultures. Photomicrographs of co-cultured cells in static culture are shown at 10x magnification for both conditions after B-C) 6 days and D-E) 10 days in culture. Oval highlights area of stratified monolayers. Scale bars represent 100 µm Figure 6.2: Gene expression and GAG production of co-cultured cells in T-flasks was determined in normoxia and hypoxia after 10 days in culture. The gene expression xi

12 relative to β-actin is shown for A) collagen I, collagen X, aggrecan, COMP, and B) collagen II. C) GAG/DNA of the co-cultured cells in normoxia and hypoxia is shown after 10 days in culture. Data was collected by Aillette Mulet-Sierra Figure 6.3: Oxygen tension was tested in hmsc and hac bioreactor co-culture. Bioreactors were inoculated in a 1 hac to 3 hmsc ratio with 20,000 cells/ml in 125 ml PPRF-msc6. Bioreactors were operated at 60 rpm for the first two days and 80 rpm for the remainder of the culture period. A) Cell densities and viabilities in normoxic (21% O2) and hypoxic conditions (3% O2) are shown. Error bars represent range of data of duplicate cell counts from duplicate cultures. B) The apparent growth rate in the exponential growth phase (days in normoxia and days 4-12 in hypoxia) are shown. Error bars show range of data from duplicate cultures. C) The GAG/DNA of the bioreactor-generated aggregates is shown after 16 days in culture. Error bars show standard error of the mean. GAG/DNA data was collected by Aillette Mulet-Sierra Figure 6.4: Aggregate morphology at the two oxygen tension levels are shown. Photomicrographs of co-cultured cells in bioreactor culture are shown at 10x magnification for normoxic and hypoxic conditions after A-B) 4 days and C-D) 10 days in culture. Scale bars represent 100 µm Figure 6.5: Gene expression and Safranin O staining of co-cultured aggregates is shown in normoxia and hypoxia after 16 days in culture. The gene expression relative to β-actin was different in the conditions. Expression of A) collagen X, aggrecan, COMP, B) collagen I, and collagen II is shown. Error bars show range of data. An asterisk (*) indicates a statistically significant difference between the conditions. Safranin O staining of co-cultured aggregates cultured in C) normoxic and D) hypoxic conditions is shown. Scale bars represent 50 µm. Data was collected by Aillette Mulet-Sierra Figure 6.6: Cumulative A) glucose consumption and B) lactic acid production are shown in normoxic and hypoxic conditions. Error bars show range of data of duplicate samples from duplicate cultures Figure 6.7: Cumulative A) glutamine consumption and B) ammonia production are shown in normoxic and hypoxic conditions. Error bars show range of data of duplicate samples from duplicate cultures Figure 6.8: Bioreactor co-culture was compared to static co-culture in duplicate. Both conditions were inoculated in a 1 hac to 3 hmsc ratio with 20,000 cells/ml in medium under hypoxia (3% O2). A) The apparent growth rate in the exponential growth phase (days 4-12 in bioreactors and days 0-6 in static culture) is shown for both conditions. Error bars show range of data from duplicate cultures. B) GAG/DNA of the bioreactor-generated aggregates and the static monolayer are shown after 10 days in culture. Error bar shows standard error of the mean. Gene expression of co-cultured cells was determined in static culture after 10 days in culture and in bioreactor culture after 16 days in culture. The gene expression xii

13 relative to β-actin is shown for C) collagen I, collagen X, aggrecan, COMP, and D) collagen II. Error bars show range of data from duplicate cultures. GAG/DNA and gene expression data was collected by Aillette Mulet-Sierra Figure 6.9: Donor-to-donor variability was tested in bioreactor co-culture. Bioreactors were inoculated in a 1 hac to 3 hmsc ratio with 20,000 cells/ml in 125 ml PPRF-msc6 under hypoxia (3% O2). Bioreactors were operated at 60 rpm for the first two days and 80 rpm for the remainder of the culture period. A) Cell densities and viabilities using donor pairs A and B are shown. Error bars show range of data of duplicate cell counts from duplicate cultures. B) The apparent growth rate in the exponential growth phase (days 4-12 in donor pair A and days 4-10 in donor pair B) are shown. Error bars show range of data from duplicate cultures. C) GAG/DNA of the aggregates in the two different donor pairs is shown. Error bars show standard error of the mean. GAG/DNA data was collected by Aillette Mulet- Sierra Figure 6.10: Aggregate morphology in the two donor pairs are shown. Photomicrographs of co-cultured cells in bioreactor culture under hypoxia are shown at 10x magnification for two donor pairs after A-B) 10 days and C-D) 16 days in culture. Scale bars represent 100 µm Figure 6.11: Safranin O staining and gene expression analysis was performed in two sets of donors in bioreactor co-culture under hypoxia. Safranin O staining of cocultured aggregates cultured in hypoxic conditions using A) donor pair A and B) donor pair B is shown after 10 days in culture. Scale bars represent 50 µm. The gene expression, relative to β-actin, in the two donor pairs is shown for A) collagen I, collagen X, aggrecan, COMP, and B) collagen II after 16 days in culture. Error bars show range of data from duplicate cultures. An asterisk (*) indicates a statistically significant difference between the conditions. Data was collected by Aillette Mulet-Sierra Figure C.1: Eccentricity of A) a circle, and B-C) ellipses. Modified from Math Open Reference (127) xiii

14 List of Symbols and Abbreviations Symbol Definition C0 CV CT Cagg D NV NX NY NX Y PD R 2 Δt td VPBS VS VH Vagg ɛ µapp Cell density upon inoculation Viable cell density Total cell density Aggregate density Aggregate diameter Number of viable cells Number of samples in first group Number of samples in second group Total number of samples Packing density Linear regression coefficient of determination Time in culture Doubling time Volume of phosphate-buffered saline Volume of cell sample Volume of haemocytometer Volume of an aggregate Aggregate eccentricity Apparent growth rate µx Mean of the first group µy Mean of the second group µx Y σ X Mean of the entire sample Standard deviation of the first group σ Y σ X Y Standard deviation of the second group Standard deviation of the entire sample xiv

15 Abbreviation 2D 3D αmem ACAN ACI ALP BM BMP COL1A2 COL2A1 COL10A1 COMP DMEM DMSO DNA ECM EDTA ESC FBS FGF-2 GAG hac HCl HEPES hmsc HSA Ihh ipsc MACI MMP Definition Two-dimensional Three-dimensional Modified Eagle s medium α Aggrecan (gene) Autologous chondrocyte implantation Alkaline phosphatase Bone marrow-derived mesenchymal stem cell Bone morphogenetic protein Collagen I (gene) Collagen II (gene) Collagen X (gene) Cartilage oligomeric matrix protein (gene) Dulbecco s modified Eagle medium Dimethyl sulfoxide Deoxyribonucleic acid Extracellular matrix Ethylenediaminetetraacetic acid Embryonic stem cell Fetal bovine serum Fibroblast growth factor-2 Glycosaminoglycan Human articular chondrocyte Hydrochloric acid 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid Human mesenchymal stem cell Human serum albumin Indian hedgehog (protein) Induced pluripotent stem cell Matrix-associated chondrocyte implantation Matrix metalloprotease xv

16 NSC Neural stem cell OA Osteoarthritis p(number) Passage (number) PBS Phosphate-buffered saline PCL Poly(3-caprolactone) PDGF Platelet-derived growth factor PenStrep Penicillin and streptomycin solution PPRF Pharmaceutical production research facility PTHrP Parathyroid hormone-related protein RNA Ribonucleic acid RT-PCR Reverse transcriptase polymerase chain reaction SA Specific aim SCM Serum-containing medium SFM Serum-free medium T-25 Tissue culture flask with growth area of 25 cm 2 T-75 Tissue culture flask with growth area of 75 cm 2 T-flask Tissue culture flask TGF-β Transforming growth factor beta xvi

17 CHAPTER 1: SCOPE OF THESIS 1.1. Motivation Cartilage is a connective tissue that is found in various places in the body. Articular cartilage lines articulating joints and aids in the low-friction movement of joints. This cartilage is largely composed of matrix, generated by resident cells called chondrocytes. The matrix contains molecules that strengthen the tissue and help it absorb shock (1). Cartilage defects can be caused by injury as well as wear and tear. These defects start a degenerative process that can cause a painful, chronic condition called osteoarthritis (OA) (2,3). OA affects 10% of adults in Canada (4) and 25% of people over 50 years old (5). Cartilage is avascular, so it has a limited capacity for self-repair (6,7). Thus, clinical cartilage repair strategies are required both to prevent the degeneration of cartilage defects into OA, and to replace the cartilage lost due to OA. Current methods used clinically to repair cartilage include microfracture and cartilage grafting. However, these therapies have major drawbacks such as the formation of mechanically inferior cartilage that lacks durability (1,6,8). Autologous chondrocyte implantation (ACI) is a cell-based therapy in which the patient s chondrocytes are: (i) isolated from the surrounding matrix, (ii) expanded in culture and (iii) transplanted back into the defect site either alone or together with a scaffold. In this treatment, there is no disruption of the subchondral plate. The cells need to be expanded, since there are not sufficient quantities to fill the defect after cell isolation (10 6 cells/cm 2 defect) (9). Larger quantities of cells are required to resurface larger areas affected by OA (>25 cm 2 ) (9), whereas the repair of smaller focal cartilage defects ( cm 2 ) (10) requires fewer cells. Despite some clinical success, ACI has limitations. Chondrocytes tend to lose the ability to produce cartilage matrix in culture (through a process called dedifferentiation) (1,6,11), so it is difficult to generate large quantities of cells that retain this characteristic. In 1

18 addition, the implantation of dedifferentiated chondrocytes produces low quality cartilage (8,11). Furthermore, the chondrocytes from patients suffering from OA have poor characteristics for cartilage repair such as: slower cell growth, less matrix production and hypertrophy (9,12 14). Hypertrophy is undesirable in implanted cells, since it is characterized by cell enlargement, as well as vascularization and calcification of the cartilage matrix (15,16). Thus, improvements to the cell culture protocol are required such that a large number of OA cells, which retain the ability to produce cartilage matrix, can be generated. Such improvements would allow the treatment of (i) larger cartilage defects, (ii) autologous treatment of OA, and (iii) more durable cartilage repair. Mesenchymal stem cells (MSCs) are cells that are capable of producing more unspecialized cells like themselves as well as differentiating into various specialized cells. MSCs proliferate extensively and can differentiate into chondrocytes, and thus also demonstrate utility in potential cartilage therapies (1,6,8,11). MSCs also have to be expanded in culture, but their ability to produce cartilage decreases after prolonged culture (8,17), and the cartilage tissues produced are hypertrophic (11). Chondrocytes as well as MSCs are typically expanded in tissue culture flasks (T-flasks) as two-dimensional (2D) monolayers. However, articular chondrocytes (ACs) and MSCs tend to lose their cartilage-forming characteristics in 2D culture (8,17,18). Three-dimensional (3D) culture has been shown to maintain ACs in a cartilage-producing state and induce chondrogenic differentiation in MSCs (6,11,19 22). 3D culture can be carried out as pellets, aggregates and using scaffolds. MSCs have also been cultured as aggregates in suspension bioreactors with greater differentiation potential towards adipogenic and osteogenic lineages (23). Recently, it has been shown that the co-culture of ACs and hmscs as 3D pellets can produce cartilage tissues that have properties that are more like native cartilage, than the culture of either cell type alone (17,24,25). In this co-culture system, hmscs experience chondrogenesis and hacs proliferate without losing their characteristics (17,26). Co- 2

19 cultures also increase cartilage matrix content and decrease hypertrophy as compared to the culture of either cell type (13,17,24). However, previous studies have been carried out on a small-scale in static T-flasks, which represents an impediment to clinical translation. Culture in T-flasks is inefficient and labour-intensive, so alternative methods are needed to generate the large quantities of cells required for ACI Thesis Overview This thesis describes the development of a bioprocess that can efficiently co-expand populations of ACs and MSCs on a large-scale using suspension bioreactors. Chapter 2 explains the background required to understand the thesis and its objectives. The third chapter explains the materials and experimental methods used, including statistical analyses. In addition, it outlines the experimental design and the challenges encountered in designing the experiments. Chapter 4 presents a paper submitted to the journal, Arthritis Research & Therapy. This paper introduces the bioprocess, screens different culture media, tests methods to improve its performance and evaluates bioreactor co-culture against static co-culture. The fifth chapter examines the effect of agitation rate on the bioprocess. The last results chapter tests the role of oxygen tension and donors on the bioreactor co-culture, and compares the performance of bioreactor co-culture to that in T-flasks. Finally, a conclusion is provided with recommendations for future work. 3

20 CHAPTER 2: BACKGROUND 2.1. Introduction Cartilage defects can cause severe pain and the loss of quality of life in patients. Current treatments are unable to produce replacement cartilage of sufficient quality. However, there is a cartilage repair strategy that has demonstrated success clinically. It involves the in vitro expansion of the patient s own cartilage cells and subsequent implantation into the defect site. The in vitro expansion of these cartilage cells yields low quantities of cells with poor cartilage matrix production characteristics. Thus, new culture methods are being explored by researchers to achieve large quantities of matrix-producing cells and thus, improve clinical results. This chapter reviews these novel culture techniques, which form the context for the development of the bioprocess described in this thesis. The proposed bioprocess aims to produce large numbers of cells, which have applicability in cartilage repair therapies, in a clinically-relevant manner using suspension bioreactors Articular Cartilage and Osteoarthritis Articular cartilage is the connective tissue that covers the articulating surfaces of bones. This tissue, together with the surrounding synovial fluid, aides in the nearly friction-free movement of joints, adapts and distributes compressive loads, and allows virtually frictionless movement. This tissue is avascular and aneural. As such, it only has a limited capacity for self-repair (1,6,8,11). Defects in articular cartilage can be caused by normal wear and tear of joints, or as a result of injury. These defects can lead to further cartilage degeneration and eventually result in OA. OA is a chronic disease, which is characterized by large surface lesions on the cartilage, as illustrated in Figure 2.1 (1). These lesions cause severe pain and reduced mobility for patients. The incidence of OA is very high, affecting 25% of adults over 50 years old (5). Thus, replacements for cartilage damaged by this disease are required. 4

21 Figure 2.1: Healthy and osteoarthritic knee joints. Modified from (27). Healthy articular cartilage contains only hyaline cartilage. Thus, ideal cartilage substitutes should only have hyaline cartilage. Fibrocartilage and hypertrophic cartilage are common but inferior substitutes for articular cartilage. Hyaline cartilage is smooth, glassy and hard. It is composed largely of water and extracellular matrix (ECM), mainly collagen II and proteoglycans, which contain sulfated glycosaminoglycans (GAGs) (1). Collagen II lends the tissue strength and the proteoglycans help it resist mechanical load. Cartilage cells, called chondrocytes, are also present, although at low cell densities of 10 8 cells/cm 3 (28). The chondrogenic phenotype of these cells is characterized by the abundant production of collagen II and GAGs. There are also reports of progenitor cells in cartilage (29,30). Fibrocartilage is a fibrous type of cartilage, which is tough, dense and elastic. It is composed mainly of collagen I fibres. Fibrocartilage is suited to tendons and ligaments, which require greater tensile strength. It is produced as a repair response to severe cartilage damage, but is undesirable in articular cartilage due to its inferior mechanical properties (8,31). During human development, cartilage forms bone in vivo in a process called endochondral ossification (see Figure 2.2). MSCs (see Section 2.4 for details) differentiate into chondrocytes, which eventually become hypertrophic. Eventually, chondrocytes die off in 5

22 favor of osteoblasts that form bone. While this process is important in forming bone in vivo, hypertrophy of cells is not desirable when the goal is to produce cartilage in vitro. In the first stage, MSCs proliferate and generate collagen I and when exposed to certain factors such as transforming growth factor beta (TGF-β) and fibroblast growth factor-2 (FGF-2) differentiate into chondrocytes. The chondrocytes secrete many matrix molecules including: collagen II, aggrecan, cartilage oligomeric matrix protein (COMP) and GAGs (1,6). Chondrocytes that are exposed to parathyroid hormone-related protein (PTHrP) maintain their phenotype, but those that are not exposed to it (due to diffusional limitations) secrete Indian hedgehog (Ihh), which induces hypertrophy (13). Hypertrophy (10 cell enlargement (13)) of chondrocytes causes vascularization and calcification of cartilage (15,16). It is characterized by the markers: collagen X, alkaline phosphatase (ALP) and a matrix metalloprotease (MMP-13). Collagen X and ALP encourage the mineralization of matrix and MMP-13 degrades collagen II and aggrecan, which are integral to the cartilage matrix (13). Thus, healthy cartilage matrix is degraded in favor of undesirable hypertrophic cartilage. 6

23 Figure 2.2: Stages in the formation of hypertrophic cartilage in vivo and in vitro. The cell types (highlighted in pink), factors (green) and matrix proteins and enzymes (blue) are also shown during the progression of this process. Adapted from (1,6,13). 7

24 2.3. Current Treatments for Cartilage Repair There are cartilage repair treatments, which attempt to use the body s regenerative capacity, that are currently used i.e. microfracture and cartilage grafting. Microfracture involves drilling small holes into the bone so that cells in the bone marrow can leak into the joint space. However, this treatment results in undesirable fibrocartilage. In addition, cartilage grafts don t integrate well with the existing cartilage (1,6,8). Autologous grafts result in donor site morbidity. Allogeneic grafts have a limited supply and pose the risk of an adverse immune reaction (8). Cell-based strategies have also been investigated for utility in repairing cartilage defects. Cell-based therapies involve the use of cells to repair or regenerate tissues. Tissue engineering is a type of cell-based therapy that uses scaffolds and signals in addition to cells to achieve tissue repair. The signals can be biochemical and/or biomechanical in nature and include cell-to-cell bonds (1,6,8,11). A clinical intervention, which is used clinically to treat cartilage defects, involves the implantation of autologous human ACs (hacs), either alone or together in a scaffold, into defect sites to prevent the cartilage from undergoing progressive degeneration that leads to OA (8). This procedure is called ACI or matrix-associated chondrocyte implantation (MACI), when a scaffold or periosteal flap is employed. ACI and MACI are established treatments for cartilage defects (32) and have shown positive results clinically, with integration into the existing cartilage in 90% of patients (1). However, challenges remain in this approach, such as generating sufficient quantities of cells that retain a chondrogenic phenotype. ACI requires large numbers of cells (10 6 /cm 2 ) to fill in a defect (9). The size of focal defects range from cm 2 (10), but osteoarthritic joints have larger areas that require resurfacing (>25 cm 2 ) (9). Thus, the clinical implementation of this cell-based therapy to treat OA necessitates the ability to generate large quantities of high-quality cells. Obtaining sufficient quantities of hacs from donor sites also remains a challenge, due to the low chondrocyte densities in native cartilage (12-16 million cells/g tissue (33)), coupled 8

25 with the substantial losses in hacs as a result of the procedure to separate the cells from the surrounding matrix (<22% hac yield after cartilage digestion (33)). Thus, the in vitro culture of chondrocytes is required to expand populations of these cells and generate sufficient quantities for a cell-based therapy. Unfortunately, cell culture of ACs using current cell culture techniques leads to a loss of phenotype. Thus, new methods of cell culture need to be investigated to address this problem Cell Culture for Cartilage Repair Current cell-based therapies, such as ACI and MACI, necessitate the cell culture of chondrocytes in an effort to obtain large numbers of cells for implantation, as part of cartilage repair strategies. However, maintaining the cartilage-forming phenotype of the chondrocytes is difficult. There is dedifferentiation (loss of phenotype) and senescence of hacs during in vitro expansion (17). The dedifferentiation of hacs on 2D surfaces represents the loss of a chondrogenic phenotype, characterized by decreased production of collagen II, increased production of collagen I and a fibroblast-like morphology (1,6,11). In addition, the formation of hypertrophic cartilage and fibrocartilage post-implantation, with the use of these dedifferentiated cells, is problematic (13). For the autologous therapy of OA, healthy ACs may not be available and osteoarthritic chondrocytes may have to be utilized instead. OA ACs do not proliferate and produce cartilage matrix to the same extent as healthy chondrocytes using traditional cell culture techniques (9,14), so attempts to expand such cells will result in limited quantities of cells and these cells will be dedifferentiated. Thus, to facilitate the development of hac-based cartilage repair strategies, new culture techniques capable of expanding osteoarthritic hacs to clinical quantities, while maintaining chondrogenic characteristics, are required. There has been considerable research interest in using human MSCs (hmscs) in the tissue engineering of cartilage, since they proliferate extensively in culture while maintaining their characteristics and are capable of differentiating into a chondrogenic phenotype. Specifically, MSCs have the ability to self-renew and differentiate along multiple lineages 9

26 (multipotency), having the capacity to make fat, cartilage and bone tissues. hmscs can be isolated from various tissues including bone marrow, adipose tissue, synovial membrane and synovial fluid (34,35). Bone marrow is an accessible source of hmscs and it can be obtained using aspiration, so it is commonly used (8). Despite the many advantages of using hmscs, the chondrogenesis of MSCs can be challenging. Specifically, hmscs tend to lose their chondrogenic potential over time and undergo hypertrophy (8,17). Both the culture of hacs as well as hmscs have limitations in terms of obtaining sufficient quantities of cells with adequate quality for utility in cartilage repair strategies. Researchers have investigated several approaches to increase the quantity of chondrogenic cells and the quality of cartilage tissues obtained, including: 3D culture, medium supplementation, coculture (see Section 2.4.1), hypoxia (see Section 2.4.2), and mechanical stimulation (see Section 2.6.1). A combination of approaches will likely be required to achieve high quality cartilage tissues. 3D culture is important in the redifferentiation of chondrocytes and the chondrogenesis of MSCs (6,11,19 21,36). Growth factors also play an important role in promoting a chondrogenic phenotype and inhibiting hypertrophy, such as TGF-β3, TGF-β1, bone morphogenetic protein-6 (BMP-6), and PTHrP (26,34). Even transient exposure to growth factors can improve cartilage formation, especially in combination with other culture conditions Co-culture of Chondrocytes and Mesenchymal Stem Cells Recently, the co-culture of chondrocytes and MSCs has been investigated in an effort to produce improved cartilage tissues. Co-cultures prevented dedifferentiation of primary hacs (37), promoted redifferentiation of cultured chondrocytes, and chondrogenesis of MSCs with reduced hypertrophy (36). These benefits were reported mostly in direct contact co-culture. The chondrocytes produced the majority of the matrix in co-cultures with MSCs. The number of MSCs decreased over time in co-culture, so there was a lower percentage of MSCs at the end of the co-culture period (17,36). The mechanisms were 10

27 primarily trophic effects of MSCs on ACs (25,36,38) and AC secretion of PTHrP (26), which has been shown to be an important factor in decreasing hypertrophy (13). Cocultures required supplementation with growth factors, but even only short exposure times led to a stable cartilage phenotype (36,39). The greatest benefit of co-culture was the reduced need for chondrocytes to be harvested and passaged, with up to 80% reduction in chondrocyte numbers required to harvested and passaged (36,40). Interestingly, increasing the percentage of chondrocytes in MSC coculture improved the chondrogenic phenotype and decreased hypertrophy in MSCs (36,37). Co-cultures of primary and cultured chondrocytes also improved cartilage tissue formation, but this co-culture system still required large quantities of chondrocytes to be harvested from the patient (25). In addition, co-cultures of meniscus cells with MSCs also decreased hypertrophy, increased GAG content and collagen II expression (12). The Mikos research group has published several papers on the co-culture of ACs and MSCs in electrospun poly(3-caprolactone) (PCL) scaffolds (25,39,41 44). They found that cocultures of ACs and MSCs had higher GAG content than MSCs or cultured ACs alone after four weeks in culture, with levels similar to that observed with primary ACs. Co-cultures reduced the need for large numbers of primary ACs by replacing initial quantities of chondrocytes required with MSCs and promoting chondrocyte proliferation. This reduction resulted in lower areas of cartilage harvested and lower culture durations necessary for AC expansion. The MSCs had a trophic effect on the ACs, but decreased in number with time (44). Co-culture-seeded scaffolds, when implanted in a rat osteochondral defect model, produced hyaline-like cartilage similar to AC-seeded scaffolds, reducing the number of chondrocytes that were required to be harvested and expanded in culture (42). Acharya et al. have demonstrated that the co-culture of ACs with MSCs in scaffold-free, 3D static cell pellet cultures initiated chondro-induction. Chondro-induction was characterized by two effects: AC proliferation without dedifferentiation (which is induced by MSCs), and improvement of MSC chondrogenesis (which is induced by ACs). The co- 11

28 culture of ACs and MSCs led to enhanced cartilage formation as compared to monoculture of either cell type, as evidenced by higher GAG content, higher collagen II expression and lower collagen X expression (17). This group also tested the effect of cell-to-cell contact and the effect of soluble factors, using a transwell model, in AC and MSC co-culture. In cell-to-cell contact systems, ACs proliferated when in co-culture with MSCs and maintained their phenotype, and MSCs adopted a chondrogenic phenotype. Data from the transwell co-culture of ACs and MSCs indicated that the effect of soluble factors did not affect proliferation or chondrogenesis and that soluble factors alone could not account for chondro-induction (17). Zuo et al. also demonstrated that gene expression of collagen II and aggrecan was higher in direct contact co-culture than in the transwell system (45). Thus, it can be concluded that cell-to-cell contact is important for chondro-induction. The co-culture of MSCs with ACs resulted in a more robust chondrogenic differentiation (25 times lower collagen I expression and 46 times higher collagen II expression) and reduced hypertrophy (171 times lower collagen X expression), with a differentiated AC phenotype (17). The MSCs, in addition to the ACs, were able to deposit type II collagen. The number of MSCs decreased (about 2-4 times), but the number of ACs increased in coculture pellets (about 2-4 times) over a 3 week culture period. The decrease in number of MSCs in co-culture has been observed elsewhere in literature (38,40). Thus, the most noteworthy outcome of co-culture was the increased proliferation of differentiated chondrocytes. Chondro-induction was observed in AC co-cultures with MSCs derived from different sources (17). Acharya and colleagues measured the GAG produced in monocultures of ACs and MSCs, and used these values to calculate the amount of GAG that would be expected in a coculture of these two cell types (assuming that the GAG production rates remained constant) (Equation 2.1). Thus, the expected amount of GAG in co-culture was the sum of the proportionate amounts of GAG produced by MSCs and ACs in monoculture (17). 12

29 GAG expected = %ACs GAG AC + %MSCs GAG MSC Equation 2.1 To determine if the co-culture of these cell types supported increased GAG production rates, an interaction ratio was calculated (Equation 2.2), in which the GAG production measured in co-culture was compared to the expected value of GAG production (17). Interaction index = GAG measured GAG expected Equation 2.2 Interaction indices greater than unity suggested that co-culture of these two cell types upregulated the overall production of GAGs, which would be desirable in cell-based cartilage repair applications. This was used as quantitative evidence of chondro-induction, i.e. if the interaction index was greater than one. The interaction index was 1.4 in co-cultures of 1 AC to 9 MSCs and 1.7 in co-cultures of 1 AC to 3 MSCs. Thus, co-culture does up-regulate GAG production, with 70% more GAG being produced in a 1 AC to 3 MSC co-culture than what would be expected if each cell type was grown in monoculture (17). A ratio of 1 AC to 3 MSC has been reported to show positive results with an interaction index of 1.4 (37). It was also demonstrated that the level of chondro-induction (as measured by the interaction index) was not correlated to the chondrogenic capacity or quality of the two cell types used in the co-culture (as indicated by the GAG to deoxyribonucleic acid (DNA) ratio of each of the two cell types). Specifically, the linear regression coefficient of determination (R 2 ) of the interaction index to the GAG/DNA was (0.09) 2 and (0.31) 2 for MSCs and ACs respectively (17). Thus, the high levels of GAG up-regulation in co-culture were not a result of the quality of the ACs or MSCs used. So, OA ACs may stand to benefit from coculture with MSCs via chondro-induction, despite their low cartilage production capacities. 13

30 The co-culture of MSCs and ACs represents an important cell culture innovation that can contribute to the improvement of cell-based therapies for cartilage repair. Co-culture can improve both cell proliferation, phenotype and matrix quality. Co-culture approaches should be investigated further to reduce the requirement of high AC quantities, and improve the quality of generated cartilage tissues Cell Culture in Hypoxia Hypoxic conditions, typically 1-5% oxygen tension, have been tested for enhancing chondrogenesis in several different culture systems including: micropellets, macropellets, microcarriers, hydrogels, and other scaffolds. Hypoxia has been shown to promote the redifferentiation of ACs (46 48), the chondrogenesis of MSCs (49 52), and enhanced cartilage tissue in the co-culture of both these cell types (43). In these studies, hypoxic conditions improved cartilage tissue formation, demonstrated by an increase in the expression of collagen II, a decrease in collagen I and an increase in the production of GAGs. There is some evidence to show that these two cell types have different responses to hypoxia (48). Hypoxic conditions also increased the proliferation of MSCs and suppressed their senescence (53,54). There are conflicting reports as to whether hypoxic conditions help to increase (43,50) or decrease (49,55) hypertrophy. Adesida et al. have shown that human bone marrow-derived MSCs isolated and expanded under hypoxia (3% oxygen tension) exhibited enhanced capacity to form cartilage as compared to MSCs isolated and expanded under normoxia (21% oxygen tension). The effect of oxygen tension on the differentiation of hmscs (that were expanded in hypoxia or normoxia) was also tested. With regard to oxygen tension during isolation, there were a greater number of hmsc colonies following hypoxic expansion (8-37% higher than normoxic depending on the donor). There was no difference between the diameters of the colonies grown in hypoxic or normoxic conditions (52). The oxygen tension during hmsc expansion had greater consequence for chondrogenesis (where hypoxic conditions induced greater GAG/DNA, collagen II expression and uniform 14

31 localization throughout the pellet) than the oxygen tension during pellet culture/chondrogenic differentiation (where hypoxia only decreased collagen X expression). This can be clearly seen in Figure 2.3, as the staining is more intense in pellets in which hmscs were isolated and expanded under hypoxia. All pellets, regardless of oxygen tension during expansion or differentiation, had a uniform distribution of collagen type I (52). hmscs isolated and expanded in both hypoxia and normoxia were left to differentiate in pellet culture in chondrogenic medium. hmscs underwent greater chondrogenesis under hypoxic differentiation conditions as opposed to normoxic differentiation conditions, regardless of oxygen tension during expansion conditions. This was evidenced by measuring GAG content using Safranin O staining and GAG/DNA following pellet culture. Following normoxic differentiation, hmscs originally expanded under hypoxic conditions had a fold increase in GAG/DNA content over normoxic-expanded hmscs. Following hypoxic differentiation, hmscs expanded under hypoxic conditions had a fold increase in GAG/DNA content over normoxic-expanded hmscs (52). Oxygen tension is an important culture parameter in the generation of cells for cartilage repair strategies. Overall, it was observed that the common practice of carrying out both the expansion and differentiation of hmscs under normoxic conditions did not promote chondrogenesis to the same extent as under hypoxic conditions. It is particularly important to isolate and expand hmscs under hypoxic conditions when the goal is to generate cartilage tissues. 15

32 Figure 2.3: Safranin O staining of pellets that were isolated and expanded in hypoxia or normoxia and subsequently, differentiated under both oxygen tension levels. Modified from (52) Serum-Free Media Serum is a commonly used component in culture medium that is used to grow mammalian cells, including hacs and hmscs. It comprises of an undefined, complex mixture of components, including: growth factors, attachment factors, nutrients, trace elements, binding proteins, buffers, protease inhibitors, protection factors and antitoxins. Serum contains a mixture of molecules that promote and inhibit cell growth, as well as induce differentiation. Serum can support the growth and normal function of most cells (56). There are many drawbacks to using serum with regards to developing a reproducible bioprocess. There is significant difficulty in standardization due to batch-to-batch variation of the performance of serum in any given cell culture system. Due to this batch-to-batch variability, it is important to screen several lots of serum and identify a suitable lot that will support cell proliferation and other desirable cell culture outcomes. Additionally, serum contains cytotoxic agents and molecules that inhibit desired cell functions. These agents 16

33 can also hinder the goals of the bioprocess, whether cell proliferation or differentiation. Furthermore, there is limited availability of serum, which limits cell production (56). Thus, the use of serum presents severe obstacles to the development of a bioprocess for the production of quality-assured cells. Apart from bioprocessing challenges, serum poses a substantial challenge to clinical translation from a regulatory standpoint. There are major safety issues with the use of animal serum including: the risk of infection with biological contaminants and zoonotic agents, and the risk of immune reaction due to xenogenic proteins. Human-derived serum has also been investigated, with conflicting cell expansion results. Furthermore, humanderived serum has many of the same drawbacks as animal-derived serum, and there is a risk that it can contain human pathogens (56). For these reasons, the use of serumcontaining medium (SCM) hinders regulatory approval of a cell-based therapy. Recent work has focused on developing serum-free media (SFM) to overcome the limitations associated with SCM. These SFM contain defined basal media, buffers, nutrients, and proteins such as attachment factors, growth factors, hormones, and binding proteins. SFM are more desirable for clinical applications, since they are defined and can give more reproducible results than SCM. Furthermore, SFM can be designed to isolate and expand certain cell types, such as hmscs, for a particular application (56). However, a drawback of SFM is that they are typically substantially more expensive than SCM Cell Culture Scale-up T-flasks are small vessels that are commonly used to expand mammalian cells. T-flasks typically have growing surfaces of up to 175 cm 2, which support the expansion of adherent cell populations (i.e. cells that need to be attached to survive and grow, such as hacs and hmscs). Small numbers of adherent cells are inoculated into a flask where they attach to the growing surface and divide. Since the contents of the flasks are not being mixed during the proliferation process, this approach to cell production is described as being static. 17

34 Once the surface of the flask is covered in a monolayer of cells, the cells are sub-cultured. Sub-culturing entails the release of the attached cells from the surface of a flask, and reinoculation of these cells in small numbers into multiple new flasks with fresh medium where they again have space to divide. By repeating this process over and over again, a very large population of cells can be generated. However, the use of increasingly greater numbers of T-flasks for cell production can quickly become labour-intensive and inefficient. Due to the large numbers of T-flasks required and the inherent variability between the flasks, this culture method can also introduce undesirable heterogeneity into the collective cell population. Bioreactors are scalable vessels that provide a suitable alternative to T-flasks for the scaleup of cell production. A single bioreactor can accommodate the same culture volume as many T-flasks, permitting the growth of large quantities of cells in a homogeneous cell culture environment. In addition, key culture parameters, such as ph, oxygen, feeding, agitation and temperature, can be monitored and computer-controlled in real time in larger bioreactor systems. Furthermore, these systems can also store culture data. Due to the larger volumes and automated nature of several operations, bioreactors can substantially reduce the manual aspects of the cell production process and thus increase reproducibility Suspension Bioreactors Stirred suspension bioreactors can be used to scale-up the production of mammalian cells. In these vessels, an impeller is used to stir the contents of the bioreactor (i.e. the conditions are not static). Agitation is an important component of this system, since it: (i) keeps the cells in suspension, (ii) helps homogenize the contents, and (iii) introduces shear in the bioreactor. Suspension bioreactors have been used for the large-scale expansion of hacs (22,57), hmscs (23,58 61), and other stem cells (62 65). An important consideration when using stirred suspension bioreactors is the agitation rate (and thus the magnitude of the associated shear forces) to which the cells are exposed. Mammalian cells lack a cell wall, and thus are shear sensitive. Therefore, it is important to 18

35 understand the effect of agitation rate on each cell type being grown, as it can affect cell growth and viability. In addition, for those cells that tend to aggregate together and form spheroids, the agitation rate can affect the characteristics of the aggregates. Low agitation rates (i.e. low shear) can encourage excessive agglomeration between aggregates, and high shear can decrease cell viabilities. The optimal shear stress depends on the cell type and method of culture (aggregates vs. single cells). These bioreactors can be scaled-up by calculating the maximum shear stress (66). Mechanical stimulation, other than shear conditioning, imparted on tissue constructs using bioreactors has also been shown to have a positive impact on cartilage tissue formation. For example, dynamic compressive loading and shear exposure of MSC-seeded scaffolds resulted in enhanced chondrogenesis, both with and without growth factor supplementation. Further investigation is required to determine the amount and type of mechanical stimulation that produces the best cartilage matrix (36) Scale-up as Aggregates In order to grow adherent cells in suspension bioreactors, microcarriers, which are small beads around µm in diameter, are typically added to the bioreactor in order to provide a surface for cell attachment. However, the use of these beads introduces an added level of difficulty in bioreactor operation and downstream bioprocessing, since the attachment and detachment of cells from microcarriers can be very inefficient. Alternatively, certain cells can be grown in suspension culture as aggregates, in which they attach to one another without the need for microcarriers. Aggregates have advantages over microcarriers in that they form spontaneously and can be optionally dissociated into single cells relatively easily using mechanical, enzymatic or chemical methods (67). In addition, cells form bonds with matrix and other cells, which represents a unique 3D culture microenvironment. These bonds can be strengthened with shear conditioning in bioreactors. Furthermore, the size of the aggregates can be controlled 19

36 by the agitation rate (shear forces) to enable nutrients to diffuse into the aggregates and reach the cells in the center (65). Whereas hmscs are known to be an adherent cell type, there is evidence to suggest that they can also grow as cell aggregates in suspension culture (23,68). Aggregates of MSCs, grown in bioreactors, display enhanced biological function such as greater differentiation potential (23) and secretion of growth factors. This is likely due to agitation in the bioreactors, which imparts shear stress and improves mass transfer. In addition, the matrix proteins and cell-to-cell bonds in the aggregates may play a role in cell survival over dissociated cells from monolayers, and help to improve differentiation potential (69). Some interesting characteristics have been reported about MSC aggregates. They can adhere to cartilage defects, without sutures or glue, thereby enhancing pre-clinical outcomes in rabbits. In addition, the proliferative capacity of MSC aggregates in long-term bioreactor culture is inversely related to the aggregate size. This may be due to mass transfer limitations with large aggregates. Furthermore, it has been observed that these aggregates tend to compact over time (69). Chondrocytes, another adherent cell type, have also been shown to be able to form aggregates in suspension culture (70). Furthermore, there is some evidence to suggest that the aggregation of ACs promotes chondrogenic differentiation and ECM deposition (71). In addition, AC aggregates have been shown to seed faster than single cells on polymeric scaffolds in spinner flasks (72). There have not been any reports of hmscs and hacs being grown in co-culture in stirred suspension bioreactors. However, there have been several reports of ACs, MSCs, neural stem cells (NSCs), embryonic stem cells (ESCs) and induced pluripotent stem cells (ipscs) grown alone as aggregates in suspension bioreactors under various culture conditions. Table 2.1 lists several studies that reported aggregate culture, in suspension bioreactors, of 20

37 different cell types from different sources. The culture conditions, aggregate size and cell fold expansion are also listed. 21

38 Table 2.1: Culture of mammalian cells as aggregates in smooth-walled spinner flask bioreactors as reported in the literature. Cell Medium Culture Agitation Cell Culture Feeding Aggregate Cell-Fold Source Density Volume Time Rate Type Medium Regimen Size (µm) Expansion (cells/ml) (ml) (days) (rpm) Reference ACs Rabbit SCM Fed-batch 5 40 <150 Not (73) reported ACs Bovine SCM and Batch (22) SFM ACs Bovine SCM Batch , 80 Not Not (71) reported reported MSCs Human SCM Fed-batch Not (23) reported NSCs Murine SFM Batch (65) ipscs Murine SCM Batch (62) ipscs Human SFM Fed-batch (<24 h) (74) 50 ESCs Human SFM Fed-batch (<24 h) (74) 50 ESCs Murine SFM Fed-batch < (63) ESCs Murine SCM Batch < (64) 22

39 2.7. Objectives The goal of this thesis was to design, develop and evaluate a 3D co-culture protocol using scalable, suspension bioreactors that can efficiently generate large quantities of cells, which may have utility in cartilage repair, under serum-free conditions. Specifically, suspension co-culture methods were developed to support the rapid and reproducible expansion of hacs and hmscs to clinically relevant numbers as well as the generation of extracellular matrix, while encouraging a chondrogenic phenotype. This work aims to facilitate the development of a clinical therapy for the repair of cartilage defects. There were three specific aims (SA) in this thesis: serum-free media screening, bioreactor protocol development and evaluation of the bioreactor co-culture protocol. SA 1: Screening of serum-free media Two SFM, originally developed for the expansion of hmscs, were tested in a hac and hmsc co-culture system against conventional SCM. The cell growth and viability were monitored over three passages in static co-culture. The serum-free medium that supported the long-term expansion of the co-cultured cells was employed in bioreactor co-culture. The serum-free media screening was the first experiment, since the choice of medium can affect most cell characteristics. SA 2: Development of the bioreactor protocol Several important variables were tested to improve the quantity and quality of cells and matrix produced by the bioreactor co-culture protocol. Once tested, the successful condition was incorporated into the protocol. The agitation rate has been shown to impact cell proliferation and aggregate characteristics in other aggregating cell types (65), so it was tested first in bioreactor development. The effect of agitation rate during the growth phase and then the inoculation phase was tested. Next, serial passaging and feeding were tested to extend the culture period within a bioreactor vessel, thereby minimizing the need to subculture cells between vessels. Lastly, the effect of oxygen tension was studied for the entire duration of the culture including: isolation, pre-culture and bioreactor co-culture to improve the chondrogenic characteristics of the aggregates. 23

40 SA 3: Evaluation of the bioreactor protocol Lastly, the bioreactor protocol was evaluated by comparison to the gold standard, i.e. the corresponding static co-culture protocol. The quality of the bioreactor-generated aggregates was compared to the static-generated monolayer. The robustness of the bioreactor protocol was evaluated by comparing different donor sets with regards to cell growth and phenotype, as well as matrix production. 24

41 CHAPTER 3: MATERIALS AND METHODS 3.1. Introduction Some of the culture methods were chosen based on the literature and past experience. Coculture of hacs and hmscs showed positive results, especially in a ratio of 1 hac to 3 hmscs (17,37), so this co-culture ratio was employed in this thesis. A low inoculation density of 20,000 cells/ml (23) was used in order to minimize time spent in pre-culture using T-flasks, since cells can lose their desirable characteristics in 2D culture (see Section 3.7.1). Hypoxic conditions at 3% oxygen tension were employed previously (52) to improve chondrogenic characteristics, so this oxygen tension level was tested in some studies Cell Lines Primary hacs were isolated from the articular cartilage of human knees after total knee arthroplasty as described previously (75). hmscs were isolated from bone marrow aspirates from the iliac crest harvested during surgical procedures using existing methods (52). The isolation protocol for hmscs was previously demonstrated to reproducibly yield MSCs, as defined by: adherence to plastic, surface antigen expression and multi-lineage potential (76). Briefly, the mono-nucleated cells from the bone marrow aspirate were counted with a haemocytometer after staining the cell nuclei with crystal violet. These cells were inoculated into T-flasks at a density of 10 5 cells/cm 2 and cultured using SCM. The first medium change was made after seven days and twice a week thereafter. The plasticadherent hmscs were harvested after reaching 70-80% confluence. hacs and hmscs were isolated and expanded under hypoxic (3% O2) or normoxic (21% O2) conditions by storing the cultures in incubators with 3% and 21% oxygen tension respectively. There were three bone marrow donors (signified by BM) and four 25

42 chondrocyte donors (signified by hac) from whom cells were procured to carry out the work described in this thesis. The details of the donors are listed in Table 3.1. The hacs were derived from osteoarthritic donors, but the hmscs were not derived from osteoarthritic donors. All cells were procured at the University of Alberta, Edmonton, Canada with approval from the local ethics committee. The committee waived the requirement of informed consent from donors of tissue samples, since the collected tissues were intended for surgical discard. There were measures taken to protect the personal information of the donors. In addition, institutional safety and ethical guidelines were respected while carrying out these experiments. Table 3.1: Donor Information Donor Age (years) Gender Oxygen Tension during Isolation BM Female 3% BM Male 3% and 21% BM Female 3% and 21% hac79 85 Female 3% hac Female 3% hac Female 3% and 21% hac Male 3% and 21% 3.3. Cell Culture Media There were five different culture media used in this thesis: three SCM and two SFM Serum-Containing Media Medium for hac monoculture: The culture medium for the hacs was based on an SCM described previously in the literature (77) and was composed of 10% (v/v) fetal bovine serum (FBS) (Lonza Group, Basel, Switzerland, Cat. No. PT-4105) 90% Dulbecco s modified Eagle medium (DMEM) (VWR International, Radnor, USA, Cat. No. CA

43 594) with 30 µg/ml Gentamicin, 15 ng/ml Amphotericin, mg/ml L-Glutamine (Lonza, Cat. No. PT-4105) and 10 mm 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) (Sigma-Aldrich, St. Louis, USA, Cat. No. H4034). Medium for hmsc monoculture: The culture medium for the hmscs was composed of 10% FBS 90% modified Eagle s medium α (αmem) (VWR, Cat. No. CA ) with 30 µg/ml Gentamicin, 15 ng/ml Amphotericin, mg/ml L-Glutamine, 1% sodium pyruvate (Sigma-Aldrich, Cat. No. S8636) and 10 mm HEPES, and was similar to SCM used to culture hmscs in other studies (52). αmem was used to grow hmscs, since it had more nucleotides than DMEM. Sodium pyruvate, which is required in the cell cycle, was added to αmem to supplement the concentration of pyruvate in the medium. Medium for hac and hmsc co-culture: The serum-containing medium for co-culture was a blend of the hac and hmsc co-culture media. It was composed of 10% FBS 22.5% DMEM and 67.5% αmem with 30 µg/ml Gentamicin, 15 ng/ml Amphotericin, mg/ml L-Glutamine, 0.75% sodium pyruvate, 10 mm HEPES and 0.1 mg/ml L-ascorbic acid (Sigma-Aldrich, Cat. No. A-4544) Serum-Free Media TheraPEAK Mesenchymal Stem Cell Chemically Defined Growth Medium (Lonza, Cat. No ) was purchased and supplemented with 1x Penicillin-Streptomycin (PenStrep) (Life Technologies, Carlsbad, USA, Cat. No ) prior to use to help prevent microbial contamination. PPRF-msc6 is a serum-free medium developed for the isolation and expansion of human MSCs (56,78,79). Supplementation with L-ascorbic acid (Sigma-Aldrich, Cat. No. A8960) is required for this medium as per the medium formulation. Since ascorbic acid degrades, fresh ascorbic acid was added upon inoculation of medium and during medium changes. The formulation of PPRF-msc6 and the procedure for preparing this culture medium can be found in Appendix A. 27

44 3.4. Static Culture Procedures Primary cell culture was carried out in the Adesida lab (University of Alberta) for both hacs and hmscs. The hmscs were cryopreserved at passage 2 and the hacs were cryopreserved at passage 0. The cryopreserved cells were couriered to the Pharmaceutical Production Research Facility (PPRF; University of Calgary). The cells were thawed at PPRF and serially passaged in static tissue culture flasks to generate sufficient quantities of cells to carry out bioreactor and static co-culture studies. Tissue culture flasks with surface areas of 25 cm 2 (T-25) and 75 cm 2 (T-75) were used. When the medium being used was PPRF-msc6, all T-flasks were coated with 0.1% gelatin prior to use Gelatin Coating The following procedure was used to coat a tissue culture surface with 0.1% gelatin in water under sterile conditions. The procedure for preparing 0.1% gelatin in outlined in Appendix A ml of 0.1% gelatin was added to a T-75 flask. 2.5 ml was added to a T-25 flask. 2. The tissue culture surface was rotated and tapped to ensure the entire surface was coated. 3. The tissue culture surface was left in the biosafety cabinet for 30 minutes for the gelatin to settle at the surface. 4. The supernatant was aspirated out. 5. The caps of the tissue culture flasks were loosened and they were left to dry in the biosafety cabinet for 1-2 hours. 6. The gelatin-coated vessels were stored at 4 C in the dark until required for cell culture. 7. Prior to cell or medium inoculation, the vessels were placed in the incubator at 37 C for 30 min to warm the tissue culture surface. 28

45 3.4.2 Cell Culture Handling of cell cultures in this thesis was performed in a sterile biosafety cabinet using aseptic techniques. The attached cells were monitored for confluence, morphology, spreading and other characteristics using light microscopy. Medium changes were performed three times a week, or every 3 days, depending on the experimental design, to replenish nutrients and remove metabolites as follows: 1. Ascorbic acid degrades quickly in medium, so if ascorbic acid is present in the medium formulation, it has to be added just prior to inoculation of fresh medium to delay this degradation. Ascorbic acid was added at a concentration of 0.1 mg/ml (0.2 mm) to the fresh medium. The supplemented fresh medium was mixed % of the medium was removed from the vessel in an aseptic manner and discarded. 3. An equal volume of fresh medium (previously warmed to 37 C) was added to the vessel. The cells were serially passaged from static tissue culture flasks to new flasks once they reached 75-80% confluence, or on the day dictated by the experimental design. 1. The necessary amounts of culture medium, trypsin solution with ethylenediaminetetraacetic acid (EDTA) (Life Technologies, Cat. No ) and Phosphate-Buffered Saline (PBS) (Life Technologies, Cat. No ) were aliquoted into conical tubes. 2. Trypsin-EDTA was placed in the water bath at 37 C for at least 30 min. The medium and PBS were placed in the incubator at 37 C and 5% CO2 for at least 30 min. 3. The supernatant in the T-flasks was removed and disposed of. 4. The cell monolayer in the T-flasks were washed with 4.0 ml of PBS/T-75 flask or 2.0 ml of PBS/T-25 flask to remove any cell debris. 29

46 ml of 0.25% trypsin-edta was deposited in each T-75 flask and 2.0 ml in each T-25 flask. 6. The T-flasks were incubated for 5 minutes in the incubator at 37 C. 7. The flasks were shaken and tapped against a bench to dislodge the cells from the surface. 8. The flasks were observed under the microscope to ensure that all the cells had detached. 9. The cell suspension was pipetted out of the flasks and deposited in a conical tube ml of medium was used to rinse each T-flask so as to remove any remaining cells. 11. This rinse was pipetted out and deposited in the conical tube containing the cells to block the action of the trypsin enzyme. 12. The conical tube was centrifuged at 400 g for 5 min. 13. The supernatant in the conical tube was removed. 14. The cells were dispersed in about ml of fresh medium for every T-75 flask harvested. 15. Two 50 μl samples were taken from the conical tube using a micropipetter for cell counts. 16. The cells were counted and the viability was determined with a haemocytometer and a trypan blue live-dead assay by counting all eight squares of the haemocytometer. 17. New T-flasks were labelled with culture conditions. 18. The cell suspension was pipetted into the flask so as to achieve the desired cell density. 19. Culture medium was deposited into each flask so as to achieve 10 ml of working volume for every T-75 flask and 4.0 ml in every T The flasks were swirled and agitated to ensure that the cells were evenly dispersed in the medium. 21. The inoculated flasks were placed in a humidified incubator at 37 C and 5% CO2. 30

47 3.4.3 Cryopreservation of Single Cells Cryopreservation is a procedure used for the long term storage of cells. The formulation of cryopreservation media was identical to the growth media, but supplemented with 10% dimethyl sulfoxide (DMSO) to aid in preserving the cells using cryopreservation. For SCM, FBS was added to the cell suspension such that the final concentration of FBS was 20% and the final concentration of basal medium was 70%, with the remainder of the medium being DMSO. For SFM, 90% of the cryopreservation medium was either PPRFmsc6 (without ascorbic acid) or TheraPEAK, and these media were also supplemented with DMSO. There was no FBS in the serum-free cryopreservation media. The following procedure was used for cryopreservation. 1. The cells were centrifuged at 400 g for 5 min in conical tubes. 2. The cells were resuspended in fresh medium such that there were 1 to 2 million cells/ml. 3. DMSO was carefully added to the cell suspension to give a final solution of 10% DMSO by volume 4. The cell suspension was distributed into cryovials (1.0 ml/vial). 5. The cryovials were kept in a Nalgene Cryo 1 C Cryostorage Container in isopropanol overnight at -80 C. 6. The cryovials were then transferred to a liquid nitrogen container for long-term storage Cell Resuscitation The cryopreserved cells were thawed from cryovials and inoculated into tissue culture flasks using the following protocol: 1. After wiping the cryovial with ethanol, the cap of the cryovial was twisted a quarter of a turn within a biosafety cabinet to release any pressure inside the cryovial, and then closed. 31

48 2. The cells in the cryovial were thawed at 37 C in a water bath. 3. After the last sliver of ice dissolved, the cryovial was wiped with ethanol and brought into a biosafety cabinet. 4. Cultures were inoculated with 10 ml of medium per T-75 flask. 5. T-75 flasks containing 10 ml of the appropriate medium were inoculated with one cell type in a specific growth medium at an approximate cell density of 10,000 cells/cm 2 for hacs and 5,000-6,000 cells/cm 2 for hmscs. 6. After 24 hours, all the cells adhered to the bottom of the flasks. A 100% medium change was performed on all flasks to remove the DMSO from the culture medium Bioreactor Culture Procedures The co-culture of hacs and hmscs was tested in the dynamic environment of a suspension bioreactor. 125 ml NDS spinner flasks with paddle-wheel impellers were used to grow the cells as aggregates. Thus, this section describes the cell culture protocols used for bioreactor co-culture Bioreactor Preparation The spinner flasks were not disposable and were reused after cell culture. Thus, the cleaning and silicon-coating protocol was essential to ensure the vessels were sterile and resistant to cell attachment. The spinner flasks were prepared before use in experiments using the following protocol. 1. Spinner flasks, including their impellers, were washed thoroughly with double distilled water and a brush. 2. Virkon (DuPont, Mississauga, Canada), a disinfectant, was used to prevent the potential spread of viruses, bacteria and fungi. Virkon powder was added to each spinner flask and the flask was filled with water to dissolve the powder. 3. After soaking overnight in Virkon, the flasks were emptied and rinsed with double distilled water three times. 32

49 ml of double distilled water was poured into each flask and kept there for 8-12 hours. 5. The flasks were emptied of double distilled water and rinsed with double distilled water three more times. 6. The flasks were left to dry for 4-8 h. 7. The flasks were coated with silicone to prevent the attachment of cells to the walls of the flasks. In a fume hood, 10 ml of SigmaCote (Sigma-Aldrich, Cat. No. SL2) was pipetted into each clean and dry 125 ml spinner flask. 8. Spinner flasks were coated with 10 ml of SigmaCote by coating each side of the impeller (including the top) twice, the inner surface of the spinner flask (until the 100 ml mark) by turning it three times and the bump in the flask by pipetting silicone on it. 9. The flasks were left to dry overnight in the fume hood. 10. The flasks were soaked with PBS and water to ensure that any silicone that could leach into the liquid has done so prior to cell culture. 10x PBS/flask was diluted in a graduated cylinder with double distilled water to make 1x PBS/flask. 300 ml of 1x PBS was poured into each spinner flask. 11. After 8-12 hours, the flasks were emptied of PBS and rinsed with double distilled water three times. 12. The same volume of double distilled water was poured into each flask and kept for 8-12 hours. 13. The flasks were emptied of double distilled water and rinsed with double distilled water three times. 14. The flasks were left to dry for 4-8 h. 15. The operation of the impeller was tested on a magnetic stir plate to ensure that its operation was smooth. Also, the lid was checked to ensure the seal ring was in place and sterility was maintained when the lid was closed. 16. The flasks were autoclaved to render them sterile. The lids and caps of the spinner flasks were covered with aluminum foil and autoclave tape was stuck on the foil. 33

50 The cap of one of the arms was cracked open to prevent pressure build-up in the flask during autoclaving. 17. The spinner flasks were sterilized using the autoclave on the dry cycle for 30 min. 18. The spinner flasks were taken out of the autoclave following the end of the cycle. The cap of the arm was closed when taking the spinner flask out of the autoclave to maintain sterility. 19. The magnetic stir plate was programmed to the desired agitation rate using a counter and a timer Bioreactor Inoculation The procedure for inoculating the cells into the spinner flasks is outlined below. Due to the large amount of time required for static pre-culture and the limited window of passage numbers that could be used (see Section 3.7.1), a relatively low inoculation density of 20,000 cells/ml was used. 1. hmsc and hac monocultures were harvested from T-75 flasks, and each cell type was resuspended in PPRF-msc6 (without ascorbic acid). 2. The cells were counted and the volume of each cell suspension required to inoculate a spinner flask at 20,000 cells/ml in a 1 hac to 3 hmsc ratio was calculated. 3. The required volumes of hac cell suspension and hmsc suspension were put into marked conical tubes. Additional medium was added such that each tube had a total of 50 ml of PPRF-msc6 (without ascorbic acid). 4. Spinner flasks were labelled with the conditions. 5. The contents of the 50 ml conical tube were triturated and the appropriate amount of cell suspension was deposited into each spinner flask ml more medium was added to each spinner flask to achieve a final working volume of 125 ml/flask μl of ascorbic acid was added to each spinner flask (the final ingredient in the formulation of PPRF-msc6). 34

51 8. The spinner flasks were placed on spinner plates in the incubator at 5% CO2 and 37 C. The spinner plates were calibrated to operate at agitation rates specific to a given experiment Bioreactor Feeding 50% medium changes were performed to replenish nutrients and remove metabolites in some experiments. The following procedure was used for feeding the bioreactor co-culture. 1. The spinner flasks were removed from the incubator and placed in a biosafety cabinet where the aggregates in the bioreactor were allowed to settle. 2. Half the medium was removed from each spinner flask using a 10 ml pipette and placed in a 50 ml conical tube. 3. The tubes were centrifuged at 400 g for 10 min. The cell pellet was very loose and could be easily disturbed. 4. All the media in the tube except for the last 2.0 ml was removed using a 10 ml pipette ml of fresh medium was added to each tube. 6. The cell suspension was agitated in the tubes with a 5 ml pipette. 7. The remaining fresh medium was added into each tube and the cell suspension was agitated again. The volume of this cell suspension was equal to the volume of spent medium that was removed from the spinner flask. 8. The cell suspension was added back to the spinner flasks from the conical tubes Serial Passaging of Cells in Bioreactors Serial passaging of bioreactor co-culture necessitates dissociation of aggregates into single cells. In this thesis, enzymatic dissociation of aggregates with trypsin-edta and mechanical trituration was used to dissociate the aggregates using the following protocol. 1. The spinner flasks were placed in the biosafety cabinet on a magnetic stir plate. 35

52 2. Using a 10 ml pipette, 70 ml of cell suspension was removed from each spinner flask into two conical tubes (with 35 ml each) for harvesting. 3. The conical tubes were centrifuged at 400 g for 10 min. The cell pellet was very loose and could be easily disturbed. 4. All the media was removed in the tube except for the last 1.0 ml using a 10 ml pipette into a conical tube ml of trypsin-edta was added to each of the tubes. 6. The cell suspension was triturated in the tubes with a 5 ml pipette. 7. More trypsin solution was added to each tube to achieve the same volume (35 ml) of trypsin solution in each tube. 8. The cell suspension was triturated in the tubes. 9. The cell suspension was incubated at 37 C for 10 minutes. 10. The cell suspension was triturated with a micropipette for 15 to 60 seconds to break apart the aggregates. 11. The aggregates were very sticky and difficult to dissociate into single cells. Furthermore, they became stickier with culture time, so greater trypsin exposure was required to dissociate the aggregates as the culture progressed. 10 µl of cell suspension was deposited into a well of a 96 well plate and observed under the microscope to check if aggregates had dissociated into single cells. If aggregates were still visible in the well, the dissociation was repeated (steps 9-10) up to two more times, for a maximum trypsin exposure time of 30 minutes. This trypsin exposure did not have a negative effect on the cells, as evidenced by the proliferation of cells, which were dissociated from aggregates, and placed in static culture. 12. The conical tubes were centrifuged at 400 g for 10 min. The cell pellet was very loose and could be easily disturbed. 13. All the medium was removed from the tube using a 10 ml pipette except for the last 3.0 ml. 36

53 14. The 3.0 ml of cell suspension in each of the two 50 ml conical tubes was transferred into a single 15 ml conical tube. This cell suspension contained the cells from the original 70 ml in one spinner flask ml of PPRF-msc6 (without ascorbic acid) was added into each tube to block the action of the trypsin enzyme. 16. The conical tubes were centrifuged at 400 g for 10 min. The cell pellet was very loose and could be easily disturbed. 17. All the media was removed in the tube except for the last 20 µl using a 10 ml pipette and a micropipetter. 18. Each tube was resuspended into 10 ml PPRF-msc6 (without ascorbic acid). The tube was triturated well so cells do not aggregate. 19. The cells were counted and used to inoculate fresh spinner flasks at a viable cell density of 20,000 cells/ml Analytical Procedures Several analytical procedures were used to determine the quality and quantity of the cells produced by bioreactor and static co-culture. The cell numbers and viabilities were measured to determine the efficiency of cell production in bioreactor and static co-culture. In addition, the size, shape and density of the bioreactor-generated aggregates were characterized with respect to time. Furthermore, the medium was analyzed to determine the nutrient uptake and metabolite production kinetics in culture. Moreover, histology, gene expression and GAG/DNA were determined to characterize the chondrogenic potential of the cells being generated Aggregate Dissociation The counting of cells from spinner flasks required the dissociation of cell aggregates into single cells. The following dissociation protocol, which is a combination of enzymatic and mechanical dissociation, was used. The incubation time in trypsin varied from min, depending on the size of the aggregates. 37

54 1. The necessary amounts of culture medium and trypsin-edta were aliquoted into conical tubes. 2. The trypsin-edta was placed in the water bath at 37 C. The medium was placed in the incubator at 37 C and 5% CO2. 3. A 4.0 ml sample of cell suspension was removed from each of the spinner flasks and placed in a labelled conical tube. Note that sampling was performed from the centre of the culture volume after the flask was agitated in order to obtain a representative sample. 4. The cell suspension was centrifuged at 400 g for 5 min. The supernatant was removed and discarded. 5. The cell pellet was resuspended in 4.0 ml of trypsin solution and then incubated at 37 C for 10 minutes. 6. The cell suspension was triturated with a micropipette for 15 to 60 seconds to break apart the aggregates µl of cell suspension was deposited into a well of a 96 well plate and observed under the microscope to check if aggregates had dissociated into single cells. 8. The dissociation was repeated (steps 7-8) up to two more times, for a maximum trypsin exposure time of 30 minutes ml of medium was deposited in each conical tube to block the action of the enzyme. 10. The conical tube was centrifuged at 400 g for 5 min. 11. The supernatant in the conical tube was removed. 12. The cells were resuspended in 1.0 ml of fresh SCM for cell counting Manual Cell Density and Viability Measurements Manual counts were more cost-effective than automated counts, and so were used in static monoculture, when there were fewer samples. Automated cell counting (See Section 3.6.3) was utilized in static and bioreactor co-culture studies, when there were many samples that needed to be counted quickly. Manual counts were also used to validate the automated cell 38

55 counting procedure. Manual counts were carried out using a haemocytometer (Hausser Scientific, Horsham, USA) and the trypan blue exclusion (Sigma-Aldrich) assay as follows. 1. Two 50 μl samples were taken from the conical tube with the dissociated cells using a micropipetter. 2. The live and dead cells were counted from these two samples using a haemocytometer and the trypan blue exclusion assay. 3. The cell density (see Equation 3.1) and viability were calculated. CV =N V V PBS+V TB +V S V S 1 8 chambers VH Equation 3.1 where: CV = viable cell density of cell suspension (cells/ml) N V = number of viable cells (cells) in all eight chambers V PBS = volume of PBS (ml) V TB = volume of trypan blue (ml) V S = volume of cell sample (ml) VH = volume of haemocytometer = 10-4 ml/chamber Automated Cell Density and Viability Measurements Following aggregate dissociation and harvest of cell monolayers, the cells were counted automatically with the ViCell Cell Viability Analyzer (Beckman Coulter Canada, Mississauga, Canada). The ViCell was used to automatically determine the cell number, cell viability and cell size distribution. The procedure for automated counts was standardized and took less operator time, so large numbers of samples could be counted quickly and reproducibly. The manual counts were used to check the calibration of the cell viability analyzer. The procedure for automated cell counting is as follows: 39

56 1. Two 0.30 ml samples were taken from the conical tube with the dissociated cells using a micropipetter and deposited into ViCell tubes ml of trypan blue was deposited into each ViCell tube. 3. The 600 µl cell suspension in the ViCell tubes was triturated to prevent the reformation of aggregates. 4. The ViCell tubes were placed on the ViCell carousel for analysis of cell density and viability. 5. The cell density in the bioreactor was calculated based on the dilution factor. The cell density data were used to determine the apparent growth rate and doubling time. The natural logarithm of the cell density during the exponential growth phase was plotted against time in culture in keeping with the linearized exponential cell growth equation (see Equation 3.2). ln(cv) = µapp Δt + ln(c0) Equation 3.2 where: µapp = apparent growth rate (h -1 ) Δt = time in culture (h) C0 = cell density upon inoculation (cells/ml) The apparent growth rate was the slope of the linear trend line, calculated using linear regression in Microsoft Excel. Equation 3.2 can be rearranged into Equation 3.3 to determine the doubling time. td = ln (2) / µapp Equation 3.3 where: td = doubling time (h) 40

57 3.6.4 Aggregate Characterization The aggregates were visualized using brightfield microscopy and characterized by measuring average and maximum diameter, diameter distribution, eccentricity, aggregate density and packing density. The average aggregate diameter (measured in µm) for the sample was calculated from the diameters of at least 20 aggregates (65). The first diameter was measured at the longest axis of the aggregate and the second diameter was measured at 90 to the first diameter. The average of these two diameters was taken as the diameter of the aggregate. Cell clusters with diameters less than 35 μm were not considered aggregates (65).The aggregate sampling procedure is as follows: 1. The spinner flasks were removed from the incubator and brought into the biosafety cabinet. 2. One 0.50 ml sample was removed from each spinner flask using a 2 ml pipette and deposited into a well of a 24 well plate. 3. Pictures were taken of all aggregates in the sample at 2.5 magnification, so that the total number of aggregates in 0.50 ml of bioreactor medium could be counted magnification pictures of at least 20 aggregates were taken per sample to measure the aggregate diameter. 5. The diameters of each of at least 20 aggregates was measured using the Zen software (Carl Zeiss, Oberkochen, Germany) of the confocal microscope. Eccentricity (µm/µm) is a measure of the circularity of an ellipse. The eccentricity of aggregates can be calculated using Equation 3.4, which assumes the 3D aggregates to be 2D projections of ellipses. Aggregate eccentricities less than 0.60 were considered to be acceptable in this bioprocess. Examples of ellipses with representative eccentricities can be found in Appendix C. 41

58 ɛ = D 1 2 D 2 2 D 1 Equation 3.4 where: ɛ = aggregate eccentricity (µm/µm) D = aggregate diameter on one axis (µm) D 1 > D 2 The aggregate concentration (aggregates/ml medium) is determined by dividing the total number of aggregates in the sample by the sample volume (0.50 ml) in duplicate. Packing density is the total number of cells per aggregate volume and be used to measure compaction in aggregates. The equation for packing density (see Equation 3.5) assumes spherical aggregates. PD = C T C agg V agg Equation 3.5 where: PD = packing density (cells/ml aggregates) C T = total cell density (cells/ml medium) Cagg = aggregate density (aggregates/ml medium) Vagg = volume of an aggregate (ml aggregate/aggregate) Cell Culture Media Analysis Fresh and spent culture media were analyzed for glucose, glutamine, ammonia and lactate concentrations using a NOVA BioProfile 100 Analyzer (Nova Biomedical, Waltham, USA). Limitations in nutrient concentration or build-up of metabolites can have an adverse effect on cell growth (80,81). The medium analysis was carried out as follows. 42

59 1. From each spinner flask and T-flask, 4.0 ml of medium was removed into conical tubes. 2. For samples removed from bioreactors, the tubes were centrifuged at 400 g for 5 min. 3. The supernatant was removed from each conical tube and pipetted into two Eppendorf tubes (with 1.5 ml each). 4. The conical tubes were stored at -20 C until analysis (58). 5. The media samples were thawed at room temperature. 6. The Eppendorf tubes were centrifuged again at 2000 rpm for 3 min to ensure that all debris was removed from the samples ml of the supernatant (spent medium) was transferred to the NOVA vials in duplicate. 8. The tray with the vials was fed into the NOVA Analyzer for analysis. Note that the Analyzer was calibrated prior to use according to the manufacturer s instructions. Various kinetic parameters were calculated to understand the cell metabolism, including: the yield of lactic acid on glucose and ammonia on glutamine (mol/mol), the uptake rates of glucose and glutamine (pmol/cell.day), and the production rates of lactic acid and ammonia (pmol/cell.day). The kinetic parameters were calculated in Microsoft Excel. The cumulative glucose consumption (mmol/l), during the exponential growth phase, was plotted against the cumulative lactic acid production (mmol/l) in the same time period. The yield of lactic acid produced per glucose consumed (mol/mol) was the slope of the linear trend line of this plot, as determined using linear regression. A similar approach was used to determine the yield of ammonia on glutamine. The values for glutamine and ammonia were adjusted for natural deamination and deamination or deamidation from free amino acids in the medium (82). Glutamine was found to naturally degrade at a rate of 0.06 mm/day and ammonia spontaneously formed at a rate of 0.13 mm/day. The nutrient consumption and waste production rates were calculated on a per cell basis. First, the cell density (cells/ml) was numerically integrated with respect to time (day) 43

60 using the trapezoid rule, to give the integral of cell density (cell.day/ml). Next, the cumulative consumption and production (mmol/l), during the exponential growth phase, was plotted against this integral (cell.day/ml) in the same time period. The slopes of the linear trend lines of these plots (pmol/cell.day) were the consumption and production rates (82) Histology The aggregates were stained with Safranin O with the following protocol (52) to reveal sulfated GAG depositions and visualize cartilage tissue ml of the cell suspension ( 250,000 cells), harvested from bioreactors and T-flasks, was aliquoted into conical tubes. 2. The tubes were centrifuged at 400 g for 10 min. 3. The aggregates were resuspended in 2.0 ml of formalin per tube (Thermo Fisher Scientific, Whitby, Canada, Cat. No ). 4. The tubes with the aggregates were stored at 4 C overnight. 5. The tubes were centrifuged at 400 g for 10 min. 6. The cell aggregates were resuspended in 1.0 ml of PBS (Life Technologies) per tube. 7. The aggregate suspension in PBS was transferred to 1 ml Eppendorf tubes. 8. The Eppendorf tubes were stored at 4 C. 9. The tubes were packaged and shipped to Dr. Adesida s lab in wet ice. 10. The samples were embedded in paraffin wax and sectioned with a microtome at 5 µm. 11. The samples were stained with 0.1% (w/v) Safranin O (Sigma-Aldrich, Cat No. S G) and counterstained with 0.01 % (w/v) fast green (Sigma-Aldrich, Cat. No. F G). 12. The stained slides were photographed using the Omano OM159T biological trinocular microscope (Microscope Store, Virginia, USA), Optixcam summit series 5MP digital camera, and Optixcam software. 44

61 3.6.7 Gene Expression The gene expression of the co-cultured cells was determined using Reverse Transcription Polymerase Chain Reaction (RT-PCR). The genes studied were collagens I, II and X, aggrecan and cartilage oligomeric matrix protein (COMP). The following protocol (52) was used to determine the gene expression ml of cell suspension ( 250,000 cells), harvested from bioreactors and T- flasks, was aliquoted in conical tubes. 1. The tubes were centrifuged at 400 g for 10 min. 2. The aggregates were resuspended in ml of Trizol solution (Life Technologies, Cat. No ) per tube. 3. The cell suspension was triturated several times and left for 1 hour at room temperature to release the DNA from the cells. 4. The tubes were frozen at -80 C. 5. The tubes were packaged and shipped to Dr. Adesida s lab in dry ice. 6. Contaminating genomic DNA was removed from the RNA solution using DNase treatment (Qiagen Mississauga, Canada; Cat No ). 7. The ribonucleic acid (RNA) was purified using of RNeasy mini-kit columns (Qiagen, Cat No ). 8. The concentration and purity of RNA was quantified using the ratio of absorbance at 260 nm over 280 nm, which was measured using a Nanodrop 1000 spectrophotometer (Thermo Fisher Scientific). Ratios of represented a high purity RNA solution ng of total RNA was reverse transcribed to cdna using GoScript reverse transcriptase (Thermo Fisher Scientific, Cat No A5004) and oligo dt primers in a 40 µl reaction. 10. Quantitative RT-PCR was performed with DNA Engine Opticon I Continuous Fluorescence Detection System (Bio-Rad, Mississauga, Canada) using hot start Taq and SYBR Green detection (Eurogentec North America, San Diego, USA; Cat No. 45

62 RT-SN10-05NR). The primers (Life Technologies) used (300 nm) are listed in Table 3.2. The amplification efficiencies were %. 11. The gene expression was normalized to human β-actin using the 2 -Δct method (83). Table 3.2: List of primer sequences used in RT-PCR Gene Sequence Reference Collagen I 5 -TTGCCCAAAGTTGTCCTCTTCT-3 (Forward) (84) COL1A2 5 -AGCTTCTGTGGAACCATGGAA-3 (Reverse) Collagen II 5 -CTGCAAAATAAAATCTCGGTGTTCT-3 (Forward) (84) COL2A1 5 -GGGCATTTGACTCACACCAGT-3 (Reverse) Collagen X 5 -CTGCAAAATAAAATCTCGGTGTTCT-3 (Forward) (85) COL10A1 5 -GGGCATTTGACTCACACCAGT-3 (Reverse) Aggrecan 5 -AGGGCGAGTGGAATGATGTT-3 (Forward) (84) ACAN 5 -GGTGGCTGTGCCCTTTTTAC-3 (Reverse) COMP 5 -CCGACAGCAACGTGGTCTT-3 (Forward) 5 -CAGGTTGGCCCAGATGATG-3 (Reverse) (85) GAG and DNA Content Analysis The GAG to DNA ratio is used to quantify matrix production. The sulfated GAG and DNA content was measured using the following protocol (52) ml of cell suspension ( 250,000 cells), harvested from bioreactors and T- flasks, was aliquoted in conical tubes. 2. The tubes were centrifuged at 400 g for 10 min. 3. The aggregates were resuspended in 1.0 ml of PBS (Life Technologies) per conical tube. 4. The aggregate suspension in PBS was transferred to 1 ml Eppendorf tubes. 5. The tubes were centrifuged at 400 g for 10 min. 6. The excess PBS was removed until only the cell pellet remained. 7. The Eppendorf tubes were frozen at -80 C. 46

63 8. The Eppendorf tubes were packaged and shipped to Dr. Adesida s lab in dry ice. 9. The samples were digested at 56 C for 16 hours using Proteinase K solution (1.0 mg/ml Proteinase K (Sigma-Aldrich, Cat. No. P6556) in 50 mm Tris (Sigma- Aldrich, Cat. No.T1503) at ph 7.6, 1 mm EDTA (Sigma-Aldrich, Cat No. E9884), 1 mm Iodoacetamide (Sigma-Aldrich, Cat. No ), and 10 µg/ml pepstatin A (Sigma-Aldrich, Cat No. P5318). 10. The samples were vortexed. 11. The DNA content was determined using the CyQuant cell proliferation kit and bacteriophage λ DNA standard (Life Technologies, Cat. No. C7026). 12. The GAG content was determined using dimethylene blue dye (Sigma-Aldrich, Cat No ) and chondroitin sulfate (Sigma-Aldrich, Cat. No. C8529) to make the standard curve. 13. A microplate reader (Dynex Technologies, Chantilly, USA) and the Microfluor1 black, flat bottom 96 well plate (Thermo Fischer Scientific) were used to measure the fluorescence (excitation: 450 nm, emission: 530 nm) of the samples and standards to determine the DNA content. In addition, the microplate reader and clear, flat bottom 96 well plates were used to read the absorbance (525 nm) of the samples to determine the GAG content. 14. The sample measurements were compared to the standard curves and adjusted with the appropriate dilution factors to determine the GAG and DNA content, and calculate the GAG/DNA Design of Experiments Biological systems are highly complex and have great inherent variability (82). The experiments in this thesis were designed to limit the effects of variability, while respecting practical limitations in time and resources. In addition, mammalian cell culture, especially culture of heterogeneous populations of stem cells, has several challenges as described below. The experimental design had to address these challenges to maintain healthy cultures. 47

64 3.7.1 Passage Level The quantities of hacs and hmscs obtained from donors after isolation were not sufficient for use in experiments. Thus, cell expansion was first required for each of these cell types in static culture flasks. There was a limit to the number of population doublings each cell type could go through before losing its characteristics or aging. Thus, each cell type had to be cultured within a certain timeframe. hacs start to lose their properties almost immediately when cultured alone (18), and should not be used to carry out experiments after passage 3. In addition, hmscs should be used within passage 6, after which they can experience senescence as well as genotypic and phenotypic variation (86). However, hmscs isolated and expanded in PPRF-msc6 can be healthy in culture for up to 12 passages (78). In this thesis, the inoculum of the co-culture experiments contained hmscs from passages 5-6 and hacs from passages 1-2. Thus, the passage numbers of the cells in bioreactor and static co-culture experiments did not exceed seven for hmscs and three for hacs Donors and Replicates The required quantities of both hacs and hmscs had to be available at the same time, since hac and hmsc co-culture necessitated the inoculation of both cell types together. Thus, each cell type had to be cultured under conditions that allowed for the required number of each cell type to be present at the time of inoculation. hmscs tended to multiply at a slower rate than hacs, so it took more time to generate large quantities of hmscs. Furthermore, there was a greater number of hmscs required when co-culturing at a ratio of 1 hac to 3 hmscs. For these reasons, generating sufficient quantities of two different cell types at the same time was very difficult. The hmscs and hacs have to be grown to sufficiently large quantities for the inoculation of static and especially bioreactor co-cultures. Given the slow growth rate of the hmscs and the limited window of passage numbers that could be used for both cell types, 48

65 achieving the quantities of each cell type required for inoculation, especially together, was quite a challenge. It was even more logistically difficult to carry out all of the experiments with the same donor. Thus, multiple donors were used for the development of the coculture protocol in suspension bioreactors. The experiments took place in duplicate, since triplicate tests were logistically difficult to execute. For example, if a variable with three conditions was tested in triplicate, there would need to be nine bioreactors, whereas, in duplicate, there would be six. Inoculating, sampling and passaging nine spinners was a long process. It would be difficult to perform these techniques within a short time for all of the bioreactors, especially with only one operator. While one vessel is being handled, biological activities, such as cell proliferation, would continue in the other vessels. Due to the time delay between handling of flasks, measurements from different flasks would correspond to different time points. Thus, the data gathered from such large experiments would have large discrepancies, even between replicates, due to the large time delays Factorial Experiments Factorial experiments were particularly difficult to execute. The effects of oxygen tension, donors and culture method were investigated in a factorial manner as illustrated in Figure 3.1. The different donors used in this experiment were: BM142, BM143, hac119 and hac120 (see Table 3.1). Cell populations from each of these donors were isolated and split, with half being expanded in T-flasks under normoxia conditions (21%), and the other half in T-flasks under hypoxic conditions (3%) (52). The cells were co-cultured in both static and bioreactor co-culture under the same oxygen tension conditions. This experiment was challenging, since it required the availability of eight specific cell pools and the static monoculture of these pools such that each pool had sufficient cell numbers, at the same time, to inoculate static and bioreactor co-culture experiments. This factorial experiment had 8 conditions, which required a total of 8 spinner flasks and 80 T- 49

66 flasks. This amounts to 2 million hacs and 8 million hmscs required from each donor and oxygen tension combination for inoculation. 50

67 BM142 hac119 hac120 BM143 Normoxia Hypoxia Normoxia Hypoxia Normoxia Hypoxia Normoxia Hypoxia hac119 + BM142 hac119 + BM142 hac120 + BM142 hac120 + BM143 Normoxia Hypoxia Normoxia Hypoxia Bioreactor-Static Comparison Bioreactor-Static Comparison Effect of Oxygen Tension Effect of Donors Figure 3.1: Factorial experiment design to test effect of oxygen tension and donors and compare static co-culture to bioreactor co-culture. There were eight cell pools of different donors, cell types and oxygen tensions. All eight conditions were tested in duplicates. 51

68 To test the effect of oxygen tension, BM142 and hac119 were grown in both bioreactor and static co-culture under normoxia and hypoxia (see Chapter 6). A comparison between static and bioreactor co-culture was also made with these donors under hypoxia (see Chapter 6). To test the effect of donors, bioreactor co-culture in hypoxia was carried out using two donor sets (BM142 and hac119, BM143 and hac120) in Chapter 6. BM143 failed to grow in in normoxia, so it was not used for experiments. Instead, the combination of BM142 and hac120 in normoxia was tested in bioreactor and static co-culture (see Chapter 4). The feeding and serial passaging experiments tested the possibility of extending bioreactor co-culture. These experiments were also tested together, being inoculated with the same donor pool (BM119 and hac79) at the same time. These experiments required a total of 4 million hacs and 11 million hmscs for the inoculation of 6 spinner flasks Statistics and Error Analysis Most of the statistical and error analyses were performed using Microsoft Excel 2013 and the Tukey s test was carried out in MATLAB R2012b. Cell densities, aggregate characteristics and metabolism data were recorded at different time points throughout the culture period. Cell densities, cumulative nutrient consumption and metabolite production, maximum aggregate diameter, aggregate density, and packing density were reported as mean ± range of data, since these data were collected in duplicates. Average aggregate diameter and eccentricity were reported as mean ± standard deviation, since there were 20 replicates (aggregates). All of these data were analyzed using two-way ANOVA with replicates, since there were two independent variables. The apparent growth rate and gene expression data were measured in duplicate and reported as mean ± range of data. The GAG/DNA was reported as mean ± standard error of the mean, since there were 2-4 replicates. In cases with two conditions, these variables were tested using a t-test. In cases with more than two conditions, they were analyzed using one-way ANOVA, since there was only one independent variable. Multiple comparisons 52

69 test was carried out using Tukey s test (honest significant difference criterion) to determine which conditions were different. Differences were considered significant at p<0.05. Forward error analysis was carried out on the uncertainties (bound of error in the approximations) in the raw cell counts and metabolism data by taking the derivative of the formulas, in order to calculate the propagated errors. For average aggregate diameter and eccentricity data, Equation 3.6 and Equation 3.7 were used to calculate the aggregated standard deviation of the non-overlapping samples (i.e. the two replicates). Equation 3.6 where: µx Y = mean of the entire sample µx = mean of the first group µy = mean of the second group NX Y = total number of samples NX = number of samples in first group NY = number of samples in second group 53

70 Equation 3.7 where: σ X Y = standard deviation of the entire sample σ X = standard deviation of the first group σ Y = standard deviation of the second group 54

71 CHAPTER 4: BIOPROCESS OVERVIEW 4.1. Statement of Contribution This chapter is composed of a manuscript titled Serum-free Co-culture of Human Mesenchymal Stem Cells and Articular Chondrocytes as Three-dimensional Aggregates in Suspension Bioreactors. This manuscript has been submitted to the journal, Arthritis Research & Therapy. The author of this thesis, Madiha Khurshid, has done the majority of the research and writing of this manuscript and is the first author of the paper. Thus, this manuscript has been included as a chapter of this thesis. Further details of each co-author s contributions can be found in Section Please note that the Introduction and Materials and Methods sections of this manuscript are similar to the Introduction, Background and Materials and Methods sections of this thesis (see Chapters 1-3). Thus, the same information is repeated in the manuscript, but in a more concise format. 55

72 4.2. Title Page SERUM-FREE CO-CULTURE OF HUMAN MESENCHYMAL STEM CELLS AND ARTICULAR CHONDROCYTES AS THREE-DIMENSIONAL AGGREGATES IN SUSPENSION BIOREACTORS Madiha Khurshid B.Sc. 1,2, Aillette Mulet-Sierra M.Sc. 3, Adetola Adesida PhD 2,3*, Arindom Sen PhD 1,2* 1. Pharmaceutical Production Research Facility, Schulich School of Engineering, University of Calgary 2. McCaig Institute for Bone and Joint Health, Cumming School of Medicine, University of Calgary 3. Department of Surgery, Faculty of Medicine and Dentistry, University of Alberta *Equally contributing senior authors Madiha Khurshid Room BI-514, Department of Chemical Engineering, University of Calgary, 2500 University Drive NW, Calgary, Alberta, Canada, T2N 1N4 Phone: Aillette Mulet-Sierra Room 3-021, Department of Surgery, Li Ka Shing Centre for Health Research Innovation, University of Alberta, Edmonton, Alberta T6G 2E1 Phone: Dr. Adetola Adesida Room 3-002E, Department of Surgery, Li Ka Shing Centre for Health Research Innovation, University of Alberta, Edmonton, Alberta T6G 2E1 56

73 Phone: Fax: Dr. Arindom Sen (corresponding author) Room ENB 204J, Department of Chemical Engineering, University of Calgary, 2500 University Drive NW, Calgary, Alberta, Canada, T2N 1N4 Phone: Fax: Submitted for consideration to Arthritis Research & Therapy; August 21,

74 4.3. Abstract Introduction: Osteoarthritis (OA) is characterized by surface erosion of articular cartilage. Chondrocytes from OA patients can recapitulate the functional matrix-forming phenotype of articular cartilage, and thus, may have utility in cell-based strategies to resurface cartilage lesions in OA joints. The development and implementation of such strategies will require large quantities of OA chondrocytes, which may be unattainable directly from OA joints, but could be generated in robust culture systems. Human articular chondrocytes (hacs) have previously been shown to proliferate without dedifferentiating when cocultured in direct contact with human mesenchymal stem cells (hmscs), and the hmscs preferentially undergo chondrogenesis. Together, the co-cultured cell populations synergistically generated neocartilage with enhanced functional matrix formation relative to that produced from either cell type alone. However, these previous studies were performed under non-scalable static culture conditions, which substantially limits the quantity of cells that can be generated for therapeutic purposes. To facilitate clinical translation, we investigated if hmscs and hacs could be co-expanded in scalable suspension bioreactors using serum-free medium (SFM). Methods: A serum-containing medium and two SFM (TheraPEAK, PPRF-msc6) were screened for their ability to support hac and hmsc proliferation. The best medium was then used to co-culture hmscs and OA hacs in static culture vessels and 125 ml stirred suspension bioreactors (inoculated at 20,000 cells/ml; 3 hmscs per hac). Cell densities and viabilities were measured using an automated cell counter. The phenotype of the cocultured cells was studied using gene expression, Safranin O staining and glycosaminoglycan (GAG) content. Results: PPRF-msc6 supported the greatest cell expansion in static co-culture, reaching cell densities of 540,000 cells/ml. Co-cultured cells spontaneously formed aggregates, when inoculated into bioreactors containing PPRF-msc6, which grew as a result of cell proliferation and matrix production. Compared to cells in static culture, the cells in bioreactors had a higher collagen I to II gene expression ratio, and six times the GAG/DNA 58

75 (8.0 micrograms/microgram versus 1.3 micrograms/microgram), but exhibited a lower apparent growth rate (0.01/h versus 0.02/h). Feeding in bioreactor co-culture more than doubled the cell yield from 41,000 cells/ml to 95,000 cells/ml, and extended the cell growth period in a single vessel from 8 to 16 days. Conclusions: Suspension bioreactors can be used to support the serum-free expansion of co-cultured hmscs and hacs as aggregates. This represents a new approach for the production of cells with the potential to contribute to cartilage repair. Keywords: cartilage tissue engineering, bioprocess development, scale-up, threedimensional culture, co-culture, serum-free medium, suspension bioreactors, aggregates, spheroids 59

76 4.4. Introduction Articulating joint surfaces have superior mechanical properties due to the presence of hyaline cartilage, a complex, three-dimensional (3D) extracellular matrix of collagen II and proteoglycans, which houses cells known as chondrocytes (1,6,8,11). Defects in articular cartilage as a result of injury, or wear and tear, can initiate a progressive, degenerative process that can eventually lead to osteoarthritis (OA), a chronic and debilitating medical condition that affects 25% of adults over 50 years of age (5). Since cartilage is avascular and has a very limited capacity for spontaneous self-repair, clinical interventions, such as microfracture and cartilage grafting, have been developed to facilitate repair. However, microfracture can result in the formation of undesirable fibrocartilage, and grafts are limited by poor integration with the native tissue (1,6,8). Thus, alternative approaches have emerged and are being explored. Cell-based therapies have been examined to determine their potential in resurfacing cartilage defects (6,8,10,11). Autologous chondrocyte implantation/transplantation is a procedure where cartilage is sampled from a non-weight bearing site and processed to isolate chondrocytes. Due to their paucity in this tissue (only million chondrocytes/g tissue (33)), and the low chondrocyte yields using current cartilage processing methods (<22% post-isolation (33)), the chondrocytes need to be cultured for 2 to 3 weeks under conditions that promote their proliferation in order to obtain clinically relevant numbers of about 2.6 to 5 million cells for typical focal defect sizes of 1.6 to 6.5 cm 2 respectively (10). In contrast, autologous treatment of OA necessitates restoration of larger surfaces, at least 25 cm 2 in size (9). This requires large numbers of OA-derived cells and matrix that are difficult to obtain using current culture techniques, since OA-derived cells have low proliferation rates and matrix production (9). Following cell culture, the expanded cell population is then reintroduced into the defect site, either in a scaffold (called matrixassociated autologous chondrocyte implantation) or together with a membrane (e.g. a periosteal flap), in an attempt to regenerate hyaline cartilage, and slow the degenerative process (1,8,11). 60

77 Whereas this approach has shown some clinical success (1), it suffers from a number of problems. First, certain long-term complications of this procedure, such as graft hypertrophy, have been linked to the prevailing cell culture methods (13). Second, harvested chondrocytes tend to dedifferentiate, undergo hypertrophy and senesce when cultured in a two-dimensional (2D) environment such as the tissue culture flasks typically used for this bioprocess. 2D expanded chondrocytes exhibit decreased production of collagen II and aggrecan, increased production of collagen I, and a fibroblastic morphology (1,6,11). For these reasons, it is difficult to obtain large numbers of differentiated chondrocytes, sufficient for a clinical intervention for OA. Furthermore, implantation of dedifferentiated human articular chondrocytes (hacs) into the defect produces fibrocartilage, which has inferior biomechanical properties and poor integration with the surrounding native tissue (8,11). Thus, improved culture methodologies are required that address the drawbacks of current hac population expansion techniques, and increase the efficacy of this otherwise promising cartilage repair strategy for OA. Human mesenchymal stem cells (hmscs) can be isolated from a number of adult tissues, including bone marrow, and can undergo chondrogenic differentiation (87). Thus, hmscs are also being studied extensively for cartilage repair applications (1,6,8,11). However, when serially passaged on a 2D surface in serum-containing medium (SCM) for an extended period of time, MSCs can lose their ability to proliferate, and their chondrogenic potential declines (88 90). In some cases, cultured MSCs have been shown to produce hypertrophic cartilage (11). It has been suggested that issues related to the chondrogenic redifferentiation of ACs and the differentiation of MSCs can be overcome by utilizing 3D culture methods (6,11,19 21). For example, when bovine chondrocytes were placed in stirred suspension culture, they proliferated within 3D aggregates without dedifferentiating (22). Moreover, when chondrocytes were seeded on 3D scaffolds and implanted into cartilage defects, these cellscaffold constructs gave rise to hyaline cartilage (91). When cultured in SCM as 3D cell aggregates, MSCs remained viable and their stem cell properties were enhanced (23). 61

78 Recently, the co-culture of ACs and MSCs has been investigated for the generation of higher quality cartilage tissues. For example, it has been shown that encasing an aggregated population of hmscs within a layer of juvenile chondrocytes can up-regulate chondrogenesis without inducing hypertrophy in the presence of fetal bovine serum (FBS) (24). Moreover, co-culturing MSCs with dedifferentiated chondrocytes in porous 3D scaffolds promoted the redifferentiation of the chondrocytes in a SCM (25). Co-culturing ACs and hmscs in a 1:3 ratio as 3D pellets in static culture has been shown to result in substantially greater glycosaminoglycan (GAG)/deoxyribonucleic acid (DNA) content than the proportional expected contribution of each cell type. Furthermore, there was a 46- fold increase in collagen II and a 171-fold decrease in collagen X expression in co-cultured pellets as compared to monoculture pellets (17). Co-culture has been shown to decrease hypertrophy in the generated tissues via the secretion of parathyroid hormone-related protein (13). Thus, there is evidence that co-culturing hacs and hmscs in 3D can produce cartilage-like tissues with higher quality compared to the cartilage-like tissues generated by each cultured cell type alone. However, greater cell expansion is required to generate sufficient numbers of cells using these techniques for the treatment of large cartilage surface defects such as in OA. Most of these studies were carried out on a small scale in static culture vessels, which need to be scaled up to larger, more efficient systems for clinical translation. These cells can be induced to form 3D aggregates in static vessels (e.g. coated plates, low attachment plates), but the use of traditional static culture vessels is manually intensive and inefficient, and thus not feasible for the generation of clinical quantities of cells. Bioreactors are scalable vessels, in which the environment can be carefully monitored and controlled to support the culture of cells. Bioreactors have been used successfully to efficiently generate large quantities of several mammalian cell types such as: hacs (22,57), hmscs (23,58 60), and other stem cells (62 65) on microcarriers and as 3D aggregates. 62

79 The aforementioned studies demonstrating successful cartilage generation utilized animalderived serum, such as FBS, which represents a regulatory hurdle. FBS is undesirable as it poses a risk of transmission of infectious zoonotic agents, and suffers from batch-to-batch variability, which can result in irreproducible cell growth, among other issues (56). In the present study, the feasibility of using suspension culture bioreactors to co-culture hmscs and hacs as 3D aggregates was investigated under serum free conditions. 63

80 4.5. Materials and Methods Cell harvest and expansion Articular cartilage specimens were obtained from donors who underwent total knee replacement surgeries. Bone marrow aspirates were obtained from the iliac crest of patients during routine orthopedic procedures. Local ethics committee approval of the University of Alberta, Edmonton, Canada was obtained for this study and institutional safety and ethical guidelines were followed. The ethics committee waived the need for written informed consent of donors as specimens used in the study were intended for surgical discard. Precautions were taken to preserve the privacy of the donors. Since this bioprocess was developed for an autologous treatment of OA, hacs were isolated from articular cartilage samples of donors diagnosed with OA (75). hmscs were isolated from the bone marrow of human iliac crest samples using protocols that have been described previously (17,52). The average age of the four donors used in these experiments (hac79, hac120, BM142 and BM119) was 55. One set of hac and hmsc donors (i.e. hac79 and BM119) was used for developing static and bioreactor co-culture protocols and a different set of hac and hmsc donors (i.e. hac120 and BM142) was utilized to evaluate the developed co-culture protocols at the end of the study. The passage level of the hacs used in experiments did not exceed three to mitigate dedifferentiation (11). hmscs were only used up to passage 7 (p7) to avoid the effects of cell senescence. It has been reported that these cells may undergo senescence as well as genotypic and phenotypic variation as early as passage 6 (86), although it has also been found that such detrimental changes do not appear until p12 when these cells are cultured in a serum-free medium (SFM) called PPRF-msc6 (78). Primary hacs and hmscs at p2 were frozen separately overnight at -80 o C in cryogenic medium (10% dimethyl sulfoxide (Sigma-Aldrich, St. Louis, USA) 20% FBS (Life Technologies, Mississauga, Canada) in basal medium) using a Mr. Frosty freezing 64

81 container (Thermo Fischer Scientific, Waltham, USA), and then stored in a cryopreserved state in the vapor phase of liquid nitrogen until needed for experiments. hacs were thawed and expanded in monolayer culture at inoculation densities of 5,000-15,000 cells/cm 2 in 25 cm 2 tissue culture flasks (T-25s) (Nunc, Penfield, USA). They were cultured using a serumcontaining chondrocyte medium composed of 10% FBS (Lonza Group, Basel, Switzerland) and 90% Dulbecco s modified Eagle s medium (DMEM) (VWR) and supplemented with Gentamicin-Amphotericin (Lonza), mg/ml L-Glutamine (Lonza) and 10 mm 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) (Sigma-Aldrich). Additionally, hmsc populations were thawed, inoculated into gelatincoated T-25 flasks at a density of 5,000 cells/cm 2 and allowed to expand as a monolayer using PPRF-msc6 (78,79). This medium had been supplemented with 100 units/ml Penicillin and 100 µg/ml Streptomycin (PenStrep) (Life Technologies) Media screening Initial studies focused on evaluating two SFM (each supplemented with PenStrep) for their ability to support the co-culture of hmscs and hacs. TheraPEAK (Lonza) is a commercially available, chemically-defined medium used for hmsc culture. PPRF-msc6 has been described in the literature, and has been shown previously to support rapid hmsc isolation and expansion (78,79). For comparison, a typical SCM reported in the literature for culturing hmscs (10% FBS 22.5% DMEM and 67.5% modified Eagle s medium α (αmem) with Gentamicin-Amphotericin, mg/ml L-Glutamine, 0.75% sodium pyruvate and 10 mm HEPES) was used as a control. Cell populations previously expanded in SCM were cultured, in duplicate, in each medium for three consecutive passages using gelatin-coated T-25s. The cells were inoculated into 4 ml of medium at a density of 5,000 cells/cm 2 at every passage and at a ratio of 1 hac to 3 hmscs in the first co-expansion passage. A hac to hmsc ratio of 1:3 has been demonstrated to result in a higher interaction index in co-culture pellets than a 1:9 ratio, meaning that it had greater GAG content than the proportionate contributions from both cell types (17). This ratio has also been used successfully in several other co-culture 65

82 systems (12,39,44). The ratio was not controlled beyond the first passage (i.e. the cell population generated in a vessel at a particular passage level was simply sub-cultured). Cell densities and viabilities were evaluated manually using a haemocytometer (Hausser Scientific, Horsham, PA) and the trypan blue exclusion assay (Sigma-Aldrich) (65) Bioreactor and static co-culture For most studies, cells were inoculated at a density of 20,000 cells/ml into 125 ml suspension culture bioreactors (NDS Technologies, Vineland, USA) that were treated with Sigmacote to prevent cell attachment. Each bioreactor contained 125 ml of PPRF-msc6 medium supplemented with PenStrep, and the inoculum consisted of 3 hmscs for each hac. The vessels were placed in a humidified 37 o C incubator containing 5% CO2 for the duration of the study. Based on preliminary experiments in our lab (data not shown), the agitation rate was maintained at 60 rpm for the first two days post-inoculation, and thereafter increased to 80 rpm to support high cell expansion in aggregates. For comparison, hacs and hmscs were also inoculated at a 1:3 ratio (20,000 cells/ml) in gelatin-coated T-25s containing 4 ml of PPRF-msc6. For the initial study on bioreactor co-culture as aggregates, hacs and hmscs were stained with long-term fluorescent dyes in contrasting colors. hacs were stained with PKH26 (red) and hmscs were stained with PKH67 (green). Cells were inoculated at a ratio of 1 hac to 2 hmscs and a viable cell density of 19,000 cells/ml in 60 ml of SCM. Bioreactors were operated at 60 rpm, and medium changes were carried out once a week. Periodically, representative 4 ml samples containing aggregates were removed aseptically from the bioreactors using a pipette. The samples were resuspended in 4 ml of trypsinethylenediaminetetraacetic acid (EDTA) (Life Technologies) and incubated for min depending on the aggregate size at 37 C with gentle mechanical trituration with a micropipetter every 10 min until the aggregates were dissociated into single cells. Cells were able to proliferate normally in static culture vessels following this dissociation protocol indicating that the period of exposure to trypsin was not detrimental to these cells 66

83 (data not shown). Subsequently, cell densities and viabilities were evaluated in duplicate using a Vi-CELL Cell Viability Analyzer (Beckman Coulter Canada, Mississauga, ON). Serial passaging of the aggregates was performed by employing enzymatic dissociation and mechanical trituration. After 10 days in culture, aggregates were exposed to two cycles of trypsin-edta at 37 C for 10 minutes followed by mechanical trituration using a micropipetter. The resulting cell suspension was observed under a microscope to ensure that the aggregates were dissociated. This single cell suspension was then used to inoculate the next set of bioreactors at 20,000 cells/ml. In bioreactor studies involving feeding, aggregates were allowed to settle by ceasing agitation, and the upper half of the spent medium was aseptically replaced by an equal volume of pre-warmed fresh medium. Note that the spent medium was centrifuged to separate any residual cells and aggregates, which were returned to the bioreactor along with the fresh medium Nutrient and metabolite analysis It is important to assess nutrient and metabolite kinetics in order to design a feeding strategy that overcomes growth inhibition due to nutrient depletion and waste accumulation (80,81). The concentrations of the nutrients, glucose and glutamine, as well as the metabolic waste products, lactic acid and ammonia, were measured in the spent medium to determine several kinetic parameters including: the yields of lactic acid on glucose and ammonia on glutamine (mol/mol), the uptake rates of glucose and glutamine (pmol/cell.day), as well as the production rates of lactic acid and ammonia (pmol/cell.day). The glutamine consumption and ammonia production were adjusted for spontaneous glutamine degradation and ammonia formation respectively. Sample sizes of 4 ml were removed from agitated bioreactors and T-25s, centrifuged to remove the cells, and stored at -20 C (58). The media samples were later thawed at 37 o C, and every replicate was analyzed twice using a BioProfiler 100 Plus (Nova Biomedical, Waltham, MA), which was calibrated according to the manufacturer s instructions. 67

84 4.5.5 Aggregate characterization Representative aggregate samples (0.5 ml) were removed from well-mixed bioreactors and deposited in a well of a 24 well plate (Nunc). Photomicrographs were taken of the aggregates using an Axio Observer light microscope (Carl Zeiss, Oberkochen, Germany) and ZEN 2011 microscope software (Carl Zeiss). As per previously published methods, the mean aggregate diameter (at least 20 aggregates analyzed per sample) was microscopically determined by averaging the length of the longest axis and the diameter perpendicular to this axis. Cell clusters with diameters less than 35 µm were not considered to be aggregates (65). The circularity of an aggregate in a 2D picture was quantified using eccentricity, which can be used to determine if the shear environment was sufficient to form smooth, spherical aggregates as opposed to elongated shapes. The eccentricity of an ellipse was calculated as the square root of the difference between the square of the semimajor and square of the semi-minor axes, divided by the length of the semi-major axis. Eccentricities less than 0.40 were nearly circular and greater than 0.60 were more elliptical. Thus, eccentricities less 0.60 were considered acceptable for the purpose of aggregate culture. For fluorescent imaging of the aggregates, representative aggregate samples (0.4 ml) were removed from well-mixed bioreactors, deposited in a glass-bottomed Petri dish (MatTek Corporation, Ashland, USA), and visualized using the LSM 700 Confocal Microscope (Carl Zeiss) and ZEN 2011 microscope software. Photomicrographs were taken of the aggregates under brightfield settings, and using the FITC (for PKH67) and Rhodamine (for PKH26) filters to discriminate hacs from hmscs Biochemical assays As an indication of extracellular matrix production, the amount of sulfated GAGs produced in comparison to the number of cells (i.e. the GAG/DNA) was measured ml of cell aggregate samples from bioreactors and cells harvested from T-25s using trypsin-edta were centrifuged. The cells were washed with phosphate-buffered saline (PBS), pelleted 68

85 and stored at -80 C until analysis. The aggregates were digested using 1 mg/ml Proteinase K in 50 mm Tris (ph 7.6), 1 mm EDTA, 1 mm Iodoacetamide, and 10 µg/ml pepstatin A (Sigma-Aldrich) at 56 C overnight. The DNA content was measured using the CyQuant cell proliferation assay kit (Life Technologies). The GAG content was measured by dimethylmethylene blue dye (Sigma-Aldrich) using chondroitin sulfate as a standard. A Dynex MRX microplate reader was used to measure the processed samples and the data were compared to standard curves to calculate DNA and GAG content Histology Aggregates were stained with Safranin O to detect the presence of cartilage. Cell aggregates were removed from bioreactors and T-25s and centrifuged. The cells were immersed overnight at 4 o C in 10% formalin (Thermo Fischer Scientific), then resuspended in PBS (Sigma-Aldrich) and stored at 4 C until analyzed. For analysis, samples were embedded in paraffin wax, sectioned at 5 μm, and stained with 0.1% (w/v) Safranin O (Sigma-Aldrich) and counterstained with 0.01 % (w/v) fast green (Sigma-Aldrich) to visualize GAG depositions Gene expression Total ribonucleic acid (RNA) was extracted from cells in flasks and aggregates in bioreactors using Trizol (Life Technologies). All the RNA were isolated by using pure Trizol (Life Technologies) for the monolayer cultures or in conjunction with molecular grinding resin (Geno Technology Inc. St Louis, USA) when extracting RNA from the aggregates. The extracted RNA was purified with the use of RNeasy mini-kit columns (Qiagen, Mississauga, Canada) after removal of contaminating genomic DNA by DNase treatment. RNA concentration and purity was determined using a Nanodrop 1000 spectrophotometer (Thermo Fisher Scientific). The purity of the extracted RNA was assessed spectrophotometrically by the ratio of absorbance at 260 nm (A260) to absorbance at 280 nm (A280). The values of the ratios ranged from 1.8 to 2.0, indicating a high level of purity. Total RNA (100 ng) in a 40 µl reaction was reverse transcribed to complementary 69

86 DNA using GoScript reverse transcriptase (Thermo Fisher Scientific) primed with oligo dt primers. Quantitative reverse transcription polymerase chain reaction was performed in a DNA Engine Opticon I Continuous Fluorescence Detection System (Bio-Rad, Mississauga, Canada) using hotstart Taq and SYBR Green detection (Eurogentec North America Inc., San Diego, USA). Primer sequences (see Table 4.1) were either taken from previously published work or custom designed by using Primer Express software (Life Technologies). All primers were obtained from Life Technologies. Gene (mrna) expression levels for each gene of interest were normalized to the expression level of human β-actin by the 2 -Δct method. All primer pairs were used at a concentration of 300 nm and the recorded amplification efficiencies were between 90% and 105% Statistics Statistical analysis was carried out in Microsoft Excel (2013). Cell density, aggregate diameter and metabolism data was reported with mean and range of data, and analyzed using two-way analysis of variance. Tukey s test was performed in MATLAB. GAG and DNA data was reported as mean ± standard error of the mean, and analyzed using a t-test. Significant differences were defined at p<

87 Table 4.1: Primer sequences used in quantitative real-time polymerase chain reaction. All primers were purchased from Life Technologies. Gene Primer Direction -Actin Accession# NM_ Aggrecan Accession# M55172 Collagen I (COL1A2) Accession# NM_ Collagen II (COL2A1) Accession# NM_ Collagen X Accession# X_60382 COMP Accession# NM_ AAGCCACCCCACTTCTCTCTAA-3 5 -AATGCTATCACCTCCCCTGTGT-3 5'-AGGGCGAGTGGAATGATGTT-3' 5'-GGTGGCTGTGCCCTTTTTAC-3 5'-TTGCCCAAAGTTGTCCTCTTCT-3' 5'-AGCTTCTGTGGAACCATGGAA-3' 5'-CTGCAAAATAAAATCTCGGTGTTCT-3' 5'-GGGCATTTGACTCACACCAGT-3' 5 -CAAGGCACCATCTCCAGGAA-3 5 -AAAGGGTATTTGTGGCAGCATATT-3 5'-CCGACAGCAACGTGGTCTT-3' 5'-CAGGTTGGCCCAGATGATG-3' (Fwd) (Rev) (Fwd) (Rev) (Fwd) (Rev) (Fwd) (Rev) (Fwd) (Rev) (Fwd) (Rev) 71

88 4.6. Results Serum-Free Media Screening in Static Co-culture First, the ability of serum-free medium to support the expansion of hac and hmsc populations was tested. hacs could be expanded separately in SCM, TheraPEAK and PPRF-msc6 for at least three passages with a doubling time of 3.5, 3.6 and 1.3 days respectively, at the end of the third passage. hmscs could be expanded separately in SCM, TheraPEAK and PPRF-msc6 for at least two passages with a doubling time of 6.6, -2.2 and 1.3 days respectively, at the end of the third passage. Based on these results, co-culture studies were then carried out in these three media. As shown in Figure 4.1A, co-cultured hmsc and hac populations had the greatest expansion in PPRF-msc6, with the cell density in this medium progressively increasing in each of three serial passages (6 days per passage). The cell densities in PPRF-msc6 (540,000 cells/ml on day 6 of the third passage) were significantly greater than those measured in TheraPEAK (9,000 cells/ml) or SCM (36,000 cells/ml). This translated to a much higher cumulative cell-fold increase of 2,100 in PPRF-msc6, compared to 0.63 in TheraPEAK and 3.7 in SCM over the 18 day experiment. In contrast, TheraPEAK and SCM were not able to support cell expansion for three serial passages, and thus were not considered to be suitable for co-culture of hacs and hmscs. The morphology of the cells in each medium was markedly different as illustrated in Figure 4.1B-D. Cells in PPRF-msc6 were smaller than cells grown in the other media and became smaller with every passage, adjusting from pre-culture in SCM. These cells also reached greater confluence, completely covering the culture surface within 6 days post-inoculation. Collagen I and II expression levels were highest in cells generated using PPRF-msc6. The gene expression ratio of collagen I to collagen II was several orders of magnitude lower in the PPRF-msc6 condition than the other conditions ( vs in SCM and 72

89 in TheraPEAK). Collagen X genes had the highest expression in TheraPEAK, and had 10 and 110 times lower expression in PPRF-msc6 and SCM respectively. Similarly, cartilage oligomeric matrix protein (COMP) gene expression was two and four times lower in PPRF-msc6 and SCM respectively. In contrast, aggrecan expression was highest in SCM, being 30 and 67 times lower in PPRF-msc6 and TheraPEAK respectively. As PPRFmsc6 best supported cell expansion, this SFM was chosen for all future bioreactor coculture studies Extension of Viable Bioreactor Co-culture Period Suspension bioreactors were used for the large-scale production of co-cultured cells as they are scalable, and can hold culture volumes that are orders of magnitude higher than static tissue culture flasks. Hence, they can provide an efficient means to drastically increase cell production. Bioreactor culture of adherent cells can take place on microcarriers or as aggregates of cells and matrix. Initially, bioreactor co-culture was tested as aggregates. In addition, hacs and hmscs were stained with two different long-term fluorescent dyes, to observe the presence of each cell type within the aggregates. The cells were able to proliferate as aggregates at a rate of h -1 under these conditions, reaching a maximum cell density of 74,000 cells/ml at day 25. The average aggregate size increased at a rate of 1.2 µm/day. In addition, the GAG/DNA was 27.2 µg/µg and the collagen I to II expression ratio was 0.14 at day 28. Also, the expression of collagen X was low at COMP and aggrecan expression was 0.11 and Notably, photomicrographs of the aggregates showed the presence of both hacs and hmscs within each aggregate (see Figure 4.2). Two approaches were evaluated for their ability to extend the culture period and increase cell production. The first approach was serial passaging, in which aggregates removed from a vessel after 10 days of culture were enzymatically dissociated into single cells and then inoculated into a new vessel with fresh medium at a density of 20,000 cells/ml. The second approach was feeding without passaging, wherein 50% of the spent medium in a given 73

90 vessel was replaced by fresh medium in order to replenish nutrients and dilute potentially harmful metabolic waste products. 74

91 Figure 4.1: Serum-free media screening. Three media (SCM, TheraPEAK and PPRFmsc6) were screened for their ability to support the proliferation of hacs and hmscs in static co-culture. All experiments were conducted in duplicate. Cells were passaged three times for 6 days each without medium changes. The first passage was inoculated at a 1 hac to 3 hmsc ratio and each passage was inoculated at 20,000 cells/ml in 4 ml of medium. A) Cell densities at day 6 in the three media are shown over three passages. Error bars show range of data for duplicate cell counts from duplicate conditions. Photomicrographs of cells in static co-culture at the third passage on day 6 are shown at 5x magnification for B) SCM, C) TheraPEAK and D) PPRF-msc6. 75

92 Figure 4.2: Photomicrographs of co-cultured aggregates. hmscs were stained with PKH67 (green) and hacs were stained with PKH26 (red) prior to inoculation in bioreactors in a ratio of 1 hac to 3 MSCs. Bioreactors were inoculated with 60 ml of SCM and agitated at 60 rpm. Representative samples of aggregates were imaged A) 4, B) 7 and C) 10 days after inoculation Serial Passaging in Bioreactor Co-Culture Serial passaging involves the transfer of cells generated in one bioreactor to one or more new bioreactors containing fresh medium. The capacity to serially passage co-cultures of hacs and hmscs in bioreactors was tested by enzymatically dissociating the bioreactorgenerated aggregates into single cells, and inoculating them into new bioreactors at a density of 20,000 cells/ml. The hac to hmsc inoculation ratio was purposefully not controlled during this passaging process to further propagate the cells using the same ratio at which they were harvested. When sub-cultured into the second bioreactor, the cell density did not increase over a span of 10 days (see Figure 4.3A). This lack of proliferation was surprising considering that, similar to the first bioreactor, the cells spontaneously formed aggregates (see Figure 4.3C- D), and maintained a relatively high viability. The apparent growth rate in the exponential growth phase of the first and second passage was 0.009/h (day 0 to day 10) and /h (day 0 to day 10) respectively. Also, the aggregates in the first passage had a GAG/DNA of 2.0 µg/µg. In addition, the co-cultured aggregates contained substantial deposits of self- 76

93 generated matrix, but no GAG deposits as evidenced by Safranin O staining (see Figure 4.3B). 77

94 Figure 4.3: Serial passaging of cells in bioreactor co-culture. Bioreactors were operated at 60 rpm for the first two days followed by 80 rpm. Cells were cultured for 10 days without medium changes for each of two passages. Aggregates harvested from the first bioreactor were dissociated with trypsin-edta and inoculated (20,000 cells/ml) into new bioreactor vessels as single cells in fresh medium. A) Cell densities in bioreactor co-culture in two consecutive passages are shown. Error bars show range of data for duplicate cell counts from duplicate conditions. B) Safranin O staining of cells co-cultured in the first passage is shown. C-D) Photomicrographs of bioreactor co-cultures for each of two passages are shown at 10x magnification after 10 days in culture. 78

95 Feeding in Bioreactor Co-Culture As serial sub-culturing into fresh medium did not appear to be an option to extend the overall period of cell expansion in culture, the effect of feeding fresh medium into bioreactor co-culture was examined. A fed-batch strategy, involving a 50% medium change performed on days 8, 12 and 16 was compared to a batch condition with no medium changes. Medium changes were carried out starting on day 8, since a four day lag phase was observed in previous studies (data not shown). As shown in Figure 4.4A, the cell density in the fed-batch condition increased steadily until day 16 (95,000 cells/ml), after which it decreased. In contrast, cells in the batch condition grew until day 8 (37,000 cells/ml) and there was no significant growth thereafter (41,000 cells/ml at day 16). The apparent growth rate in the exponential growth phase of the batch condition (days 2 to 8) was 0.007/h, whereas in the fed-batch condition, the growth rate over the same time period was 0.008/h. Although there was no further growth in the batch condition past day 8 (the day of the first feeding), there was cell growth in the fed-batch condition at a rate of 0.005/h (days 8 to 16). These results were corroborated visually by observing the aggregates from day 10 (see Figure 4.4B-C) and day 16 (see Figure 4.4D- E). Both conditions had similar amounts of GAG, and the average GAG/DNA of the batch condition was not significantly higher than that of the fed-batch condition (see Figure 4.5A- C). Furthermore, both conditions were negative for Safranin O (see Figure 4.5D-E). Aggregate diameter is an important consideration as cells within the core of large aggregates have limited access to nutrients in the medium (65). The average aggregate diameter appeared to increase over the culture period from approximately 50 µm (day 4) to 150 µm in both conditions as shown in Figure 4.5F. However, the aggregate diameter distribution (Figure 4.5G) showed smaller aggregates in the fed-batch condition (62% of aggregates were 50 to 150 µm) than the batch (65% of aggregates were 100 to 200 µm). Also, there was a higher aggregate density (95 aggregates/ml at day 16) in the fed-batch condition than the batch (20 aggregates/ml). 79

96 Figure 4.4: Feeding cells in bioreactor co-culture cell density and cell morphology. Effect of feeding was tested in bioreactor co-culture with regards to cell density and morphology. A) Cell densities are shown in bioreactor co-culture in the batch and fed-batch conditions. Error bars show range of data. Green arrows indicate time points for 50% medium change for the fed-batch condition. Photomicrographs of hmsc and hac aggregates at B-C) 10 days and D-E) 16 days in culture are shown at 10x magnification. 80

97 Figure 4.5: Feeding cells in bioreactor co-culture GAG levels and aggregate morphology. A) GAG, B) DNA and C) GAG/DNA of the aggregates are shown in the batch and fed-batch conditions after 19 days in culture. Error bars show standard error of the mean. Safranin O staining of cells co-cultured in the D) batch and E) fed-batch conditions are shown. F) Average aggregate diameter is shown over the culture period. Error bars show standard deviation. Green arrows indicate time points for 50% medium change for the fed-batch condition. G) Aggregate diameter distribution after 16 days in culture is shown. 81

98 Medium analyses revealed that the cumulative glutamine consumption and waste production were higher in the fed-batch condition (p<0.0005), as shown in Figure 4.6A-D. However, the cumulative glucose consumption was not different in the two conditions. In addition, the glucose uptake rates ( pmol/cell day) and lactic acid production rates (20 pmol/cell day) were quite similar in both conditions as shown in Table 4.2. The glutamine uptake rates (8.0 vs. 4.0 pmol/cell day) and the ammonia production rates (4.0 vs. 2.0 pmol/cell day) were double in the batch condition as compared to the fed-batch condition. The yields of lactate over glucose and ammonia over glutamine were 57% and 32% lower respectively in the batch condition versus the fed-batch. The maximum ammonia concentration appeared lower in the fed-batch condition (1.9 mm at days 8, 12, and 15-16) than the batch condition (2.5 mm at day 12 to 3.1 mm at day 19). Similarly, the maximum lactic acid concentration was 6.9 mm in the fed-batch condition at day 16 (versus 7.2 mm at day 12 to 9.1 mm at day 19 in the batch condition). Based on these results, the bioreactor protocol was modified to incorporate feeding at day 8 and 12 during a 16 day culture period. 82

99 Figure 4.6: Feeding cells in bioreactor co-culture nutrient consumption and waste production. The cumulative A) glucose consumption, B) lactic acid production, C) glutamine consumption and D) ammonia production are shown in both conditions. Error bars show range of data for duplicate samples from duplicate cultures. Green arrows indicate time points for 50% medium change in the fed-batch condition. 83

100 Table 4.2: Kinetic parameters for hmsc and hac bioreactor co-culture for batch and fed-batch operation modes. Parameter Batch Fed-batch Value R 2 Value R 2 Maximum Viable Cell Density (cells/ml) 4.1E4 N/A 9.5E4 N/A Maximum Apparent Growth Rate (h -1 ) N/A N/A Doubling Time (h) 96 N/A 83 N/A Glucose Uptake Rate (pmol/cell day) Glutamine Uptake Rate (pmol/cell day) Lactic Acid Production Rate (pmol/cell day) Ammonia Production Rate (pmol/cell day) Yield of Lactic Acid over Glucose (mol/mol) Yield of Ammonia over Glutamine (mol/mol) Comparison of Bioreactor and Static Co-culture Protocols Co-culture of hacs and hmscs is typically carried out in static culture vessels, but bioreactors have several advantages over static vessels including: scalability, control, and efficiency. Thus, the cell productivity using the new bioreactor protocol was compared to the corresponding static protocol, i.e. under serum-free conditions and with feeding. Since the static condition acted as the control in this experiment, the feeding regimen developed for the bioreactor condition was applied to the static condition. The static condition was stopped at day 12, since data from previous studies (not shown) demonstrated that cell growth in static co-culture (as a batch process) ceases at day 8. There was only one feeding (day 8) scheduled in the static condition (day 0 to day 12) whereas there were two feedings (day 8 and 12) in the bioreactor condition, since the bioreactor co-culture took place over a longer period of time (day 0 to day 16). 84

101 Growth curves of the cells in the static condition and in bioreactors are shown in Figure 4.7A and B respectively. The growth curve of the static condition is displayed in units of cells/cm 2, since it represents cell growth on a 2D surface. In contrast, the growth curve of the bioreactor condition is shown in units of cells/ml, since it represents growth as 3D aggregates in suspension. The cell density in static co-culture peaked at 113,000 cells/cm 2 (which translates to 707,000 cells/ml given the volume of medium being used) whereas in bioreactor co-culture, a typical growth curve was observed, with a lower peak density of 91,000 cells/ml. The apparent growth rate of co-cultured cells in the bioreactor condition (0.010/h from day 4 to day 14) was significantly lower than in the static condition (0.021/h from day 0 to day 6). It was interesting to note that in the static culture condition, whereas a near confluent monolayer was observed by day 4, feeding on day 8 induced a second layer of cells to grow on top of the existing monolayer (i.e. stratified monolayers) (Figure 4.7C), representing a second period of growth (as noted in previously in Figure 4.7A). This linear growth period following an exponential growth phase resulted in a peak density of 113,000 cells/cm 2 by day 12. In bioreactors, the co-cultured cells once again grew as aggregates (Figure 4.7D) with average aggregate diameter ranging from about 75 µm to 150 µm (Figure 4.8A). The average aggregate eccentricity did not change significantly across the culture period (Figure 4.8B). The growth rate of the average aggregate diameter was 4.9 µm/day. As shown in the aggregate size distribution (Figure 4.8C), most aggregates were between 50 µm to 100 µm on day 16, which represents a narrow range of aggregate sizes. The static condition had more of both total DNA and GAG than the bioreactor condition (see Figure 4.8D-F). However, the cellular matrix production rate in the bioreactorgenerated aggregates (5.8±0.6 µg GAG/µg DNA at day 10) was more than double that of the static condition (2.5 µg GAG/µg DNA), as shown in Figure 4.8F. In addition, there was a trend towards lower collagen I and higher collagen II in the bioreactor condition (see 85

102 Figure 4.9A-B), with a collagen I to collagen II gene expression ratio of 275 versus 59,500 for the static condition. There was also a suggestion of higher collagen X and aggrecan and lower COMP in the bioreactor condition (see Figure 4.9A). Safranin O staining revealed extensive matrix deposition, but did not show sulfated GAGs in the bioreactor condition (see Figure 4.9C-D). 86

103 Figure 4.7: Comparison of bioreactor and static culture protocols cell density and cell morphology. Cell densities in A) static co-culture and B) bioreactor co-culture are shown. Arrows indicate time points for 50% medium change. Photomicrographs of cocultures of hmscs and hacs are shown at 10x magnification after 10 days in culture in C) static culture flasks and D) bioreactors. White ovals highlight areas of stratified monolayers. 87

104 Figure 4.8: Comparison of bioreactor and static culture protocols GAG levels and aggregate morphology. A) Average aggregate diameter and B) average aggregate eccentricity are shown. Error bars show standard deviation. Green arrows indicate time points for 50% medium change. C) Aggregate diameter distribution after 16 days in culture is also shown. D) GAG, E) DNA and F) GAG/DNA in the bioreactor and static conditions are shown after 10 days in culture. Error bars show standard error of the mean. 88

105 Figure 4.9: Comparison of bioreactor and static culture protocols gene expression levels and GAG deposition. The gene expression, relative to β-actin, for the static and bioreactor conditions is shown after 10 and 16 days in culture respectively for C) collagen I, collagen X, aggrecan, COMP, and D) collagen II. Error bars show range of data. Safranin O staining of cells co-cultured in the bioreactor condition is shown after C) 10 and D) 16 days in culture. As shown in Figure 4.10, the cumulative glucose consumption, glutamine consumption, lactic acid production and ammonia production were significantly greater in the static condition (p<0.0001). Additionally, the uptake rates of glucose (10 vs. 7.0 pmol/cell day) and glutamine (4.0 vs. 2.0 pmol/cell day) appeared higher in the bioreactor condition on a per cell basis as tabulated in Table 4.3. Interestingly, the production rates of lactic acid (20 pmol/cell day) and ammonia production (2.0 pmol/cell day) were identical in both conditions. The yields of lactic acid over glucose and ammonia over glutamine seemed lower (15% and 58% lower respectively) in the bioreactor condition. Also, the maximum ammonia (2.7 mm at day 6 to 4.1 mm at day 12 vs. 1.9 mm at day 16) and lactic acid 89

106 concentrations (22.0 mm at day 12 vs. 7.0 mm at day 16) trended towards higher values in the static condition than the bioreactor condition. 90

107 Figure 4.10: Comparison of bioreactor and static culture protocols nutrient consumption and waste production. The cumulative A) glucose consumption, B) lactic acid production, C) glutamine consumption and D) ammonia production are shown in both conditions. Error bars show range of data for duplicate samples from duplicate cultures. Arrows indicate time points for 50% medium change. 91

108 Table 4.3: Kinetic parameters for hmsc and hac co-cultures in static and bioreactor conditions. Literature values for bovine ACs (92), hmscs (80,93,94) and murine MSCs (95) in static culture are included for comparison. Note that literature values for hac/hmsc co-culture in static conditions were not available, and literature values for bioreactor culture of hacs or hmscs were also not available. Parameter Static Bioreactor Static Value R 2 Value R 2 Literature Values (Reference Number) Maximum Viable Cell Density (cells/ml) 707,000 N/A 91,000 N/A N/A Maximum Apparent Growth Rate (h -1 ) N/A N/A 0.02 (93), (80) Doubling Time (h) 33 N/A 71 N/A 35.5 (93), (80), (94) Glucose Uptake Rate (pmol/cell day) (93), (80), (94) Glutamine Uptake Rate (pmol/cell day) (93), (80) Lactic Acid Production Rate (pmol/cell day) (80), (94) Ammonia Production Rate (pmol/cell day) Yield of Lactic Acid over Glucose (mol/mol) (92), 1.96 (93), (80), (95) Yield of Ammonia over Glutamine (mol/mol) (93) 92

109 4.7. Discussion In this study, we examined the feasibility of expanding populations of hacs and hmscs, both considered to be adherent cell types, together in suspension culture bioreactors. Bioreactors have previously been shown to support the culture of adherent cell populations using microcarriers (57 60,96,97) and as aggregates (22,23,62 65,73,74). However, the co-expansion of hacs and hmscs in bioreactors is novel and has yet to be reported. Our results clearly indicate that the 3D co-expansion of these two cell types as aggregates under serum free conditions in suspension culture is feasible. It has previously been shown that the co-culture of hacs and hmscs in static culture results in greater expansion of hacs without inducing dedifferentiation, and also promotes the differentiation of hmscs towards a chondrogenic fate (17). Here, we show that we can accomplish direct contact co-culture of these cell types in bioreactors as aggregates, as evidenced by fluorescent microscopy. These aggregates, composed of cells and accumulated matrix molecules, allow for the cells to exist in a tissue-like 3D structure, an environment that cells in a 2D monolayer culture cannot experience. The 3D culture of hacs and hmscs as aggregates in bioreactors resulted in several times greater matrix production than the same cells in monolayer culture within tissue culture flasks. It has been shown that 3D cell culture environments can affect cell behavior (98 101), and promote the generation of tissues with desired functional characteristics (6,11,19 21). Tissue culture flasks are typically used for the expansion of cell populations, but represent an inefficient means of generating clinical numbers of cells due to the large numbers of flasks required. In addition, handling large numbers of these vessels is manually intensive and leads to a lack of reproducibility in the cell populations generated. Moreover, the 2D nature of adherent cell culture in static vessels does not in any way resemble the 3D in vivo tissue environment in which cells normally reside. Bioreactors, on the other hand, can be used to grow cells as 3D aggregates. Moreover, they are scalable, meaning that vessel size can be increased to generate sufficient quantities of 93

110 cells for a given application. Thus, even if cell growth rates in bioreactors are lower than in static culture flasks, the ability to increase bioreactor culture volume can result in substantially greater numbers of cells. In addition to being scalable, the culture environment in bioreactors can be monitored and closely controlled to a much greater degree than in tissue culture flasks, which in turn can promote the generation of more uniform cell populations. In fact, larger bioreactor systems can accommodate many sensors, and the culture environment can be controlled in real time with computers using robust control strategies. Bioprocess control and product reproducibility are important considerations when the goal is to generate cell populations for application in a clinical setting (86). Whereas the spinner flask studies described here did not result in cell expansion rates as high as those seen in static culture, it should be noted that the culture environment did not go through a rigorous optimization regimen. Rather, this study lays the groundwork and shows proof-of-concept for using bioreactor technology to generate cell populations with utility in engineering functional cartilage tissues (17) and cartilage-like tissues such as nucleus pulposus (102). Notably, even under non-optimized conditions, the GAG/DNA was much higher than that measured in static culture, suggesting that the aggregate microenvironment may be conducive towards extracellular matrix production, although collagen II production will have to be improved. Furthermore, it is noteworthy that the bioreactor-generated, co-expanded cell aggregates in this study were not subjected to chondrogenic differentiation media containing transforming growth factor β (TGF-β), dexamethasone and ascorbic acid, as typically practiced in the art (9,17,20). Despite the up-regulation in the synthesis of GAGs, the measured gene expression of the co-cultured hmscs and OA hacs did not indicate a fully chondrogenic phenotype. The trends observed in our experiments have been reported previously in the literature. Hardingham and colleagues showed that when culturing hacs, collagen II and aggrecan expression decrease sharply (by p2) while collagen I expression increases immediately (18). In addition, the hacs were cultured for a few passages prior to use in experiments 94

111 and derived from patients suffering from OA, so high collagen X expression and decreased GAG were anticipated in favor of higher cell growth as shown previously (9,12 14). However, these experiments have demonstrated that OA chondrocytes can generate large quantities of cell and matrix as aggregates, which could be employed to resurface large areas of cartilage erosion, such as those found in OA. Thus, autologous cell therapy for OA patients may be possible using these techniques. The higher matrix production in bioreactors (using PPRF-msc6) was accompanied by lower cell growth, which was expected, as cartilage matrix production is known to be antagonistic to cell proliferation (18). The GAG/DNA values obtained in the serum-free bioreactor studies described here ( µg/µg) were similar to some reported co-culture studies (1-5 µg/µg) (12,103,104), but lower than others (9-25 µg/µg) (37 40,105,106). It is noteworthy that bioreactor co-culture in SCM resulted in much higher GAG/DNA values (27.2 µg/µg) and lower collagen I to II expression ratios than bioreactor co-culture in PPRF-msc6. The use of serum resulted in enhanced chondrogenic expression at the cost of a lower growth rate. Enzymatic dissociation and mechanical trituration of the aggregates disrupts cell-to-cell and cell-to-matrix associations. Dissociation of the aggregates from bioreactors was more difficult than detaching and dissociating adherent layers of cells in a tissue culture flask, due to the larger quantities of accumulated matrix produced by the cells and the increased number of intercellular-matrix interactions that would exist in a 3D environment. In addition, the cell-matrix interaction formed in aggregates may be more mature and stronger, thereby making them more capable of withstanding the shear stress caused by agitation of the culture environment. For these reasons, it was not surprising that dissociation of the aggregates into single cells required longer trypsin exposure and more rigorous trituration than cells harvested from static culture. The aggregate dissociation procedure used did not immediately decrease the viability of cells when sub-cultured into a second bioreactor. The cells retained the ability to form 95

112 aggregates, and the GAG/DNA found during the first culture period was maintained, even after two additional serial passages (data not shown). It is not immediately clear why these cells did not proliferate after being sub-cultured in bioreactors, especially considering that they were able to grow after being inoculated into static conditions (data not shown). The inoculation cell density in the second bioreactor was much lower than cell density at end of the culture period in the first bioreactor, which may have hampered cell proliferation in suspension. Moreover, the spent medium at the end of the first bioreactor culture period may have contained secreted factors that promoted cell division in aggregates, and moving the cells to fresh medium may have negatively impacted their proliferation in these structures. This may explain, in part, why the culture feeding strategy was more successful than the serial passaging approach. In addition, the vigorous enzymatic dissociation method may have had a negative impact on the cells from the first bioreactor culture period, cleaving the cell-to-cell bonds that form in 3D tissue-like structures (67). Thus, different dissociation protocols should be investigated for serial passaging. The enzymatic digestion protocol could be improved by decreasing the trypsin concentration, decreasing the period of exposure and using another enzyme such as collagenase (22), TrypLE, or Accumax cell aggregate dissociation medium (74). Other methods of dissociation that can be explored include chemical dissociation by modulating the medium ph (67) and mechanical dissociation (67,107). For other cell types, it has been demonstrated that the aggregate diameter needs to be maintained below 300 µm to avoid mass transfer limitations, and to minimize difficulties in aggregate dissociation for cell counting (65). The average aggregate diameter in our study was maintained at 150 µm, which is well below this limit. In addition, the average aggregate diameter did not change significantly after 8 days in culture even though the cell densities in culture were increasing, meaning that the shear field was able to control aggregate diameter (64,65). Furthermore, the aggregate size distribution was maintained within a range of 50 µm to 150 µm, which represents a narrow diameter distribution, resulting in more homogenous cell products, and suggesting that there was sufficient shear within the culture to impact aggregate size. However, the shear levels were not high enough 96

113 to prevent the coalescence of smaller aggregates into larger aggregates during the later phases of culture. This issue can be addressed in the current system by ending the culture prior to the onset of aggregate coalescence. Alternatively, coalescence could also be minimized by decreasing the calcium concentration in the medium, and increasing the agitation rate (and thus the magnitude of the shear) in the bioreactor (108,109). The inoculated single cells formed aggregates that had a circular profile on photomicrographs (average eccentricity between 0.50 and 0.60), likely due to the shear field in the bioreactors. Later in culture, non-circular aggregates were found to be the result of aggregate coalescence. In the present study, it can be inferred that the rate at which these coalesced aggregates reorganized themselves into spheres was rapid, since the average eccentricity measured was low (<0.60) throughout the culture period. Rapid coalescence of aggregates has been observed in other cell types such as neural precursor cells (110). Following the studies to screen the two SFM in static culture, PPRF-msc6 was chosen due to the high cell densities achieved. The accommodation of a larger number of cells per unit area (i.e. high cell densities) at confluence using PPRF-msc6 was due to the smaller average cell size. Small MSC size has previously been reported in the literature, and it has been suggested that the smaller cell size may be correlated to the degree of stemness exhibited by a cell, including reduced senescence, greater proliferation and multi-lineage potential (78,111,112). Furthermore, larger cell sizes have been linked to hypertrophy in chondrocytes (16). PPRF-msc6 has also been reported to be successful in the isolation of hmscs (78) and can support the long-term growth of hacs (data not shown). This means that PPRF-msc6 can also be used for serum-free isolation and pre-culture prior to bioreactor co-culture. In static culture, most adherent cell populations grow as a monolayer and stop proliferating due to contact-inhibition at confluence. However, in static co-culture with feeding, the cells were found to form an additional layer on top of the monolayer that was adhered to the flask. This phenomenon has previously been reported (113). It is possible that the high 97

114 levels of matrix produced by the cells within the first monolayer served as a substrate to which new cells could bond (i.e. it acted as a new surface), thereby supporting a second layer of cells. Furthermore, the stratified layers formed after the cultures were fed fresh medium, suggesting that the increased nutrient levels and/or decreased waste concentrations promoted the additional growth. As previously shown (113), medium consumption was the same in stratified culture (after day 8) as monolayer culture (before day 8), although the cell density was higher. The nutrient and waste concentrations were modulated by performing medium changes in the fed-batch condition. The feeding strategy employed here was able to maintain metabolite concentrations below inhibitory levels (35.4 mm lactic acid and 2.4 mm ammonia) reported in the literature (93), whereas the ammonia concentration climbed past inhibitory levels in the batch condition after day 12. The feeding strategy developed for bioreactor co-culture was not able to maintain ammonia concentrations below inhibitory levels in static co-culture past day 6. In addition to limiting metabolite concentrations, feeding provided a means to extend the culture period, and obtain greater cell productivity out of the same culture vessel. It is important to note that the metabolism values in the bioreactor condition (with feeding) were similar to that in the fed-batch condition of the feeding experiment (see Table 4.2 and Table 4.3). Literature values for the metabolism of hacs and hmscs in co-culture or in bioreactors have not been reported. However, the glucose and glutamine uptake rates as well as lactic acid production rates in the present study were comparable to literature values reported for AC and MSC monocultures (see Table 4.3). The yield of lactic acid on glucose was high and this is consistent with literature values reported in Table 4.3. Cartilage exists in a low oxygen environment, which encourages anaerobic respiration in chondrocytes. Thus, glucose is preferentially broken down into lactic acid, and a high yield was expected due to the presence of chondrocytes. 98

115 This study provides proof of concept that adherent cartilage-forming cell types can be coexpanded to clinically relevant quantities in suspension culture bioreactors, a technology that has been shown to support the rapid, large-scale expansion of non-adherent stem cell types. The 125 ml volume used in this study was sufficient to generate 6-11 million cells per bioreactor; treatment of a surface defect in load bearing cartilage is estimated to require 1 million cells/cm 2 (9). Moreover, we have shown that this can be accomplished under serum-free conditions. Furthermore, OA chondrocytes were able to generate matrix and GAGs in these studies, which have the potential to be used for autologous therapies for OA after isolation and expansion in bioreactors. Future studies should aim to optimize growth in suspension culture, with an emphasis on key factors known to influence hacs and hmscs, such as agitation rate and oxygen tension. Scale-up into larger, computer-controlled bioreactors should also be pursued as a means to produce greater numbers of cells in a better controlled culture environment, which should translate to more reproducible results. Furthermore, the aggregates should be differentiated using chondrogenic factors such as TGF-β1 (20,114), TGF-β3 (39), and dexamethasone (115) to investigate their capacity to undergo chondrogenesis. In the event that these cells prove useful for repair of cartilage, combining multiple aggregates into a tissue-engineered construct to treat a single defect site may be beneficial as it would preserve existing cell-cell as well as cell-matrix interactions within the 3D structure (116) Conclusions We have demonstrated for the first time that hmsc and hac populations can be coexpanded as aggregates in stirred suspension bioreactors. Moreover, this was accomplished using a defined, serum-free medium, which is an important consideration for any bioprocess in which the product has the potential to be used clinically. This study lays the foundation for future work aimed at optimizing a bioreactor-based process for generating clinical quantities of chondrogenic cells that can be used in therapies aimed at repairing cartilage or cartilage-like tissues. 99

116 4.9. Competing Interests The authors report that they do not have any conflict of interest related to the work presented in this publication Author s Contributions MK designed the bioprocess and studies, carried out the experiments and wrote the manuscript. In addition, MK performed the cell density and growth analyses, nutrient and metabolite analysis, and aggregate characterization, as well as the corresponding data analysis and statistics. AMS performed biochemical assays, histology and gene expression studies, as well as data analysis. AA conceived the experimental study, performed gene expression, data analysis, provided funding, participated in writing the manuscript, and cosupervised the study. AS conceptualized and designed the experiments, performed data analysis, participated in writing the manuscript, provided funding for the project, and cosupervised the study Acknowledgements Funding for this work is gratefully acknowledged as coming from the Natural Sciences and Engineering Research Council of Canada (NSERC) to AS, University Hospital Foundation of Alberta (UHF) to AA, the Canadian Institutes of Health Research (CIHR) MOP to AA, and Alberta Innovates Health Solutions (AIHS). We thank Dr. Nadr Jomha, Orthopaedic Surgeon-Scientist at the University of Alberta Hospital, for assistance with the procurement of hacs and hmscs Ethical considerations All studies were performed with cells taken from either surgical discards of patients undergoing total knee arthroplasty or during routine orthopedic surgery after approval and a waiver of informed consent of the local ethics committee of the University of Alberta (Edmonton, Canada). 100

117 4.13. List of Abbreviations Abbreviation Meaning 2D Two-dimensional 3D Three-dimensional αmem Modified Eagle s medium α COMP Cartilage oligomeric matrix protein DMEM Dulbecco s modified Eagle s medium DNA Deoxyribonucleic acid EDTA Ethylenediaminetetraacetic acid FBS Fetal bovine serum GAG Glycosaminoglycan hac Human articular chondrocyte hmsc Human mesenchymal stem cell HEPES 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid OA Osteoarthritis p(number) Passage (number) PBS Phosphate-buffered saline PenStrep Penicillin and streptomycin solution RNA Ribonucleic acid SCM Serum-containing medium SFM Serum-free medium T cm 2 tissue culture flasks TGF-β Transforming growth factor β 101

118 CHAPTER 5: EFFECT OF AGITATION RATE ON CO-CULTURED AGGREGATES IN BIOREACTORS 5.1. Introduction It was demonstrated that hmscs and hacs can be co-cultured on a large-scale in bioreactors using SFM (see Chapter 4). This bioprocess yielded 3D aggregates of cells and matrix that had a higher GAG/DNA than the 2D monolayer generated in T-flasks. In this chapter, the bioprocess was developed further by examining the effect of agitation rate on the co-cultured aggregates. Agitation is essential in stirred suspension bioreactors, since it serves to keep the cells in suspension, mixes the medium, increases oxygen diffusion into the medium, and improves convective flow inside aggregates (117). It is important to study the effect of agitation rate on the culture, since different cell types respond differently to shear and excessive agitation could damage the cells (66). Furthermore, agitation rate has an effect on cell growth rates and can control aggregate diameter (64,65,74,107). There are several reasons to control and limit aggregate diameters in bioreactor culture. Controlling aggregate size can prevent the necrotic cores, which can form in larger aggregates, caused by diffusional gradients that limit mass transfer of nutrient and wastes (65). In addition, it is desirable to have a narrow distribution of aggregate diameters to facilitate reproducibility in these cell products. Moreover, larger aggregates have lower contractility, cell growth and cell viabilities (69,118), and smaller aggregates can also pack into an irregularly shaped cartilage defect site with greater ease. Thus, agitation rate was tested in co-culture to determine its effect on several culture parameters including: cell growth, cell viability, cell phenotype, matrix production and 102

119 aggregate characteristics. Agitation rate was investigated both during the inoculation phase, in which the cells are forming aggregates, and the growth phase Growth Phase Agitation Rate Agitation rate in bioreactors was manipulated to determine the effect of shear stress on the co-culture of hmsc and hac aggregates. Three agitation rates were tested: 60 rpm, 80 rpm and 100 rpm for the entire culture period. In particular, the cell growth, cell viability and aggregate characteristics were examined for the effect of agitation rate during the entire culture period. The bioreactors were inoculated with 20,000 cells/ml at a 1 hac to 3 hmsc ratio in 125 ml PPRF-msc6 supplemented with PenStrep (see Appendix A). All bioreactors were inoculated at the same time with the same cell suspension, and thus had identical cell densities. The experiment was carried out as a batch process (without medium changes) to observe the cell growth phases (lag phase, exponential growth phase, plateau and death phase). Samples were taken every other day for evaluation. BM119 and hac79 were available and so were used in this experiment (see Table 3.1 for donor details) Cell Quantity There were several measures that were used to determine the effect of agitation rate on the quantity of viable cells obtained including: the apparent growth rate, length of the lag phase and the exponential growth phase, cell densities and viabilities. The apparent growth rate of the cells during the exponential growth phase was similar in the three tested agitation rates (0.011 h -1 at 60 rpm, h -1 at 80 rpm and h -1 at 100 rpm). However, the exponential growth phase took place at different times at each agitation rate (days 4-10 at 60 rpm, days 2-10 at 80 rpm and days 4-12 at 100 rpm). The growth curves obtained for each agitation rate appeared distinct (see Figure 5.1). For example, the peak cell density was different in each condition and took place at different times. Specifically, the peak cell density was on day 10 for 60 rpm (48,000 cells/ml), but day 12 for the 80 rpm (89,000 cells/ml) and the 100 rpm condition (68,000 cells/ml). 103

120 Furthermore, the average cell densities over the course of the culture period at each agitation rate were significantly different from the cell densities at other agitation rates. The 80 rpm and 100 rpm conditions had similar cell densities until day 8, after which there were greater cell densities in 80 rpm. The cell densities in the 60 rpm condition were lower throughout the culture period. The effect of agitation rate on cell densities may be due to the shear forces experienced by the aggregates and their influence on cell fate via mechanotransduction (101). The exact mechanism by which shear stress affects cell proliferation has not yet been determined. There was a two to four day lag phase that was observed in all conditions. Also, there were cell losses of 13-61% observed two days after the inoculation of 20,000 cells/ml medium in the bioreactors, as has been shown previously (74). Although some of the cells died initially, the remainder survived and proliferated. Despite the cell losses early in the culture period, most of the cell viabilities in all conditions were greater than 80%, especially later in the culture period. There were higher cell densities, but a trend towards lower cell viabilities in the 100 rpm condition, as compared to other conditions. The lower cell losses at 100 rpm were likely due to less cell death in that condition. Additionally, the viabilities at 60 and 80 rpm appeared higher than at 100 rpm in the first two to four days, since the non-viable cells in these conditions likely disintegrated in culture and could not be enumerated in the cell counts. Thus, the 60 and 80 rpm conditions have higher viabilities reported. The aggregate dissociation protocol may have lowered the cell viabilities during cell counting in the first two counts (day 2 and day 4), since the aggregates were small and the cells were more vulnerable to damage from the trypsin enzyme. Even so, the cell counts by and large reflected the reality of cells in culture, since the counts were verified with photomicrographs. Trends with increasing or decreasing cell densities could be verified with pictures of the aggregate samples. 104

121 Figure 5.1: Growth phase agitation rate was tested in bioreactor co-culture with respect to cell densities and viabilities. Bioreactors were inoculated in a 1 hac to 3 hmsc ratio with 20,000 cells/ml in 125 ml medium and operated at 60, 80 and 100 rpm. Cell densities in each agitation rate were different from cell densities at other agitation rates (p<0.05). Error bars show range of data of duplicate cell counts from duplicate cultures Aggregate Characteristics There were many single cells in the bioreactor 24 hours after inoculation as shown in Figure 5.2A-C. However, some cells had started agglomerating to form cell clusters. Aggregates were defined as agglomerations of cells and matrix that were at least 35 µm in diameter (65). From day 4 to day 16, the aggregate diameters steadily increased in the 60 and 80 rpm condition (see Figure 5.3A). However, the diameters remained the same in the 100 rpm condition after day 10 as with previous studies (65), likely due to the increased shear at 100 rpm. Average aggregate diameters in the 80 rpm condition were significantly different from the 100 rpm condition. The rate of growth of the average aggregate diameter was not different in the three conditions (5.6 µm/day at 60 rpm, 8.1 µm/day at 80 rpm and 2.2 µm/day at 100 rpm). The maximum aggregate diameter showed similar trends as the average aggregate diameter (see Figure 5.3B). Maximum aggregate diameters in 100 rpm 105

122 were different from 60 and 80 rpm. Note that the maximum aggregate diameter data was sensitive to outliers. 106

123 Figure 5.2: Aggregate morphology is shown for three agitation rates. Photomicrographs of co-cultured aggregates A-C) 1 day and D-F) 10 days in culture are shown at 10x magnification. 107

124 Figure 5.3: Aggregate diameters are shown at several time points during the bioreactor co-culture for three agitation rates. A) Average aggregate diameters and B) maximum aggregate diameters in the three agitation rates is shown. Error bars on average diameters show standard deviation of 20 samples from duplicate cultures. Error bars on maximum diameters show range of data from duplicate cultures. There were several factors that affected aggregate diameter: cell proliferation, matrix production, shear, aggregate coalescence and aggregate compaction. The amount of shear in the 100 rpm condition was sufficient to control the average aggregate diameter at about 80 µm at day 10, even though the cell densities in this condition increased until day 12. In the 80 rpm condition, the aggregate diameter leveled off below 150 µm at day 14, whereas cell densities dropped dramatically after day 12. However, the shear level in the 60 rpm condition was not sufficient to control aggregate size, so the aggregate size increased. It 108

125 was anticipated that the largest aggregate diameter would result from using 60 rpm. However, that condition had the least cell growth and there were not enough cells to increase the aggregate size sufficiently over time, so the aggregate size was lower than the 80 rpm condition. Both the 80 rpm and 100 rpm conditions could limit the average aggregate diameter to a maximum value. Thus, an agitation rate of 80 rpm can be used in circumstances in which a higher cell growth or larger aggregate size is required. Alternatively, the 100 rpm rate can be employed when smaller aggregates are desired. A possible mechanism for shear to impact cell proliferation is mechanotransduction. Shear stress can be sensed by the cells at the periphery of the aggregate, transmitted to the cells in the centre via cell-to-cell contacts and cell-to-matrix contacts, and transduced into biochemical signals that influence cell fate. The exact mechanism is not well-understood, but certain mechanical forces have been shown to impact cell proliferation and production of cartilage tissues (1,31). The aggregate size distribution was narrower at higher agitation rates as shown in Figure 5.4A. 83% of the aggregates cultured at 100 rpm were in a 50 µm range ( µm), whereas 70% of the aggregates grown at 80 rpm and only 51% of aggregates cultured at 60 rpm were in that range ( µm and µm respectively) at day 10. By day 14, the effects of agitation rate were similar, except that the aggregate diameter distributions at 60 and 80 rpm became wider (see Figure 5.4B). Furthermore, the aggregate diameter distribution in the 60 rpm condition at 14 days had a bimodal distribution, with many aggregates within µm as well as above 175 µm. Similar trends (i.e. multiple peaks in aggregate diameter distributions) have been shown previously (65). The size of the aggregate diameter distribution was clearly related to the agitation rate and thus, was controlled by shear. A narrower range of aggregate sizes, if desired, can be obtained by employing a higher agitation rate, which can be correlated to a maximum shear stress. 109

126 Figure 5.4: Aggregate diameter distribution is shown for three agitation rates after A) 10 and B) 14 days in culture. The aggregate density (measured in aggregates/ml medium) sharply increased in culture in the 80 rpm and 100 rpm conditions from day 4 to 6 (see Figure 5.5A), which coincided with the onset of the exponential growth phase. This increase can be attributed to the onset of cell proliferation. The peak aggregate density was greater at higher agitation rates, meaning there were more aggregates at higher shear levels. Cell densities increased from day 6 to day 10 (see Figure 5.1), but there was a substantial drop in aggregate densities after day 6, due to widespread aggregate coalescence. The matrix secreted by the cells made the aggregates very sticky, which may have encouraged aggregate coalescence. Also, there were similar decreases in aggregate density after day 10 in all conditions. Note that there was cell death after day 10-12, so aggregate density also dropped due to the disintegration of the cells during this time period. The aggregate density at 60 rpm was 110

127 significantly lower than at 80 and 100 rpm. Aggregate coalescence was observed in all conditions and thus, the level of shear did not seem to discourage coalescence. This topic will be discussed further in Section Aggregate coalescence takes places when two aggregates merge together to form one large aggregate. Single aggregates are usually more spherical, but aggregates form odd shapes during a merging event. Eccentricity is a measure of the circularity of an ellipse, but it was used to determine how spherical the 3D aggregates were and thus, measure the degree of aggregate coalescence taking place in culture. The eccentricity was quite high when the merging event was taking place, but it was lowered again when the newly merged, larger aggregate was formed. Eccentricity was calculated by measuring the two perpendicular diameters of the 2D projections of the aggregates in photomicrographs. Eccentricities less than 0.40 are nearly circular and greater than 0.60 are more elliptical, so aggregates with eccentricities less than 0.60 were desirable (see Appendix C for examples). The eccentricity of the aggregates did not increase significantly over the culture period (see Figure 5.5B). The average eccentricity in all three conditions hovered between 0.40 and 0.60 for the entire culture period. There were some deviations from this trend, which may be due to sampling errors. Thus, it can be concluded that aggregate coalescence occurred quickly and the newly merged aggregates became spherical within a short period of time. Additionally, photomicrographs did not show many aggregates coalescing at any given time, although the number of aggregates (aggregate density) decreased in culture. Although aggregate coalescence took place at all agitation rates, aggregate size was limited at 80 rpm and 100 rpm. Aggregate coalescence did not seem to affect aggregate size, likely due to another phenomenon, i.e. compaction of the aggregates, which will be discussed in more detail in Section The aggregates coalesced, which would imply increases in 111

128 size, but the 3D aggregates also compacted due to contraction of the actin filaments, which decreased their size. Overall, aggregate size was controlled using 80 and 100 rpm. Figure 5.5: A) Aggregate densities and B) eccentricities are shown at three agitation rates. Error bars on aggregate densities show range of data from duplicate cultures. Error bars on eccentricities show standard deviation of 20 samples from duplicate cultures. 112

129 5.2.3 Cell Quality The matrix production per cell, characterized by the GAG/DNA ratio, was an important characteristic in this study, since it provided a measure of the quality of the bioreactorgenerated aggregates (higher GAG/DNA values are considered better). There were no significant differences between the GAG/DNA of the aggregates at the different agitation rates (see Figure 5.6A). Note that matrix production tended to be higher in conditions that produced lower cell densities in this experiment, since matrix production is an antagonist to cell proliferation (18). The amount of cartilage matrix production can vary based on the cell type, cell source (different locations of cartilage), donor (including manifestation of osteoarthritis), and culture conditions (including oxygen tension and substrate types) (9,12,14,52,75,119), so it is difficult to compare values from literature. The GAG/DNA values of the aggregates in this study ( µg/µg) were similar to the GAG/DNA of the co-cultured pellets in some studies (1-5 µg/µg) (12,103,104), but were lower compared to other studies (9-25 µg/µg) (37 40,105,106). The Safranin O staining of the aggregates did not indicate the presence of GAGs (absence of pink), although there was a well-developed collagen matrix present (presence of green) (see Figure 5.6B). Expression of collagen II and aggrecan seemed low (<0.002 relative to β-actin) and collagen I expression seemed higher (>0.5) in the aggregates (see Figure 5.7A and C). This shows that the aggregates did not have a chondrogenic phenotype. There was also some collagen X expression (<0.02) (see Figure 5.7C), which indicates some hypertrophy. Hypertrophy is undesirable in cartilage tissues, but is expected during chondrogenic differentiation of MSCs (13,16) and thus, was expected in this study as well. None of the genes had differential expression between the three agitation rates with the exception of aggrecan, which was significantly lower in 60 rpm than 80 or 100 rpm (p<0.05). The ratio of collagen I to collagen II expression, the inverse of which is commonly used in literature to quantify chondrogenic gene expression (25,41,43), was lower at 100 rpm (5,800), but it was not significantly different from the ratios at other agitation rates (37,000 at 80 rpm and 8,400 at 60 rpm). 113

130 Figure 5.6: Growth phase agitation rate was tested with regards to GAG deposition. A) GAG/DNA of the aggregates at different agitation rates is shown after 16 days in culture. Error bars show standard error of the mean. B) Safranin O staining of aggregates cultured at 80 rpm is shown after 16 days in culture. Scale bar represents 100 µm. Data was collected by Aillette Mulet-Sierra. These results indicate that the cells in the aggregates were not fully differentiated towards a chondrogenic lineage, but produced high levels of matrix and GAGs. Osteoarthritic chondrocytes were employed in this study and these are known to display lower levels of collagen II and GAG, as well as higher levels of collagen X compared to healthy chondrocytes (9,12,13). Moreover, chondrocytes quickly lose their phenotype in culture, within 1-2 passages (120). In this study, the aggregates contained osteoarthritic chondrocytes that had been cultured for 2-3 passages. Thus, these gene expression patterns were to be expected. The purpose of this bioprocess was to generate large numbers of aggregates that have potential in cartilage repair. Specifically, the primary goal of these studies was to determine if hmscs and hacs can be co-cultured as aggregates on a large-scale in suspension bioreactors. The resulting aggregates could then be induced to produce cartilage tissues. These results indicate that 3D culture in bioreactors alone was not sufficient to differentiate the cells. In the future, it will be necessary to differentiate the aggregates, by culturing the aggregates with chondrogenic factors. This will stimulate the cells to produce greater amounts of collagen II and mature into more cartilage-like tissue (39,113,114). 114

131 There was no clear trend on the effect of agitation rate on matrix production or cell phenotype. However, agitation rate had an important effect on the cell densities. The greatest cell densities (89,000 cells/ml) and the longest exponential growth phase (10 days) were found at 80 rpm with high viabilities. Furthermore, aggregate diameter was controlled below 150 µm at 80 rpm, such that mass transfer into the aggregate was not limited like it is in larger aggregates. Aggregate diameters were also controlled at 100 rpm at about 75 µm, but there were lower cell densities in that condition. Thus, an agitation rate of 80 rpm was employed for future studies. 115

132 Figure 5.7: Growth phase agitation rate was tested with regards to gene expression. Gene expression of co-cultured cells relative to β-actin at different agitation rates is shown for A) collagen I, B) collagen X, COMP, C) collagen II and aggrecan. Error bars show range of data from duplicate cultures. An asterisk (*) indicates a statistically significant difference from the 60 rpm condition. Data was collected by Aillette Mulet-Sierra. 116

133 5.3. Inoculation Phase Agitation Rate There are two phases of culture in suspension bioreactors: the inoculation phase (day 0 to day 2), when cells initially form aggregates, and the growth phase, during which the cells predominantly proliferate. A lower agitation rate of 60 rpm in the first day of culture produced bigger clumps of cells (see Figure 5.2A-C) in the previous study (see Section 5.2.2), and thus was thought to favor cell agglomeration. It was hypothesized that a lower agitation rate in the first two days could improve cell aggregation and thereby, increase cell viabilities and densities. A reduced agitation of 60 rpm was tested in the inoculation phase (first two days) with 80 rpm for the remainder of the culture period (60/80 rpm condition) and compared to agitation at 80 rpm for the entire culture period (80/80 rpm condition). This study employed donors BM119 and hac109 (see Table 3.1) Cell Quantity Even though the agitation rate was only 20 rpm less in the 60/80 rpm condition and only for the first two days, the cell densities in that condition were significantly higher than the 80/80 rpm condition (see Figure 5.8). The reduced agitation was employed at a time of little to no cell growth, but was able to affect cell growth later in the culture period. The reduced shear in the first two days may have reduced cell damage or triggered different biomechanical signals, which encouraged cell proliferation later in the culture period. Both conditions had a lag phase of 4 days and had an exponential growth phase of 6 days (day 4 to 10). The apparent growth in the two conditions was not significantly different either (0.011 h -1 at 60/80 rpm and h -1 at 80/80 rpm). Remarkably, cell viabilities remained high in culture even when cell densities dropped (after day 10). This may be due to the rapid disintegration of non-viable cells in culture, such that they did not affect the cell viabilities. 117

134 Figure 5.8: Inoculation phase agitation rate was tested in bioreactor co-culture with respect to cell densities and viabilities. Bioreactors were operated at either 60 rpm or 80 rpm for the first two days, followed by 80 rpm for the remainder of the culture. Error bars show range of data of duplicate cell counts from duplicate cultures Aggregate Characteristics The cells start to stick together as early as 2 hours after inoculation in both conditions (see Figure 5.9A and E). However, it took more than two days for most of the single cells to form aggregates, since there were many single cells left after two days in culture. The aggregate density stayed constant from day 2 to 6 (see Figure 5.10A). Following day 6, aggregate coalescence occurred in both conditions (see Figure 5.9C-D, G-H), as aggregate density decreased exponentially (see Figure 5.10A). The shear environment in the bioreactor was insufficient to stop aggregate coalescence, since the cell-secreted matrix made the aggregates very sticky. 118

135 Figure 5.9: Aggregate morphology is shown for two inoculation agitation rates. Photomicrographs of aggregates cultured at A-D) 60/80 rpm and E-H) 80/80 rpm are shown at 10x magnification after 1, 6, 10 and 12 days in culture. Scale bars represent 200 µm. 119

136 Figure 5.10: A) Aggregate densities and B) eccentricities are shown for two inoculation agitation rates. Error bars on aggregate densities show range of data from duplicate cultures. Error bars on eccentricities show standard deviation of 20 samples from duplicate cultures. The aggregate eccentricity remained around as can be seen in Figure 5.10B. This observation was consistent with the previous experiment regarding growth phase agitation rate. This uniform eccentricity values mean that aggregate coalescence occurred too quickly to affect this measure. The average and maximum aggregate diameters were similar in both conditions (see Figure 5.11A-B). The average aggregate diameter increased at a rate of 9.4 µm/day and 7.4 µm/day in the 60 rpm and 80 rpm inoculation conditions respectively. The maximum 120

137 aggregate diameter increased at 20 and 14 µm/day in the 60 rpm and 80 rpm inoculation conditions respectively. After day 6, aggregate density decreased due to aggregate coalescence and cell densities increased due to cell proliferation. These two phenomenon accounted for the increase in average and maximum aggregate size from about 75 and 100 µm respectively after day 6. The aggregate size distribution was similar in both conditions at day 10. Specifically, 56% and 49% of aggregates were within µm for the 60 and 80 rpm conditions respectively (see Figure 5.12A), which represents a narrow distribution of aggregate sizes. Figure 5.11: A) Average and B) maximum aggregate diameters are shown at several time points during the bioreactor co-culture for two inoculation agitation rates. Error bars on average diameters show standard deviation of 20 samples from duplicate cultures. Error bars on maximum diameters show range of data from duplicate cultures. 121

138 Figure 5.12: Aggregate characteristics are shown at two inoculation agitation rates. A) Aggregate diameter distribution is shown after 10 days in culture for both conditions. B) Packing density is shown at days 2, 6, 8 and 10 in the two conditions. Error bars show range of data from duplicate cultures. The packing density of aggregates is used as a measure of compaction of the aggregates. This compaction is commonly observed in aggregate cultures of MSCs (69) and may be due to actin-mediated contractility in the aggregates, which can impact aggregation, coalescence, and motility (118). The packing density of the aggregates was expressed as cells per volume of aggregate (cells/ml aggregate). The packing density was similar in both conditions and ranged from cells/ml aggregate (see Figure 5.12B), which is lower than the packing densities in literature ( cells/ml aggregate (65)) and 122

139 most mammalian tissue ( cells/ml (65)). However, chondrocytes make up only 1-2% of the volume of native cartilage (1), so low packing densities are acceptable for cartilage repair applications. Since the packing density at day 8 was double that of day 6 (see Figure 5.12B), it can be concluded that the aggregates were experiencing compaction. This means that there were more cells packed into the same volume of aggregate. Therefore, an increase in the number of cells in the aggregate (due to cell proliferation) would not result in a proportional increase in aggregate volume (and thus diameter). Note that the cells also produced matrix, which accounted for some of the aggregate volume. Thus, aggregate coalescence (as opposed to cell proliferation) was likely the dominant mechanism that caused an increase in aggregate diameters. This conclusion is supported by photomicrographs of the aggregates (see Figure 5.9C-D and G-H), which show the formation of larger aggregates that were clearly formed due to aggregate coalescence. Such aggregates have irregular shapes that illustrate the merging of multiple aggregates Cell Quality The GAG/DNA was not significantly different between the conditions (see Figure 5.13A). The values were consistent with the results in the preceding agitation rate experiment. The Safranin O staining of the co-cultured aggregates was again negative for sulfated GAGs, given by an absence of pink staining (see Figure 5.13B-C), meaning no substantial GAG deposition was observed in the aggregates. Even so, there was a well-developed collagen matrix surrounding the cells, stained in green. Gene expression was similar in both conditions (see Figure 5.14A-B), with high collagen I expression, as expected. There was lower expression for collagen II, aggrecan and COMP, genes that signify a chondrogenic phenotype. Collagen X was also present, indicating some level of hypertrophy. 123

140 Figure 5.13: Inoculation phase agitation rate was tested with regards to GAG deposition. A) GAG/DNA is shown in both conditions after 13 days in culture. Error bars show standard error of the mean. Safranin O staining of aggregates cultured at inoculation agitation rates of B) 60 rpm and C) 80 rpm are shown after 13 days in culture. Scale bars represent 100 µm. Data was collected by Aillette Mulet-Sierra. Figure 5.14: Inoculation phase agitation rate was tested with regards to gene expression. Expression of A) collagen I, B) collagen II and X, aggrecan and COMP, relative to β-actin, is shown in both conditions after 13 days. Error bars show range of data from duplicate cultures. Data was collected by Aillette Mulet-Sierra. 124

141 Furthermore, the collagen I to collagen II expression ratios of these aggregates were also similar in both conditions (55 at 60/80 rpm and 71 at 80/80 rpm), although these ratios were much lower than in the previous experiment (see Section 5.2.3). This indicates that the co-cultured aggregates in this experiment had an enhanced chondrogenic phenotype than the aggregates in the previous experiment. This difference in chondrogenic gene expression could arise from donor-to-donor variability, since there was a different chondrocyte donor in this experiment. The goal of this study was to produce large quantities of aggregates that could be later induced to form cartilage tissues. The GAG deposition and gene expression were studied to determine what effect, if any, agitation rate had on the aggregates in this regard. These aggregates had not been stimulated with chondrogenic factors for the purposes of this study, so a chondrogenic phenotype was not expected. The aim of this agitation study was to increase cell densities by examining the effect of lowering the inoculation phase growth rate. Interestingly, a reduced inoculation agitation rate of 60 rpm increased cell growth significantly later in the culture period. However, it did not impact cell phenotype, matrix production or aggregate characteristics. Thus, bioreactors were agitated at 60 rpm in the first two days of culture for all future experiments Conclusions The growth phase agitation rate had a significant impact on cell densities and some aggregate characteristics. In addition, the agitation rate during the first two days (inoculation phase) impacted cell densities later in the culture period. Agitation rates of 80 and 100 rpm were able to control average aggregate size to an upper limit. There was also widespread coalescence of multiple aggregates in culture, which led to greater aggregate diameters. The GAG deposition and chondrogenic phenotype were lacking in the aggregates and were largely unaffected by agitation rate. However, there was substantial matrix deposition in the aggregates, which made them very sticky. Overall, manipulation 125

142 of the agitation rate led to the production of higher cell densities in the aggregates. This supported the end goal of this thesis, i.e. the large-scale production of co-cultured aggregates in bioreactors. 126

143 CHAPTER 6: EFFECT OF OXYGEN TENSION AND DONORS IN BIOREACTOR AND STATIC CO- CULTURE 6.1. Introduction This final results chapter examines the effect of oxygen tension in the bioreactor and static co-culture systems described earlier in this thesis. In addition, the bioreactor co-culture protocol, developed in this chapter, was evaluated by comparing it against the static coculture protocol. Furthermore, the robustness of the bioreactor co-culture protocol was explored by testing it with cells from different donors. The experiments described in this chapter were carried out using factorial design (see Section and Figure 3.1). This means that they were carried out at the same time, so that they can be compared directly. Oxygen tension has been shown to have an effect on both hmscs and hacs in culture. Specifically, hypoxia has been shown to aid in the redifferentiation of hacs in 3D culture by increasing the GAG/DNA (46) and collagen II expression (43,47). Furthermore, hypoxic conditions have been shown to enhance increase collagen II expression in hmscs (49,50,52). However, the effect of hypoxic conditions on hypertrophy in MSCs or ACs is more controversial, with conflicting results from different studies. Hypoxia has been shown to both delay and reduce hypertrophy (49,55) as well as increase it (43,50,51) in MSCs. Additionally, hypoxia has reduced (46) and had no effect (47) on hypertrophy in hacs. Hypoxic preconditioning may affect the chondrogenic response of cells. For example, hypoxic-expanded cells and normoxic-expanded cells have different responses in hypoxic and normoxic differentiation conditions (52) and the length of exposure to hypoxic conditions can impact matrix deposition (43,46). The effect of hypoxia was tested in static and bioreactor co-culture. The oxygen tension was held constant at either 3% or 21% throughout the culture period. It should be noted 127

144 that all cells were actually kept in their respective oxygen tension during cell isolation, static pre-culture, and in both static and bioreactor co-culture. This was accomplished by culturing two sets of cells for each hmsc and hac donor (i.e. BM142 and hac119). One set was kept in an incubator with 21% oxygen tension and the other set was kept in an incubator with 3% oxygen tension. At the end of static pre-culture, there were 4 sets of cells (two cell types each grown in hypoxia and normoxia). The hacs and hmscs that were cultured in hypoxia were used to inoculate the hypoxic conditions in bioreactor and static co-culture. Similarly, the hacs and hmscs that were cultured in normoxia were used to inoculate the normoxic conditions in bioreactor and static co-culture. So, the cells that were co-cultured in the bioreactor and static conditions had been exposed to either hypoxia or normoxia for the entire duration of in vitro culture. Thus, the cumulative effect of oxygen tension (as opposed to just the oxygen tension during the co-culture period) was tested, and the exposure time to normoxia or hypoxia was identical in both conditions. A comparison to static co-culture was required to determine the merits of the bioreactor co-culture protocol developed in this thesis. The serum-free co-culture in bioreactors was compared to static co-culture in Chapter 4. It is compared again in this chapter under hypoxic conditions at 3% O2 (52). The same donor pair (i.e. BM142 and hac119) was used in both static and bioreactor co-culture, so that the results in this study were not confounded by donor-to-donor variability. Donor-to-donor variability is important in the evaluation of a bioprocess, since the choice of donor can affect cell productivity, including cell proliferation and matrix production (9,77). In a co-culture system, the donor from each cell type can affect the culture endpoints. So, another set of bioreactors was added to this factorial study to evaluate how a different pair of donors would affect this bioprocess. Cells produced from the bioreactor co-culture with one donor pair (BM142 and hac119) were compared to the cells obtained from the co-culture of an additional donor pair (BM143 and hac120) to determine the effect of donor-to-donor variability. 128

145 The bioreactors were inoculated at 20,000 cells/ml in a 1:3 hac to hmsc ratio using 125 ml PPRF-msc6 in 125 ml suspension bioreactors. hmscs at passage 5 and hacs at passage 2 were used to inoculate the bioreactors. 50% medium changes were carried out on days 8 and 12 based upon results from a previous study (see Chapter 4). The culture was carried out for 16 days based on expected culture time from those experiments. Bioreactors were agitated at 60 rpm for the first two days and then at 80 rpm for the remainder of the culture period based upon results from Chapter 5. The static co-culture was carried out by inoculating 20,000 cells/ml in a 1:3 hac to hmsc ratio using 4 ml of PPRF-msc6 in T-25s. hmscs at passage 5 and hacs at passage 2 were used to inoculate the static co-cultures. A medium change was carried out on day 8 to correspond to the medium changes required in bioreactor co-culture. Thus, a direct comparison could be made between the static and bioreactor protocols. The co-culture was carried out for a period of 12 days, since the cells were expected to reach confluence and stop growing after 8 days in culture based on previous experiments (see Chapter 4) Role of Oxygen Tension in Static Co-culture The effect of hypoxic conditions on static co-culture in serum-free medium was studied. After a two day lag phase, exponential cell growth occurred in both conditions (see Figure 6.1A). Both cultures had reached confluence by day 6 as shown in Figure 6.1B-C. Furthermore, the cell growth curves were very similar until day 8. The apparent growth rate in the exponential growth phase (days 0-6) of the two conditions were not different (0.022/h in normoxia and 0.024/h in hypoxia). The cells in the hypoxic condition started to die after day 10 (see Figure 6.1A and E), as seen in previous experiments (data not shown). However, the cell density in the normoxic condition increased again after day 8, and thus the cell densities in the normoxic condition were greater than the hypoxic condition (see Figure 6.1A). Remarkably, the cells grew in stratified monolayers (see Figure 6.1D), reaching cell densities of 102,000 cells/cm 2 on day 12. This contrasts sharply with the maximum cell density in the hypoxic condition (48,000 cells/cm 2 ) on day

146 Stratified monolayers, in which cells grow on top of the confluent layer of cells and matrix, have been reported previously in the literature (113). Specifically, hacs have been shown to proliferate in the exponential phase for a longer period of time in stratified cultures and produce a higher cell yield, as confirmed with these results. Cell densities at confluent culture ( cells/cm 2 (113)) reported here are within published values for hac culture, and cell densities obtained in stratified culture here were lower than literature values ( cells/cm 2 (113)). This may be due to the fact that only parts of the static culture in normoxia displayed stratification. Note that the second growth phase took place only in normoxic condition, which had higher oxygen levels. So, oxygen was able to diffuse past the second layer of cells and support both layers of cells. Also, the second growth phase coincided with the medium change on day 8, so it is possible that the higher nutrient concentrations caused by the feeding allowed the additional cell growth. It has also been previously demonstrated that the nutrient consumption in stratified and monolayer cultures are similar, although there are more cells produced per volume of medium in stratified cultures (113). Thus, the abundance of nutrients and oxygen in the normoxic condition after feeding may have allowed the cells to form another layer on top of the monolayer, possibly adhering to the self-generated matrix. Due to the onset of stratified culture, the cell density in the normoxic and hypoxic conditions were significantly different. Interestingly, the GAG/DNA of the co-cultured monolayers was almost identical in both conditions at day 10 (see Figure 6.2C). There was more GAG and DNA produced in the normoxic condition than the hypoxic condition, but the ratio of GAG to DNA was identical. In addition, all genes seemed to be up-regulated under hypoxia in static culture, especially COMP, collagen I, and II (see Figure 6.2A-B). The collagen I to collagen II gene expression ratio was 5000 in the normoxic condition, but only 560 for the hypoxic condition, which indicates that hypoxia may have helped to up-regulate the chondrogenic 130

147 phenotype even without the addition of exogenous factors known to induce chondrogenesis. Although there was higher cell growth in normoxia, hypoxia encouraged a more chondrogenic phenotype in static co-culture, as compared to normoxia. The greater number of dedifferentiated cells produced in normoxia may not be as helpful towards a cartilage repair therapy. For this reason, static co-culture should employ hypoxia and have a shorter culture time, passaging when cells are near confluence. 131

148 Figure 6.1: Oxygen tension was tested in hmsc and hac static co-culture in duplicates. T-flasks were inoculated in a 1 hac to 3 hmsc ratio with 20,000 cells/ml in 4 ml PPRF-msc6. A) Cell densities and viabilities in normoxic (21% O2) and hypoxic conditions (3% O2) are shown. Error bars show range of data of duplicate cell counts from duplicate cultures. Photomicrographs of co-cultured cells in static culture are shown at 10x magnification for both conditions after B-C) 6 days and D-E) 10 days in culture. Oval highlights area of stratified monolayers. Scale bars represent 100 µm. 132

149 Figure 6.2: Gene expression and GAG production of co-cultured cells in T-flasks was determined in normoxia and hypoxia after 10 days in culture. The gene expression relative to β-actin is shown for A) collagen I, collagen X, aggrecan, COMP, and B) collagen II. C) GAG/DNA of the co-cultured cells in normoxia and hypoxia is shown after 10 days in culture. Data was collected by Aillette Mulet-Sierra. 133

150 6.3. Role of Oxygen Tension in Bioreactor Co-culture The effect of hypoxia on bioreactor co-culture was also tested. The cell densities measured in the hypoxic condition were significantly higher than the normoxic condition until the last day of culture (day 16) (see Figure 6.3A). The reason for this was the particularly long ten day lag phase in the normoxic condition, as opposed to a two day lag phase in hypoxia. After the long lag phase in normoxia, the cell densities climbed up and the maximum cell density (71,000 cells/ml at day 16) ended up being higher than in hypoxia (63,000 cells/ml at day 12). The apparent growth rate in the exponential growth phase of the normoxic condition (0.013 h -1 from days 10-16) was not significantly different from that of the hypoxic condition (0.011 h -1 for days 4 to 10) (see Figure 6.3B). Normoxic conditions increased the lag phase in bioreactor co-culture compared to hypoxic conditions, with the donor pair that was tested. There was a two day lag phase in the hypoxic condition, but a long ten day lag phase in the normoxic condition. Note that BM142 and hac120 were pre-cultured and co-cultured in normoxia at the same time and in the same conditions as this experiment, but had a lag phase of 4 days (see Section 4.6.3). In this experiment, the cells (BM142 and hac119) in the normoxic condition were also pre-cultured and co-cultured in normoxia, but there was a lag phase of 10 days. Since all other culture conditions were the same in these two experiments including passage level of each cell type, this difference in lag phase may be due to the different chondrocyte donor. The effect of a different chondrocyte donor will be discussed further in Section 6.5. Even so, it is important to note that in the hypoxic condition with the same donor pair (BM142 and hac119), there was a shorter lag phase relative to the normoxic condition. 134

151 Figure 6.3: Oxygen tension was tested in hmsc and hac bioreactor co-culture. Bioreactors were inoculated in a 1 hac to 3 hmsc ratio with 20,000 cells/ml in 125 ml PPRF-msc6. Bioreactors were operated at 60 rpm for the first two days and 80 rpm for the remainder of the culture period. A) Cell densities and viabilities in normoxic (21% O2) and hypoxic conditions (3% O2) are shown. Error bars represent range of data of duplicate cell counts from duplicate cultures. B) The apparent growth rate in the exponential growth phase (days in normoxia and days 4-12 in hypoxia) are shown. Error bars show range of data from duplicate cultures. C) The GAG/DNA of the bioreactor-generated aggregates is shown after 16 days in culture. Error bars show standard error of the mean. GAG/DNA data was collected by Aillette Mulet-Sierra. 135

152 Most single cells in both conditions formed aggregates within four days (see Figure 6.4A- B), despite the long lag phase in the normoxic condition. Thus, aggregate formation was independent of cell growth. At day 10, the aggregates generated in hypoxia were all welldefined and appeared healthy, as opposed to the aggregates in the normoxic condition, which were misshapen and did not have well-defined boundaries (see Figure 6.4C-D). Thereafter, the cells in both conditions coalesced into larger aggregates, as described in Chapter 5. The cells in hypoxia coalesced more than in hypoxia, since there were likely stickier due to greater matrix production. Figure 6.4: Aggregate morphology at the two oxygen tension levels are shown. Photomicrographs of co-cultured cells in bioreactor culture are shown at 10x magnification for normoxic and hypoxic conditions after A-B) 4 days and C-D) 10 days in culture. Scale bars represent 100 µm. 136

153 After 16 days in culture, both conditions had reached relatively high cell densities. Despite this similarity, the GAG/DNA in the hypoxic condition was about double that in the normoxic condition, although there were no significant differences (see Figure 6.3C). In Chapter 4, it was discussed that cell growth is an antagonist to matrix production (18), but hypoxia was able to encourage both cell proliferation (with a shorter lag phase) and GAG production. Furthermore, the hypoxic condition appeared to have higher expression of all genes tested (see Figure 6.5A-B), with the expression of COMP being significantly higher in hypoxia. Interestingly, the collagen I to collagen II gene expression ratio was only 2.0 in hypoxia (compared to 13 in normoxia). However, the Safranin O staining of the aggregates in both conditions was negative for sulfated GAGs, meaning that there were not enough GAGs in the aggregates to visualize using this stain. The stain did display substantial collagen matrix deposition (see Figure 6.5C-D). These results indicate that hypoxia encouraged a more chondrogenic phenotype. The concentrations of nutrients and waste products were analyzed to better understand the metabolism of the cells. The co-culture exhibited greater cumulative glucose consumption in hypoxia than in normoxia (see Figure 6.6A). The values in hypoxia surpassed the values in normoxia past day 6, which coincided with greater cell densities in hypoxia. However, the glucose uptake rates per cell over the exponential growth phase were identical in both conditions (see Table 6.1) and to values from previous experiments (see Tables 4.2 and 4.3). The glucose uptake rates was also similar to values in literature (see Table 6.1). This is due to the greater cell growth earlier in the culture period in the hypoxic condition. There was steady production of lactic acid in both conditions. Medium changes decreased the lactic acid levels, and enabled the concentrations to be kept below inhibitory levels of 35 mm (93). The overall lactic acid production (see Figure 6.6B) mirrored the overall glucose consumption described above, as expected in anaerobic metabolism. The overall amount of lactic acid produced, taking medium changes into account, was significantly higher in the hypoxic condition (see Figure 6.6B), since hypoxia encourages anaerobic respiration. This is also due to the longer lag phase in normoxia, which caused lower cell 137

154 densities early on in the culture period. However, the lactic acid production rate (on a per cell basis), over the exponential growth period, in hypoxia was the same as in normoxia (see Table 6.1). The lactic acid production rate was higher than those reported in Chapter 4 (see Tables 4.2 and 4.3) and literature values (see Table 6.1). As an alternative to aerobic respiration, cells can take an anaerobic pathway, in which one molecule of glucose results in two molecules of lactic acid. A higher yield of lactic acid suggests that the metabolic state of the cells in both conditions was anaerobic. This was expected since cartilage exists in an environment with low oxygen levels of encourages anaerobic respiration in chondrocytes, a cell type that was present in the aggregates. The yield of lactic acid over glucose was greater than two in both conditions (see Table 6.1), which is consistent with the results in Chapter 4. Yields greater than 2.0, and as high as 2.69, have also been reported in literature for MSCs in both 2D and 3D culture systems (95). In this case, the yield of lactic acid over glucose was 2.38 in the normoxic condition, but the R 2 value was quite low, being The R 2 value was low due to a single data point, which increased the value of the yield above 2.0. The predominant metabolic pathway in both conditions was anaerobic respiration, similar to literature values (see Table 6.1). The overall amounts of glutamine consumed and ammonia produced were significantly higher in the hypoxic condition (see Figure 6.7A-B), although the trends looked very similar. The overall glutamine consumption mimicked the overall ammonia production (see Figure 6.7A-B), which was expected, since glutamine was required for ammonia production. Feeding maintained the ammonia concentrations below the inhibitory level of 2.4 mm (93), as demonstrated in Chapter 4. The glutamine and ammonia uptake rates were the same in both conditions (see Table 6.1). These values were slightly higher than values from other experiments (see Chapter 4), due to the different donors used. The yield of ammonia over glutamine was higher in normoxia and similar to values in Chapter 4. Thus, glutamine metabolism was important in both conditions. 138

155 The cells in the hypoxic condition were able to reach maximum cell densities earlier than the normoxic condition. In addition, hypoxia up-regulated a chondrogenic phenotype, increased GAG/DNA production and yielded healthy-looking aggregates. For these reasons, hypoxia was incorporated into the bioreactor co-culture protocol. 139

156 Figure 6.5: Gene expression and Safranin O staining of co-cultured aggregates is shown in normoxia and hypoxia after 16 days in culture. The gene expression relative to β-actin was different in the conditions. Expression of A) collagen X, aggrecan, COMP, B) collagen I, and collagen II is shown. Error bars show range of data. An asterisk (*) indicates a statistically significant difference between the conditions. Safranin O staining of co-cultured aggregates cultured in C) normoxic and D) hypoxic conditions is shown. Scale bars represent 50 µm. Data was collected by Aillette Mulet-Sierra. 140

157 Figure 6.6: Cumulative A) glucose consumption and B) lactic acid production are shown in normoxic and hypoxic conditions. Error bars show range of data of duplicate samples from duplicate cultures. 141

158 Figure 6.7: Cumulative A) glutamine consumption and B) ammonia production are shown in normoxic and hypoxic conditions. Error bars show range of data of duplicate samples from duplicate cultures. 142

159 Table 6.1: Kinetic parameters for hmsc and hac bioreactor co-cultures in normoxic and hypoxic conditions. Literature values are shown for bovine ACs (92), hmscs (80,93,94) and murine MSCs (95) in static culture only under normoxia, because values for static co-culture, bioreactor culture and hypoxic conditions are not available. Parameter Normoxia Hypoxia Normoxia (Static Culture) Value R 2 Value R 2 Literature Values (Reference Number) Maximum Viable Cell Density (cells/ml) 71,000 N/A 63,000 N/A N/A Maximum Apparent Growth Rate (h -1 ) N/A N/A 0.02 (93), (80) Doubling Time (h) 55 N/A 62 N/A (80), 35.5 (93), (94) Glucose Uptake Rate (pmol/cell day) (80), (94), 9.19 (93) Glutamine Uptake Rate (pmol/cell day) (93), (80) Lactic Acid Production Rate (pmol/cell day) (80), (94) Ammonia Production Rate (pmol/cell day) Yield of Lactic Acid over Glucose (mol/mol) (93), 1.98 (92), (80), (95) Yield of Ammonia over Glutamine (mol/mol) (93) 143

160 6.4. Comparison of Bioreactor Co-culture to Static Co-culture In Chapter 4, comparisons between bioreactor co-culture and static co-culture were made under normoxic conditions. However, the newly developed bioreactor protocol included hypoxia. Thus, a comparison between the static and bioreactor protocols was required under hypoxia (3% O2). The apparent growth rate in the exponential growth phase was significantly higher in static co-culture (days 0-6) than bioreactor co-culture (days 4-12) (see Figure 6.8A). These results were consistent with the results in Chapter 4. The GAG/DNA of the aggregates generated in bioreactor co-culture was more than double that of the stratified monolayer generated in static co-culture (see Figure 6.8B), similar to results in Chapter 4. The bioreactor protocol displayed a trend towards higher collagen II and aggrecan than the static protocol, whereas the static protocol suggested higher collagen I and COMP (see Figure 6.8C-D). In addition, the collagen I to collagen II gene expression ratio was 1.9 in bioreactor co-culture on day 16 (vs. 560 in static co-culture at day 10). Collagen X had similar expression in both conditions. Thus, the bioreactor co-culture seemed to encourage expression of chondrogenic genes. This result is expected, due to the 3D co-culture as aggregates in bioreactors. 3D culture conditions have been to encourage a chondrogenic phenotype in literature (6,11,20 23). Even though higher cell densities can be reached in static co-culture, bioreactor co-culture encouraged chondrogenic traits. The bioreactor-generated aggregates produced more matrix as evidenced by GAG/DNA. In addition, the bioreactor protocol could efficiently generate co-cultured cells on a large-scale, since bioreactors are more scalable. Furthermore, it yielded three-dimensional aggregates of cells and self-generated matrix. These aggregates had cell-matrix bonds, as they were attached to matrix. They may also have had cell-to-cell connections. These bonds help to transmit biomechanical signals generated by forces, such as fluid shear stress, compression and hydrostatic pressure, from one part of the tissue to another. These important signals can then influence cell fate and behaviour. In this way, cell bonds can be functionally important in their niche (1,101). 144

161 Moreover, the cells generated in bioreactors were conditioned to an environment with shear forces. Thus, the co-culture of hacs and hmscs should be carried out in bioreactors as aggregates. 145

162 Figure 6.8: Bioreactor co-culture was compared to static co-culture in duplicate. Both conditions were inoculated in a 1 hac to 3 hmsc ratio with 20,000 cells/ml in medium under hypoxia (3% O2). A) The apparent growth rate in the exponential growth phase (days 4-12 in bioreactors and days 0-6 in static culture) is shown for both conditions. Error bars show range of data from duplicate cultures. B) GAG/DNA of the bioreactor-generated aggregates and the static monolayer are shown after 10 days in culture. Error bar shows standard error of the mean. Gene expression of co-cultured cells was determined in static culture after 10 days in culture and in bioreactor culture after 16 days in culture. The gene expression relative to β-actin is shown for C) collagen I, collagen X, aggrecan, COMP, and D) collagen II. Error bars show range of data from duplicate cultures. GAG/DNA and gene expression data was collected by Aillette Mulet-Sierra. 146

163 6.5. Effect of Different Donors on Bioreactor Co-culture The effect of donor-to-donor variability was tested in bioreactor co-culture. Two pairs of hac and hmsc donors were tested under hypoxic conditions. The oxygen tension was kept at 3% throughout isolation, pre-culture and bioreactor co-culture for all four donors. Donor pair A was composed of BM142 (44 year old male) and hac119 (29 year old female), and donor pair B was composed of cells from BM143 (44 year old female) and hac120 (50 year old male). After a two day lag phase in both conditions, the cells grew exponentially until day 12 for donor pair A and until day 10 for donor pair B (see Figure 6.9A). The cell densities of the two donor pairs were similar until day 10 (see Figure 6.9A). The aggregates also looked healthy in both cases, displaying well-defined boundaries, which indicated that the cells with the secreted matrix were adhered strongly within the aggregates (see Figure 6.10A- B). They looked similar to healthy aggregates shown previously in Chapters 4 and 5. However, the cell densities in donor pair B dropped rapidly after day 10, due to the onset of cell death. Thus, the cell densities at the end of the culture period were significantly different and the aggregates in donor pair B were smaller and fewer than in donor pair A (see Figure 6.10C-D). The apparent growth rates in the exponential growth phase of the two donor pairs (days 4-12 in donor pair A and days 4-10 in donor pair B) were not significantly different (see Figure 6.9B). The apparent growth rates of the bioreactor co-culture protocol were similar across experiments, regardless of the culture conditions and donors. The apparent growth rates in bioreactor co-culture in this experiment was 0.010/h using BM142 and hac119 and 0.012/h using BM143 and hac120, which are similar to values in Chapter 4 (0.009/h using BM142 and hac120; 0.007/h using BM119 and hac79) and Chapter 5 ( /h using BM119 and hac79). 147

164 Figure 6.9: Donor-to-donor variability was tested in bioreactor co-culture. Bioreactors were inoculated in a 1 hac to 3 hmsc ratio with 20,000 cells/ml in 125 ml PPRF-msc6 under hypoxia (3% O2). Bioreactors were operated at 60 rpm for the first two days and 80 rpm for the remainder of the culture period. A) Cell densities and viabilities using donor pairs A and B are shown. Error bars show range of data of duplicate cell counts from duplicate cultures. B) The apparent growth rate in the exponential growth phase (days 4-12 in donor pair A and days 4-10 in donor pair B) are shown. Error bars show range of data from duplicate cultures. C) GAG/DNA of the aggregates in the two different donor pairs is shown. Error bars show standard error of the mean. GAG/DNA data was collected by Aillette Mulet-Sierra. 148

165 Figure 6.10: Aggregate morphology in the two donor pairs are shown. Photomicrographs of co-cultured cells in bioreactor culture under hypoxia are shown at 10x magnification for two donor pairs after A-B) 10 days and C-D) 16 days in culture. Scale bars represent 100 µm. Despite the early drop in cell densities in donor pair B, the GAG/DNA was similar in both donors at the end of culture (see Figure 6.9C). Additionally, the GAG/DNA did not change from day 10 to day 16 in both conditions (data not shown). The Safranin O staining did not indicate chondrogenic differentiation for both donor pairs (see Figure 6.11A-B). Collagen I was significantly lower in donor pair A than B. Collagen X expression appeared lower and COMP expression appeared higher in donor pair A than B, whereas aggrecan and collagen II expression was similar in both conditions (see Figure 6.11C-D). Also, the gene expression ratio of collagen I to II was 2 in donor pair A (vs. 135 in donor pair B). Thus, donor pair A had greater chondrogenic expression than donor pair B. 149

166 Figure 6.11: Safranin O staining and gene expression analysis was performed in two sets of donors in bioreactor co-culture under hypoxia. Safranin O staining of cocultured aggregates cultured in hypoxic conditions using A) donor pair A and B) donor pair B is shown after 10 days in culture. Scale bars represent 50 µm. The gene expression, relative to β-actin, in the two donor pairs is shown for A) collagen I, collagen X, aggrecan, COMP, and B) collagen II after 16 days in culture. Error bars show range of data from duplicate cultures. An asterisk (*) indicates a statistically significant difference between the conditions. Data was collected by Aillette Mulet-Sierra. There is substantial donor-to-donor variability in hmscs. In particular, the proliferative capacity of MSCs has been shown to decrease with age in several studies (90,121). hmscs derived from younger, female donors proliferate at a higher rate (122). There are large 150

167 inter-donor variations in the hypertrophic expression and chondrogenic differentiation of MSCs under hypoxia (49). Note that the chondrogenic potential of cells derived from even a single MSC clonal subpopulation (from one donor) has substantial variation (only 2 out of 14 clones were shown to have chondrogenic potential) (123). It has been shown that there is no correlation between the quality of chondrogenic differentiation of hmscs, and the age and gender of the hmsc donor (121,122). However, contradictory results have also been reported of reduced multi-lineage differentiation potential due to MSC aging (88,89). This contradiction may be due to different culture techniques employed in the isolation and culture of MSCs (90). In this study, the two hmsc donors had the same age and were cultured identically. Thus, differences due to age and culture techniques were not expected, but there may be differences due to the gender of the donor. It is also important to account for differences in the proliferative potential of hacs due to donor-to-donor variability. Barbero et al. have shown that there are large variations in the redifferentiation capacity of clonal populations of hacs (only 7 out of 20 clones were shown to be able to redifferentiate) (124). It has also been shown that younger AC donors result in higher cell yields. They also result in higher GAG/DNA, when cultured with TGFβ1, FGF-2 and PDGF-BB (77). In contrast, another study demonstrated that while chondrocytes are a heterogeneous population, age has no effect on the proliferation rate (14). The lower cell densities in donor pair B could be explained by the age difference between the chondrocyte donors. However, the MSC donor was likely responsible for the decrease in cell densities in donor pair B. BM143 did not grow well in culture. Under hypoxic conditions, there was some cell expansion using this donor. However, when this MSC donor was isolated and cultured under normoxia, the cells died. BM142 grew well in culture under both normoxia and hypoxia. Moreover, BM142 displayed higher cell growth than BM143 in the hypoxic condition, with a doubling time of 1.1 vs. 1.7 days at the end of passage five. Thus, BM143 was inferior in terms of cell growth. 151

168 Donor-to-donor variability can impact cell densities and cell phenotype to a large degree. Thus, the selection of donors for experiments can affect the development and evaluation of a bioprocess. For example, the developed bioprocess could have had a shorter culture time of 10 days, instead of 16 days, if donor pair B was used (under these culture conditions). The clinical implication of this study is that the cell densities and chondrogenic phenotype obtained could vary from patient to patient. Thus, the aggregates from some donors may not be sufficient in quantity and quality for a clinical therapy Conclusions Oxygen tension throughout the culture period had a substantial impact on culture endpoints in both static and bioreactor co-culture. The isolation, static pre-culture and co-culture of hacs and hmscs should be carried out in low oxygen tension (3%) for a shorter lag phase, enhanced chondrogenic phenotype and healthier aggregate culture. Bioreactor co-culture has many advantages over static co-culture for the large-scale production of cells that are capable of cartilage repair. Specifically, bioreactors are scalable vessels that can be used to generate large quantities of aggregates of hacs and hmscs. It is important to note that donor-to-donor variability had an effect on cell densities and phenotype, and thus had an impact on the bioreactor protocol. The bioreactor protocol can be scaled up to larger vessels for cells from donors, which are anticipated as having lower proliferation potential. 152

169 CHAPTER 7: CONCLUSIONS AND RECOMMENDATIONS 7.1. Conclusions The body of work in this thesis details the development and evaluation of a bioprocess that can co-expand hmscs and hacs in suspension bioreactors under serum-free conditions. The bioprocess was able to support the generation of both cells and matrix as aggregates. The three main specific aims of this thesis were addressed as described below. SA 1: Screening of serum-free media The production of chondrogenic cells under serum-free conditions represents an important step towards the clinical translation of a cell-based cartilage repair therapy. SFM were tested to support the co-expansion of hmscs and hacs. PPRF-msc6 consistently supported the greatest cell proliferation during serial passaging. It outperformed a commercially-available SFM and SCM, which is commonly used by researchers. It was clearly demonstrated that a SFM, PPRF-msc6, can be utilized to co-expand populations of hmscs and hacs. This finding can facilitate the clinical translation of cell-based therapies involving in vitro-expanded hmscs and hacs. SA 2: Development of a bioreactor protocol The agitation rate during the growth phase had a substantial impact on cell densities and aggregate characteristics. In addition, a reduced agitation rate during the inoculation phase also increased cell densities. Serial passaging of the cells resulted in aggregate formation, but no cell growth. However, feeding extended the period of co-culture and doubled the cell yields. Oxygen tension had an important effect on bioreactor co-culture, as the length of the lag phase was reduced in hypoxic conditions. There was also greater matrix production and chondrogenic phenotype under hypoxia. These studies demonstrate that many end-points of this bioprocess can be improved by examining key culture conditions. 153

170 This work can help to inform the scale-up of other combinations of cell types that may have utility in cartilage repair strategies. SA 3: Evaluation of the bioreactor protocol The co-culture of hacs and hmscs was conducted on a large-scale in suspension bioreactors as aggregates. Large amounts of matrix (3.3 to 7.5 µg GAG/µg DNA) was produced in this bioprocess on a large scale. Large numbers of co-cultured cells were also generated, around 63,000 to 95,000 cells/ml in 125 ml spinner flasks. This represents a 3.2 to 4.8 cell fold expansion in under 16 days. Each bioreactor produced around 1 million cells, capable of filling a 1 cm 2 defect (9). The bioreactor co-culture protocol had lower cell yields (3-fold expansion in bioreactors and 10-fold expansion in T-flasks), but had a greater GAG/DNA (3.9 vs. 1.5 µg GAG/µg DNA) than the corresponding static co-culture protocol. In addition, the bioreactor protocol was able to generate 3D aggregates on a large-scale, as opposed to 2D monolayers on a small-scale. Additionally, this bioprocess is considerably more scalable than static culture flasks, and represents a means of cell production in which there can be a much higher degree of scale-up for greater production if required. The effect of donor sets on the bioreactor co-culture was significant with regard to the cell density. To conclude, this work provides a proof-of-concept that demonstrates that hmsc and hac co-expansion can be carried out as aggregates in suspension bioreactors. The co-culture was carried out under serum-free conditions, facilitating potential clinical translation. This represents a novel contribution to the field. It also sets up a knowledge base on which future optimization studies can be carried out to improve the performance of this bioprocess Recommendations There are some modifications that may increase the reproducibility, productivity and product quality of this bioprocess. Importantly, further processing of the aggregates should be carried out to enhance the chondrogenic phenotype of the cells before application in 154

171 cartilage repair strategies. In addition, the bioprocess can be scaled-up to a larger bioreactor system. Optionally, future studies should be carried out to better define the inoculation conditions in an effort to increase the cell productivity. Also, the dynamic cell behaviour in the co-cultured aggregates should be investigated further to understand and possibly improve the aggregate microenvironment. The bioprocess can be scaled-up from the 125 ml spinner flasks employed here to larger, computer-controlled bioreactors. The larger volume would allow for the production of larger amounts of cells and aggregates. In addition, the larger vessel size allows the addition of sensors, thereby permitting computer control and monitoring of culture parameters. The greater control of culture conditions could lead to more reproducible cell expansion. Optionally, adjusted continuous feeding could be carried out in bioreactors, which may help to increase cell densities. Inoculation conditions in this bioprocess can be improved as well. The cell inoculation density and the inoculation ratio of cell types should be tested for its effect on co-culture endpoints such as cell growth, aggregate characteristics and GAG production. In addition, aggregates can be generated using AggreWell technology, in which cells are forced to aggregate by centrifugation into specialized trays. Inoculation of these AggreWellgenerated aggregates, as opposed to single cells, could decrease the lag phase. Aggregates formed using AggreWell plates are also more uniform in size and so the use of this technology would result in a more uniform inoculum and possibly more uniform aggregate sizes later in culture. The cell behaviour in the co-culture should be studied in more detail. Several studies have determined that MSCs promote ACs proliferation in co-culture (17,36), but these findings should be confirmed in the bioreactor co-culture protocol described here. The number of hacs and hmscs in culture at different time points should be quantified using GFP labelling (17), to understand which cell types are proliferating or dying in the co-culture. In addition, cell localization within the aggregates, aggregate coalescence and the possible 155

172 fusion of the cells in co-culture should be studied over the culture period using long-term dyes in contrasting colors for each cell type (102,110). These proposed studies can give important insights about the dynamic behaviour of populations of hacs and hmscs in this suspension co-culture system. These insights may eventually lead to modifications to improve the efficacy of aggregates as part of a cartilage repair strategy. The co-cultured aggregates should be processed further to have utility in a cartilage repair therapy. Further chondrogenic development of the aggregates should be carried out prior to testing in pre-clinical trials. Chondrogenic factors such as: FGF-2, TGF-β1 (114), TGFβ3 (39) and platelet-derived growth factor (PDGF) (113), can be used to induce brute-force chondrogenic differentiation, increasing GAG/DNA and collagen II production. Subsequently, the aggregates can be seeded into collagen-based scaffolds (125), combined to form a single large tissue-engineered construct (116,126), or injected into a defect as a paste of aggregates through a syringe (73). The co-cultured aggregates, produced in the bioprocess described in this thesis, are versatile and may have utility in a range of different cartilage repair strategies. 156

173 CHAPTER 8: REFERENCES 1. Huselstein, C., Li, Y., and He, X. Mesenchymal stem cells for cartilage engineering. Biomed. Mater. Eng. 22(1-3), 69, Gao, J., Yao, J.Q., and Caplan, A.I. Stem cells for tissue engineering of articular cartilage. J. Eng. Med. 221(5), 441, Ochi, M., Adachi, N., Nobuto, H., Yanada, S., Ito, Y., and Agung, M. Articular cartilage repair using tissue engineering technique-novel approach with minimally invasive procedure. Artif. Organs. 28(1), 28, The Arthritis Society. Osteoarthritis [Internet]. Available from: Schulz, R.M., and Bader, A. Cartilage tissue engineering and bioreactor systems for the cultivation and stimulation of chondrocytes. Eur. Biophys. J. 36(4-5), 539, Vinatier, C., Mrugala, D., Jorgensen, C., Guicheux, J., and Noël, D. Cartilage engineering: a crucial combination of cells, biomaterials and biofactors. Trends Biotechnol. 27(5), 307, Schumann, D., Kujat, R., Nerlich, M., and Angele, P. Mechanobiological conditioning of stem cells for cartilage tissue engineering. Biomed. Mater. Eng. 16(4 Suppl), S37, Dhinsa, B., and Adesida, A.B. Current clinical therapies for cartilage repair, their limitation and the role of stem cells. Curr. Stem Cell Res. Ther. 7(2), 143, Tallheden, T., Bengtsson, C., Brantsing, C., Sjögren-Jansson, E., Carlsson, L., Peterson, L., Brittberg, M., and Lindahl, A. Proliferation and differentiation potential of chondrocytes from osteoarthritic patients. Arthritis Res. Ther. 7(3), R560, Brittberg, M., Lindahl, A., Nilsson, A., Ohlsson, C., Isaksson, O., and Peterson, L. Treatment of deep cartilage defects in the knee with autologous chondrocyte transplantation. N. Engl. J. Med. 331(14), 889, Csaki, C., Schneider, P., and Shakibaei, M. Mesenchymal stem cells as a potential pool for cartilage tissue engineering. Ann. Anat. 190(5), 395,

174 12. Saliken, D.J., Mulet-Sierra, A., Jomha, N.M., and Adesida, A.B. Decreased hypertrophic differentiation accompanies enhanced matrix formation in co-cultures of outer meniscus cells with bone marrow mesenchymal stromal cells. Arthritis Res. Ther. 14(3), R153, Studer, D., Millan, C., Oeztuerk, E., Maniura-Weber, K., and Zenobi-Wong, M. Molecular and biophysical mechanisms regulating hypertrophic differentiation in chondrocytes and mesenchymal stem cells. Eur. Cells Mater. 24, 118, Yang, K.G.A., Saris, D.B.F., Geuze, R.E., van Rijen, M.H.P., van der Helm, Y.J.M., Verbout, A.J., Creemers, L.B. and Dhert, W.J.A. Altered in vitro chondrogenic properties of chondrocytes harvested from unaffected cartilage in osteoarthritic joints. Osteoarthr. Cartil. 14(6), 561, Richardson, S.M., Hoyland, J.A., Mobasheri, R., Csaki, C., Shakibaei, M., and Mobasheri, A. Mesenchymal stem cells in regenerative medicine: opportunities and challenges for articular cartilage and intervertebral disc tissue engineering. J. Cell. Physiol. 222(1), 23, Pelttari, K., Steck, E., and Richter, W. The use of mesenchymal stem cells for chondrogenesis. Injury. 39(Suppl 1), S58, Acharya, C., Adesida, A., Zajac, P., Mumme, M., Riesle, J., Martin, I., and Barbero, A. Enhanced chondrocyte proliferation and mesenchymal stromal cells chondrogenesis in coculture pellets mediate improved cartilage formation. J. Cell. Physiol. 227(1), 88, Hardingham, T., Tew, S., and Murdoch, A. Tissue engineering: chondrocytes and cartilage. Arthritis Res. 4(Suppl 3), S63, Bonaventure, J., Kadhom, N., Cohen-Solal, L., Ng, K.H., Bourguignon, J., Lasselin, C., and Freisinger, P. Reexpression of cartilage-specific genes by dedifferentiated human articular chondrocytes cultured in alginate beads. Exp. Cell Res. 212(1), 97, Johnstone, B., Hering, T.M., Caplan, A.I., Goldberg, V.M., and Yoo, J.U. In vitro chondrogenesis of bone marrow-derived mesenchymal progenitor cells. Exp. Cell Res. 238(1), 265,

175 21. Malda, J., van Blitterswijk, C.A., Grojec, M., Martens, D.E., Tramper, J., and Riesle, J. Expansion of bovine chondrocytes on microcarriers enhances redifferentiation. Tissue Eng. 9(5), 939, Gigout, A., Buschmann, M.D., and Jolicoeur, M. Chondrocytes cultured in stirred suspension with serum-free medium containing pluronic-68 aggregate and proliferate while maintaining their differentiated phenotype. Tissue Eng. 15(8), 2237, Frith, J.E., Thomson, B., and Genever, P.G. Dynamic three-dimensional culture methods enhance mesenchymal stem cell properties and increase therapeutic potential. Tissue Eng. 16(4), 735, Cooke, M.E., Allon, A.A., Cheng, T., Kuo, A.C., Kim, H.T., Vail, T.P., Marcucio, R.S., Schneider, R.A., Lotz, J.C., and Alliston, T. Structured three-dimensional co-culture of mesenchymal stem cells with chondrocytes promotes chondrogenic differentiation without hypertrophy. Osteoarthr. Cartil. 19(10), 1210, Meretoja, V.V., Dahlin, R.L., Wright, S., Kasper, F.K., and Mikos, A.G. Articular chondrocyte redifferentiation in 3D co-cultures with mesenchymal stem cells. Tissue Eng. 20(6), 514, Fischer, J., Dickhut, A., Rickert, M., and Richter, W. Human articular chondrocytes secrete parathyroid hormone-related protein and inhibit hypertrophy of mesenchymal stem cells in coculture during chondrogenesis. Arthritis Rheum. 62(9), 2696, Eustics, C. Osteoarthritis: Explained With Pictures [Internet]. About.com Osteoarthritis. Available from: Chiang, H., and Jiang, C.C. Repair of articular cartilage defects: review and perspectives. J. Formos. Med. Assoc. 108(2), 87, Ozbey, O., Sahin, Z., Acar, N., Ozcelik, F.T., Ozenci, A.M., Koksoy, S., and Ustunel, I. Characterization of colony-forming cells in adult human articular cartilage. Acta Histochem. 116(5), 763, Williams, R., Khan, I.M., Richardson, K., Nelson, L., McCarthy, H.E., Analbelsi, T., Singhrao, S.K., Dowthwaite, G.P., Jones, R.E., Baird, D.M., Lewis, H., Roberts, S., Shaw, H.M., Dudhia, J., Fairclough, J., Briggs, T., and Archer, C.W. Identification and 159

176 clonal characterisation of a progenitor cell sub-population in normal human articular cartilage. PLoS One. 5(10), e13246, LeBaron, R.G., and Athanasiou, K.A. Ex vivo synthesis of articular cartilage. Biomaterials. 21(24), 2575, Dewan, A.K., Gibson, M.A., Elisseeff, J.H., and Trice, M.E. Evolution of autologous chondrocyte repair and comparison to other cartilage repair techniques. Biomed Res. Int. 2014, , Jakob, M., Démarteau, O., Schäfer, D., Stumm, M., Heberer, M., and Martin, I. Enzymatic digestion of adult human articular cartilage yields a small fraction of the total available cells. Connect. Tissue Res. 44(3-4), 173, Danisovic, L., Varga, I., Zamborsky, R., and Böhmer, D. The tissue engineering of articular cartilage: cells, scaffolds and stimulating factors. Exp. Biol. Med. 237(1), 10, Harvanova, D., Tothova, T., Sarissky, M., Amrichova, J., and Rosocha, J. Isolation and characterization of synovial mesenchymal stem cells. Folia Biol. 57(3), 119, Hubka, K.M., Dahlin, R.L., Meretoja, V.V., Kasper, F.K., and Mikos, A.G. Enhancing chondrogenic phenotype for cartilage tissue engineering: monoculture and coculture of articular chondrocytes and mesenchymal stem cells. Tissue Eng. Part B Rev Sabatino, M.A., Santoro, R., Gueven, S., Jaquiery, C., Wendt, D.J., Martin, I., Moretti, M., and Barbero, A. Cartilage graft engineering by co-culturing primary human articular chondrocytes with human bone marrow stromal cells. J. Tissue Eng. Regen. Med Wu, L., Prins, H.J., Helder, M.N., van Blitterswijk, C.A., and Karperien, M. Trophic effects of mesenchymal stem cells in chondrocyte co-cultures are independent of culture conditions and cell sources. Tissue Eng. Part A. 18(15-16), 1542, Dahlin, R.L., Ni, M., Meretoja, V.V., Kasper, F.K., and Mikos, A.G. TGF-β3- induced chondrogenesis in co-cultures of chondrocytes and mesenchymal stem cells on biodegradable scaffolds. Biomaterials. 35(1), 123,

177 40. Wu, L., Leijten, J.C., Georgi, N., Post, J.N., van Blitterswijk, C.A., and Karperien, M. Trophic effects of mesenchymal stem cells increase chondrocyte proliferation and matrix formation. Tissue Eng. Part A. 17(9-10), 1425, Dahlin, R.L., Meretoja, V.V., Ni, M., Kasper, F.K., and Mikos, A.G. Chondrogenic phenotype of articular chondrocytes in monoculture and co-culture with mesenchymal stem cells in flow perfusion. Tissue Eng. Part A Dahlin, R.L., Kinard, L.A., Lam, J., Needham, C.J., Lu, S., Kasper, F.K., and Mikos, A.G. Articular chondrocytes and mesenchymal stem cells seeded on biodegradable scaffolds for the repair of cartilage in a rat osteochondral defect model. Biomaterials. 35(26), 7460, Meretoja, V.V., Dahlin, R.L., Wright, S., Kasper, F.K., and Mikos, A.G. The effect of hypoxia on the chondrogenic differentiation of co-cultured articular chondrocytes and mesenchymal stem cells in scaffolds. Biomaterials. 34(17), 4266, Meretoja, V.V., Dahlin, R.L., Kasper, F.K., and Mikos, A.G. Enhanced chondrogenesis in co-cultures with articular chondrocytes and mesenchymal stem cells. Biomaterials. 33(27), 6362, Zuo, Q., Cui, W., Liu, F., Wang, Q., Chen, Z., and Fan, W. Co-cultivated mesenchymal stem cells support chondrocytic differentiation of articular chondrocytes. Int. Orthop. 37(4), 747, Babur, B.K., Ghanavi, P., Levett, P., Lott, W.B., Klein, T., Cooper-White, J.J., Crawford, R., and Doran, M.R. The interplay between chondrocyte redifferentiation pellet size and oxygen concentration. PLoS One. 8(3), e58865, Schrobback, K., Klein, T.J., Crawford, R., Upton, Z., Malda, J., and Leavesley, D.I. Effects of oxygen and culture system on in vitro propagation and redifferentiation of osteoarthritic human articular chondrocytes. Cell Tissue Res. 347(3), 649, Buckley, C.T., Vinardell, T., and Kelly, D.J. Oxygen tension differentially regulates the functional properties of cartilaginous tissues engineered from infrapatellar fat pad derived MSCs and articular chondrocytes. Osteoarthr. Cartil. 18(10), 1345,

178 49. Gawlitta, D., van Rijen, M.H., Schrijver, E.J., Alblas, J., and Dhert, W.J. Hypoxia impedes hypertrophic chondrogenesis of human multipotent stromal cells. Tissue Eng. Part A. 18(19-20), 1957, Müller, J., Benz, K., Ahlers, M., Gaissmaier, C., and Mollenhauer, J. Hypoxic conditions during expansion culture prime human mesenchymal stromal precursor cells for chondrogenic differentiation in three-dimensional cultures. Cell Transplant. 20(10), 1589, Markway, B.D., Tan, G.K., Brooke, G., Hudson, J.E., Cooper-White, J.J., and Doran, M.R. Enhanced chondrogenic differentiation of human bone marrow-derived mesenchymal stem cells in low oxygen environment micropellet cultures. Cell Transplant. 19(1), 29, Adesida, A.B., Mulet-Sierra, A., and Jomha, N.M. Hypoxia mediated isolation and expansion enhances the chondrogenic capacity of bone marrow mesenchymal stromal cells. Stem Cell Res. Ther. 3(2), 9, Haque, N., Rahman, M.T., Abu Kasim, N.H., and Alabsi, A.M. Hypoxic culture conditions as a solution for mesenchymal stem cell based regenerative therapy. Sci. World J. 2013, , Abdollahi, H., Harris, L.J., Zhang, P., McIlhenny, S., Srinivas, V., Tulenko, T., and DiMuzio, P.J. The role of hypoxia in stem cell differentiation and therapeutics. J. Surg. Res. 165(1), 112, Lee, H.H., Chang, C.C., Shieh, M.J., Wang, J.P., Chen, Y.T., Young, T.H., and Hung, S.C. Hypoxia enhances chondrogenesis and prevents terminal differentiation through PI3K/Akt/FoxO dependent anti-apoptotic effect. Sci. Rep. 3, 2683, Jung, S., Panchalingam, K.M., Rosenberg, L., and Behie, L.A. Ex vivo expansion of human mesenchymal stem cells in defined serum-free media. Stem Cells Int. 2012, , Schrobback, K., Klein, T.J., Schuetz, M., Upton, Z., Leavesley, D.I., and Malda, J. Adult human articular chondrocytes in a microcarrier-based culture system: expansion and redifferentiation. J. Orthop. Res. 29(4), 539,

179 58. Yuan, Y., Kallos, M.S., Hunter, C., and Sen, A. Improved expansion of human bone marrow-derived mesenchymal stem cells in microcarrier-based suspension culture. J. Tissue Eng. Regen. Med. 8(3), 210, Jorgenson, K. Development of suspension culture protocols to expand populations of adult human stem cells from synovial fluid [M.Sc. thesis]. Department of Chemical and Petroleum Engineering, University of Calgary, Calgary, AB, Canada, Dry, H. Development of culture protocols for the production of synovium-derived precursor cells [M.Sc. thesis]. Department of Biomedical Engineering, University of Calgary, Calgary, AB, Canada, Eibes, G., dos Santos, F.G., Andrade, P.Z., Boura, J.S., Abecasis, M.M., da Silva, C.L., and Cabral, J.M. Maximizing the ex vivo expansion of human mesenchymal stem cells using a microcarrier-based stirred culture system. J. Biotechnol. 146(4), 194, Shafa, M., Sjonnesen, K., Yamashita, A., Liu, S., Michalak, M., Kallos, M.S., and Rancourt, D.E. Expansion and long-term maintenance of induced pluripotent stem cells in stirred suspension bioreactors. J. Tissue Eng. Regen. Med. 6(6), 462, Alfred, R., Gareau, T., Krawetz, R., Rancourt, D., and Kallos, M.S. Serum-free scaled up expansion and differentiation of murine embryonic stem cells to osteoblasts in suspension bioreactors. Biotechnol. Bioeng. 106(5), 829, Cormier, J.T., zur Nieden, N.I., Rancourt, D.E., and Kallos, M.S. Expansion of undifferentiated murine embryonic stem cells as aggregates in suspension culture bioreactors. Tissue Eng. 12(11), 3233, Sen, A., Kallos, M.S., and Behie, L.A. Effects of hydrodynamics on cultures of mammalian neural stem cell aggregates in suspension bioreactors. Ind. Eng. Chem. Res. 40(23), 5350, King, J.A., and Miller, W.M. Bioreactor development for stem cell expansion and controlled differentiation. Curr. Opin. Chem. Biol. 11(4), 394, Sen, A., Kallos, M.S., and Behie, L.A. New tissue dissociation protocol for scaledup production of neural stem cells in suspension bioreactors. Tissue Eng. 10(5-6), 904,

180 68. Ferrari, C., Balandras, F., Guedon, E., Olmos, E., Chevalot, I., and Marc, A. Limiting cell aggregation during mesenchymal stem cell expansion on microcarriers. Biotechnol. Lett. 28(3), 780, Sart, S., Tsai, A.C., Li, Y., and Ma, T. Three-dimensional aggregates of mesenchymal stem cells: cellular mechanisms, biological properties, and applications. Tissue Eng. Part B Rev. 20(5), 365, Gigout, A., Jolicoeur, M., Nelea, M., Raynal, N., Farndale, R., and Buschmann, M.D. Chondrocyte aggregation in suspension culture is GFOGER-GPP- and beta1 integrin-dependent. J. Biol. Chem. 283(46), 31522, Bueno, E.M., Bilgen, B., Carrier, R.L., and Barabino, G.A. Increased rate of chondrocyte aggregation in a wavy-walled bioreactor. Biotechnol. Bioeng. 88(6), 767, Vunjak-Novakovic, G., Obradovic, B., Martin, I., Bursac, P.M., Langer, R., and Freed, L.E. Dynamic cell seeding of polymer scaffolds for cartilage tissue engineering. Biotechnol. Prog. 14(2), 193, Lee, T.J., Bhang, S.H., La, W.G., Yang, H.S., Seong, J.Y., Lee, H., Im, G.I., Lee, S.H., and Kim, B.S. Spinner-flask culture induces redifferentiation of de-differentiated chondrocytes. Biotechnol. Lett. 33(4), 829, Abbasalizadeh, S., Larijani, M.R., Samadian, A., and Baharvand, H. Bioprocess development for mass production of size-controlled human pluripotent stem cell aggregates in stirred suspension bioreactors. Tissue Eng. Part C Methods. 18(11), 831, Barbero, A., Grogan, S.P., Mainil-Varlet, P., and Martin, I. Expansion on specific substrates regulates the phenotype and differentiation capacity of human articular chondrocytes. J. Cell. Biochem. 98(5), 1140, Dominici, M., Le Blanc, K., Mueller, I., Slaper-Cortenbach, I., Marini, F.C., Krause, D.S., Deans, R.J., Keating, A., Prockop, D.J., and Horwitz, E.M. Minimal criteria for defining multipotent mesenchymal stromal cells. The International Society for Cellular Therapy position statement. Cytotherapy. 8(4), 315,

181 77. Barbero, A., Grogan, S., Schäfer, D., Heberer, M., Mainil-Varlet, P., and Martin, I. Age related changes in human articular chondrocyte yield, proliferation and postexpansion chondrogenic capacity. Osteoarthr. Cartil. 12(6), 476, Jung, S., Sen, A., Rosenberg, L., and Behie, L.A. Human mesenchymal stem cell culture: rapid and efficient isolation and expansion in a defined serum free medium. J. Tissue Eng. Regen. Med. 6(5), 391, Jung, S., Sen, A., Rosenberg, L., and Behie, L.A. Identification of growth and attachment factors for the serum-free isolation and expansion of human mesenchymal stromal cells. Cytotherapy. 12(5), 637, Higuera, G., Schop, D., Janssen, F., van Dijkhuizen-Radersma, R., van Boxtel, T., and van Blitterswijk, C.A. Quantifying in vitro growth and metabolism kinetics of human mesenchymal stem cells using a mathematical model. Tissue Eng. 15(9), 2653, Sart, S., Agathos, S.N., and Li, Y. Process engineering of stem cell metabolism for large scale expansion and differentiation in bioreactors. Biochem. Eng. J. 84, 74, Sen, A. The development of media for the in vitro expansion of mammalian neural stem cells [M.Sc. thesis]. Department of Chemical and Petroleum Engineering, University of Calgary, Calgary, AB, Canada, Livak, K.J., and Schmittgen, T.D. Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) Method. Methods. 25(4), 402, Adesida, A.B., Grady, L.M., Khan, W.S., and Hardingham, T.E. The matrixforming phenotype of cultured human meniscus cells is enhanced after culture with fibroblast growth factor 2 and is further stimulated by hypoxia. Arthritis Res. Ther. 8(3), R61, Murdoch, A.D., Grady, L.M., Ablett, M.P., Katopodi, T., Meadows, R.S., and Hardingham, T.E. Chondrogenic differentiation of human bone marrow stem cells in transwell cultures: generation of scaffold-free cartilage. Stem Cells. 25(11), 2786, Thirumala, S., Goebel, W.S., and Woods, E.J. Manufacturing and banking of mesenchymal stem cells. Expert Opin. Biol. Ther. 13(5), 673,

182 87. Pittenger, M.F., Mckay, A.M., Beck, S.C., Jaiswal, R.K., Douglas, R., Mosca, J.D., Moorman, M.A., Simonetti, D.W., Craig, S., and Marshak, D. R. Multilineage potential of adult human mesenchymal stem cells. Science. 284(5411), 143, Banfi, A., Bianchi, G., Notaro, R., Luzzatto, L., Cancedda, R., and Quarto, R. Replicative aging and gene expression in long-term cultures of human bone marrow stromal cells. Tissue Eng. 8(6), 901, Banfi, A., Muraglia, A., Dozin, B., Mastrogiacomo, M., Cancedda, R., and Quarto, R. Proliferation kinetics and differentiation potential of ex vivo expanded human bone marrow stromal cells: implications for their use in cell therapy. Exp. Hematol. 28(6), 707, Sethe, S., Scutt, A., and Stolzing, A. Aging of mesenchymal stem cells. Ageing Res. Rev. 5(1), 91, Ronga, M., Grassi, F.A., and Bulgheroni, P. Arthroscopic autologous chondrocyte implantation for the treatment of a chondral defect in the tibial plateau of the knee. Arthroscopy. 20(1), 79, Suits, J.M.T. Relationship between glucose metabolism, waste product accumulation and matrix synthesis of chondrocytes in high-density static and perfusion bioreactor culture systems. Department of Chemical Engineering, Queen s University, Kingston, ON, Canada, Schop, D., Janssen, F.W., van Rijn, L.D., Fernandes, H., Bloem, R.M., de Bruijn, J.D., and van Dijkhuizen-Radersma, R. Growth, metabolism, and growth inhibitors of mesenchymal stem cells. Tissue Eng Part A. 15(8), 1877, Pattappa, G., Heywood, H.K., de Bruijn, J.D., and Lee, D.A. The metabolism of human mesenchymal stem cells during proliferation and differentiation. J. Cell. Physiol. 226(10), 2562, Deorosan, B., and Nauman, E.A. The role of glucose, serum, and three-dimensional cell culture on the metabolism of bone marrow-derived mesenchymal stem cells. Stem Cells Int. 2011, ,

183 96. Malda, J., van den Brink, P., Meeuwse, P., Grojec, M., Martens, D.E., Tramper, J., Riesle, J., and van Blitterswijk, C.A. Effect of oxygen tension on adult articular chondrocytes in microcarrier bioreactor culture. Tissue Eng. 10(7-8), 987, Frauenschuh, S., Reichmann, E., Ibold, Y., Goetz, P.M., Sittinger, M., and Ringe, J. A microcarrier-based cultivation system for expansion of primary mesenchymal stem cells. Biotechnol. Prog. 23(1), 187, Saleh, F.A., and Genever, P.G. Turning round: multipotent stromal cells, a threedimensional revolution? Cytotherapy. 13(8), 903, Zhang, H., Dai, S., Bi, J., and Liu, K.K. Biomimetic three-dimensional microenvironment for controlling stem cell fate. Interface Focus. 1(5), 792, Lund, A.W., Yener, B., Stegemann, J.P., and Plopper, G.E. The natural and engineered 3D microenvironment as a regulatory cue during stem cell fate determination. Tissue Eng. 15(3), 371, Chen, G., Lv, Y., Guo, P., Lin, C., Zhang, X., Yang, L., and Xu, Z. Matrix mechanics and fluid shear stress control stem cells fate in three dimensional microenvironment. Curr. Stem Cell Res. Ther. 8(4), 313, Niu, C.C., Yuan, L.J., Lin, S.S., Chen, L.H., and Chen, W.J. Mesenchymal stem cell and nucleus pulposus cell coculture modulates cell profile. Clin. Orthop. Relat. Res. 467(12), 3263, Xu, L., Wang, Q., Xu, F., Ye, Z., Zhou, Y., and Tan, W.S. Mesenchymal stem cells downregulate articular chondrocyte differentiation in noncontact coculture systems: implications in cartilage tissue regeneration. Stem Cells Dev. 22(11), 1657, Bekkers, J.E., Tsuchida, A.I., van Rijen, M.H., Vonk, L.A., Dhert, W.J., Creemers, L.B., and Saris, D.B. Single-stage cell-based cartilage regeneration using a combination of chondrons and mesenchymal stromal cells: comparison with microfracture. Am. J. Sports Med. 41(9), 2158, Giovannini, S., Diaz-Romero, J., Aigner, T., Heini, P., Mainil-Varlet, P., and Nesic, D. Micromass co-culture of human articular chondrocytes and human bone marrow mesenchymal stem cells to investigate stable neocartilage tissue formation in vitro. Eur. Cells Mater. 20, 245,

184 106. Aung, A., Gupta, G., Majid, G., and Varghese, S. Osteoarthritic chondrocytesecreted morphogens induce chondrogenic differentiation of human mesenchymal stem cells. Arthritis Rheum. 63(1), 148, Sen, A., Kallos, M.S., and Behie, L.A. Expansion of mammalian neural stem cells in bioreactors: effect of power input and medium viscosity. Dev. Brain Res. 134(1-2), 103, Malda, J., and Frondoza, C.G. Microcarriers in the engineering of cartilage and bone. Trends Biotechnol. 24(7), 299, Shikani, A.H., Fink, D.J., Sohrabi, A., Phan, P., Polotsky, A., Hungerford, D.S., and Frondoza, C.G. Propogation of human nasal chondrocytes in microcarrier spinner culture. Am. J. Rhinol. 18(2), 105, Wang, T.Y., Sen, A., Behie, L.A., and Kallos, M.S. Dynamic behavior of cells within neurospheres in expanding populations of neural precursors. Brain Res. 1107(1), 82, Zscharnack, M., Poesel, C., Galle, J., and Bader, A. Low oxygen expansion improves chondrogenesis of ovine bone-marrow-derived mesenchymal stem cells in collagen type I hyrdrogel. Cells Tissues Organs. 190(2), 81, Colter, D., Sekiya, I., and Prockop, D.J. Identification of a subpopulation of rapidly self-renewing and multipotential adult stem cells in colonies of human marrow stromal cells. Proc. Natl. Acad. Sci. 98(14), 7841, Hendriks, J., Riesle, J., and van Blitterswijk, C.A. Effect of stratified culture compared to confluent culture in monolayer on proliferation and differentiation of human articular chondrocytes. Tissue Eng. 12(9), 2397, Jakob, M., Démarteau, O., Schäfer, D., Hintermann, B., Dick, W., Heberer, M., and Martin, I. Specific growth factors during the expansion and redifferentiation of adult human articular chondrocytes enhance chondrogenesis and cartilaginous tissue formation in vitro. J. Cell. Biochem. 81(2), 368, Sekiya, I., Koopman, P., Tsuji, K., Mertin, S., Harley, V., Yamada, Y., Shinomiya, K., Nifuji, A., and Noda, M. Dexamethasone enhances SOX9 expression in chondrocytes. J. Endocrinol. 169(3), 573,

185 116. Ando, W., Tateishi, K., Katakai, D., Hart, D.A., Higuchi, C., Nakata, K., Hashimoto, J., Fujie, H., Shino, K., Yoshikawa, H., and Nakamura, N. In vitro generation of a scaffold-free tissue-engineered construct (TEC) derived from human synovial mesenchymal stem cells: biological and mechanical properties and further chondrogenic potential. Tissue Eng Part A. 14(12), 2041, Hunt, M.M. Bioprocessing for expansion and standardization of embryonic stem cell cultures [Ph.D. thesis]. Department of Biomedical Engineering, University of Calgary, Calgary, AB, Canada, Tsai, A.C., Liu, Y., and Ma, T. Actin-mediated contractility in three-dimensional aggregates of human mesenchymal stem cells. Abstract presented at the 4th International Conference on Stem Cell Engineering, Coronado, CA, Stenhamre, H., Slynarski, K., Petrén, C., Tallheden, T., and Lindahl, A. Topographic variation in redifferentiation capacity of chondrocytes in the adult human knee joint. Osteoarthr. Cartil. 16(11), 1356, auw Yang, K.G., Saris, D.B.F., Geuze, R.E., van der Helm, Y.J.M., van Rijen, M.H.P., Verbout, A.J., Dhert, W.J.A., and Creemers, L.B. Impact of expansion and redifferentiation conditions on chondrogenic capacity of cultured chondrocytes. Tissue Eng. 12(9), 2435, Dexheimer, V., Mueller, S., Braatz, F., and Richter, W. Reduced reactivation from dormancy but maintained lineage choice of human mesenchymal stem cells with donor age. PLoS One. 6(8), e22980, Siegel, G., Kluba, T., Hermanutz-Klein, U., Bieback, K., Northoff, H., and Schäfer, R. Phenotype, donor age and gender affect function of human bone marrow-derived mesenchymal stromal cells. BMC Med. 11(146), Prins, H.J., Braat, A.K., Gawlitta, D., Dhert, W.J.A., Egan, D.A., Tijssen-Slump, E., Yuan, H., Coffer, P.J., Rozemuller, H., and Martens, A.C. In vitro induction of alkaline phosphatase levels predicts in vivo bone forming capacity of human bone marrow stromal cells. Stem Cell Res. 12(2), 428, Barbero, A., Ploegert, S., Heberer, M., Martin, I. Plasticity of clonal populations of dedifferentiated adult human articular chondrocytes. Arthritis Rheum. 48(5), 1315,

186 125. Kerker, J.T., Leo, A.J., and Sgaglione, N.A. Cartilage repair: synthetics and scaffolds: basic science, surgical techniques, and clinical outcomes. Sports Med. Arthrosc. 16(4), 208, Ando, W., Tateishi, K., Hart, D.A., Katakai, D., Tanaka, Y., Nakata, K., Hashimoto, J., Fujie, H., Shino, K., Yoshikawa, H., and Nakamura N. Cartilage repair using an in vitro generated scaffold-free tissue-engineered construct derived from porcine synovial mesenchymal stem cells. Biomaterials. 28(36), 5462, Math Open Reference. Eccentricity an ellipse [Internet]. Available from: BioMed Central Ltd. Access to articles [Internet]. Stem Cell Research and Therapy. Available from:

187 APPENDIX A: PREPARATION OF PPRF-MSC6 The instructions for making PPRF-msc6 have been published previously (79). The following instructions add supplementary detail to making the components of PPRF-msc6 and assembling the medium. A.1. Preparation of DMEM Stock Solution DMEM is a basal medium. A 5x stock solution was made for PPRF-msc6 as follows: ml of cell-culture grade water (Lonza, Cat. No Q) was measured into a graduated cylinder. 2. Half the water was poured into a beaker. 3. On a magnetic stir plate, two packets of DMEM (Life Technologies, Cat. No ) were emptied into a clean beaker with a magnetic stir bar. 4. Some of the water was used to wash out the packets. 5. The remaining water was added into the beaker. 6. The solution was stirred until the powder dissolved completely. 7. In a biosafety cabinet, a 0.22 μm bottle-top filter was used to filter the DMEM stock solution into a sterile 500 ml Pyrex glass bottle. 8. The stock solution was labelled and stored at 4 C in the dark for two months at the most. A.2. Preparation of F-12 Stock Solution F-12 is a basal medium. A 10x stock solution was made for PPRF-msc6 as follows: ml of cell-culture grade water (Lonza, Cat. No Q) was measured into a graduated cylinder. 2. Half the water was poured into a beaker. 171

188 3. On a magnetic stir plate, two packets of F-12 (Life Technologies, Cat. No ) were emptied into a clean beaker with a magnetic stir bar. 4. Some of the water was used to wash out the packets. 5. The remaining water was added into the beaker. 6. The solution was stirred until the powder dissolved completely. 7. In a biosafety cabinet, a 0.22 μm bottle-top filter was used to filter the F-12 stock solution into a sterile 250 ml Pyrex glass bottle. 8. The stock solution was labelled and stored at 4 C in the dark for two months at the most. A.3. Preparation of L-Glutamine Stock Solution L-Glutamine is an important amino acid for cells. The stock solution was aliquoted as follows: 1. The sterile 200 mm glutamine solution (Life Technologies, Cat. No ) was thawed to a liquid. 2. The solution was aliquoted into twenty-five 15 ml conical tubes with 4 ml solution each. 3. The tubes were labelled and stored at -20 C. A.4. Preparation of Lipid Concentrate Chemically defined lipids are important for cell metabolism and are required components in this serum-free medium. The stock solution was aliquoted as follows: 1. The sterile lipid concentrate (Life Technologies, Cat. No ) was mixed well to ensure homogeneity. 2. The solution was aliquoted into multiple 1.5 ml Eppendorf tubes with 1.7 ml solution each such that there was almost no air left in the tubes that could oxidize the fatty acids. 3. The tubes were labelled and stored at 4 C in the dark. 172

189 A.5. Preparation of Hydrocortisone Hydrocortisone was one of the components in PPRF-msc6. The stock solution was aliquoted as follows: 1. The sterile 50 µm hydrocortisone solution (Sigma, Cat. No. H6909) was thawed to a liquid. 2. The solution was aliquoted into ten 1.5 ml Eppendorf tubes with 1 ml solution each. 3. The tubes were labelled and stored in a box at -20 C. A.6. Human Serum Albumin Sterile 100 mg/ml (w/v) Human Serum Albumin (HSA) (In Vitro Care, Cat. No. 2101) was purchased in 5 ml glass bottles, and thus, did not need to be aliquoted. It was stored at 4 C. A.7. Preparation of Sodium Bicarbonate Stock Solution Sodium bicarbonate is a buffer that is commonly used in cell culture. A 7.5% stock solution was made for PPRF-msc6 as follows: ml of cell-culture grade water (Lonza, Cat. No Q) was measured into a graduated cylinder. 2. Some of the water was poured into a beaker g of sodium bicarbonate (Sigma, Cat. No. S5761) was weighed using weighing paper on a digital scale. 4. On a magnetic stir plate, the powder was emptied into a clean beaker with a magnetic stir bar. 5. Some of the water was used to wash out the weighing paper. 6. The remaining water was added into the beaker. 7. The solution was stirred until the powder dissolved completely. 173

190 8. In a biosafety cabinet, a 0.22 μm bottle-top filter was used to filter the sodium bicarbonate stock solution into a sterile 250 ml Pyrex glass bottle. 9. The solution was aliquoted into ten 15 ml conical tubes with 12 ml solution each. 10. The tubes were labelled and stored at -20 C. A.8. Preparation of HEPES Stock Solution HEPES is a buffer that is commonly used in cell culture. A 1.0 M stock solution was made for PPRF-msc6 as follows: g of sodium bicarbonate (Sigma, Cat. No. S5761) was weighed using weighing paper on a digital scale. 2. The powder was emptied into a 50 ml conical tube ml of cell-culture grade water (Lonza, Cat. No Q) was pipetted into the conical tube. 4. The contents of the tube were triturated until the HEPES powder had dissolved completely. 5. A 0.22 μm syringe filter was used to filter the HEPES stock solution into a sterile 50 ml conical tube. 6. The solution was aliquoted into ten 15 ml conical tubes with 2.5 ml solution each. 7. The tubes were labelled and stored at 4 C. A.9. Preparation of Human Recombinant Insulin Insulin is a hormone that is required in this serum-free medium. The 10 mg/ml insulin stock solution was made in 0.1 N HCl in cell-culture grade water ml of 12 N HCl was added to 5.5 ml of cell-culture grade water (Lonza, Cat. No Q) in a 15 ml conical tube to make 6 ml of 1 N HCl in water (12x dilution). 2. A 0.22 μm syringe filter was used to filter the 1 N HCl solution into a sterile 15 ml conical tube. 174

191 ml of the sterile 1 N HCl solution was added to 22.5 ml of cell-culture grade in a 50 ml conical tube to make 25 ml of 0.1 N HCl in water (10x dilution). 4. The sterile 1 N HCl solution was stored at room temperature in the acids cabinet. 5. The sterile 0.1 N HCl was used to dissolve 250 mg of insulin (Sigma, Cat. No C). The contents of the tube were triturated until the powder had dissolved completely. 6. The solution was transferred from its original bottle into a sterile 50 ml conical tube. 7. A 0.22 μm syringe filter was used to filter the insulin stock solution into a sterile 50 ml conical tube. 8. The solution was aliquoted into twenty-one 1.5 ml Eppendorf tubes with 1.15 ml solution each. 9. The tubes were labelled and stored at 4 C. A.10. Preparation of Human Apo-Transferrin Stock Solution Apo-transferrin is a blood plasma glycoprotein that is required in this serum-free medium. The 10 mg/ml apo-transferrin stock solution was made in cell-culture grade water ml cell-culture grade water (Lonza, Cat. No Q) was used to dissolve 200 mg of apo-transferrin (Sigma, Cat. No. T2252) and transfer it from its original bottles into a sterile 50 ml conical tube. 2. The contents of the tube were triturated until the powder had dissolved completely. 3. A 0.22 μm syringe filter was used to filter the apo-transferrin stock solution into a sterile 50 ml conical tube. 4. The solution was aliquoted into fifteen 1.5 ml Eppendorf tubes with 1.25 ml solution each. 5. The tubes were labelled and stored at -20 C. 175

192 A.11. Preparation of Putrescine Stock Solution Putrescine is a growth factor required in PPRF-msc6. The 10 mg/ml putrescine stock solution was made in cell-culture grade water mg of putrescine (Sigma, Cat. No. P7505) was weighed using weighing paper on a digital scale. 2. The powder was emptied into a sterile 15 ml conical tube ml cell-culture grade water (Lonza, Cat. No Q) was added to 60 mg of putrescine in the conical tube. 4. The contents of the tube were triturated until the powder had dissolved completely. 5. A 0.22 μm syringe filter was used to filter the putrescine stock solution into a sterile 15 ml conical tube. 6. The solution was aliquoted into twelve 1.5 ml Eppendorf tubes with 0.5 ml solution each. 7. The tubes were labelled and stored at -20 C. A.12. Preparation of Progesterone Stock Solution Progesterone is a hormone that is required in this serum-free medium. The 20 µg/ml progesterone stock solution was made in cell-culture grade water ml of 95% ethanol added to the 1 mg of progesterone (Sigma, Cat. No. P6149) in its original bottle. 2. The ethanol was swirled in the bottle to dissolve the powder ml of cell-culture grade water (Lonza, Cat. No Q) was added to the original bottle and used to dissolve the progesterone. 4. The contents of the bottle were triturated until the powder had dissolved completely. 5. A 0.22 μm syringe filter was used to filter the progesterone stock solution into a sterile 50 ml conical tube. 176

193 6. The solution was aliquoted into fifty 1.5 ml Eppendorf tubes with 1 ml solution each. 7. The tubes were labelled and stored at -20 C. A.13. Preparation of Human Recombinant bfgf Stock Solution Human recombinant basic Fibroblast Growth Factor (bfgf) is an important growth factor in PPRF-msc6. The 20 µg/ml bfgf stock solution was made in a 1.0 mg/ml HSA in PBS µl of 100 mg/ml (w/v) HSA was added to 9.9 ml of Phosphate-Buffered Saline (PBS) (Life Technologies, Cat. No ) in a 15 ml conical tube to make 1.0 mg/ml HSA in PBS (100x dilution). 2. A 0.22 μm syringe filter was used to filter the HSA solution into a sterile 15 ml conical tube µl of sterile HSA in PBS was used to dissolve 250 µg of bfgf (R&D Systems, Cat. No. 233-FB) in its original vial. This was repeated five times. 4. The final 250 µl of sterile HSA in PBS was added for a final volume of 1.25 ml and the solution was carefully triturated to avoid spillage. 5. The solution was aliquoted into twenty-three 0.65 ml Eppendorf tubes with 50 µl solution each. 6. The Eppendorf tubes were labelled and stored at -80 C. 7. The sterile HSA solution in PBS was stored at 4 C. A.14. Preparation of Human Recombinant TGF-β1 Stock Solution Human recombinant Transforming Growth Factor-beta1 (TGF-β1) is an important growth factor in PPRF-msc6. The 10 µg/ml TGF-β1 stock solution was made in a 1.0 mg/ml HSA and 4 mm HCl in cell-culture grade water µl of 1 M HCl was added to 600 µl of cell-culture grade water (Lonza, Cat. No Q) in a 15 ml conical tube to make 1.0 ml 0.4 M HCl in water (2.5x dilution). 177

194 µl of 0.4 M HCl in water was added to 900 µl of cell-culture grade water in a 15 ml conical tube to make 1.0 ml 40 mm HCl in water (10x dilution) ml of 40 mm HCl in water was added to 9 ml of cell-culture grade water in a 15 ml conical tube to make 10 ml 4 mm HCl in water (10x dilution) µl of 100 mg/ml (w/v) HSA was added to 9.9 ml of 4 mm HCl in water in a 15 ml conical tube to make 10 ml of 1.0 mg/ml HSA and 4 mm HCl in water (100x dilution). 5. A 0.22 μm syringe filter was used to filter the HSA and HCl solution into a sterile 15 ml conical tube µl of sterile HSA and HCl in water was used to dissolve the 2 µg in the original vial of TGF-β1 (R&D Systems, Cat. No. 240-B). 7. Another 100 µl of sterile HSA and HCl in water was added to the vial and triturated. This was repeated for each vial. 8. The solution was aliquoted into four 0.65 ml Eppendorf tubes for every vial of TGF-β1 with 50 µl solution each. 9. The Eppendorf tubes were labelled and stored at -20 C. 10. The sterile HSA and HCl solution in water was stored at 4 C. A.15. Preparation of PenStrep Although 10,000 units/ml Penicillin 10,000 µg/ml Streptomycin (PenStrep) was not in the original formulation of PPRF-msc6, antibiotics were added to the medium as a precaution. The PenStrep solution was aliquoted as follows: 1. The sterile PenStrep solution (Life Technologies, Cat. No ) was thawed to a liquid. 2. The solution was aliquoted into twenty 15 ml conical tubes with 5 ml solution each. 3. The tubes were labelled and stored at -20 C. 178

195 A.16. Preparation of Ascorbic Acid Stock Solution L-Ascorbic acid 2-phosphate (Vitamin C) was required in PPRF-msc6. However, it could only be added to the medium just prior to cell culture. A 50 mg/ml ascorbic acid stock solution was prepared by dissolving ascorbic acid powder in cell-culture grade water at a concentration of 57.2 mg/ml mg of ascorbic acid powder (Sigma, Cat. No. A8960) was weighed using weighing paper on a digital scale. 2. The powder was emptied into a sterile 50 ml conical tube ml cell-culture grade water (Lonza, Cat. No Q) was added to the ascorbic acid powder in the conical tube. 4. The contents of the tube were vortexed extensively until the powder had dissolved completely. Typically min of voxtexing was required to achieve homogeneity. 5. A 0.22 μm syringe filter was used to filter the ascorbic acid stock solution into a sterile 15 ml conical tube. 6. Some of the solution was aliquoted into several 1.5 ml Eppendorf tubes with µl solution for use in spinner flasks. 7. The remainder of the solution was aliquoted into fifty 0.65 ml Eppendorf tubes with 100 µl solution each for use in static culture. 8. The tubes were labelled and stored in a box at -20 C. A.17. Preparation of Gelatin Stock Solution Gelatin was used to coat static tissue culture flasks (T-flasks) to aid cell adhesion and growth. The 0.1% (0.1 g per 100 ml) gelatin working solution was reconstituted from powdered gelatin type b from bovine skin in cell-culture grade water and subsequently sterilized mg of gelatin pellets (Sigma, Cat. No. G6650) was weighed using weighing paper on a digital scale. 179

196 2. The gelatin pellets were emptied into a sterile 500 ml Pyrex glass bottle ml of cell-culture grade water (Lonza, Cat. No Q) was added to the bottle. The extra headspace was required so that the solution did not spill during steam sterilization in the autoclave. 4. The cap of the bottle was left partly open to relieve pressure during steam sterilization in the autoclave. Aluminum foil was used to cover the cap of the bottle. The bottle was sterilized in the autoclave at 120 C and 1.55 kg/cm3 for 30 minutes. 5. During autoclaving, the gelatin pellets dissolved in the water. 6. Following sterilization, the cap of the bottle was closed tightly and left at room temperature to cool. 7. After the gelatin solution had cooled to room temperature, it was stored at 4 C. A.18. Assembling PPRF-msc6 (-) The PPRF-msc6 (-) constitutes all the components in the formulation of PPRF-msc6 except ascorbic acid. The frozen components were thawed in a water bath at 37 C or at room temperature. Each of the components that were thawed from storage at -20 C could not be re-frozen or re-used. The only exceptions were bfgf and TGF-β1, which could be stored at 4 C for at most one month after being thawed. Table A.1 lists the quantities of all the components required to make 500 ml or 1.0 L of PPRF-msc6 (-). Ascorbic acid was added just prior to medium inoculation at 1 µg ascorbic acid/ml PPRF-msc6 (-). 1. All components were gathered in a biosafety cabinet. 2. Cell-culture grade water (Lonza, Cat. No Q) was measured with a graduated cylinder and added to a sterile Pyrex glass bottle. 3. All other components except fetuin and ascorbic acid were added aseptically to the bottle. 4. Bovine fetuin (Sigma, Cat. No. F3385, Lot. No. 071M7401V or SLBD2554V) was weighed using weighing paper on a digital scale such that 500 mg was deposited into each cup of weighing paper. 180

197 5. The fetuin was carefully deposited into the glass bottle with the aid of a weighing utensil. 6. The cap of the glass bottle was closed tightly and the bottle was agitated several times until the fetuin was completely dissolved. 7. In a biosafety cabinet, a 0.22 μm bottle-top filter was used to filter the growth medium into a new sterile Pyrex glass bottle with one filter per 500 ml medium. 8. The medium was labelled and stored at 4 C in the dark for one month at the most. Table A.1: Formulation of PPRF-msc6 (-). Component/Volume 1000 ml 500 ml DMEM 100 ml 50 ml F ml 25 ml L-Glutamine 7.5 ml 3.75 ml Lipid concentrate 1 ml 0.5 ml Hydrocortisone 2 ml 1 ml Human serum albumin 40 ml 20 ml Sodium bicarbonate 23 ml 11.5 ml HEPES 4.9 ml 2.45 ml Human recombinant insulin 2.3 ml 1.15 ml Human apo-transferrin 2.5 ml 1.25 ml Putrescine 0.9 ml 0.45 ml Progesterone ml 0.14 ml Human recombinant bfgf 0.1 ml 0.05 ml Human recombinant TGF-β1 0.1 ml 0.05 ml PenStrep 10 ml 5 ml Water ml ml Bovine fetuin 1.0 g 0.50 g 181

198 APPENDIX B: FACTORIAL EXPERIMENT RESULTS There was a factorial experiment described in this thesis (see Chapter 3.6.3). The results of the factorial experiment are tabulated in Table B.1. Table B.1: Results of Factorial Experiment Culture Vessel T-flask T-flask T-flask T-flask Bioreactor Bioreactor Bioreactor Bioreactor Oxygen Tension Normoxic Normoxic Hypoxic Hypoxic Normoxic Normoxic Hypoxic Hypoxic Donor Pairs Apparent Growth Rate (h -1 ) Lag Phase Length (days) GAG/DNA (µg/µg at day 10) Collagen I/II Expression Ratio hac120 + hac119 + hac119 + hac120 + hac120 + hac119 + BM142 BM142 BM142 BM143 BM142 BM142 hac119 + hac120 + BM142 BM E4 5.0E (day 10) (day 10) (day 10) (day 10) (day 16) (day 16) (day 16) (day 16) Details Chapter 4 Chapter 6 Chapter 6 N/A Chapter 4 Chapter 6 Chapter 6 Chapter 6 182

199 APPENDIX C: AGGREGATE ECCENTRICITY The eccentricity of an ellipse was calculated using Equation 3.4 (see Chapter 3). Representative ellipses of eccentricities of 0.0, 0.4 and 0.6 are shown in Figure C.1. Eccentricities between 0.4 and 0.6 were considered acceptable in this bioprocess. Figure C.1: Eccentricity of A) a circle, and B-C) ellipses. Modified from Math Open Reference (127). 183