Tissue Engineering of Multi-Zonal, Osteochondral-Like Constructs with Bone Marrow Stromal Cells

Whitaik David Lee

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Institute of Biomaterials and Biomedical Engineering University of Toronto

Whitaik David Lee

Doctor of Philosophy

Institute of Biomaterials and Biomedical Engineering University of Toronto

2015 Abstract

Articular cartilage facilitates weight bearing and movement in a synovial joint. As articular cartilage does not spontaneously heal, damaged cartilage leads to loss of cartilage and subchondral bone remodelling. In previous studies we developed biphasic constructs with articular chondrocytes and porous calcium polyphosphate (CPP) substrate that could replace both damaged cartilage and bone. However, articular chondrocytes are scarce and cannot be expanded in vitro without losing their functional attributes. Since bone marrow stromal cells (BMSCs) can be expanded to large quantities and differentiated into chondrocytes, it was hypothesized that, using BMSCs as the cell source, cartilage with a multi-zonal architecture can be formed on porous CPP substrate to produce osteochondral-like constructs with biomechanical properties that approach native articular cartilage. A protocol was devised to differentiate BMSCs on collagen-coated porous membranes and harvest the cells, which were subsequently cultured on porous CPP substrates to form biphasic constructs with high compressive mechanical properties.

Implantation of BMSC-derived biphasic constructs in sheep joints revealed insufficient mechanical integration between the cartilage and the substrate. To address this limitation, a strategy was devised to generate a zone of calcified cartilage at the interface, mimicking the

multi-zonal osteochondral architecture in healthy joints. A thin calcium hydroxyapatite film was formed on the surfaces of CPP using organic sol-gel processing method to limit the release of inorganic polyphosphates (polyP), which inhibit mineralization near the interface. Also, sequential seeding of BMSC-derived chondrocytes, in which only the interfacial layer was treated with triiodothyronine, yielded scaffold-free cartilage tissues on coated CPP substrates with distinct hyaline and calcified zones. The interfacial shear strength of constructs with the distinct calcified zone was significantly higher compared to non-calcified control. In vivo studies are warranted to investigate whether the multi-zonal constructs could effectively repair defects in a joint. Generation of anatomically shaped multi-zonal constructs for whole joint replacement are ongoing.

iii

Acknowledgments

In full gratitude and sincerity, I acknowledge and thank those who encouraged, inspired, supported, assisted and sacrificed themselves to enable my pursuit of the doctoral degree.

First, my thesis supervisors Dr. Rita Kandel and Dr. William Stanford: thank you for your leadership, guidance, support and patience. Under your watch I grew so much as a scholar, as a scientist and as a person. It was an honour and a privilege to be your trainee.

Dr. Bob Pilliar, thank you for your guidance and support on the sol-gel project. I am grateful for introducing me to the exciting field of biomaterials. Dr. Craig Simmons and Dr. Benjamin Alman, thank you for providing valuable feedback as members of my PhD committee through thick and thin. Dr. Mark Hurtig, Dr. Paul Zalzal, Dr. Ehsan Toyserkani and Dr. Mihaela Vlasea, thank you for the opportunity to work with you in collaboration.

I was blessed with great colleagues from both the Stanford Lab and the Kandel Lab. I enjoyed the company of everyone and I thank all of them for being helpful and supportive. In particular, thank you Tammy Reid and Dr. Lu Gan for your help in starting my project. Also, I thank Dr. Wing Chang, Dr. Kamal Garcha, Dr. Nicole Anderson, Dr. Paul Cassar, Dr. Jeffrey Kiernan and Sarah Kwon of the Stanford Lab and Dr. Amritha de Croos, Dr. Jean- Philippe St-Pierre, Dr. Nazish Ahmed, Dr. Rahul Gawri, Justin Parreno, Jonathan Iu and Vanessa Bianchi of the Kandel Lab for their valuable contributions, mentorship and friendship. I also acknowledge Steven Tong, Cheryl Cui and Nicole Tavares for their valuable contributions to the project as undergraduate students while putting up with me.

I was fortunate to receive so much encouragement and support from many mentors, friends and colleagues in the community. Every single bit was truly needed. Dr. Jim Friesen, Dr. Paul Santerre, Dr. Chul Bum Park and Dr. Cristina Amon, thank you for being my role models and generously providing your words of wisdom and support. Sandy Walker, Elizabeth Flannery, Carolyn Bryant and Jeffrey Little, thank you for all your help and support from the Institute. I also thank all those who made it possible to play music throughout my graduate studies, which provided a respite from research and help me carry on: Capt. Michael Lawson (ret.) and the 48th Highlanders of Canada military band; Capt. Graziano Brascacin (ret.), Lt.

Nick Arrigo and the 7th Toronto Regiment, Royal Canadian Artillery military band; J-M Erlendson and Countermeasure A Cappella; Geoffrey Siu, Alessandro Ariza and the Skule Orchestra; as well as John Edward Liddle and the Etobicoke Community Concert Band.

Finally, I thank my family in our mother tongue: 어머니, 아버지, 그리고 동생 예택, 항상 격려와 위로를 아끼지 않으시고 그 많은 희생을 묵묵히 감당하셔서 제가 긴 학위의 길을 마칠 수 있게 해 주신 그 큰 은혜는 말이나 글로 형용할 수 없습니다. 이 논문은 가족의 도움 없이는 저 혼자 힘으로 완성할 수 없었을 것입니다. 감사합니다 그리고 사랑합니다. And to my fiancée wife Esther Lau, thank you for your love, support and perseverance. I love you. I dedicate this accomplishment to you and my family.

Table of Contents

Abstract ...... ii

Acknowledgments ...... iv

Table of Contents ...... vi

List of Tables ...... xi

List of Figures ...... xii

Chapter 1 Introduction ...... 1 1.1 Articular Cartilage ...... 1 1.1.1 Structure and Function of Articular Cartilage ...... 1 1.1.2 Origin of Articular Cartilage: Articular vs. Growth Plate Cartilage ...... 4 1.1.3 Barriers to the Repair of Articular Cartilage ...... 7 1.1.4 Current Cartilage Repair Strategies ...... 9 1.2 Cells for Cartilage Repair ...... 12 1.2.1 Articular Chondrocytes ...... 12 1.2.2 Pluripotent Stem Cells ...... 12 1.2.3 Mesenchymal Stromal Cells and Bone Marrow Stromal Cells ...... 13 1.2.4 Chondrogenesis of Bone Marrow Stromal Cells ...... 15 1.3 Osteochondral-Like Constructs for Cartilage Repair ...... 18 1.3.1 A Case for Scaffold-Free Tissue Engineering of Cartilage ...... 18 1.3.2 Tissue Engineering of Osteochondral Constructs ...... 20 1.3.3 Cartilage Biomineralization ...... 22 1.3.4 Porous calcium polyphosphate (CPP) substrate and inorganic polyphosphate ...... 23 1.3.5 Effect of polyphosphates on biomineralization ...... 24 1.4 Conclusion ...... 25 1.5 Hypothesis ...... 26 1.6 Specific Aims ...... 26 1.7 References ...... 27

Chapter 2 Membrane Culture of Bone Marrow Stromal Cells Yields Better Tissue than Pellet Culture for Engineering Cartilage-Bone Substitute Biphasic Constructs in a Two-Step Process...... 50 2.1 Abstract ...... 50 2.2 Introduction ...... 51 2.3 Material and Methods ...... 53 2.3.1 Isolation and expansion of BMSCs ...... 53 2.3.2 Chondrogenic Pre-differentiation of BMSCs ...... 53 2.3.3 Cartilage-CPP biphasic construct culture ...... 54 2.3.4 Histological and immunohistological evaluation ...... 55 2.3.5 Biochemical analysis ...... 55 2.3.6 Gene expression analysis ...... 56 2.3.7 Stress relaxation assay for compressive modulus ...... 56 2.3.8 Statistical analysis ...... 56 2.4 Results ...... 57 2.4.1 Chondrogenic predifferentiation of BMSCs in membrane and pellet cultures ...... 57 2.4.2 Predifferentiated cells from membrane cultures form better cartilage tissue in biphasic constructs ...... 59 2.4.3 Optimization of biphasic construct tissue culture protocol ...... 61 2.5 Discussion ...... 64 2.6 Author Contributions ...... 66 2.7 Acknowledgements ...... 66 2.8 References ...... 66

Chapter 3 Sol gel-derived hydroxyapatite films over porous calcium polyphosphate substrates for improved tissue engineering of osteochondral-like constructs ...... 73 3.1 Abstract ...... 73 3.2 Introduction ...... 74 3.3 Materials and Methods ...... 76 3.3.1 Sol-gel processing of porous CPP disks ...... 76 3.3.2 Characterization of HAp coating on CPP disks ...... 76 3.3.3 Quantification of released polyP, calcium and phosphate by CPP disks ...... 77 3.3.4 Tissue culture of biphasic constructs ...... 78 3.3.5 Histological and radiological analysis of constructs ...... 78 3.3.6 Biochemical analysis of extracellular matrix and mineral accumulation ...... 78

vii

3.3.7 Quantification of inorganic polyphosphates accumulated in tissue ...... 79 3.3.8 Alkaline phosphatase (ALP) activity assay ...... 79 3.3.9 Interfacial shear testing ...... 79 3.3.10 Statistical analysis ...... 80 3.4 Results ...... 81 3.4.1 Characterization of sol-gel-derived coatings on porous CPP disks ...... 81 3.4.2 Characterization of polyP and mineral release from coated CPP substrates ...... 81 3.4.3 Characterization of cartilage tissue formed on coated CPP substrates ...... 84 3.4.4 HAp barrier coating on the CPP substrate affected polyP accumulation in cartilage ...... 86 3.4.5 Interfacial shear strength of cartilage-CPP substrate constructs ...... 87 3.5 Discussion ...... 88 3.6 Conclusion ...... 91 3.7 Author Contributions ...... 91 3.8 Acknowledgements ...... 91 3.9 References ...... 93

Chapter 4 Engineering of Hyaline Cartilage with a Calcified Zone Using Bone Marrow Stromal Cells ...... 97 4.1 Abstract ...... 97 4.2 Introduction ...... 98 4.3 Materials and Methods ...... 100 4.3.1 Chondrogenic predifferentiation of BMSCs in membrane cultures ...... 100 4.3.2 Preparation of porous calcium polyphosphate substrates with a hydroxyapatite coating ... 100 4.3.3 Optimizing the mineralizing culture condition for predifferentiated chondrocytes ...... 101 4.3.4 Tissue culture of multiphasic constructs ...... 101 4.3.5 Visualization and quantification of alkaline phosphate (ALP) activity ...... 101 4.3.6 Gene expression analysis ...... 102 4.3.7 Micro-computed tomography (µCT) imaging ...... 103 4.3.8 Histological analysis of whole constructs ...... 103 4.3.9 Histological analysis of cartilaginous tissues ...... 103 4.3.10 Biochemical analysis of extracellular matrix and mineral accumulation ...... 103 4.3.11 Mechanical testing of multiphasic constructs ...... 104 4.3.12 Statistical testing ...... 104 4.4 Results ...... 105

viii

4.4.1 Generation of calcified cartilage at the calcium polyphosphate interface with predifferentiated chondrocytes ...... 105 4.4.2 Short-term treatment of predifferentiated chondrocytes with T3 was sufficient to stimulate terminal differentiation ...... 107 4.4.3 T3-treated predifferentiated chondrocytes did not induce ALP activity in non-T3-treated predifferentiated chondrocytes ...... 107 4.4.4 Characterization of the multiphasic constructs ...... 109 4.5 Discussion ...... 113 4.6 Conclusion ...... 116 4.7 Author Contributions ...... 116 4.8 Acknowledgments ...... 116 4.9 Supplementary Material ...... 117 4.9.1 Isolation and expansion of sheep BMSCs using autologous serum ...... 117 4.9.2 Immunofluorescence of histological sections ...... 117 4.9.3 Analysis of interfacial shear strength ...... 117 4.9.4 List of primers used ...... 118 4.10 References ...... 118

Chapter 5 Discussion & Conclusion ...... 123 5.1 Discussion ...... 123 5.2 Future Experiments ...... 132 5.3 Conclusion ...... 134 5.4 References ...... 136

Appendix A In Vivo Validation of Biphasic Constructs for Repair of Focal Osteochondral Defects ...... 142 A.1 Development of FBS-free protocol for biphasic construct culture ...... 142 A.2 In vivo evaluation: repair of focal defects ...... 145 A.2.1 Materials and Methods ...... 145 A.2.2 Results ...... 145 A.2.3 Discussion ...... 147 A.3 Conclusion ...... 147 A.4 References ...... 148

Appendix B Development of a High-Sensitivity Method for Isolating and Quantifying Inorganic Polyphosphates from Cartilage Tissues Using Silica Spin Columns ...... 150

B.1 Abstract ...... 150 B.2 Introduction ...... 151 B.3 Materials and Methods ...... 153 B.3.1 Preparation of polyP standard solutions ...... 153 B.3.2 Silica spin column isolation and fluorometric quantification of polyP ...... 153 B.3.3 Determination of the polyP recovery ratios from silica spin columns ...... 154 B.3.4 Determination of potential sources of interference for the polyP quantification assay ...... 154 B.3.5 Elimination of DAPI fluorescence from DNA and RNA ...... 155 B.3.6 Elimination of DAPI fluorescence from chondroitin sulfate ...... 155 B.3.7 Proteinase K pre-treatment of samples ...... 156 B.3.8 PolyP quantification in in vitro-grown and native cartilage samples ...... 156 B.3.9 Statistical Analysis ...... 157 B.3.10 Safety Considerations ...... 157 B.4 Results ...... 158 B.4.1 Binding and recovery of polyP from silica spin columns ...... 158 B.4.2 Elimination of non-polyP interference for the polyP quantification assay ...... 160 B.4.3 Quantification of polyP added to in vitro-grown cartilage samples ...... 165 B.5 Discussion ...... 167 B.6 Author Contributions ...... 168 B.7 Acknowledgements ...... 168 B.8 References ...... 169 B.9 Additional Figures ...... 173 B.9.1 Development of the silica spin column protocol ...... 173 B.9.2 Optimizing the recovery ratios of different chain length polyP ...... 178 B.9.3 Nuclease protocol development ...... 184

List of Tables

Table 4.1: List of Primers for Chapter 4 ...... 118

Table B.1: Potential sources of extraneous 415 nm/558 nm DAPI fluorescence present in mammalian tissues, per 106 chondrocytes in 140µL volume...... 155

Table B.2: PolyP recovery ratios of silica spin column isolation method were higher than those reported in other studies and independent of initially loaded polyP quantities...... 159

List of Figures

Figure 2.1: The CPP disks...... 54

Figure 2.2: Histological appearance of tissue derived from membrane (left) and pellet cultures (right) of BMSCs cultured for 2 and 3 weeks in defined chondrogenic media...... 57

Figure 2.3: Accumulation of extracellular matrix in tissue derived from membrane and pellet cultures of BMSCs cultured for up to 3 weeks in defined chondrogenic media...... 58

Figure 2.4: Gene expression of chondrogenic markers by tissue derived from membrane and pellet cultures of BMSCs cultured for up to 3 weeks in defined chondrogenic media...... 59

Figure 2.5: BMSCs predifferentiated in membrane culture yielded better cartilage on CPP than those predifferentiated in pellet culture...... 60

Figure 2.6. BMSCs must be pre-differentiated for at least 2 weeks in membrane culture to form tissue on CPP after isolation...... 62

Figure 2.7. Optimal cartilage tissue was obtained by predifferentiating the BMSCs in membrane culture for 2 weeks and culturing the differentiated cells on CPP for 8 weeks...... 63

Figure 3.1: Porous CPP disks were coated using two different sol-gel thin film processing methods...... 82

Figure 3.2: Coating deposited by sol-gel thin film processing methods was HAp...... 83

Figure 3.3: In an aqueous environment, both OSG-derived and ISG-derived coating inhibited polyP release while releasing calcium and phosphate...... 83

Figure 3.4: Deep zone chondrocytes grown on OSG- and ISG-coated CPP substrates formed a ZCC in direct contact with the substrate...... 85

Figure 3.5: Sol-gel-derived coating did not affect the extracellular matrix accumulation of cartilage tissue grown on coated porous CPP substrates...... 85

xii

Figure 3.6: Net polyP accumulation (A) and alkaline phosphatase (ALP) activity (B) of tissues grown on coated and non-coated CPP substrates for up to 14 days...... 86

Figure 3.7: Constructs formed with OSG-coated CPP substrates withstood a greater interfacial shear load than those formed with ISG-coated or non-coated CPP substrates...... 87

Figure 4.1: Line diagram of the tissue culture protocol used for forming scaffold-free multiphasic osteochondral-like constructs...... 102

Figure 4.2: Predifferentiated chondrocytes formed cartilaginous tissues on porous calcium polyphosphate substrates with a zone of mineralized cartilage...... 106

Figure 4.3: Short term T3 treatment of predifferentiated chondrocytes was sufficient to stimulate terminal differentiation...... 108

Figure 4.4: Layer of predifferentiated chondrocytes treated with T3 did not induce the overlaid layer of non-treated predifferentiated chondrocytes to activate ALP activity...... 109

Figure 4.5. A two-step culture protocol of predifferentiated chondrocytes on porous calcium polyphosphate substrates produced cartilage tissue with a mineralized zone at the interface. ... 110

Figure 4.6: Accumulation of extracellular matrix in T3-treated (+T3) tissues was less compared to those in untreated (–T3) tissues, but comparable to native articular cartilage...... 111

Figure 4.7: T3-treated constructs (+T3) exhibited comparable compressive strength as native sheep osteochondral explants and stronger shear strength than untreated constructs (–T3)...... 112

Figure A.1: Blebbistatin treatment did not affect chondrogenic differentiation of BMSCs in membrane predifferentiation cultures...... 143

Figure A.2: Predifferentiated chondrocytes derived from FBS-free BMSCs gave rise to tissue on porous CPP substrates in serum-free media...... 144

Figure A.3: In a preliminary study, addition of L-proline in media increased extracellular matrix accumulation by 4 weeks of culture...... 144

xiii

Figure A.4: Gross and histological images of biphasic constructs after 3 month of implantation...... 146

Figure B.1: Quantity and chain length affected the polyP recovery ratios of the silica spin column protocol at nanomole and sub-nanomole quantities...... 159

Figure B.2: DNA, chondroitin sulfate and FBS affected the DAPI fluorescence signal of polyP isolated by using silica spin columns...... 161

Figure B.3: Both nuclease treatment and Tris were required to eliminate DAPI fluorescence signals produced by DNA and RNA...... 162

Figure B.4: DAPI fluorescence signal produced by residual chondroitin sulfate (CS) in post- isolation samples were reduced by the addition of Tris...... 163

Figure B.5: Proteinase K pre-treatment enhances polyP recovery...... 164

Figure B.6: Line diagram of the complete polyP quantification method...... 165

Figure B.7: Quantification of polyP in the in vitro-formed cartilage and native cartilage tissues with correction for recovery rate...... 166

Figure B.8: Initial silca spin column trial...... 173

Figure B.9: Optimization of buffers used in the silica spin columns...... 174

Figure B.10: PolyP was consistently bound and eluted from Epoch EconoSpin DNA Mini columns ...... 176

Figure B.11: Elution volume did not significantly affect the recovery ratio ...... 176

Figure B.12: Protocol optimization for minimizing the background signal ...... 177

Figure B.13: Minor differences in quantity were observed between polyP standards with different chain lengths ...... 178

xiv

Figure B.15: Short-chain polyP binding is enhanced by increasing the ethanol concentration at the binding step ...... 180

Figure B.16: Presence of abundant proteoglycans limited the maximum concentration of ethanol at 40%...... 181

Figure B.17: GuSCN concentration in the desalting buffer also affected the recovery ratios shorter chain length polyP...... 182

Figure B.18: Washing the silica spin columns before loading reduced the false positive signals in longer-chain polyP samples...... 183

Figure B.19: Incubation of polyP with calcium in 37ºC for even 15 minutes results in signal degradation...... 184

Figure B.20: Preliminary data on the use of DNase I...... 185

Figure B.21: Slope of the polyP standard curve is affected by the addition of DNase and EDTA...... 186

Figure B.22: Validation of nuclease treatments in polyP and DNA/RNA samples ...... 186

xv 1

Chapter 1 Introduction

1.1 Articular Cartilage

1.1.1 Structure and Function of Articular Cartilage

Articular cartilage lines the articulating surfaces of bones in a synovial joint. In humans, the thickness of articular cartilage ranges from 1 to 5 millimetres, depending on the particular joint as well as the locations within the joint [1]. The surface of articular cartilage is low in friction [2], which allows for painless and unhindered joint movement. In addition to reducing the friction, articular cartilage must bear weight and withstand shock from joint movement, for which a high compressive strength is required.

The principal constituent of articular cartilage is water, comprising up to 80% of its wet weight [3]. Articular cartilage is noted for its extracellular matrix content, which is primarily composed of collagens and proteoglycans (60% and 25% respectively [4]). Collagen type II is the principal collagen component in cartilage, accounting for 90% of the cartilage matrix collagen mass [5]. Collagen types IX and XI are also present throughout, while collagen type I, VI and X can also be found in specific locations within cartilage. The collagens form a tight meshwork to provide the tensile stiffness and strength of articular cartilage, and contribute to the cohesiveness of the tissue by mechanically entrapping the large proteoglycans [6].

A proteoglycan consists of a protein core and one or more glycosaminoglycan (GAG) chains, which are long, unbranched polysaccharide chains that consist of repeating disaccharides that contain an amino sugar. Each disaccharide unit has at least one negatively charged carboxylate or sulfate group, which causes the polysaccharide chains to repel one another and attracts cations and water. Aggrecan is the principal proteoglycan component in cartilage. It is composed of many chondroitin sulfate and keratan sulfate chains attached to the protein core. It binds non-covalently to hyaluronan and link protein to form proteoglycan aggregates, which do not dissociate nor displace under physiological conditions [6]. They provide the osmotic 2 resistance necessary for cartilage to resist compressive loads. Many other proteoglycans have been identified in cartilage and found to have various functions [7].

Articular cartilage is neither innervated nor vascularized. Gas and nutrition exchange in articular cartilage relies on diffusion and interstitial fluid flow from the synovial fluid [8]. Mechanical loading facilitates both diffusion and the fluid flow [9, 10]. The synovial fluid is an ultrafiltrate of serum to which hyaluronan and lubricin are secreted [11]. It occupies the synovial cavity in a joint, and functions to lubricate the articular cartilage, lowering its static and dynamic friction coefficients [12]. The synovium that surrounds the synovial fluid contains dense capillary networks that are particularly close to the surface in areas adjacent to cartilage [13]. These capillary networks supply the oxygen and nutrients to the synovial fluid [14].

Articular cartilage is sparsely populated with cells. Chondrocytes, the constituent cells of articular cartilage, comprise 10% of the tissue dry weight [5]. They are characterized by their round morphology, quiescence and high metabolic activity. Articular chondrocytes are the effectors of tissue homeostasis, as they secrete the extracellular matrix components and the enzymes that remodel the cartilage tissue in response to various cytokines and chemokines [15]. Articular chondrocytes sense mechanical forces via integrins and ion channels expressed in the primary cilia, non-motile organelles that project from cells [15]. Compression of articular cartilage causes deformation of cells and matrix, hydrostatic pressure gradients, electrical current via movement of ions and various other physiochemical changes in tissue [16]. Dynamic compression stimulates the anabolic response of chondrocytes, which is dependent on its amplitude and frequency [17, 18].

Surrounding each articular chondrocyte is the pericellular matrix, whose composition differs in its composition as compared to the rest of the tissue. The pericellular matrix uniquely contains collagen type VI [19], but also possesses a high concentration of proteoglycans, including aggrecan, hyaluronan, and decorin, as well as fibronectin, and types II and IX collagen [20]. This difference in composition, and its resultant biomechanical properties, significantly alters the stress and the strain experienced by the enclosed chondrocyte, acting as a mechanical transducer [21]. The biomechanical properties of the pericellular matrix were shown to be affected by osteoarthritis, a degenerative disease of articular cartilage [22]. 3

Articular cartilage possesses a highly ordered, depth-dependent architecture, which can be divided into four zones. These zones differ in the extracellular matrix composition, collagen fibre alignment and the biomechanical properties, as well as the morphology and gene expression profiles of chondrocytes. The superficial zone (SZ) occupies the top 10-20% of the articular cartilage, which forms the gliding surface of the joint. The collagen fibrils are aligned parallel to the joint surface and crosslinked more extensively compared to other zones, which confers its high tensile strength [23, 24]. Chondrocytes in the SZ have an elongated, flatter morphology and express lubricin, also known as PRG4, that confers the tribological properties of articular cartilage [25]. Damage in the SZ caused by loss and degradation of matrix components, as well as changes in collagen organization, precedes all other degenerative events in cartilage, suggesting the importance of SZ in the overall homeostasis of articular cartilage [26, 27]. The middle zone (also known as transitional zone) occupies the next 40-60% of the articular cartilage. Its extracellular matrix has a higher accumulation of proteoglycans compared to the SZ and collagen fibrils are not oriented parallel to the joint surface [28]. Chondrocytes in the middle zone have a more round morphology compared to those of the SZ. The deep zone (DZ; also known as the radial zone) occupies the innermost 25-35% of the articular cartilage. Its collagen fibrils are oriented orthogonally to the joint surface. DZ chondrocytes are spheroidal in morphology and arranged in a columnar fashion. They uniquely express alkaline phosphatase (ALP) and type X collagen [29]. The zone of calcified cartilage (ZCC) is about 100-200µm thick [30], situated at the cartilage-bone interface. The collagen fibrils continue from the DZ running orthogonally into the ZCC. Accumulation of mineral in the cartilaginous extracellular matrix stiffens the tissue, similar to how stiff a component of an engineered composite material increases the overall stiffness of the material [31]. While it was initially reported that the elastic modulus of the calcified cartilage is an order of magnitude less than that of subchondral bone [32], nanoindentation studies of calcified cartilage revealed a depth-dependent distribution with its deepest aspect matching that of the subchondral bone [31]. The ZCC provides both a mechanical transition from cartilage to bone, as well as fixation between the two tissues [30]. The DZ and ZCC are divided by a tidemark, a demarcation that is visible when stained with hematoxylin and eosin. The tidemark is believed to serve as a barrier to vascular invasion and calcification of the hyaline cartilage, but this is controversial [30]. 4

Articular cartilage is distinct from other types of cartilage in the body. Fibrocartilage is a tissue whose structural and functional properties are intermediate between those of dense fibrous connective tissue and hyaline cartilage [33]. Fibrocartilaginous structures are found in the bone- connective tissue interface, menisci of the knee and the annulus fibrosis of the intervertebral discs. Fibrocartilage also forms as a response to injury in the articular cartilage. Owing to the collagenous makeup that includes both types I and II collagens and differences in proteoglycans, fibrocartilage and hyaline cartilage have different biomechanical properties [34]. Growth plate cartilage is found in the limbs of skeletally immature body. Growth plate cartilage undergoes endochondral ossification, in which chondrocytes proliferate, become hypertrophic, mineralize the matrix and undergo apoptosis as blood vessels and osteoprogenitors invade to lay down bone [35]. While the ZCC in articular cartilage shares some of these characteristics such as mineralized matrix with growth plate cartilage, the healthy ZCC of the articular cartilage persists.

In conclusion, articular cartilage is an extracellular matrix-rich, poorly cellularized tissue that is distinct from other cartilaginous structures of the body. The complex and intricate architecture of articular cartilage confers its capacity for handling the compressive and shear stress present in the joint.

1.1.2 Origin of Articular Cartilage: Articular vs. Growth Plate Cartilage

In development, limbs first begin as primordial buds that consist of undifferentiated mesenchymal cells from the lateral plate and the somitic mesoderm, covered by an ectodermal layer [36]. The mesenchymal cells proliferate and the buds extend distally. The limb patterning is coordinated by a multitude of signalling pathways, soluble growth factors and gene expression: BMP [37], FGF [38] and hedgehog [35] pathways and Hox genes [39] all participate in limb patterning.

The mesenchymal cells condense and express the transcription factor Sox9, causing the mesenchymal cells to differentiate to chondrocytes [40] and form the cartilage anlagen. At this time, layers of compact, avascular mesenchymal tissues also emerge at prospective joint sites. Termed interzone, this layer is composed of flat-shaped cells that align perpendicular to the 5 length of the limb and bind to each other by gap junctions. The mechanism underlying interzone formation is a highly controversial topic. Conventionally, expression of type II collagen and Sox9 throughout the anlage occurs prior to the emergence of interzone supported the understanding that a single, continuous cartilage anlage forms first, and interzones emerge at specific prospective joint sites by an unknown mechanism as cells are transdifferentiated or replaced [40-43]. This hypothesis is challenged by the observation that the previous study did not distinguish between two splice forms of collagen type II. Type IIA collagen is expressed in many other tissues during development, and therefore its expression cannot identify the tissue as articular cartilage [41, 44]. Type IIB collagen is specific to articular cartilage, but direct localization of type IIB collagen has not yet been done as the type IIB-specific collagen antibody has only been recently characterized [45]. Further, as with type IIA collagen, Sox9 expression is not specific to cartilage either [46]. Therefore, others argue that the mesenchymal cells condense in discrete units that are separated by distinct, non-cartilaginous interzones [47].

Regardless of how the interzone is formed, it is clear that articular chondrocytes originate from the interzone, while non-articular chondrocytes originate from the cartilage anlagen [47, 48]. Interzone cells express GDF5 and Wnt9a [43]. Interzone cells do not express matrilin-1: it is expressed only in non-articular chondrocytes, and this is evident from E13.5 in mouse, which demonstrates that articular chondrocytes are distinct from the rest of the chondrocytes in early stages of joint formation [49]. Interzone cells initially do not express Col2a1, but a Col2a1- expressing cell population emerges along the prospective articulating sites. The joint cavity emerges by cavitation in the presence of mechanical stimulus, and interzone cells go on to give rise to synovial tissues and ligaments that persist to adulthood [43]. A recent report identified that Prg4- expressing cells located at the joint surface in the embryo serve as a progenitor population for all deeper layers of the mature articular cartilage [48].

In contrast to articular chondrocytes that persist into adulthood and establish permanent articular cartilage, non-articular chondrocytes are transient in nature: they form the long bones of the limbs by a process called endochondral ossification [35]. Endochondral ossification takes place in the epiphyseal plate, also known as growth plate, and the resident chondrocytes are termed growth plate chondrocytes. Growth plate chondrocytes proliferate as development progresses. At the centre of each anlage, cells stop proliferating, enlarge (i.e. hypertrophy) and start synthesizing type X collagen. In addition, hypertrophic chondrocytes express MMP-9 and 6

MMP-13, alkaline phosphatase, vascular endothelial growth factor (VEGF) and osteopontin. Then, the hypertrophic chondrocytes undergo apoptosis as the matrix is vascularized, enabling the osteoprogenitors to migrate and form calcified bone. This site, situated in the middle of the anlage, is termed the primary centre of ossification. Both the proliferating chondrocytes at the ends and the ossified center extend out appositionally, lengthening the long bones. Postnatally, secondary ossification centres appear at each end of the bone. The growth plate between the primary and the secondary ossification centres continues to proliferate, lengthening the bone. Bone growth ceases in adulthood when the epiphyseal plate calcifies.

Comparing the gene expression profiles of articular chondrocytes with growth plate chondrocytes, as well as other tissues taken from 6-week-old minipigs have revealed genes such as Thbs4 and Six1 that are unique to articular chondrocytes [50]. In another study, articular chondrocytes were found to have decreased Wnt signalling, with increased expressions of Wnt inhibitors Frizzled-related protein (FRP) and Dkk1 [51]. Endochondral ossification is regulated by a feedback loop between parathyroid hormone-related protein (PTHrP) and Indian hedgehog (Ihh) [52]. PTHrP is secreted by perichondral cells during the foetal stage and early proliferative chondrocytes, and is demonstrated to be an inhibitor of chondrocyte hypertrophy. On the other hand, Ihh is secreted by post-proliferative chondrocytes and early hypertrophic chondrocytes, and is a pro-hypertrophy factor. Other signalling pathways such as FGF, Wnt and TGF-β are involved in the regulation of endochondral ossification [36, 53]. In a diseased state such as osteoarthritis, articular chondrocytes become altered and can also express growth plate chondrocyte markers [54].

In summary, chondrocytes arise from condensed mesenchymal cells. Articular chondrocytes and growth plate chondrocytes are distinct in their developmental origins, role and cell fate. The articular chondrocytes are derived from interzone. They form the articular cartilage that persists into adulthood. The growth plate chondrocytes are derived from the condensed mesenchyme and are transient in nature, giving rise to the long bones via endochondral ossification. Insight on the development of articular cartilage provides clues on how stable, persistent cartilage capable of replacing damaged articular cartilage may be generated.

1.1.3 Barriers to the Repair of Articular Cartilage

Under normal conditions, chondrocytes in articular cartilage maintain a dynamic equilibrium state between the synthesis and degradation of extracellular matrix components, resulting in a strictly regulated matrix turnover. However, this balance is disturbed in a diseased state such as osteoarthritis, where upregulation of matrix-degrading enzymes, inhibition of matrix synthesis and excessive production of pro-inflammatory cytokines can be observed [4]. While many risk factors for osteoarthritis have been identified, its etiology is still not entirely elucidated [55].

The progressive loss of tissue starts with damage in the superficial zone (SZ). The SZ is exposed to perturbations such as mechanical insult and inflammatory cytokines. Indeed, loss of tensile strength, surface fibrillation and denaturation of type II collagen in the SZ are observed in early degenerated cartilage [26]. In a canine osteoarthritis model, SZ had decreased tensile strength and lower glysaminoglycan and collagen content as compared to healthy control [24]. Other studies have shown that collagen content remains unchanged [56] but the collagen fibrils were disorganized [57]. Further degenerative progression is characterized by deeper fibrillation of tissue past the SZ, loss of extracellular matrix and clustering of chondrocytes, as well as calcification of cartilage, chondrocyte hypertrophy, tidemark advancement, duplication and vascular penetration from the subchondral bone [15].

Other tissues in the joint can also impact cartilage damage and degeneration. Synovium plays a critical role in cartilage homeostasis by facilitating gas and nutrient exchange. Inflammation of synovium, driven by macrophages, plays a role in regulating both inflammatory and catabolic responses in osteoarthritis [58]. Many osteoarthritis patients exhibit various degrees of synovitis, whose macrophages exhibit an activated phenotype and produce both proinflammatory and angiogenic cytokines [59, 60]. Improper load-bearing on cartilage due to damage in ligaments, tendons and menisci as well as bone alignment have an impact on the rate of degeneration. Half of those patients who undergo anterior cruciate ligament reconstruction will develop symptomatic knee osteoarthritis 10-20 years later [61]. Varus knee (bow-leg) increases the risk of incident knee osteoarthritis [62]. Successful repair of the cartilage will therefore require systemic manipulation of both damaged cartilage itself and the joint.

Many tissues and organs respond to injury by invoking a wound healing process – hemostasis, inflammation, proliferation and remodelling. However, as articular cartilage is 8 avascular, the wound healing process cannot take place until the degeneration of tissue progresses so far that injury reaches the vascularized subchondral bone. The reparative response is the accumulation of fibrocartilage, which has inferior biomechanical properties compared to hyaline cartilage. This is in contrast to some animal models such as rabbits that are known to be capable of regenerating hyaline cartilage as a response to small focal cartilage defects less than 3mm in size [63].

Articular cartilage is profoundly affected by aging; for example, the biomechanical properties are altered during aging via changes in their proteoglycan content and collagen organization [24, 64]. Aged articular cartilage also has reduced capacity to respond to growth factors involved in cartilage homeostasis, due to decreased cell number and changes in cell signalling [65, 66]. Changes in TGF-β cell surface receptors favoured the aged chondrocytes’ catabolic response by increasing the ALK1 to ALK5 ratio, which results in an increased MMP- 13 expression [67]. Further, when chondrocytes undergo apoptosis, the lack of vascularization and tissue macrophages result in the accumulation of apoptotic bodies in the extracellular matrix, which can provide a nucleation point for ectopic mineralization [68]. Such accumulation of mineral in the cartilaginous matrix affects the mechanical properties of the tissue [69].

The limited number of resident articular chondrocytes implies that the remodelling and regeneration would take a long time. As previously noted, the number of cells also decreases with aging, making it difficult to replace in situ the lost chondrocytes to injury or disease [70]. Chondrocytes in their diseased state may proliferate, e.g. osteoarthritic chondrocytes, which is consistent with their display of growth plate chondrocyte characteristics [54]. A subpopulation of chondroprogenitors is found throughout the cartilage tissue [71], but they are concentrated in the superficial zone [72]. However, since the superficial zone is also the most prone to injury as well as loss due to aging and disease, the ability to manipulate these progenitors for in situ regeneration of chondrocytes is insufficient for clinically relevant cartilage repair [73].

The complex organization of the articular cartilage is also a barrier to cartilage repair. Each zone of articular cartilage has a specific mechanical role that, in concert, enables the articular cartilage’s function. The alignment of collagen fibres is important for the tensile strength of the SZ and the interfacial shear strength of the cartilage-subchondral bone interface: successful cartilage repair must be able to recreate this organization in order to regain function 9

[74]. Recapitulation of the articular cartilage’s organization is also critical for its self- maintenance, particularly as the zones of cartilage appear to influence each other. The SZ expresses many growth factors and receptors at their cell surface, indicating its role as a regulatory component [75, 76]. Aged/diseased ZCC expands and calcifies more densely over time [30], affecting the dynamics of the interaction between articular cartilage and subchondral bone [77].

1.1.4 Current Cartilage Repair Strategies

In adults, damaged cartilage does not spontaneously heal. The current clinical methods aim to repair the damaged tissues to prolong the use of the existing joint until the ongoing degeneration renders the replacement of the joint necessary. The methods to repair damaged cartilage can be divided into biological and non-biological approaches.

Biological approaches to cartilage repair are generally best suited for focal lesions. The marrow stimulation technique is based on the principle of penetrating the subchondral bone to induce the outflow of blood and marrow, and can take on the forms of abrasion, subchondral drilling and microfracture. This promotes the accumulation of fibrocartilage at the treatment site. These techniques are used for localized defects typically less than 3cm2 in size [78]. Clinical results after microfracture in the knee are age dependent: most favourable clinical results are seen in younger patients less than 40 years of age with isolated traumatic lesions [78]. The deterioration of the repaired cartilage begins by 18 months following the procedure and is significantly more pronounced in older patients [79]. A 10-14 year follow-up study found that, while the outcome scores of treated knees were significantly higher than the baseline scores, normal knee function was not achieved and most patients required further surgery [80]. Limitations of marrow stimulation include the inferior quality of the repair tissue, incomplete defect filling and new bone formation: in an animal model, defects such as residual unfilled gaps, cysts, and bony overgrowth were frequently observed after marrow stimulation [81].

Another biological approach is the harvest of healthy articular cartilage from non- damaged areas of the joint and transferred to the damaged site either as cells or intact tissue. Autologous chondrocyte implantation (ACI) is a surgical procedure by which articular 10 chondrocytes are isolated from the harvested cartilage, expanded ex vivo for up to 3 weeks, and implanted using a periosteal flap, bio-absorbable scaffolds or bioactive scaffolds [4]. ACI is suited for treating symptomatic ICRS grade III and IV lesions greater than 2mm2 but less than 12mm2 in size. A recent systematic review has identified a low overall failure rate of 5.8%, but a high overall re-operation rate of 33%, the most problematic of which was the hypertrophy of the periosteal flap [82]. Aside from this, delamination of neocartilage was cited as the next common mode of failure. Further, ex vivo cultured articular chondrocytes dedifferentiate as they expand [83], losing their collagen type II expression and gaining collagen type I expression [84], among many other changes. The matrix-induced autologous chondrocyte implantation, or MACI (Genzyme Biosurgery, Cambridge, MA, USA), employs a collagen type I/III scaffold on which chondrocytes are seeded. The resulting cartilage tissue is fixed to the lesion by fibrin glue [85]. MACI eliminates the use of the periosteal cover used in the conventional ACI, which undergoes hypertrophy and causes the repair tissue to fail [86]. Indeed, the overall failure and complication rates were lower, but those that failed did so by delamination [82].

Intact articular cartilage tissue may be transferred to the damage site as osteochondral grafts, also known as mosaicplasty [87] or osteochondral transfer (OCT). Unlike ACI, the pericellular and extracellular matrices of articular cartilage is preserved, thereby minimizing the disturbance to chondrocytes. The zonal architecture is also preserved. The source of the articular cartilage being transferred may be autologous, taken from non-weight-bearing locations of the joint, or allogeneic. Fixation is provided through subchondral bone ingrowth. However, transplanted cartilage tissue integrates poorly with surrounding native cartilage tissue, and lesions are partially filled with fibrocartilage, as marrow is stimulated during surgery [88]. Since the osteochondral plugs come from non-weight-bearing locations, the biomechanical properties of the cartilage tissues on the plugs are inferior from the surrounding, weight-bearing tissue [89]. Further, harvesting osteochondral plugs leads to donor site morbidity, which do not spontaneously heal [90].

Arthroplasty is a surgical procedure by which the damaged joint is replaced by non- biological, synthetic implants [91]. It may be total or partial (only parts of the damaged joint – femoral, tibial or patellar). The synthetic implants are made of metal, organic polymers, ceramic, or a combination of such material, and bone cement may be used to enhance the adhesion. While synthetic joint replacements are successful in restoring joint functionality, their lifespans are 11 limited. Arthroplasty is considered to be a major surgical procedure, and there exists a significant risk for complications. The overall clinical success rate, patient satisfaction rate and the gains in quality life were reported to be high [92], but more recent studies dispute this figure, with patient dissatisfaction rate reaching as high as 20% [93]. Failures in arthroplasty occur at an average of 5.9 years primarily due to loosening, instability, infections and arthrofibrosis and malalignment [94]. The increasing patient dissatisfaction rate and the myriad of possible complications present a significant opportunity for introducing innovative designs for devices used in the procedure [95].

Thus, all current approaches to biological cartilage repair have significant limitations. For biological approaches, there are three significant major issues to overcome: cell source to create the neocartilage, recreating the native cartilage architecture and biomechanical properties, and integration of the neocartilage with surrounding cartilage tissues. Marrow stimulation and ACI techniques cannot regenerate hyaline cartilage with a correct zonal architecture, resulting in poor biomechanical properties and failure to prevent further degeneration. OCT techniques also suffer from donor site morbidity, and although the zonal architecture of cartilage is preserved, the cartilage integrates poorly with the surrounding cartilage tissue. The subsequent sections examine the current research to overcome these issues.

1.2 Cells for Cartilage Repair

1.2.1 Articular Chondrocytes

To overcome the previously discussed limitations on the ex vivo expanded autologous chondrocytes used in the ACI, the use of alternate sources of cells may be considered. Since the joint is an immunoprivileged site [96], allografts can be used to treat the defects [97]. Therefore, harvesting articular chondrocytes from allogeneic sources could overcome the scarcity of cells in theory. Allogeneic chondrocytes were used with or without scaffold to repair chondral defects in rabbits [97, 98], and in humans, knee osteochondral allografts have yielded satisfactory, long- term functional results [99]. However, for all transplantable organs, demand outstrips supply, and the low cell density of cartilage may require many joints to be harvested to treat a single joint, further exacerbating the supply problem. T-cell reactivity against chondrocytes were observed in patients with osteoarthritis [100], suggesting that diseased joints may no longer be immunoprivileged. Finally, as with other transplant strategies, allografts carry the risk of transmitting diseases.

Articular chondrocytes lose their phenotype after being expanded in vitro [83]. Attempts to preserve the chondrocytic phenotype while expanding in vitro were not successful [101, 102]. Instead, these expanded chondrocytes could be cultured in three-dimensional cultures to regain parts of their chondrocytic phenotype [103-106]. Expression of type I collagen, acquisition of stress fibres and spindle-shape morphology after adhering to tissue culture plastic have been likened to progenitor cells in culture, which led to them being termed “dedifferentiated” chondrocytes [103, 107, 108]. Dedifferentiated chondrocytes co-cultured with primary chondrocytes can result in cartilage tissue formation, paving a way for amplification of the cell number [109, 110]. Further work is necessary to understand how to better manipulate them for optimal tissue growth.

1.2.2 Pluripotent Stem Cells

Many researchers have turned their attention to stem cells as a source of chondrocytes. Embryonic stem cells (ESCs) are derived from the inner cell mass of a blastocyst, and can give rise to three germ layers – ectoderm, mesoderm and endoderm. Compared to somatic cells, 13 embryonic stem cells can proliferate rapidly in vitro. Since they appear much earlier in development than interzone cells that give rise to articular chondrocytes, they must have the potential to yield bona fide articular chondrocytes under the right culture conditions. Indeed, ESCs have been demonstrated to differentiate to chondrocytes through a mesodermal intermediate that recapitulate selected properties of articular chondrocytes, rather than non- articular/growth plate chondrocytes in a mouse model [111]. Currently, ESC-derived retinal pigment epithelium are undergoing clinical trials for treating various forms of macular degeneration [112], whose outcome will establish the safety of ESCs for therapeutic use.

Induced pluripotent stem cells (iPSCs) are cells that are manipulated, commonly by reprogramming, to exhibit ESC-like phenotype. In a pair of seminal papers by Takahashi and Yamanaka, iPSCs were derived from mouse and human fibroblasts by transduction of Oct3/4, Sox2, c-Myc and Klf4 [113, 114]. A number of studies have induced the expression of chondrogenic markers from iPSCs by differentiation, either directly or via mesodermal/mesenchymal intermediate, or co-culture with primary chondrocytes, but a cartilage tissue has yet to have been generated from them [115]. iPSCs have the potential to substitute ESCs for use in tissue engineering without the ethical issues associated with its procurement. In September 2014, the first iPSC clinical trial was approved in Japan, in which patient-specific iPSC-derived retinal pigment epithelium were implanted. The study is currently underway.

1.2.3 Mesenchymal Stromal Cells and Bone Marrow Stromal Cells

Mesenchymal stromal cells (MSCs) were first discovered by Friedenstein et al. in 1966 [116]. Using in vitro cultures of mice bone marrow, the authors observed that a subpopulation of highly mitotic cells within the bone marrow could form bone tissue, not hematopoietic cells. The requirement of a certain number of cells to form such osteogenic colonies were likened to those of hematopoietic stem cells (HSCs) described by Till & McCulloch a few years earlier [117], and postulated that a separate stem cell population from HSCs is responsible for the “stromal elements” of bone marrow. The multipotency of MSCs was demonstrated in the 1980s [118]. In vitro chondrogenesis was first reported in 1998 by Johnstone et al. [119], whom described the pellet culture method of rabbit MSCs in a serum-free defined media that contained 14 dexamethasone and transforming growth factor-beta 1 (TGF-β1). Hence, MSCs are said to be multipotent.

MSCs also proliferate rapidly in vitro, and the expanded cells maintain their multipotency. This was interpreted as evidence for self-renewal, which led to the term mesenchymal stem cells. However, unlike HSCs, MSCs undergo senescence in culture and lose their multipotency [120]. MSCs are defined by the demonstration of the following characteristics: adherence to plastic, specific surface antigen expression and multipotential differentiation potential [121]. Populations of cells that satisfy this set of criteria were isolated from many tissues, including bone marrow, synovium, adipose tissue, peripheral blood vessels, teeth and many others, but they appear to vary in their efficacy at differentiation and the stimuli to which they respond to differentiate – for example, adipose-derived MSCs undergo chondrogenesis in the presence of BMP6, but bone marrow-derived MSCs do not [122]. MSCs of different origins have varied expression of genes [123] and surface antigens [124]. Chondrocytes derived from MSCs of different origins also have disparate gene expression patterns [125] and efficacy in cartilage repair [126-129]. Since bone marrow-derived MSCs (bone marrow stromal cells, or BMSCs) are a de facto gold standard of MSCs [125] subsequent discussion will focus on BMSCs.

The developmental origin of BMSCs is not entirely clear. Because bone is mesodermal in origin (except craniofacial bone, which is neuroectodermal), BMSCs were also presumed to be mesodermal. In bone, BMSCs exist as a rare cell subpopulation of the marrow stroma. Using a colony forming unit-fibroblastic (CFU-F) assay, one BMSC exists for every 104–105 nucleated cells in bone marrow. Combined with the current lack of reliable MSC-specific markers, it is difficult to characterize its niche in vivo [130].

BMSCs have immunomodulative properties and are themselves immunoprivileged [131]: that is, they can manipulate the host immune system to decrease their inflammatory response, and the cells themselves will not elicit an immune response [132]. For this reason, BMSCs have been said to be the “safe” cells for transplantation in tissue engineering, and in their undifferentiated form, they are used to treat graft-versus-host disease [133], septic shock [134], and they can be co-transplanted with HSCs to augment its grafting [135]. However, BMSCs can internalize antigens such as those from the fetal bovine serum during its in vitro culture [136], 15 and they can elicit an immune response after transplantation [137]. Moreover, previous studies reported that osteoblasts differentiated from BMSCs lose their immunomodulative property while keeping its immunoprivileged nature [138], while chondrocytes differentiated from BMSCs lose both their immunomodulative and immunoprivileged properties [139, 140]. The use of autologously sourced BMSCs would minimize the cells’ immunogenic potential; for this reason, autologous BMSCs are preferred in cell therapies [141]. To mitigate the batch-to-batch variability that rises using autologously sourced BMSCs, standardizing the tissue culture conditions [120] and eliminate inherently variable reagents such as serum [142] would be beneficial.

1.2.4 Chondrogenesis of Bone Marrow Stromal Cells

The simplest example of BMSC chondrogenesis is an in situ chondrogenesis of undifferentiated BMSCs in a chondral defect. Porcine BMSCs were implanted in chondral defects without any growth factors under a collagen types I/III membrane, similar to matrix-induced ACI [143]. Patches of cartilaginous matrix were observed after 8 weeks in both BMSC implant and the sham control, demonstrating that the microenvironment of the joint itself is insufficient for BMSC chondrogenesis.

In vitro, pellet culture of BMSCs is the most common method for inducing chondrogenesis [144, 145]. To set up the pellet culture, BMSCs are simply centrifuged in a conical cube to produce a closely packed cell mass. This creates a 3-dimensional niche in which cells exist in a round morphology, with neighbouring cells as the substrate that possess a Young’s modulus in the order of kilopascals [146]. Both the stiffness of substrate and the cell shape were previously demonstrated to be determinants of BMSC differentiation [147, 148]. BMSCs in pellets are then cultured in a serum-free, defined media that consists of high-glucose Dulbecco’s modified Eagle’s medium (hgDMEM), insulin-selenium-transferrin culture supplement (ITS), ascorbic acid, dexamethasone and a pro-chondrogenic cytokine, transforming growth factor (TGF-β) isoforms 1-3 or bone morphogenic protein (BMP) isoforms 2, 4, 6 or 7.

BMSCs in chondrogenic pellet culture expresses Sox9, similar to the condensed mesenchymal cell mass in limb buds. After that, cells will express Col2a1 (type II collagen), Agc 16

(aggrecan), COMP (cartilage oligomeric matrix protein) and others, while also accumulating collagenous extracellular matrix and proteoglycans, leading to a hyaline appearance. The pellets also grow in size as matrix is accumulated. At the end of 3 weeks of culture, histological examination reveals the round morphology of cells and the presence of extracellular matrix rich in collagen type II and proteoglycans.

After the 2nd and 3rd weeks of differentiation culture, cells undergo terminal differentiation: that is, cells enlarge while also expressing genetic markers of hypertrophy – Runx2 (runt-related transcription factor isoform 2, an osteogenic transcription factor), Col10a1 (type X collagen), ALP (alkaline phosphatase), MMP13 (matrix metallopeptidase 13) and VEGF (vascular endothelial growth factor), displaying a hypertrophic chondrocyte phenotype [149]. Withdrawal of pro-chondrogenic factors TGF-β and dexamethasone, as well as introduction of β- glycerophosphate in the culture condition, produces mineralized matrix similar to growth plate undergoing calcification [150]. PTHrP, which inhibits the hypertrophic progression of growth plate chondrocytes, also prevents BMSC-derived chondrocytes from displaying this phenotype [151-153]. However, withdrawal of PTHrP from the culture media restores the hypertrophic phenotype [152].

The propensity of BMSC-derived chondrocytes to undergo this terminal differentiation is a critical limitation for cartilage tissue engineering using this cell source. Hif-1α, the principal transcription factor expressed in response to hypoxia (pO2 < 5%), was shown to regulate the differentiation of limb bud mesenchyme and joint development [154]. Indeed, BMSCs undergo chondrogenesis under hypoxia more effectively [155, 156], and their expression of Col10a1 and Runx2 were downregulated, suggesting that hypoxia may suppress terminal chondrocyte differentiation [157]. Also, decreasing the size of the pellets resulted in increased expression levels of chondrogenic markers [158], as delivery of nutrients and signalling molecules to cells in the core of larger pellets may be limited by diffusion. A recent study demonstrated the meta- aggregation of pellets to create a large-sized cartilage on joint-shaped surfaces [159]. Still, this may not be sufficient: in a combinatorial study across five pro-chondrogenic soluble factors, expression of terminal differentiation markers concomitant with in vitro chondrogenesis was unavoidable [160]. This is not limited to BMSC-derived chondrogenesis, but also for all MSCs [161, 162], as well as ESC-derived cartilage that, in principle, should be capable of differentiation to bona fide articular chondrocytes. A pharmacologic inhibitor of retinoic acid 17 receptor successfully induced chondrogenesis of BMSCs in a Sox9-independent manner that also avoided expression of terminal differentiation markers [163], but this study was not followed up further. However, multiple in vivo studies using BMSC-derived cartilage have now shown the cartilage mineralization [164-166], suggesting the importance of the orthotopic environment in preventing the terminal differentiation of these cells.

1.3 Osteochondral-Like Constructs for Cartilage Repair

1.3.1 A Case for Scaffold-Free Tissue Engineering of Cartilage

Scaffolds are materials that act as matrices in which cells grow. An ideal scaffold must be biocompatible, biodegradable, non-cytotoxic and non-immunogenic. It must possess appropriate mechanical strength, support cell attachment, proliferation and differentiation, and allow for the transportation of nutrients and metabolic waste [167]. Scaffolds are considered by some to be an essential element of tissue engineering: scaffolds can be engineered, both macroscopically to specify the size, shape and the biochemical/biomechanical properties of the tissue-engineered implant, and microscopically to fine-tune the microenvironment presented to the resident cells. Scaffolds can also provide the fixation for the cells both to the implant site and in relation to each other, which is critical for recapitulating the zonal organization of the engineered tissue. Scaffolds can also provide spatiotemporally controlled delivery of growth factors and drugs.

Scaffolds used for BMSC chondrogenesis can be derived from natural or synthetic sources. Collagen is a naturally occurring component of joint tissues, and therefore is highly biocompatible [168]. Collagen supports the attachment and chondrogenic differentiation of BMSCs [169]. Collagen can be extracted from many different sources without adverse effects, and collagen-based materials are already in clinical use such as MACI. However, cells attached to collagen elongate [170]. Hyaluronan is a naturally occurring GAG abundant in cartilage. However, hyaluronan in its natural form has poor structural integrity, requiring it to be crosslinked or combined with other materials such as collagen [171, 172]. Other naturally scaffolds include chitosan, a cationic linear polysaccharide derived from crustacean shells, and alginate, an anionic polysaccharide derived from seaweed [173, 174]. As for synthetic scaffolds, they can be made to be biocompatible, bioresorbable and fine-tuned for various important parameters such as their mechanical properties and the pore size. Synthetic materials such as polylactic acid (PLA), polyglycolic acid (PGA), poly(lactic-co-glycolic) acid (PLGA), polycaprolactone (PCL) as well as polyethylene glycol (PEG) can also be combined with naturally occurring materials, as well as peptides for enhanced attachment, to enhance bioactivity, biocompatibility and mechanical properties [173]. Combined with techniques to spatially vary the composition, multi-layered hydrogel-based construct was used to recreate the zonal architecture of articular cartilage [175]. Hydrogel has a high affinity for water as with 19 articular cartilage, affords uniform distribution of cells throughout, and diffusivity that can be tightly controlled by the synthesis process [176].

Decellularized tissues have been successfully used as scaffolds for a range of different tissues [177]. Decellularization aims to remove all cells and antigens while retaining the bioactive cues that reside in the extracellular matrix, which supports and encourages specific tissue formation at the implantation site via constructive remodeling [178]. Such material can be harvested not only from the allogeneic human donors but also xenogeneic (e.g. porcine heart valves) or in vitro sources (e.g. Matrigel). However, the detergent-based methods for decellularization can also alter protein structures and remove proteoglycans that provide bioactive cues [179, 180]. Also, the biomechanical requirements and the zonal architecture of osteochondral implants remain unsolved challenges for this technique to be successfully applied to cartilage tissue engineering [177].

While the scaffolds facilitate the direct control of the cell niche and the organization of provisional matrix that regulate cellular organization, there are many drawbacks to their application for cartilage tissue engineering. First, scaffolds must be capable of not only withstanding the mechanical loads present in a joint, but do so while exhibiting similar biomechanical characteristics as hyaline cartilage [181]. This is a critical drawback for matrix- induced ACI techniques, because the scaffold will be exposed to mechanical stress as soon as it is implanted. To avoid this, cartilage could be cultured in vitro for cells to accumulate and assemble the cartilaginous matrix free of such mechanical stress.

The scaffold must be biodegradable to allow the space for the cartilaginous extracellular matrix to accumulate and not hinder tissue regeneration [182]. However, this raises several drawbacks: the rate of scaffold degradation must be carefully matched with the rate of ECM accumulation [183]. Mismatching these rates would compromise the mechanical integrity of the tissue. The removal of degradation by-products must also be considered [184], as they may also have unwanted side effects if not removed or metabolized appropriately. Harsh and often toxic chemicals are frequently employed in the synthesis of scaffolds, which must be thoroughly removed prior to both in vitro and in vivo manipulation [184].

Articular chondrocytes in hyaline cartilage possess a spherical morphology, and loss of this morphology is indicative of its dedifferentiation, which is in turn associated with its 20 production of fibrocartilaginous matrix. Therefore, scaffolds must promote appropriate cell shape. Also, scaffolds must provide the right environment for the cells to properly sense biomechanical cues, both intrinsic and extrinsic [185]. Since pericellular matrix acts as a transducer of mechanical signals, the scaffolds must be able to permit the accumulation of pericellular matrix that integrate with the surrounding scaffold material to allow the transmission of mechanical signals.

In development, articular cartilage arises from condensation of mesenchymal cells, getting cues from cell-cell contact, growth factors and mechanical forces, without scaffolds [184]. Therefore, applying our understanding of these factors on chondroprogenitors capable of undergoing the same developmental program should yield neocartilage with proper accumulation and organization of extracellular and pericellular matrix, as well as the zonal architecture, without the use of scaffolds. Pellet culture is the simplest approach, and mimics the mesenchyme condensation present in the limb bud generation, to induce chondrogenesis of mesenchymal stromal cells [144].

1.3.2 Tissue Engineering of Osteochondral Constructs

Tissue engineering of osteochondral constructs would alleviate one of the major drawbacks of osteochondral transfer discussed earlier: the donor site morbidity. The advantages of osteochondral transfer are also applicable to osteochondral constructs engineering: the cartilage tissue would already be established before implantation for joint repair. It could also replace the damaged subchondral bone, which appears in the late-stage osteoarthritis (ICRS grade IV) [186, 187].

There exists a multitude of strategies for tissue engineering constructs that would replace both damaged cartilage and its subchondral bone. They can be categorized by whether scaffolds or cells were used in which of the cartilage or bone phases (reviewed in [188]). Efforts to create tissue-engineered bone replacement have been met with numerous successes [184], which shifts the main foci of osteochondral tissue engineering to two aspects: creating zonally organized, biomechanically competent cartilage tissue; and good mechanical integration between the cartilage and bone phases. 21

As discussed, scaffolds can be used to create zonally organized cartilage. Nguyen et al. used multilayered hydrogel-based constructs with spatially varying materials to recreate the zonal architecture of articular cartilage [175]. Marquass et al. used a homogenous collagen type I scaffold for the cartilage phase, yet the 6-month postimplantation histology showed a zonal architecture within the cartilage phase [166]. Scaffold-free approaches are also applicable: Tuli et al. pelleted BMSCs on PLA scaffold in chondrogenic media, then on the flip side, BMSCs pre-cultured in osteogenic media were seeded to generate a biphasic construct [189]. In the Kandel lab, we created a biphasic construct that consists of scaffold-free, tissue-engineered cartilage and a cell-free bone substitute biomaterial substrate. We first demonstrated this biphasic construct by culturing primary bovine articular chondrocytes on porous calcium polyphosphate (CPP) substrate [190]. In vivo studies of the biphasic construct in sheep model has demonstrated that not only does tissue-engineered cartilage survive up to 9 months post-implantation, but its mechanical properties improved over time [191].

The cartilage-bone interface is important for the function of articular cartilage. The zone of calcified cartilage allows for a gradual transition in stiffness between the articular cartilage and the much stiffer bone. This allows the shear stress to be dissipated across a larger volume, making the tissue as a whole more resistant to delamination. Since delamination is the most common mode of failure in chondrocyte transfer strategies [82], being able to engineer a mechanically competent cartilage-bone interface is crucial to the success of cartilage repair. Cheng et al. encapsulated BMSCs in collagen type I microspheres, cultured them in either chondrogenic or osteogenic conditions, then assembled them together in more collagen type I in an osteochondral arrangement [192]. Remarkably, a layer of calcified cartilage spontaneously arose at the cartilage-bone interface.

Because BMSCs can generate calcified cartilage in vitro due to their propensity for terminal differentiation, they are suitable for engineering the biomimetic cartilage-bone interface. Thus far, approaches to induce selective mineralization of BMSC-derived cartilage for creating this interface include delivery of bioactive signals (discussed in the next section) either locally via microspheres [193] or by using gradients generated by a bioreactor [194] to undifferentiated BMSCs seeded in scaffolds. In another approach, BMSCs were predifferentiated to chondrocytes and osteoblasts and seeded into appropriate locations within scaffolds [192, 195]. For both strategies, use of exogenous scaffolds was necessary. Further, a number of strategies aim to 22 induce osteoblastic differentiation at the interface [194-196]. Calcified cartilage possesses mechanical properties that allow for a more gradual transition of stiffness from cartilage to subchondral bone [32, 69]. Hence, an interface that incorporates calcified cartilage is more desirable.

1.3.3 Cartilage Biomineralization

Mineralization is both a physiological and a pathological phenomenon. Physiological mineralization is observed in bone, teeth as well as calcified cartilage and others, but pathological mineralization may also occur in articular cartilage, arteries and aortic valves [197]. While the mechanism by which biomineral is formed is poorly understood, the dogma of biomineralization states that the nucleation, growth, morphology and assembly of the inorganic crystals are regulated by organized assemblies of organic matrix [198]. Cells synthesize the mineral crystals by concentrating calcium and phosphate ions in matrix vesicles that also have annexins and alkaline phosphatase [199], or releasing apoptotic bodies [68]. Matrix molecules could also provide nucleation sites [200].

Hydroxyapatite is a calcium-phosphate crystal found in bone and calcified cartilage. It is poorly crystalline. Hydroxyapatite is tolerant of substitution by other functional groups, and carbonates are often found in the physiological hydroxyapatite [201]. Hydroxyapatite has a characteristic Ca:P ratio of 1:1.67, and can be identified by the characteristic electron diffraction pattern observed by transmission electron microscopy (TEM).

Knockout studies in mice have demonstrated that core binding factor-β [202], MMP-9 [203] and the molecular chaperone Hsp47 [204] are necessary for cartilage mineralization, while the ablation of cathespin K activity caused hypermineralization of the growth plate [205]. Conditional deletion of β-catenin in the growth plate also reduced the calcification of cartilage [206]. On the other hand, Col10a1-null mice exhibit normal long bone growth and development [207], suggesting that type X collagen itself does not play a role in cartilage biomineralization.

In vitro chondrogenesis of BMSCs yield a cartilaginous tissue whose cells undergo spontaneous terminal differentiation. However, only under specific culture conditions does the tissue mineralize: human BMSC pellets pre-cultured in TGF-β3-containing media accumulated 23 mineral when β-glycerophosphate and 1,1’,3-triiodothyronine (T3) were added and the concentration of dexamethasone was reduced [150]. β-glycerophosphate was known to increase the rate of mineralization in calcifying cartilage culture in vitro for a long time [208], and is now used as a source of phosphates for studying culture systems that mineralize. T3 is a thyroid hormone that, among other functions, stimulates the growth and maturation of skeleton, and can induce the hypertrophy and the subsequent matrix calcification in cartilage explants in vitro more potently than thyroxine (T4) also synthesized by the thyroid [209]. Other studies that induced hypertrophy and calcification of cartilage in vitro used 1α,25-dihydroxyvitamin D3 [210] and retinoic acid [211].

1.3.4 Porous calcium polyphosphate (CPP) substrate and inorganic polyphosphate

Porous calcium polyphosphate (CPP) is a synthetic bone substitute biomaterial, first developed by the Lagow group [212]. Porous CPP can be machined, moulded [212] or “3D printed” via additive manufacturing [213], and its pore size and density can be controlled [214]. Porous CPP presents an alternative to crystalline hydroxyapatite, which possesses weak bulk strength in both compression and tension [212, 214].

Porous CPP can be formed by the gravity sintering of calcium polyphosphate powders [214]. CPP is synthesized by calcining calcium phosphate monobasic monohydrate powder at 500ºC for 10 hours. This CPP is melted at 1100ºC to form amorphous glass, which is subsequently quenched in distilled water (called frit). This frit is then dehydrated, milled and sifted to collect CPP powders with a specific range of particle size. The CPP powders are then either packed into a platinum container or bound together via additive manufacturing, and sintered to produce the porous CPP structure.

Porous CPP is well suited for use in engineering osteochondral constructs. First, the use of this synthetic material obviates the need for bone allografts, avoiding donor site morbidity. Second, porous CPP has been demonstrated to exhibit excellent osseointegrative properties. When porous CPP rods of 4mm in diameter were implanted in rabbit femurs, extensive bone ingrowth was observed 6 weeks after implantation [215]. When implanted in load-bearing sites within rabbit femurs, more extensive bone ingrowth was observed with porous CPP with 30% 24 porosity than those with 20% [216]. Porous CPP fabricated via additive manufacturing process exhibited the same extent of bone ingrowth as those conventionally fabricated by packing in platinum containers [217]. Third, CPP is biodegradable: it releases calcium and inorganic polyphosphate ions in an aqueous environment [214]. Inorganic polyphosphates are polymers of phosphate groups, linked by high-energy phosphoanhyride bonds. Inorganic polyphosphates are ubiquitous in all living organisms [218], including mammalian cells [219]. Effects of polyphosphates are well studied in bacteria, and bacterial kinases and phosphatases for inorganic polyphosphates were identified [220]. Long-chain polyphosphates were reported to have antimicrobial properties, which may be useful in preventing bacterial contamination when CPP is implanted [221]. Less is known about polyP metabolism and regulations in mammalian systems. Enzymes such as calf intestinal alkaline phosphatase (CIAP) and human metastasis regulator protein H-prune were shown to have exopolyphosphatase activity [222, 223] and mammalian homologues of yeast protein DDP1 (DIPP1, DIPP2 and DIPP3) were shown to have endopolyphosphatase activity [224]. A mammalian polyphosphate kinase has yet to be identified.

1.3.5 Effect of polyphosphates on biomineralization

Polyphosphates have been shown to inhibit biomineralization, but at a certain concentration and in the presence of polyphosphatases, enhance it [225]. The inhibitory effect of polyphosphates on deep zone articular chondrocyte mineralization in vitro has been directly demonstrated [226]. This posed a significant hurdle in recreating a zone of calcified cartilage at the cartilage-bone interface: deep zone articular chondrocytes cultured on porous CPP substrate in mineralizing tissue culture conditions show a band of mineral accumulation some distance away from the cartilage-CPP substrate interface [227]. To explain this, it was suggested that the polyphosphate accumulation in the cartilaginous matrix was a gradient. That is, the polyphosphate accumulation decreases with distance from polyphosphate-releasing CPP substrate. Given sufficient distance away from the substrate, the ALP activity of deep zone articular chondrocytes could utilize polyphosphates to enhance the mineralization. By coating the porous CPP substrate to prevent the release of polyphosphate, mineralization of the cartilage-CPP interface was achieved [228]. However, the interfacial shear strength did not improve, leading to the speculation that the coating process blocked the micron-sized pores of the CPP substrate and prevented cartilage 25 tissue ingrowth. Therefore, an alternative method to coat the CPP is required if the mineralized cartilage layer were to occur at the cartilage-bone substitute interface.

1.4 Conclusion

Articular cartilage is a tissue with a poor regenerative capacity. There are many barriers to overcome for successful cartilage repair, both intrinsic and extrinsic. Left untreated, damaged cartilage undergoes a progressive degenerative process, ultimately leading to a dysfunction of the joint that inflicts pain and limited mobility. Currently, treatments in the clinic aim to delay this degenerative process, but do so poorly due to the formation of neocartilage lacking the proper biochemical and biomechanical properties. This could be addressed by using cells and/or scaffolds to create replacement cartilage, which can be extended to creating osteochondral constructs. Scaffold-free approaches provide a viable alternative to scaffolds, and can be applied to create osteochondral constructs. The success of tissue-engineered osteochondral constructs would depend on the recapitulation of depth-dependent zonal architecture, such as the calcified cartilage-bone interface. BMSCs are a source of cells that can create both hyaline and calcified cartilage due to its propensity for terminal differentiation, and therefore could be utilized to create this osteochondral construct.

1.5 Hypothesis

Using BMSCs as a cell source, cartilage with a multi-zonal architecture can be formed on porous CPP substrate to produce osteochondral-like constructs with biomechanical properties that approach native articular cartilage.

1.6 Specific Aims

Specific Aim 1: Generate and characterize cartilage tissue formed by sheep BMSCs on the CPP substrate.

Specific Aim 2: Develop a mechanically competent interface between the BMSC-derived cartilage tissue and the porous CPP substrate.

Specific Aim 2.1: Investigate the use of organic-route sol gel thin film processing in coating the porous CPP substrate for limiting the inorganic polyphosphate accumulation in tissue.

Specific Aim 2.2: Induce the mineralization of BMSC-derived cartilage tissue selectively at the cartilage-CPP interface while preserving the hyaline nature of the rest of the cartilage tissue.

1.7 References

[1] Shepherd DE, Seedhom BB. Thickness of human articular cartilage in joints of the lower limb. Ann Rheum Dis (1999) 58:27-34.

[2] Wang H, Ateshian GA. The normal stress effect and equilibrium friction coefficient of articular cartilage under steady frictional shear. J Biomech (1997) 30:771-6.

[3] Sophia Fox AJ, Bedi A, Rodeo SA. The basic science of articular cartilage: structure, composition, and function. Sports Health (2009) 1:461-8.

[4] Mollon B, Kandel R, Chahal J, Theodoropoulos J. The clinical status of cartilage tissue regeneration in humans. Osteoarthritis Cartilage (2013) 21:1824-33.

[5] Mollenhauer JA. Perspectives on articular cartilage biology and osteoarthritis. Injury (2008) 39 Suppl 1:S5-12.

[6] Hunziker EB, Lippuner K, Shintani N. How best to preserve and reveal the structural intricacies of cartilaginous tissue. Matrix Biol (2014).

[7] Vynios DH. Metabolism of cartilage proteoglycans in health and disease. Biomed Res Int (2014):452315.

[8] Levick JR, Knight AD. Osmotic flows across the blood-joint barrier. Ann Rheum Dis (1987) 46:534-9.

[9] Fischer AE, Carpenter TA, Tyler JA, Hall LD. Visualisation of mass transport of small organic molecules and metal ions through articular cartilage by magnetic resonance imaging. Magn Reson Imaging (1995) 13:819-26.

[10] O'Hara BP, Urban JP, Maroudas A. Influence of cyclic loading on the nutrition of articular cartilage. Ann Rheum Dis (1990) 49:536-9.

[11] McCarty WJ, Luan A, Siddiqui M, Hansen BC, Masuda K, Sah RL. Biomechanical properties of mixtures of blood and synovial fluid. J Orthop Res (2011) 29:240-6. 28

[12] Schmidt TA, Sah RL. Effect of synovial fluid on boundary lubrication of articular cartilage. Osteoarthritis Cartilage (2007) 15:35-47.

[13] Stevens CR, Blake DR, Merry P, Revell PA, Levick JR. A comparative study by morphometry of the microvasculature in normal and rheumatoid synovium. Arthritis Rheum (1991) 34:1508-13.

[14] Blake DR, Merry P, Unsworth J, Kidd BL, Outhwaite JM, Ballard R, et al. Hypoxic- reperfusion injury in the inflamed human joint. Lancet (1989) 1:289-93.

[15] Houard X, Goldring MB, Berenbaum F. Homeostatic mechanisms in articular cartilage and role of inflammation in osteoarthritis. Curr Rheumatol Rep (2013) 15:375.

[16] Grodzinsky AJ, Levenston ME, Jin M, Frank EH. Cartilage tissue remodeling in response to mechanical forces. Annu Rev Biomed Eng (2000) 2:691-713.

[17] Sah RL, Kim YJ, Doong JY, Grodzinsky AJ, Plaas AH, Sandy JD. Biosynthetic response of cartilage explants to dynamic compression. J Orthop Res (1989) 7:619-36.

[18] Kurz B, Jin M, Patwari P, Cheng DM, Lark MW, Grodzinsky AJ. Biosynthetic response and mechanical properties of articular cartilage after injurious compression. J Orthop Res (2001) 19:1140-6.

[19] Poole CA, Ayad S, Schofield JR. Chondrons from articular cartilage: I. Immunolocalization of type VI collagen in the pericellular capsule of isolated canine tibial chondrons. J Cell Sci (1988) 90 (Pt 4):635-43.

[20] Poole CA. Articular cartilage chondrons: form, function and failure. J Anat (1997) 191 (Pt 1):1-13.

[21] Guilak F, Mow VC. The mechanical environment of the chondrocyte: a biphasic finite element model of cell-matrix interactions in articular cartilage. J Biomech (2000) 33:1663-73.

[22] Alexopoulos LG, Williams GM, Upton ML, Setton LA, Guilak F. Osteoarthritic changes in the biphasic mechanical properties of the chondrocyte pericellular matrix in articular cartilage. J Biomech (2005) 38:509-17. 29

[23] Mow VC, Guo XE. Mechano-electrochemical properties of articular cartilage: their inhomogeneities and anisotropies. Annu Rev Biomed Eng (2002) 4:175-209.

[24] Guilak F, Ratcliffe A, Lane N, Rosenwasser MP, Mow VC. Mechanical and biochemical changes in the superficial zone of articular cartilage in canine experimental osteoarthritis. J Orthop Res (1994) 12:474-84.

[25] Swann DA, Silver FH, Slayter HS, Stafford W, Shore E. The molecular structure and lubricating activity of lubricin isolated from bovine and human synovial fluids. Biochem J (1985) 225:195-201.

[26] Hollander AP, Pidoux I, Reiner A, Rorabeck C, Bourne R, Poole AR. Damage to type II collagen in aging and osteoarthritis starts at the articular surface, originates around chondrocytes, and extends into the cartilage with progressive degeneration. J Clin Invest (1995) 96:2859-69.

[27] Temple-Wong MM, Bae WC, Chen MQ, Bugbee WD, Amiel D, Coutts RD, et al. Biomechanical, structural, and biochemical indices of degenerative and osteoarthritic deterioration of adult human articular cartilage of the femoral condyle. Osteoarthritis Cartilage (2009) 17:1469-76.

[28] McLeod MA, Wilusz RE, Guilak F. Depth-dependent anisotropy of the micromechanical properties of the extracellular and pericellular matrices of articular cartilage evaluated via atomic force microscopy. J Biomech (2013) 46:586-92.

[29] Miao D, Scutt A. Histochemical localization of alkaline phosphatase activity in decalcified bone and cartilage. J Histochem Cytochem (2002) 50:333-40.

[30] Hoemann CD, Lafantaisie-Favreau CH, Lascau-Coman V, Chen G, Guzman-Morales J. The cartilage-bone interface. J Knee Surg (2012) 25:85-97.

[31] Ferguson VL, Bushby AJ, Boyde A. Nanomechanical properties and mineral concentration in articular calcified cartilage and subchondral bone. J Anat (2003) 203:191-202.

[32] Mente PL, Lewis JL. Elastic modulus of calcified cartilage is an order of magnitude less than that of subchondral bone. J Orthop Res (1994) 12:637-47. 30

[33] Benjamin M, Evans EJ. Fibrocartilage. J Anat (1990) 171:1-15.

[34] Newman AP. Articular cartilage repair. Am J Sports Med (1998) 26:309-24.

[35] Kronenberg HM. Developmental regulation of the growth plate. Nature (2003) 423:332-6.

[36] Shum L, Coleman CM, Hatakeyama Y, Tuan RS. Morphogenesis and dysmorphogenesis of the appendicular skeleton. Birth Defects Res C Embryo Today (2003) 69:102-22.

[37] Hoffman LM, Garcha K, Karamboulas K, Cowan MF, Drysdale LM, Horton WA, et al. BMP action in skeletogenesis involves attenuation of retinoid signaling. J Cell Biol (2006) 174:101-13.

[38] Ornitz DM. FGF signaling in the developing endochondral skeleton. Cytokine Growth Factor Rev (2005) 16:205-13.

[39] Davis AP, Witte DP, Hsieh-Li HM, Potter SS, Capecchi MR. Absence of radius and ulna in mice lacking hoxa-11 and hoxd-11. Nature (1995) 375:791-5.

[40] Wright E, Hargrave MR, Christiansen J, Cooper L, Kun J, Evans T, et al. The Sry-related gene Sox9 is expressed during chondrogenesis in mouse embryos. Nat Genet (1995) 9:15-20.

[41] Craig FM, Bentley G, Archer CW. The spatial and temporal pattern of collagens I and II and keratan sulphate in the developing chick metatarsophalangeal joint. Development (1987) 99:383- 91.

[42] Pacifici M, Koyama E, Shibukawa Y, Wu C, Tamamura Y, Enomoto-Iwamoto M, et al. Cellular and molecular mechanisms of synovial joint and articular cartilage formation. Ann N Y Acad Sci (2006) 1068:74-86.

[43] Decker RS, Koyama E, Pacifici M. Genesis and morphogenesis of limb synovial joints and articular cartilage. Matrix Biol (2014) 39:5-10.

[44] Sandell LJ, Nalin AM, Reife RA. Alternative splice form of type II procollagen mRNA (IIA) is predominant in skeletal precursors and non-cartilaginous tissues during early mouse development. Dev Dyn (1994) 199:129-40. 31

[45] Patra D, DeLassus E, McAlinden A, Sandell LJ. Characterization of a murine type IIB procollagen-specific antibody. Matrix Biol (2014) 34:154-60.

[46] Kawakami Y, Rodriguez-Leon J, Izpisua Belmonte JC. The role of TGFβs and Sox9 during limb chondrogenesis. Curr Opin Cell Biol (2006) 18:723-9.

[47] Pitsillides AA, Ashhurst DE. A critical evaluation of specific aspects of joint development. Dev Dyn (2008) 237:2284-94.

[48] Kozhemyakina E, Zhang M, Ionescu A, Ayturk UM, Ono N, Kobayashi A, et al. Identification of a Prg4-expressing articular cartilage progenitor cell population in mice. Arthritis Rheumatol (2015) 67:1261-73.

[49] Hyde G, Dover S, Aszodi A, Wallis GA, Boot-Handford RP. Lineage tracing using matrilin- 1 gene expression reveals that articular chondrocytes exist as the joint interzone forms. Dev Biol (2007) 304:825-33.

[50] Hissnauer TN, Baranowsky A, Pestka JM, Streichert T, Wiegandt K, Goepfert C, et al. Identification of molecular markers for articular cartilage. Osteoarthritis Cartilage (2010) 18:1630-8.

[51] Leijten JC, Emons J, Sticht C, van Gool S, Decker E, Uitterlinden A, et al. Gremlin 1, frizzled-related protein, and Dkk-1 are key regulators of human articular cartilage homeostasis. Arthritis Rheum (2012) 64:3302-12.

[52] Stevens DA, Hasserjian RP, Robson H, Siebler T, Shalet SM, Williams GR. Thyroid hormones regulate hypertrophic chondrocyte differentiation and expression of parathyroid hormone-related peptide and its receptor during endochondral bone formation. J Bone Miner Res (2000) 15:2431-42.

[53] Minina E, Kreschel C, Naski MC, Ornitz DM, Vortkamp A. Interaction of FGF, Ihh/Pthlh, and BMP signaling integrates chondrocyte proliferation and hypertrophic differentiation. Dev Cell (2002) 3:439-49.

[54] Staines KA, Pollard AS, McGonnell IM, Farquharson C, Pitsillides AA. Cartilage to bone transitions in health and disease. J Endocrinol (2013) 219:R1-R12. 32

[55] Lee AS, Ellman MB, Yan D, Kroin JS, Cole BJ, van Wijnen AJ, et al. A current review of molecular mechanisms regarding osteoarthritis and pain. Gene (2013) 527:440-7.

[56] Saarakkala S, Julkunen P, Kiviranta P, Makitalo J, Jurvelin JS, Korhonen RK. Depth-wise progression of osteoarthritis in human articular cartilage: investigation of composition, structure and biomechanics. Osteoarthritis Cartilage (2010) 18:73-81.

[57] Rivers PA, Rosenwasser MP, Mow VC, Pawluk RJ, Strauch RJ, Sugalski MT, et al. Osteoarthritic changes in the biochemical composition of thumb carpometacarpal joint cartilage and correlation with biomechanical properties. J Hand Surg Am (2000) 25:889-98.

[58] Bondeson J, Blom AB, Wainwright S, Hughes C, Caterson B, van den Berg WB. The role of synovial macrophages and macrophage-produced mediators in driving inflammatory and destructive responses in osteoarthritis. Arthritis Rheum (2010) 62:647-57.

[59] Bondeson J. Are we moving in the right direction with osteoarthritis drug discovery? Expert Opin Ther Targets (2011) 15:1355-68.

[60] Haywood L, McWilliams DF, Pearson CI, Gill SE, Ganesan A, Wilson D, et al. Inflammation and angiogenesis in osteoarthritis. Arthritis Rheum (2003) 48:2173-7.

[61] Lohmander LS, Englund PM, Dahl LL, Roos EM. The long-term consequence of anterior cruciate ligament and meniscus injuries: osteoarthritis. Am J Sports Med (2007) 35:1756-69.

[62] Sharma L, Song J, Dunlop D, Felson D, Lewis CE, Segal N, et al. Varus and valgus alignment and incident and progressive knee osteoarthritis. Ann Rheum Dis (2010) 69:1940-5.

[63] Mizuta H, Kudo S, Nakamura E, Otsuka Y, Takagi K, Hiraki Y. Active proliferation of mesenchymal cells prior to the chondrogenic repair response in rabbit full-thickness defects of articular cartilage. Osteoarthritis Cartilage (2004) 12:586-96.

[64] Desrochers J, Amrein MW, Matyas JR. Viscoelasticity of the articular cartilage surface in early osteoarthritis. Osteoarthritis Cartilage (2012) 20:413-21.

[65] Lotz M, Loeser RF. Effects of aging on articular cartilage homeostasis. Bone (2012) 51:241- 8. 33

[66] Mobasheri A, Matta C, Zakany R, Musumeci G. Chondrosenescence: definition, hallmarks and potential role in the pathogenesis of osteoarthritis. Maturitas (2015) 80:237-44.

[67] Hui W, Young DA, Rowan AD, Xu X, Cawston TE, Proctor CJ. Oxidative changes and signalling pathways are pivotal in initiating age-related changes in articular cartilage. Ann Rheum Dis (2014).

[68] Hashimoto S, Ochs RL, Rosen F, Quach J, McCabe G, Solan J, et al. Chondrocyte-derived apoptotic bodies and calcification of articular cartilage. Proc Natl Acad Sci U S A (1998) 95:3094-9.

[69] Kandel R, Hurtig M, Grynpas M. Characterization of the mineral in calcified articular cartilagenous tissue formed in vitro. Tissue Eng (1999) 5:25-34.

[70] Aigner T, Hemmel M, Neureiter D, Gebhard PM, Zeiler G, Kirchner T, et al. Apoptotic cell death is not a widespread phenomenon in normal aging and osteoarthritis human articular knee cartilage: a study of proliferation, programmed cell death (apoptosis), and viability of chondrocytes in normal and osteoarthritic human knee cartilage. Arthritis Rheum (2001) 44:1304-12.

[71] Yu Y, Zheng H, Buckwalter JA, Martin JA. Single cell sorting identifies progenitor cell population from full thickness bovine articular cartilage. Osteoarthritis Cartilage (2014) 22:1318-26.

[72] Silver FH, Bradica G, Tria A. Relationship among biomechanical, biochemical, and cellular changes associated with osteoarthritis. Crit Rev Biomed Eng (2001) 29:373-91.

[73] Stoddart MJ, Bara J, Alini M. Cells and secretome - towards endogenous cell re-activation for cartilage repair. Adv Drug Deliv Rev (2015) 84:135-45.

[74] Iwamoto M, Ohta Y, Larmour C, Enomoto-Iwamoto M. Toward regeneration of articular cartilage. Birth Defects Res C Embryo Today (2013) 99:192-202.

[75] Archer CW, Morrison H, Pitsillides AA. Cellular aspects of the development of diarthrodial joints and articular cartilage. J Anat (1994) 184 (Pt 3):447-56. 34

[76] Hayes AJ, MacPherson S, Morrison H, Dowthwaite G, Archer CW. The development of articular cartilage: evidence for an appositional growth mechanism. Anat Embryol (Berl) (2001) 203:469-79.

[77] Sharma AR, Jagga S, Lee SS, Nam JS. Interplay between cartilage and subchondral bone contributing to pathogenesis of osteoarthritis. Int J Mol Sci (2013) 14:19805-30.

[78] Steinwachs MR, Guggi T, Kreuz PC. Marrow stimulation techniques. Injury (2008) 39 Suppl 1:S26-31.

[79] Goyal D, Keyhani S, Lee EH, Hui JH. Evidence-based status of microfracture technique: a systematic review of level I and II studies. Arthroscopy (2013) 29:1579-88.

[80] Solheim E, Hegna J, Inderhaug E, Oyen J, Harlem T, Strand T. Results at 10-14 years after microfracture treatment of articular cartilage defects in the knee. Knee Surg Sports Traumatol Arthrosc (2014) [Epub ahead of print].

[81] Chen H, Chevrier A, Hoemann CD, Sun J, Ouyang W, Buschmann MD. Characterization of subchondral bone repair for marrow-stimulated chondral defects and its relationship to articular cartilage resurfacing. Am J Sports Med (2011) 39:1731-40.

[82] Harris JD, Siston RA, Brophy RH, Lattermann C, Carey JL, Flanigan DC. Failures, re- operations, and complications after autologous chondrocyte implantation--a systematic review. Osteoarthritis Cartilage (2011) 19:779-91.

[83] Holtzer H, Abbott J, Lash J, Holtzer S. The loss of phenotypic traits by differentiated cells in vitro, I. dedifferentiation of cartilage cells. Proc Natl Acad Sci U S A (1960) 46:1533-42.

[84] Darling EM, Athanasiou KA. Rapid phenotypic changes in passaged articular chondrocyte subpopulations. J Orthop Res (2005) 23:425-32.

[85] Basad E, Ishaque B, Bachmann G, Sturz H, Steinmeyer J. Matrix-induced autologous chondrocyte implantation versus microfracture in the treatment of cartilage defects of the knee: a 2-year randomised study. Knee Surg Sports Traumatol Arthrosc (2010) 18:519-27. 35

[86] Niemeyer P, Pestka JM, Kreuz PC, Erggelet C, Schmal H, Suedkamp NP, et al. Characteristic complications after autologous chondrocyte implantation for cartilage defects of the knee joint. Am J Sports Med (2008) 36:2091-9.

[87] Hangody L, Feczko P, Bartha L, Bodo G, Kish G. Mosaicplasty for the treatment of articular defects of the knee and ankle. Clin Orthop Relat Res (2001):S328-36.

[88] Goyal D, Keyhani S, Goyal A, Lee EH, Hui JH, Vaziri AS. Evidence-based status of osteochondral cylinder transfer techniques: a systematic review of level I and II studies. Arthroscopy (2014) 30:497-505.

[89] Athanasiou KA, Rosenwasser MP, Buckwalter JA, Malinin TI, Mow VC. Interspecies comparisons of in situ intrinsic mechanical properties of distal femoral cartilage. J Orthop Res (1991) 9:330-40.

[90] LaPrade RF, Botker JC. Donor-site morbidity after osteochondral autograft transfer procedures. Arthroscopy (2004) 20:e69-73.

[91] Nikolaou VS, Chytas D, Babis GC. Common controversies in total knee replacement surgery: Current evidence. World J Orthop (2014) 5:460-8.

[92] Ranawat CS, Flynn WF, Jr., Saddler S, Hansraj KK, Maynard MJ. Long-term results of the total condylar knee arthroplasty. A 15-year survivorship study. Clin Orthop Relat Res (1993):94- 102.

[93] Carr AJ, Robertsson O, Graves S, Price AJ, Arden NK, Judge A, et al. Knee replacement. Lancet (2012) 379:1331-40.

[94] Lombardi AV, Jr., Berend KR, Adams JB. Why knee replacements fail in 2013: patient, surgeon, or implant? Bone Joint J (2014) 96-b:101-4.

[95] Suter LG, Paltiel AD, Rome BN, Solomon DH, Thornhill TS, Abrams SK, et al. Placing a price on medical device innovation: the example of total knee arthroplasty. PLoS One (2013) 8:e62709. 36

[96] Sampat SR, O'Connell GD, Fong JV, Alegre-Aguaron E, Ateshian GA, Hung CT. Growth factor priming of synovium-derived stem cells for cartilage tissue engineering. Tissue Eng Part A (2011) 17:2259-65.

[97] Cheuk YC, Wong MW, Lee KM, Fu SC. Use of allogeneic scaffold-free chondrocyte pellet in repair of osteochondral defect in a rabbit model. J Orthop Res (2011) 29:1343-50.

[98] Wakitani S, Goto T, Young RG, Mansour JM, Goldberg VM, Caplan AI. Repair of large full-thickness articular cartilage defects with allograft articular chondrocytes embedded in a collagen gel. Tissue Eng (1998) 4:429-44.

[99] Salai M, Ganel A, Horoszowski H. Fresh osteochondral allografts at the knee joint: good functional results in a follow-up study of more than 15 years. Arch Orthop Trauma Surg (1997) 116:423-5.

[100] Alsalameh S, Mollenhauer J, Hain N, Stock KP, Kalden JR, Burmester GR. Cellular immune response toward human articular chondrocytes. T cell reactivities against chondrocyte and fibroblast membranes in destructive joint diseases. Arthritis Rheum (1990) 33:1477-86.

[101] Brodkin KR, Garcia AJ, Levenston ME. Chondrocyte phenotypes on different extracellular matrix monolayers. Biomaterials (2004) 25:5929-38.

[102] Malpeli M, Randazzo N, Cancedda R, Dozin B. Serum-free growth medium sustains commitment of human articular chondrocyte through maintenance of Sox9 expression. Tissue Eng (2004) 10:145-55.

[103] Bonaventure J, Kadhom N, Cohen-Solal L, Ng KH, Bourguignon J, Lasselin C, et al. Reexpression of cartilage-specific genes by dedifferentiated human articular chondrocytes cultured in alginate beads. Exp Cell Res (1994) 212:97-104.

[104] Martin I, Vunjak-Novakovic G, Yang J, Langer R, Freed LE. Mammalian chondrocytes expanded in the presence of fibroblast growth factor 2 maintain the ability to differentiate and regenerate three-dimensional cartilaginous tissue. Exp Cell Res (1999) 253:681-8.

[105] Jakob M, Demarteau O, Schafer D, Hintermann B, Dick W, Heberer M, et al. Specific growth factors during the expansion and redifferentiation of adult human articular chondrocytes 37 enhance chondrogenesis and cartilaginous tissue formation in vitro. J Cell Biochem (2001) 81:368-77.

[106] Grigolo B, Lisignoli G, Piacentini A, Fiorini M, Gobbi P, Mazzotti G, et al. Evidence for redifferentiation of human chondrocytes grown on a hyaluronan-based biomaterial (HYAff 11): molecular, immunohistochemical and ultrastructural analysis. Biomaterials (2002) 23:1187-95.

[107] Tallheden T, Dennis JE, Lennon DP, Sjogren-Jansson E, Caplan AI, Lindahl A. Phenotypic plasticity of human articular chondrocytes. J Bone Joint Surg Am (2003) 85-A Suppl 2:93-100.

[108] Ahmed N, Taylor DW, Wunder J, Nagy A, Gross AE, Kandel RA. Passaged human chondrocytes accumulate extracellular matrix when induced by bovine chondrocytes. J Tissue Eng Regen Med (2010) 4:233-41.

[109] Gan L, Kandel RA. In vitro cartilage tissue formation by co-culture of primary and passaged chondrocytes. Tissue Eng (2007) 13:831-42.

[110] Taylor DW, Ahmed N, Gan L, Gross AE, Kandel RA. Proteoglycan and collagen accumulation by passaged chondrocytes can be enhanced through side-by-side culture with primary chondrocytes. Tissue Eng Part A (2010) 16:643-51.

[111] Craft AM, Ahmed N, Rockel JS, Baht GS, Alman BA, Kandel RA, et al. Specification of chondrocytes and cartilage tissues from embryonic stem cells. Development (2013) 140:2597- 610.

[112] Ratcliffe E, Glen KE, Naing MW, Williams DJ. Current status and perspectives on stem cell-based therapies undergoing clinical trials for regenerative medicine: case studies. Br Med Bull (2013) 108:73-94.

[113] Takahashi K, Yamanaka S. Induction of pluripotent stem cells from mouse embryonic and adult fibroblast cultures by defined factors. Cell (2006) 126:663-76.

[114] Takahashi K, Tanabe K, Ohnuki M, Narita M, Ichisaka T, Tomoda K, et al. Induction of pluripotent stem cells from adult human fibroblasts by defined factors. Cell (2007) 131:861-72. 38

[115] Tsumaki N, Okada M, Yamashita A. iPS cell technologies and cartilage regeneration. Bone (2015) 70:48-54.

[116] Friedenstein AJ, Piatetzky S, I. I., Petrakova KV. Osteogenesis in transplants of bone marrow cells. J Embryol Exp Morphol (1966) 16:381-90.

[117] Till JE, McCulloch EA, Siminovitch L. A stochastic model of stem cell proliferation, based on the growth of spleen colony-forming cells. Proc Natl Acad Sci U S A (1964) 51:29-36.

[118] Owen M. Marrow stromal stem cells. J Cell Sci Suppl (1988) 10:63-76.

[119] Johnstone B, Hering TM, Caplan AI, Goldberg VM, Yoo JU. In vitro chondrogenesis of bone marrow-derived mesenchymal progenitor cells. Exp Cell Res (1998) 238:265-72.

[120] Prockop DJ. Repair of tissues by adult stem/progenitor cells (MSCs): controversies, myths, and changing paradigms. Mol Ther (2009) 17:939-46.

[121] Dominici M, Le Blanc K, Mueller I, Slaper-Cortenbach I, Marini F, Krause D, et al. Minimal criteria for defining multipotent mesenchymal stromal cells. The International Society for Cellular Therapy position statement. Cytotherapy (2006) 8:315-7.

[122] Hennig T, Lorenz H, Thiel A, Goetzke K, Dickhut A, Geiger F, et al. Reduced chondrogenic potential of adipose tissue derived stromal cells correlates with an altered TGFβ receptor and BMP profile and is overcome by BMP-6. J Cell Physiol (2007) 211:682-91.

[123] Wagner W, Wein F, Seckinger A, Frankhauser M, Wirkner U, Krause U, et al. Comparative characteristics of mesenchymal stem cells from human bone marrow, adipose tissue, and umbilical cord blood. Exp Hematol (2005) 33:1402-16.

[124] Lv FJ, Tuan RS, Cheung KM, Leung VY. Concise review: the surface markers and identity of human mesenchymal stem cells. Stem Cells (2014) 32:1408-19.

[125] Monaco E, Bionaz M, Rodriguez-Zas S, Hurley WL, Wheeler MB. Transcriptomics comparison between porcine adipose and bone marrow mesenchymal stem cells during in vitro osteogenic and adipogenic differentiation. PLoS One (2012) 7:e32481. 39

[126] Shirasawa S, Sekiya I, Sakaguchi Y, Yagishita K, Ichinose S, Muneta T. In vitro chondrogenesis of human synovium-derived mesenchymal stem cells: optimal condition and comparison with bone marrow-derived cells. J Cell Biochem (2006) 97:84-97.

[127] Huang JI, Kazmi N, Durbhakula MM, Hering TM, Yoo JU, Johnstone B. Chondrogenic potential of progenitor cells derived from human bone marrow and adipose tissue: a patient- matched comparison. J Orthop Res (2005) 23:1383-9.

[128] Bernardo ME, Emons JA, Karperien M, Nauta AJ, Willemze R, Roelofs H, et al. Human mesenchymal stem cells derived from bone marrow display a better chondrogenic differentiation compared with other sources. Connect Tissue Res (2007) 48:132-40.

[129] Seda Tigli R, Ghosh S, Laha MM, Shevde NK, Daheron L, Gimble J, et al. Comparative chondrogenesis of human cell sources in 3D scaffolds. J Tissue Eng Regen Med (2009) 3:348-60.

[130] Rasini V, Dominici M, Kluba T, Siegel G, Lusenti G, Northoff H, et al. Mesenchymal stromal/stem cells markers in the human bone marrow. Cytotherapy (2013) 15:292-306.

[131] Nasef A, Ashammakhi N, Fouillard L. Immunomodulatory effect of mesenchymal stromal cells: possible mechanisms. Regen Med (2008) 3:531-46.

[132] Rasmusson I, Ringden O, Sundberg B, Le Blanc K. Mesenchymal stem cells inhibit the formation of cytotoxic T lymphocytes, but not activated cytotoxic T lymphocytes or natural killer cells. Transplantation (2003) 76:1208-13.

[133] Tian Y, Deng YB, Huang YJ, Wang Y. Bone marrow-derived mesenchymal stem cells decrease acute graft-versus-host disease after allogeneic hematopoietic stem cells transplantation. Immunol Invest (2008) 37:29-42.

[134] Mei SH, Haitsma JJ, Dos Santos CC, Deng Y, Lai PF, Slutsky AS, et al. Mesenchymal stem cells reduce inflammation while enhancing bacterial clearance and improving survival in sepsis. Am J Respir Crit Care Med (2010) 182:1047-57.

[135] Balan A, Lucchini G, Schmidt S, Schneider A, Tramsen L, Kuci S, et al. Mesenchymal stromal cells in the antimicrobial host response of hematopoietic stem cell recipients with graft- versus-host disease-friends or foes? Leukemia (2014). 40

[136] Spees JL, Gregory CA, Singh H, Tucker HA, Peister A, Lynch PJ, et al. Internalized antigens must be removed to prepare hypoimmunogenic mesenchymal stem cells for cell and gene therapy. Mol Ther (2004) 9:747-56.

[137] Horwitz EM, Gordon PL, Koo WK, Marx JC, Neel MD, McNall RY, et al. Isolated allogeneic bone marrow-derived mesenchymal cells engraft and stimulate growth in children with osteogenesis imperfecta: Implications for cell therapy of bone. Proc Natl Acad Sci U S A (2002) 99:8932-7.

[138] Liu H, Kemeny DM, Heng BC, Ouyang HW, Melendez AJ, Cao T. The immunogenicity and immunomodulatory function of osteogenic cells differentiated from mesenchymal stem cells. J Immunol (2006) 176:2864-71.

[139] Chen X, McClurg A, Zhou GQ, McCaigue M, Armstrong MA, Li G. Chondrogenic differentiation alters the immunosuppressive property of bone marrow-derived mesenchymal stem cells, and the effect is partially due to the upregulated expression of B7 molecules. Stem Cells (2007) 25:364-70.

[140] Ryan AE, Lohan P, O'Flynn L, Treacy O, Chen X, Coleman C, et al. Chondrogenic differentiation increases antidonor immune response to allogeneic mesenchymal stem cell transplantation. Mol Ther (2014) 22:655-67.

[141] Martin I, Ireland H, Baldomero H, Passweg J. The survey on cellular and engineered tissue therapies in Europe in 2012. Tissue Eng Part A (2014).

[142] Jung S, Sen A, Rosenberg L, Behie LA. Identification of growth and attachment factors for the serum-free isolation and expansion of human mesenchymal stromal cells. Cytotherapy (2010) 12:637-57.

[143] Steck E, Fischer J, Lorenz H, Gotterbarm T, Jung M, Richter W. Mesenchymal stem cell differentiation in an experimental cartilage defect: restriction of hypertrophy to bone-close neocartilage. Stem Cells Dev (2009) 18:969-78. 41

[144] Mackay AM, Beck SC, Murphy JM, Barry FP, Chichester CO, Pittenger MF. Chondrogenic differentiation of cultured human mesenchymal stem cells from marrow. Tissue Eng (1998) 4:415-28.

[145] Bosnakovski D, Mizuno M, Kim G, Ishiguro T, Okumura M, Iwanaga T, et al. Chondrogenic differentiation of bovine bone marrow mesenchymal stem cells in pellet cultural system. Exp Hematol (2004) 32:502-9.

[146] Bao G, Suresh S. Cell and molecular mechanics of biological materials. Nat Mater (2003) 2:715-25.

[147] McBeath R, Pirone DM, Nelson CM, Bhadriraju K, Chen CS. Cell shape, cytoskeletal tension, and RhoA regulate stem cell lineage commitment. Dev Cell (2004) 6:483-95.

[148] Engler AJ, Sen S, Sweeney HL, Discher DE. Matrix elasticity directs stem cell lineage specification. Cell (2006) 126:677-89.

[149] Mueller MB, Fischer M, Zellner J, Berner A, Dienstknecht T, Kujat R, et al. Effect of parathyroid hormone-related protein in an in vitro hypertrophy model for mesenchymal stem cell chondrogenesis. Int Orthop (2013) 37:945-51.

[150] Mueller MB, Tuan RS. Functional characterization of hypertrophy in chondrogenesis of human mesenchymal stem cells. Arthritis Rheum (2008) 58:1377-88.

[151] Kim YJ, Kim HJ, Im GI. PTHrP promotes chondrogenesis and suppresses hypertrophy from both bone marrow-derived and adipose tissue-derived MSCs. Biochem Biophys Res Commun (2008) 373:104-8.

[152] Mueller MB, Fischer M, Zellner J, Berner A, Dienstknecht T, Prantl L, et al. Hypertrophy in mesenchymal stem cell chondrogenesis: effect of TGF-β isoforms and chondrogenic conditioning. Cells Tissues Organs (2010) 192:158-66.

[153] Kafienah W, Mistry S, Dickinson SC, Sims TJ, Learmonth I, Hollander AP. Three- dimensional cartilage tissue engineering using adult stem cells from osteoarthritis patients. Arthritis Rheum (2007) 56:177-87. 42

[154] Provot S, Zinyk D, Gunes Y, Kathri R, Le Q, Kronenberg HM, et al. Hif-1α regulates differentiation of limb bud mesenchyme and joint development. J Cell Biol (2007) 177:451-64.

[155] Robins JC, Akeno N, Mukherjee A, Dalal RR, Aronow BJ, Koopman P, et al. Hypoxia induces chondrocyte-specific gene expression in mesenchymal cells in association with transcriptional activation of Sox9. Bone (2005) 37:313-22.

[156] Krinner A, Zscharnack M, Bader A, Drasdo D, Galle J. Impact of oxygen environment on mesenchymal stem cell expansion and chondrogenic differentiation. Cell Prolif (2009) 42:471- 84.

[157] Hirao M, Tamai N, Tsumaki N, Yoshikawa H, Myoui A. Oxygen tension regulates chondrocyte differentiation and function during endochondral ossification. J Biol Chem (2006) 281:31079-92.

[158] Markway BD, Tan GK, Brooke G, Hudson JE, Cooper-White JJ, Doran MR. Enhanced chondrogenic differentiation of human bone marrow-derived mesenchymal stem cells in low oxygen environment micropellet cultures. Cell Transplant (2010) 19:29-42.

[159] Bhumiratana S, Eton RE, Oungoulian SR, Wan LQ, Ateshian GA, Vunjak-Novakovic G. Large, stratified, and mechanically functional human cartilage grown in vitro by mesenchymal condensation. Proc Natl Acad Sci U S A (2014) 111:6940-5.

[160] Jakobsen RB, Ostrup E, Zhang X, Mikkelsen TS, Brinchmann JE. Analysis of the effects of five factors relevant to in vitro chondrogenesis of human mesenchymal stem cells using factorial design and high throughput mRNA-profiling. PLoS One (2014) 9:e96615.

[161] Dickhut A, Pelttari K, Janicki P, Wagner W, Eckstein V, Egermann M, et al. Calcification or dedifferentiation: requirement to lock mesenchymal stem cells in a desired differentiation stage. J Cell Physiol (2009) 219:219-26.

[162] Hellingman CA, Koevoet W, van Osch GJ. Can one generate stable hyaline cartilage from adult mesenchymal stem cells? A developmental approach. J Tissue Eng Regen Med (2012) 6:e1-e11. 43

[163] Kafienah W, Mistry S, Perry MJ, Politopoulou G, Hollander AP. Pharmacological regulation of adult stem cells: chondrogenesis can be induced using a synthetic inhibitor of the retinoic acid receptor. Stem Cells (2007) 25:2460-8.

[164] Zscharnack M, Hepp P, Richter R, Aigner T, Schulz R, Somerson J, et al. Repair of chronic osteochondral defects using predifferentiated mesenchymal stem cells in an ovine model. Am J Sports Med (2010) 38:1857-69.

[165] Shimomura K, Moriguchi Y, Ando W, Nansai R, Fujie H, Hart DA, et al. Osteochondral repair using a scaffold-free tissue-engineered construct derived from synovial mesenchymal stem cells and a hydroxyapatite-based artificial bone. Tissue Eng Part A (2014) 20:2291-304.

[166] Marquass B, Somerson JS, Hepp P, Aigner T, Schwan S, Bader A, et al. A novel MSC- seeded triphasic construct for the repair of osteochondral defects. J Orthop Res (2010) 28:1586- 99.

[167] Hunziker EB. Biologic repair of articular cartilage. Defect models in experimental animals and matrix requirements. Clin Orthop Relat Res (1999):S135-46.

[168] Ryu YM, Hah YS, Park BW, Kim DR, Roh GS, Kim JR, et al. Osteogenic differentiation of human periosteal-derived cells in a three-dimensional collagen scaffold. Mol Biol Rep (2011) 38:2887-94.

[169] Ng KK, Thatte HS, Spector M. Chondrogenic differentiation of adult mesenchymal stem cells and embryonic cells in collagen scaffolds. J Biomed Mater Res A (2011) 99:275-82.

[170] Qi Y, Zhao T, Xu K, Dai T, Yan W. The restoration of full-thickness cartilage defects with mesenchymal stem cells (MSCs) loaded and cross-linked bilayer collagen scaffolds on rabbit model. Mol Biol Rep (2012) 39:1231-7.

[171] Chung C, Burdick JA. Influence of three-dimensional hyaluronic acid microenvironments on mesenchymal stem cell chondrogenesis. Tissue Eng Part A (2009) 15:243-54.

[172] Matsiko A, Levingstone TJ, O'Brien FJ, Gleeson JP. Addition of hyaluronic acid improves cellular infiltration and promotes early-stage chondrogenesis in a collagen-based scaffold for cartilage tissue engineering. J Mech Behav Biomed Mater (2012) 11:41-52. 44

[173] Kerker JT, Leo AJ, Sgaglione NA. Cartilage repair: synthetics and scaffolds: basic science, surgical techniques, and clinical outcomes. Sports Med Arthrosc (2008) 16:208-16.

[174] Marijnissen WJ, van Osch GJ, Aigner J, van der Veen SW, Hollander AP, Verwoerd- Verhoef HL, et al. Alginate as a chondrocyte-delivery substance in combination with a non- woven scaffold for cartilage tissue engineering. Biomaterials (2002) 23:1511-7.

[175] Nguyen LH, Kudva AK, Saxena NS, Roy K. Engineering articular cartilage with spatially- varying matrix composition and mechanical properties from a single stem cell population using a multi-layered hydrogel. Biomaterials (2011) 32:6946-52.

[176] Kessler MW, Ackerman G, Dines JS, Grande D. Emerging technologies and fourth generation issues in cartilage repair. Sports Med Arthrosc (2008) 16:246-54.

[177] Benders KE, van Weeren PR, Badylak SF, Saris DB, Dhert WJ, Malda J. Extracellular matrix scaffolds for cartilage and bone regeneration. Trends Biotechnol (2013) 31:169-76.

[178] Badylak SF. The extracellular matrix as a biologic scaffold material. Biomaterials (2007) 28:3587-93.

[179] Crapo PM, Gilbert TW, Badylak SF. An overview of tissue and whole organ decellularization processes. Biomaterials (2011) 32:3233-43.

[180] Hill RC, Calle EA, Dzieciatkowska M, Niklason LE, Hansen KC. Quantification of extracellular matrix proteins from a rat lung scaffold to provide a molecular readout for tissue engineering. Mol Cell Proteomics (2015) 14:961-73.

[181] Moutos FT, Guilak F. Composite scaffolds for cartilage tissue engineering. Biorheology (2008) 45:501-12.

[182] Castro NJ, Hacking SA, Zhang LG. Recent progress in interfacial tissue engineering approaches for osteochondral defects. Ann Biomed Eng (2012) 40:1628-40.

[183] Dhote V, Vernerey FJ. Mathematical model of the role of degradation on matrix development in hydrogel scaffold. Biomech Model Mechanobiol (2014) 13:167-83. 45

[184] Huey DJ, Hu JC, Athanasiou KA. Unlike bone, cartilage regeneration remains elusive. Science (2012) 338:917-21.

[185] Mouw JK, Connelly JT, Wilson CG, Michael KE, Levenston ME. Dynamic compression regulates the expression and synthesis of chondrocyte-specific matrix molecules in bone marrow stromal cells. Stem Cells (2007) 25:655-63.

[186] Dore D, Quinn S, Ding C, Winzenberg T, Cicuttini F, Jones G. Subchondral bone and cartilage damage: a prospective study in older adults. Arthritis Rheum (2010) 62:1967-73.

[187] Radin EL, Rose RM. Role of subchondral bone in the initiation and progression of cartilage damage. Clin Orthop Relat Res (1986):34-40.

[188] Martin I, Miot S, Barbero A, Jakob M, Wendt D. Osteochondral tissue engineering. J Biomech (2007) 40:750-65.

[189] Tuli R, Nandi S, Li WJ, Tuli S, Huang X, Manner PA, et al. Human mesenchymal progenitor cell-based tissue engineering of a single-unit osteochondral construct. Tissue Eng (2004) 10:1169-79.

[190] Waldman SD, Grynpas MD, Pilliar RM, Kandel RA. Characterization of cartilagenous tissue formed on calcium polyphosphate substrates in vitro. J Biomed Mater Res (2002) 62:323- 30.

[191] Kandel RA, Grynpas M, Pilliar R, Lee J, Wang J, Waldman S, et al. Repair of osteochondral defects with biphasic cartilage-calcium polyphosphate constructs in a sheep model. Biomaterials (2006) 27:4120-31.

[192] Cheng HW, Luk KD, Cheung KM, Chan BP. In vitro generation of an osteochondral interface from mesenchymal stem cell-collagen microspheres. Biomaterials (2011) 32:1526-35.

[193] Dormer NH, Singh M, Wang L, Berkland CJ, Detamore MS. Osteochondral interface tissue engineering using macroscopic gradients of bioactive signals. Ann Biomed Eng (2010) 38:2167-82. 46

[194] Chen K, Ng KS, Ravi S, Goh JC, Toh SL. In vitro generation of whole osteochondral constructs using rabbit bone marrow stromal cells, employing a two-chambered co-culture well design. J Tissue Eng Regen Med (2013).

[195] Grayson WL, Bhumiratana S, Grace Chao PH, Hung CT, Vunjak-Novakovic G. Spatial regulation of human mesenchymal stem cell differentiation in engineered osteochondral constructs: effects of pre-differentiation, soluble factors and medium perfusion. Osteoarthritis Cartilage (2010) 18:714-23.

[196] Chen K, Teh TK, Ravi S, Toh SL, Goh JC. Osteochondral interface generation by rabbit bone marrow stromal cells and osteoblasts coculture. Tissue Eng Part A (2012) 18:1902-11.

[197] Kirsch T. Determinants of pathological mineralization. Curr Opin Rheumatol (2006) 18:174-80.

[198] Mann S, Archibald DD, Didymus JM, Douglas T, Heywood BR, Meldrum FC, et al. Crystallization at inorganic-organic interfaces: biominerals and biomimetic synthesis. Science (1993) 261:1286-92.

[199] Balcerzak M, Hamade E, Zhang L, Pikula S, Azzar G, Radisson J, et al. The roles of annexins and alkaline phosphatase in mineralization process. Acta Biochim Pol (2003) 50:1019- 38.

[200] Boskey AL. Matrix proteins and mineralization: an overview. Connect Tissue Res (1996) 35:357-63.

[201] Rey C, Collins B, Goehl T, Dickson IR, Glimcher MJ. The carbonate environment in bone mineral: a resolution-enhanced Fourier Transform Infrared Spectroscopy Study. Calcif Tissue Int (1989) 45:157-64.

[202] Wu M, Li YP, Zhu G, Lu Y, Wang Y, Jules J, et al. Chondrocyte-specific knockout of Cbf β reveals the indispensable function of Cbfβ in chondrocyte maturation, growth plate development and trabecular bone formation in mice. Int J Biol Sci (2014) 10:861-72. 47

[203] Kojima T, Hasegawa T, de Freitas PH, Yamamoto T, Sasaki M, Horiuchi K, et al. Histochemical aspects of the vascular invasion at the erosion zone of the epiphyseal cartilage in MMP-9-deficient mice. Biomed Res (2013) 34:119-28.

[204] Masago Y, Hosoya A, Kawasaki K, Kawano S, Nasu A, Toguchida J, et al. The molecular chaperone Hsp47 is essential for cartilage and endochondral bone formation. J Cell Sci (2012) 125:1118-28.

[205] Boskey AL, Gelb BD, Pourmand E, Kudrashov V, Doty SB, Spevak L, et al. Ablation of cathepsin k activity in the young mouse causes hypermineralization of long bone and growth plates. Calcif Tissue Int (2009) 84:229-39.

[206] Golovchenko S, Hattori T, Hartmann C, Gebhardt M, Gebhard S, Hess A, et al. Deletion of β-catenin in hypertrophic growth plate chondrocytes impairs trabecular bone formation. Bone (2013) 55:102-12.

[207] Rosati R, Horan GS, Pinero GJ, Garofalo S, Keene DR, Horton WA, et al. Normal long bone growth and development in type X collagen-null mice. Nat Genet (1994) 8:129-35.

[208] Robison R, Rosenheim AH. Calcification of hypertrophic cartilage in vitro. Biochem J (1934) 28:684-98 1.

[209] Alini M, Kofsky Y, Wu W, Pidoux I, Poole AR. In serum-free culture thyroid hormones can induce full expression of chondrocyte hypertrophy leading to matrix calcification. J Bone Miner Res (1996) 11:105-13.

[210] Boyan BD, Schwartz Z, Swain LD. In vitro studies on the regulation of endochondral ossification by vitamin D. Crit Rev Oral Biol Med (1992) 3:15-30.

[211] Li X, Schwarz EM, Zuscik MJ, Rosier RN, Ionescu AM, Puzas JE, et al. Retinoic acid stimulates chondrocyte differentiation and enhances bone morphogenetic protein effects through induction of Smad1 and Smad5. Endocrinology (2003) 144:2514-23.

[212] Nelson SR, Wolford LM, Lagow RJ, Capano PJ, Davis WL. Evaluation of new high- performance calcium polyphosphate bioceramics as bone graft materials. J Oral Maxillofac Surg (1993) 51:1363-71. 48

[213] Shanjani Y, De Croos JN, Pilliar RM, Kandel RA, Toyserkani E. Solid freeform fabrication and characterization of porous calcium polyphosphate structures for tissue engineering purposes. J Biomed Mater Res B Appl Biomater (2010) 93:510-9.

[214] Pilliar RM, Filiaggi MJ, Wells JD, Grynpas MD, Kandel RA. Porous calcium polyphosphate scaffolds for bone substitute applications -- in vitro characterization. Biomaterials (2001) 22:963-72.

[215] Grynpas MD, Pilliar RM, Kandel RA, Renlund R, Filiaggi M, Dumitriu M. Porous calcium polyphosphate scaffolds for bone substitute applications in vivo studies. Biomaterials (2002) 23:2063-70.

[216] Pilliar RM, Kandel RA, Grynpas MD, Hu Y. Porous calcium polyphosphate as load- bearing bone substitutes: in vivo study. J Biomed Mater Res B Appl Biomater (2013) 101:1-8.

[217] Shanjani Y, Hu Y, Toyserkani E, Grynpas M, Kandel RA, Pilliar RM. Solid freeform fabrication of porous calcium polyphosphate structures for bone substitute applications: in vivo studies. J Biomed Mater Res B Appl Biomater (2013) 101:972-80.

[218] Kulaev IS, Vagabov V, Kulakovskaya T. The biochemistry of inorganic polyphosphates: Wiley; 2005.

[219] Kumble KD, Kornberg A. Inorganic polyphosphate in mammalian cells and tissues. J Biol Chem (1995) 270:5818-22.

[220] Kornberg A, Rao NN, Ault-Riche D. Inorganic polyphosphate: a molecule of many functions. Annu Rev Biochem (1999) 68:89-125.

[221] Maier SK, Scherer S, Loessner MJ. Long-chain polyphosphate causes cell lysis and inhibits Bacillus cereus septum formation, which is dependent on divalent cations. Appl Environ Microbiol (1999) 65:3942-9.

[222] Lorenz B, Schroder HC. Mammalian intestinal alkaline phosphatase acts as highly active exopolyphosphatase. Biochim Biophys Acta (2001) 1547:254-61. 49

[223] Tammenkoski M, Koivula K, Cusanelli E, Zollo M, Steegborn C, Baykov AA, et al. Human metastasis regulator protein H-prune is a short-chain exopolyphosphatase. Biochemistry (2008) 47:9707-13.

[224] Lonetti A, Szijgyarto Z, Bosch D, Loss O, Azevedo C, Saiardi A. Identification of an evolutionarily conserved family of inorganic polyphosphate endopolyphosphatases. J Biol Chem (2011) 286:31966-74.

[225] Omelon S, Ariganello M, Bonucci E, Grynpas M, Nanci A. A review of phosphate mineral nucleation in biology and geobiology. Calcif Tissue Int (2013) 93:382-96.

[226] St-Pierre JP, Pilliar RM, Grynpas MD, Kandel RA. Calcification of cartilage formed in vitro on calcium polyphosphate bone substitutes is regulated by inorganic polyphosphate. Acta Biomater (2010) 6:3302-9.

[227] Allan KS, Pilliar RM, Wang J, Grynpas MD, Kandel RA. Formation of biphasic constructs containing cartilage with a calcified zone interface. Tissue Eng (2007) 13:167-77.

[228] St-Pierre JP, Gan L, Wang J, Pilliar RM, Grynpas MD, Kandel RA. The incorporation of a zone of calcified cartilage improves the interfacial shear strength between in vitro-formed cartilage and the underlying substrate. Acta Biomater (2012) 8:1603-15.

Chapter 2 Membrane Culture of Bone Marrow Stromal Cells Yields Better Tissue than Pellet Culture for Engineering Cartilage-Bone Substitute Biphasic Constructs in a Two-Step Process.

This chapter is reprinted from Tissue Engineering Part C: Methods, September 1, 2011. Whitaik David Lee, Mark B. Hurtig, Rita A. Kandel, William L. Stanford. Membrane Culture of Bone Marrow Stromal Cells Yields Better Tissue Than Pellet Culture for Engineering Cartilage-Bone Substitute Biphasic Constructs in a Two-Step Process. Volume 17, Number 9, pages 939-948. Copyright © 2011 Mary Ann Liebert, Inc.

2.1 Abstract

Our long-term goal is to treat osteochondral lesions with bioengineered biphasic constructs. We have previously demonstrated that biphasic constructs, created in vitro with primary chondrocytes harvested from healthy joints and a porous calcium polyphosphate (CPP) substrate bone substitute, could successfully repair a focal defect in sheep joints. However, primary chondrocytes are limited in supply and cannot be used in engineering constructs large enough for clinical use. Thus, we developed a robust protocol to predifferentiate sheep bone marrow-derived stromal cells to chondrocytes on collagen-coated polytetrafluoroethane membrane inserts, and harvest the chondrocytes that develop and subsequently culturing these predifferentiated cells scaffold-free on the intended articulation surface of the CPP. Chondrocytes predifferentiated on membrane culture accumulated similar matrix as those in conventional pellet culture, but expressed less Col1a1 RNA. Membrane culture predifferentiated cells gave rise to a functionally superior hyaline cartilage tissue compared to pellet culture predifferentiated cells. Studies demonstrated that 2 weeks of membrane predifferentiation culture followed by 8 weeks of biphasic construct culture was the optimal culture period at which the compressive mechanical strength and the accumulation of extracellular matrix were maximized while avoiding tissue 51 mineralization. This protocol will be used to generate implants for preclinical study to determine their ability to repair osteochondral lesions.

2.2 Introduction

Intact synovial joints are critical to painless movement and for full mobility. Damage either by trauma or disease to the articular cartilage, a dense connective tissue that covers the articulating surfaces of the bones in a synovial joint, does not heal spontaneously in adults. The avascular nature of the tissue and the tight meshwork of the extracellular matrix limit cell migration, preventing chondrocytes and other chondroprogenitor cells from migrating to the site of injury, thereby greatly reducing the tissue’s regenerative potential [1]. This damage results in pain, deformity, and limitation of mobility that drastically reduces the quality of life.

We have previously developed a method to engineer a biological substitute for damaged cartilage by reforming cartilage tissue from chondrocytes in vitro on top of calcium polyphosphate (CPP), a porous bone substitute biomaterial with a high compressive strength that exhibits excellent osseointegrative properties as demonstrated in vivo [2, 3]. These tissue- engineered cartilage-CPP biphasic constructs could replace the osteochondral plugs currently used in mosaicplasty to treat focal joint defects, eliminating the need for harvesting the plugs from healthy joints. These constructs have been characterized [4], and used successfully in an in vivo focal defect study in sheep [5] – a clinically relevant model, as their knees’ biomechanics are similar to those of humans [6]. However, in those studies, the cartilage tissue on CPP was formed using autologous articular chondrocytes harvested from healthy cartilage. One major problem limiting the clinical application of bioengineered cartilage for joint repair is identifying a source of sufficient numbers of differentiated chondrocytes to form enough articular cartilage to repair the large defects that occur in patients. Chondrocytes de-differentiate when passaged (to expand cell number) even once in monolayer culture [7-9]. A variety of approaches have been developed to circumvent this [10-18]. Re-differentiation of passaged chondrocytes by culturing cells in a three-dimensional environment allows the cells to assume a spherical morphology [14, 18, 19] but under these conditions the cartilage phenotype is not fully restored. Alternatively, human embryonic stem cells can be induced to differentiate into chondrocytes in the presence of growth factors such as BMP-6 or when grown on scaffolds, suggesting that they may be another 52 potential source of cells for tissue engineering cartilage [20-24]. However, these cells are allogeneic and thus may require immunosuppressants or other strategies to overcome histocompatibility barriers.

Mesenchymal stromal cells (MSCs) could be an alternative cell source: they can be readily obtained from autologous donors, expanded, and differentiated into chondrocytes in vitro using pellet culture or on a biomaterial [20, 25-36]. Deriving chondrocytes from bone marrow- derived MSCs (BMSCs) obviates the need to harvest healthy cartilage, and the rapid proliferation of BMSCs in vitro expedites the process for obtaining a large number of cells to create cartilage tissue.

Thus, the aim of this study was to engineer biphasic implants consisting of cartilage formed in vitro from sheep BMSCs [37-39] on and integrated with the intended articulation surface of porous CPP substrates. We found that it was necessary to employ an intermediate step in which BMSCs were cultured on cell culture inserts [40] to induce MSC differentiation to chondrocytes before seeding on the CPP. The optimal in vitro protocol was determined to yield tissue with sufficient biochemical and biomechanical properties for use in biological repair.

2.3 Material and Methods

2.3.1 Isolation and expansion of BMSCs

Bone marrow samples were aspirated from the humerus of male sheep into heparinzed Vacutainers (Becton Dickinson, Mississauga, ON, Canada). Aspirates were filtered through a cell strainer (70 µm pore size) (Becton Dickenson). The filtrate was centrifuged (300 × g) for 25 min at 4°C, and the pellet was subjected to red blood cell lysis using a solution of 0.144 M ammonium chloride (Sigma-Aldrich, Oakville, ON, Canada) in 17 mM Tris-HCl buffer, pH 7.7. The nucleated cells were washed and plated on monolayer at a cellular density of 2.5×105 cells/cm2 in expansion media (XM) composed of minimum essential media α (Invitrogen, Burlington, ON, Canada), 10% fetal bovine serum (FBS; Wisent Inc., St-Bruno, QC, Canada), 1 mM sodium pyruvate and 1× penicillin/streptomycin (Invitrogen). Non-adherent cells were discarded after 24 h. Media was changed every 3 days until the monolayer cultures were 80–90% confluent, at which the cells were enzymatically harvested with 0.05% trypsin-EDTA (Invitrogen) and re-plated at a cell density of 5.0×103 cells/cm2 in XM. These cells were again cultured until 90% confluency was attained, followed by enzymatic harvesting by trypsin-EDTA, washing and cryopreservation in 50% FBS, 40% XM and 10% DMSO (Sigma-Aldrich) until later use.

2.3.2 Chondrogenic Pre-differentiation of BMSCs

Cryopreserved BMSCs were thawed, plated in monolayer at a cell density of 7.5×103 cells/cm2 in XM and expanded to 90% confluence. The BMSCs were harvested and suspended in a defined chondrogenic media (CM) composed of high-glucose Dulbecco’s modified Eagle medium (DMEM), 1× ITS cell culture supplement (BD Biosciences, Bedford, MA, USA), 2 mM GlutaMAX (Invitrogen), 1 mM sodium pyruvate, 100 nM dexamethasone (Sigma-Aldrich), 100 µg/mL ascorbic acid 2-phosphate (Sigma-Aldrich) and 10 ng/mL transforming growth factor-β3 (R&D Systems, Minneapolis, MN, USA). The MSC were then cultured either as pellet or membrane culture to induce chondrogenesis (predifferentiation step). For pellet cultures, 5×105 cells were centrifuged in wells of a 96-well round-bottom nontissue culture-treated polypropylene plate (Corning, Corning, NY, USA) with 250 µL of CM and placed in culture in the incubator. At day 3, pellets were transferred to a 96-well tube rack and cultured in 500 µL of CM per tube. The medium was changed twice a week. For membrane cultures, 12 mm-diameter 54 cell culture insert membranes (0.2 µm pore size, Millipore, Billerica, MA, USA) were coated with human collagen type IV (Sigma-Aldrich) in 0.1 N acetic acid overnight. The membranes were then incubated at 37°C with 100 µL of FBS for 2 h. A 400 µL cell suspension containing 2.0×106 BMSCs was placed on the membrane, left for 3 h in the incubator before additional CM was added to a total volume of 2 mL. Medium changes were performed every 2–3 days and the cultures were grown for various times up to 3 weeks.

2.3.3 Cartilage-CPP biphasic construct culture

Cylindrical CPP rods of 4 mm diameter were prepared by gravity sintering 75–106 µm CPP powder particles in platinum containment tubes at 950ºC as previously described [4], and disks of 2 mm thickness were cut from these rods (Figure 2.1).

Figure 2.1: The CPP disks. Gravity sintering of CPP powder yielded porous, biodegradable material on which BMSC-derived cartilage was grown. Gross appearance (A) and the scanning electron microscopy image (B) are shown.

The disks were placed in Tygon tubing to create a well-like structure and subsequently γ-irradiated (2.5 MRad). The cartilage tissue generated by predifferentiation was digested in 0.5% w/v collagenase A (Roche Diagnostics, Indianapolis, IN, USA) in F12 media with periodic agitation for 90 min at 37°C. The cells were washed twice, and then placed on the top surface of CPP disks (2×106 cells/disk in 30 µL). The biphasic constructs were cultured in DMEM-F12

(50:50) media supplemented with L-glutamine (1:1; Invitrogen) and 5% FBS, 1 mM sodium pyruvate and 100 µg/mL ascorbic acid 2-phosphate. After day 4, the FBS concentration was increased to 20% and the media changed every other day. The tubing was removed at 1 week and the cultures harvested at either 4 or 8 weeks for analysis.

2.3.4 Histological and immunohistological evaluation

In vitro-formed tissue was removed from their substrates, either the membrane or the CPP, fixed in 10% neutral formalin buffer, and embedded in paraffin. Four-micron sections were cut and stained with hematoxylin and eosin, toluidine blue (pH 3.0), or von Kossa stain with fast red counterstain. For collagen type I immunostaining [41] paraffin-embedded sections were rehydrated and digested with 2.5 mg/mL trypsin and 25 mg/mL hyaluronidase, blocked with 20% goat serum, and incubated with antibody reactive with collagen type I (CalBioChem, La Jolla, CA, USA) overnight at 4°C. Subsequently, samples were incubated with goat anti-mouse secondary antibody labelled with Alexa Fluor 488 fluorophore (Invitrogen) and counterstained with DAPI. Collagen type II immunostaining was carried out as previously described [42]. Paraffin-embedded sections were rehydrated and digested with 10 mg/mL pepsin for 6 min at 37°C, blocked with 2% (v/v) horse serum, and incubated with an antibody reactive to collagen type II (mouse monoclonal, Labvision, Fremont, CA). Immunoreactivity was detected using biotinylated horse anti-mouse secondary antibody (Vector Laboratories, Burlington, ON, Canada), Vectastain Elite ABC kit (Vector Laboratories) and diaminobenzidine with hematoxylin counterstain.

2.3.5 Biochemical analysis

Biochemical properties of tissues were assayed as previously described [42]. Briefly, tissues were detached from their substrates and snap-frozen at –80ºC. Frozen samples were digested with 40 µg/mL papain (Sigma-Aldrich) in a buffer containing 20 mM ammonium acetate, 1 mM EDTA and 1 µM DTT for 48 h at 65°C and stored at –30°C until further analysis. The DNA content of the digest was determined by using the Hoechst 33258 dye and fluorometry with the emission wavelength of 458 nm and the excitation wavelength of 365 nm. A standard curve was generated using calf thymus DNA (Sigma-Aldrich) in PBS. The proteoglycan content of the digest was estimated by quantifying the amount of sulfated glycosaminoglycans (GAGs), using the dimethylmethlene blue dye and spectrophotometry with a wavelength of 525 nm. The standard curve for the proteoglycan content assay was generated using chondroitin sulfate (Sigma-Aldrich). Collagen content of the digest was estimated by quantifying the hydroxyproline content after acid hydrolysis at 110ºC using chloramine-T/Ehrlich’s reagent assay and spectrophotometry with a wavelength of 561 nm. The standard curve for the collagen content 56 was generated using hydroxyproline (Sigma-Aldrich). To calculate the collagen content it was assumed that hydroxyproline comprises approximately 10% of the weight of collagen.

2.3.6 Gene expression analysis

Tissues were homogenized by glass bead milling (Cole Parmer, Laval, QC, Canada). Total RNA was isolated using the Nucleospin II RNA isolation kit (Mackerey-Nagel, Duren, Germany), treated with DNase (Ambion, Austin, TX, USA) and quantified with Nanodrop (ThermoFisher, Wilmington, DE, USA). cDNA was synthesized from 500 ng of total RNA using SuperScript II (Invitrogen) and random hexamers. Quantitative PCR (qPCR) was performed using Roche LightCycler 480 (Roche Diagnostics) and a standard protocol in a reaction volume of 10 µL containing SYBR master mix (Roche Diagnostics) and 0.5 µM primers [43]. Standard curves were generated using serial dilution of cDNA to determine the efficiency of each primer pair. Products of qPCR primers were sequenced to verify the amplification of intended gene targets.

2.3.7 Stress relaxation assay for compressive modulus

The compressive modulus of cartilaginous tissue of the biphasic construct was determined using stress relaxation testing and the Mach-1 mechanical testing apparatus (BioMomentum, Laval, QC, Canada) with a 0.65 mm-diameter indenter as previously described [5]. The thickness of the cartilage tissue was estimated by measuring the thickness from the lateral aspect of the construct with a calliper. At each step, 1% strain was applied while allowing unconstrained lateral deformation, and the compressive force was allowed to relax until an equilibrium force level, defined as a change of 2 dynes/min, was reached. This was repeated 20 times, and the compressive modulus was estimated from the best-fit linear regression of stress-strain relationship containing at least 10 data points.

2.3.8 Statistical analysis

Two-way analysis of variance (ANOVA) was used to analyze the effects of culture conditions and variance among donor animals. Unbalanced two-way ANOVA with general linear tests was employed where appropriate. In all cases, outcome was attributed much more strongly to variation in culture condition than to variation among animals. Hence, biochemical and biomechanical data from various CPP culture conditions were evaluated using one-way ANOVA and Tukey post hoc testing. Significance was assigned at p < 0.05. 57

2.4 Results

2.4.1 Chondrogenic predifferentiation of BMSCs in membrane and pellet cultures

We first compared the effectiveness of membrane and pellet culture differentiation of sheep BMSCs to chondrocytes. Histological examination of 2- and 3-week cultures revealed cartilage tissues rich in extracellular matrix, the bulk of which contained cells with round morphology, whereas peripheral portions of the tissues interfacing the culture media contained cells with more flattened morphology (Figure 2.2). In membrane cultures, the distribution of cells and extracellular matrix appeared uniform through its thickness. Toluidine blue staining showed that the accumulated matrix was rich in sulfated proteoglycan, although less staining was observed at 2 weeks than 3 weeks of culture (Figure 2.2 E & F). The pellet cultures yielded a heterogeneous sphere of matrix-rich tissue in which the cells had a round morphology. However, in contrast to the membrane cultures, there was a distinct thin outer layer that was more cellular and showed weaker staining with toluidine blue suggesting that these cells were less chondrocytic (Figure 2.2 C & D).

Figure 2.2: Histological appearance of tissue derived from membrane (left) and pellet cultures (right) of BMSCs cultured for 2 and 3 weeks in defined chondrogenic media. Hematoxylin & eosin staining (H&E; A–D) showed round cell morphology and hyaline extracellular matrix. Toluidine blue staining (Tol. Blue; E–H) verified the accumulation of proteoglycans in the matrix, with different degrees of metachromasia observed between the two time points in both membrane and pellet cultures. 58

The accumulation of proteoglycan and total collagen in membrane and pellet cultures was quantified over the 3-week period (Figure 2.3). By 3 weeks, the amount of GAG and collagen in the tissues were similar in both pellet and membrane cultures when normalized to their respective DNA content. However, while membrane cultures exhibited a steadily increasing accumulation of GAG and collagen over the 3-week period, the pellet cultures exhibited a more rapid increase that reached its maximum by 2 weeks of culture and then decreased (3 weeks).

Figure 2.3: Accumulation of extracellular matrix in tissue derived from membrane and pellet cultures of BMSCs cultured for up to 3 weeks in defined chondrogenic media. Sulfated proteoglycan (A) and collagen (B) accumulation were normalized by corresponding DNA content. Tissue in both membrane and pellet cultures accumulated similar amount of matrix by 3 weeks of culture; however, while tissue in membrane cultures accumulated the matrix in a steady increasing trend over time, in pellet cultures matrix accumulation stabilized at 2 weeks. n = 6 per time point. GAG, glycosaminoglycan. Data shown as mean ± SEM. † p ≤ 0.05, * p < 0.01 between membrane and pellet culture.

Transcript levels of chondrogenic markers were assayed over a 3-week period using qPCR (Figure 2.4). Cells in membrane culture expressed a significantly lower level of Col1a1 than those in pellet cultures from day 7 onwards (Figure 2.4 A). Meanwhile, expression levels of Col2a1 and aggrecan (Agc1; Figure 2.4 B–D) increased over time up to 2 weeks in culture in both pellet and membrane culture, after which time expression levels of both began to decrease. Sox9 expression showed a biphasic peak in both cultures with levels significantly higher in pellet cultures. Expression levels of chondrocyte hypertrophic markers, Runx2 and Col10a1, also 59 increased in both culture types, suggesting that the method of culture did not significantly alter their potential to terminally differentiate as previously observed [44].

Figure 2.4: Gene expression of chondrogenic markers by tissue derived from membrane and pellet cultures of BMSCs cultured for up to 3 weeks in defined chondrogenic media. Transcript levels were assayed by qPCR with 18s rRNA as the reference gene and compared to those of native ovine chondrocytes. Cells in membrane culture expressed a lower level of Col1a1 (A) than those in pellet cultures from day 7 onward; however, only minor differences were observed in both the expression levels of other genes and the trends they displayed over time (B–F). Data shown as mean ± SEM. † p ≤ 0.05, * p < 0.01 between membrane and pellet culture. n = 6 per time point.

2.4.2 Predifferentiated cells from membrane cultures form better cartilage tissue in biphasic constructs

At first, BMSCs were seeded directly on the CPP substrate and cultured in CM to create the cartilage-CPP biphasic construct; however, this did not result in cartilage tissue formation (data not shown). Therefore, we investigated whether cells that were “predifferentiated” in membrane and pellet cultures could form tissue on the CPP substrate. After 3 weeks of differentiating sheep 60

BMSCs in pellet and membrane cultures to chondrocytes, tissues were digested using collagenase to isolate the predifferentiated cells. The cells isolated from pellet and membrane cultures were then cultured scaffold-free on the CPP substrate for 4 weeks and the resulting tissues characterized (Figure 2.5).

Figure 2.5: BMSCs predifferentiated in membrane culture yielded better cartilage on CPP than those predifferentiated in pellet culture. Cells were differentiated to chondrocytes in membrane and pellet cultures, enzymatically isolated and cultured on the CPP substrate for 4 weeks. Histology of the tissues on CPP (A) revealed that cells from both culture systems gave rise to cartilaginous tissue with similar accumulation of proteoglycans (B), but tissues made by membrane-differentiated cells accumulated more collagen type II and less collagen type I (A, lower half), they had less total collagen content (C). Stress relaxation test revealed that tissues made by membrane-differentiated cells are more mechanically competent (D). The CPP- cartilage interface is at the bottom of each histological image. Data shown as mean ± SEM. * p < 0.01. n = 8 for membrane, n = 6 for pellet. Col., collagen. 61

Histological examination showed that cartilage tissues formed by membrane predifferentiated cells yielded were comparable in thickness to those formed by pellet predifferentiated cells, but accumulated more proteoglycans and collagen type II, and less collagen type I, than the pellet predifferentiated counterpart (Figure 2.5A). However, when normalized by cellularity as reflected by the tissues’ DNA content, the difference in proteoglycan accumulation between the two types of tissues was not statistically significant (Figure 2.5B), while tissues formed by pellet pre-differentiated cells accumulated significantly more total collagen (Figure 2.5C). Stress relaxation testing showed that the tissue formed by membrane predifferentiated cells had a significantly higher compressive modulus (Figure 2.5D). This, combined with the more desirable accumulation pattern of collagen, suggested that chondrocytes predifferentiated from BMSCs in membrane culture was best suited for creating cartilage-CPP biphasic constructs than either the undifferentiated BMSCs or pellet culture-differentiated chondrocytes.

2.4.3 Optimization of biphasic construct tissue culture protocol

To optimize the membrane culture protocol for creating biphasic constructs, we first varied the duration of pre-differentiation membrane culture and histologically examined the resulting cartilage tissue formed on the CPP substrate after 3 weeks of culture. BMSCs predifferentiated for 1 week formed a thin, fibrous layer of tissue with spindle-like cell morphology, indicating the absence of cartilage tissue (Figure 2.6A). However, BMSCs predifferentiated for 2 or 3 weeks formed thick tissues rich in extracellular matrix containing cells with round morphology, suggesting that these tissues were cartilaginous (Figure 2.6 B & C, data not shown).

Subsequently, we varied the duration of post-differentiation culture on CPP and characterized the cartilage tissue histologically, biochemically and mechanically. As the difference between tissues produced by 2-week and 3-week predifferentiated BMSCs was not clear, use of cells from both predifferentiation conditions were also explored. All culture combinations yielded tissues that accumulated extracellular matrix rich in proteoglycan and collagen, which appeared to increase in thickness with time in culture (Figure 2.7A). Within the tissue, two layers could be observed: a top layer sparsely populated with cells and exhibiting weaker toluidine blue and stronger collagen type II staining, whereas the bottom layer was 62

Figure 2.6. BMSCs must be pre-differentiated for at least 2 weeks in membrane culture to form tissue on CPP after isolation. BMSCs were pre-differentiated in membrane cultures at various length of time, enzymatically isolated and cultured on the CPP substrate for 3 weeks. BMSCs pre-differentiated for 1 week yielded fibrous tissue (C); however, cells pre-differentiated for 2 or 3 weeks yielded cartilage tissue on CPP. H&E = hematoxylin & eosin.

densely populated with cells and exhibited strong toluidine blue and weaker collagen type II staining. Of the different conditions, tissues derived from cells predifferentiated for 2 weeks and cultured on CPP for 8 weeks (Figure 2.7 A, second column) appeared to be the best condition because the least accumulation of collagen type I was observed among them (Figure 2.7 A, 3rd row). The cartilage tissues cultured under this condition also had the highest proteoglycan and total collagen accumulation (Figure 2.7 C & D), as well as the highest compressive elastic modulus and tissue thickness (Figure 2.7 E & F), even when compared with the tissues formed by pellet pre-differentiated cells (Figure 2.7 B–D, data not shown). In particular, the thickness of the cartilage tissues cultured under this condition approached the thickness of sheep native cartilage measured from the femoral condyle (Figure 2.7 F; p = 0.032). Given a long in vitro culture period and the expression of Runx2 and Col10a1 in predifferentiation cultures (Figure 2.4 E & F), we confirmed that mineralization, a consequence of terminal differentiation [45], did not occur at the end of the 8-week culture period, by staining the tissues with Von Kossa, as no mineralization was detected (Figure 2.7 B). 63

Figure 2.7. Optimal cartilage tissue was obtained by predifferentiating the BMSCs in membrane culture for 2 weeks and culturing the differentiated cells on CPP for 8 weeks. To optimize the culture conditions, BMSCs were predifferentiated on membranes for 2 or 3 weeks, cells isolated and cultured on CPP for 4 or 8 weeks. Histology (A) revealed tissues that accumulated proteoglycans and collagen type II. Minimum accumulation of collagen type I was observed in tissues derived from cells predifferentiated for 2 weeks and cultured on CPP for 8 weeks (A, second column). Von kossa staining of cartilage tissue (B) revealed no evidence of mineralization. Quantification of sulfated proteoglycan (C) and collagen (D) accumulation, as well as the compressive elastic moduli (E) and tissue thickness (F) all confirmed that the aforementioned 2 week-8 week culture yielded the most biochemically and mechanically optimum tissue. The cartilage-CPP interface is at the bottom of each histological image. Data shown as mean ± SEM. † p ≤ 0.05, * p < 0.01. n as indicated.

2.5 Discussion

We have successfully engineered cartilage-CPP biphasic constructs in vitro using sheep BMSCs by predifferentiating them to chondrocytes in high-density membrane culture. The cells predifferentiated in membrane cultures yield better cartilaginous tissue on CPP than those grown in conventional pellet cultures, and the length of time in predifferentiation and CPP culture affects the quality of cartilage tissue formed on the CPP substrate. Accumulation of extracellular matrix and the resulting compressive mechanical strength were maximized when BMSCs were predifferentiated in membrane culture for 2 weeks, isolated, and then cultured on CPP for an additional 8 weeks.

We noted that although membrane cultures and pellet cultures of BMSCs both yielded tissues with comparable matrix accumulation and gene expression profiles by 3 weeks of culture, the predifferentiated chondrocytes isolated from these tissues subsequently formed different cartilage tissues on the CPP substrate under the same culture condition. Chondrogenesis of BMSCs are conventionally carried out in micromass or pellet cultures to mimic in vitro the cellular microenvironment of condensed mesenchyme in skeletogenesis [26-28]. While previous studies comparing alternate culture methods to conventional pellet culture [35, 40] contrasted the gene expression of the cells and the matrix they accumulated while being differentiated in different culture methods, we performed an additional step to isolate these cells and compare the cartilaginous matrix they accumulate in a subsequent culture. In doing so, we observed that predifferentiated cells isolated from the membrane culture formed a more collagen type II-rich, mechanically competent cartilage tissue than predifferentiated cells from the pellet culture: this was an unexpected observation, given that we found the tissues formed by differentiating BMSCs in membrane cultures and pellet cultures were comparable.

The cartilage tissue generated with our BMSC-derived chondrocytes did not mineralize despite the extended length of culture. Although BMSCs are well known for their chondrogenic potential, their tendency to mineralize has limited their use clinically for joint repair [44, 45]. In our protocol, no mineralized tissue developed even though cells were cultured using predifferentiated cells on CPP for up to 8 weeks in serum-supplemented media in the absence of pro-differentiation growth factors. This is not likely to be a culture artefact as previous work in our laboratory showed that deep-zone chondrocytes cultured on CPP will form a zone of 65 calcified cartilage [46]. Further, mineralization can be induced in the tissues formed by predifferentiated BMSCs on CPP under the conditions described in this study if β-glycerophosphate is added to the culture media (data not shown). This is consistent with the in vitro hypertrophy model previously described [47]. Importantly when tissue-engineered cartilage formed by BMSCs was grafted orthotopically, Zscharnack et al. showed that these tissues do not undergo mineralization, suggesting that if the tissue does not mineralize in vitro it is unlikely to mineralize in vivo [48].

Although there still exists a disparity between the quality of the tissue formed in vitro with the currently developed protocol and the native articular cartilage, we have previously demonstrated that when placed in the in vivo environment, the biochemical and biomechanical properties of tissue-engineered cartilage will be improved [5]. This suggests that implantation of cartilage with properties less robust than native cartilage may still be sufficient; however, in vivo analyses will be required to confirm this. In contrast to scaffold-based cartilage tissue engineering strategies [29, 49, 50], the cartilage of the construct is free of artificial scaffolds as the tissue was formed by culturing a layer of cells on top of the CPP substrate.

Interestingly, while attempts to induce chondrogenesis of BMSCs directly on the CPP substrate failed, BMSCs predifferentiated to chondrocytes using either culture methods could readily establish tissues, albeit of different quality. The microenvironment of the substrate has a profound influence on the cells as has been shown for chondrocytes grown on biomaterials [51- 53]. During culture, the CPP substrate is known to release polyphosphates into the culture as it degrades hydrolytically [3]. It is not yet clear how they would affect the cells. It has been suggested that undifferentiated MSCs exposed to polyphosphates can influence the differentiation process in vitro [54]. Furthermore it may be that this release inhibits cartilage mineralization as we have shown previously in cartilage formed by deep-zone chondrocytes [46].

In summary, a method was developed that used BMSC-derived chondrocytes to substitute for articular chondrocytes in engineering a cartilage-CPP biphasic construct. This approach eliminates one of the limitations preventing clinical application of biphasic constructs. Preclinical studies are required to determine the efficacy of these constructs to repair osteochondral lesions. 66

2.6 Author Contributions The authors of the study were Whitaik David Lee, Mark B. Hurtig, Rita A. Kandel and William L. Stanford.

• Study conception and design: WDL, WLS, RAK. • Provision of study materials: MBH, WLS, RAK. • Analysis and interpretation of data: WDL, WLS, RAK. • Drafting of article: WDL, WLS, RAK. • Critical revision of the article: WDL, MBH, WLS, RAK. • Final approval of the article: WDL, WLS, RAK.

2.7 Acknowledgements

We thank Dr. Bob Pilliar for preparing the CPP substrates and also Dr. Jian Wang and Cheryl Cui for technical assistance. The research was supported by a CIHR Team grant to R.A.K. and a CIHR operating grant to W.L.S. (FRN 62788). W.D.L. is a recipient of the Ontario Graduate Scholarship and the Ontario Graduate Scholarship for Science and Technology. W.L.S. is supported by a Canadian Research Chair.

2.8 References

[1] Mankin HJ. The response of articular cartilage to mechanical injury. J Bone Joint Surg Am (1982) 64:460-6.

[2] Grynpas MD, Pilliar RM, Kandel RA, Renlund R, Filiaggi M, Dumitriu M. Porous calcium polyphosphate scaffolds for bone substitute applications in vivo studies. Biomaterials (2002) 23:2063-70.

[3] Pilliar RM, Filiaggi MJ, Wells JD, Grynpas MD, Kandel RA. Porous calcium polyphosphate scaffolds for bone substitute applications -- in vitro characterization. Biomaterials (2001) 22:963-72. 67

[4] Waldman SD, Grynpas MD, Pilliar RM, Kandel RA. Characterization of cartilagenous tissue formed on calcium polyphosphate substrates in vitro. J Biomed Mater Res (2002) 62:323-30.

[5] Kandel RA, Grynpas M, Pilliar R, Lee J, Wang J, Waldman S, et al. Repair of osteochondral defects with biphasic cartilage-calcium polyphosphate constructs in a sheep model. Biomaterials (2006) 27:4120-31.

[6] Martini L, Fini M, Giavaresi G, Giardino R. Sheep model in orthopedic research: a literature review. Comp Med (2001) 51:292-9.

[7] Holtzer H, Abbott J, Lash J, Holtzer S. The loss of phenotypic traits by differentiated cells in vitro, I. dedifferentiation of cartilage cells. Proc Natl Acad Sci U S A (1960) 46:1533-42.

[8] Dell'Accio F, De Bari C, Luyten FP. Microenvironment and phenotypic stability specify tissue formation by human articular cartilage-derived cells in vivo. Exp Cell Res (2003) 287:16- 27.

[9] Darling EM, Athanasiou KA. Rapid phenotypic changes in passaged articular chondrocyte subpopulations. J Orthop Res (2005) 23:425-32.

[10] Mandl EW, van der Veen SW, Verhaar JA, van Osch GJ. Multiplication of human chondrocytes with low seeding densities accelerates cell yield without losing redifferentiation capacity. Tissue Eng (2004) 10:109-18.

[11] Goldberg AJ, Lee DA, Bader DL, Bentley G. Autologous chondrocyte implantation. Culture in a TGF-β-containing medium enhances the re-expression of a chondrocytic phenotype in passaged human chondrocytes in pellet culture. J Bone Joint Surg Br (2005) 87:128-34.

[12] Mandl EW, van der Veen SW, Verhaar JA, van Osch GJ. Serum-free medium supplemented with high-concentration FGF2 for cell expansion culture of human ear chondrocytes promotes redifferentiation capacity. Tissue Eng (2002) 8:573-80.

[13] Jakob M, Demarteau O, Schafer D, Hintermann B, Dick W, Heberer M, et al. Specific growth factors during the expansion and redifferentiation of adult human articular chondrocytes enhance chondrogenesis and cartilaginous tissue formation in vitro. J Cell Biochem (2001) 81:368-77. 68

[14] Grigolo B, Lisignoli G, Piacentini A, Fiorini M, Gobbi P, Mazzotti G, et al. Evidence for redifferentiation of human chondrocytes grown on a hyaluronan-based biomaterial (HYAff 11): molecular, immunohistochemical and ultrastructural analysis. Biomaterials (2002) 23:1187-95.

[15] Yaeger PC, Masi TL, de Ortiz JL, Binette F, Tubo R, McPherson JM. Synergistic action of transforming growth factor-β and insulin-like growth factor-I induces expression of type II collagen and aggrecan genes in adult human articular chondrocytes. Exp Cell Res (1997) 237:318-25.

[16] Chaipinyo K, Oakes BW, van Damme MP. Effects of growth factors on cell proliferation and matrix synthesis of low-density, primary bovine chondrocytes cultured in collagen I gels. J Orthop Res (2002) 20:1070-8.

[17] Martin I, Vunjak-Novakovic G, Yang J, Langer R, Freed LE. Mammalian chondrocytes expanded in the presence of fibroblast growth factor 2 maintain the ability to differentiate and regenerate three-dimensional cartilaginous tissue. Exp Cell Res (1999) 253:681-8.

[18] Bonaventure J, Kadhom N, Cohen-Solal L, Ng KH, Bourguignon J, Lasselin C, et al. Reexpression of cartilage-specific genes by dedifferentiated human articular chondrocytes cultured in alginate beads. Exp Cell Res (1994) 212:97-104.

[19] Malda J, Kreijveld E, Temenoff JS, van Blitterswijk CA, Riesle J. Expansion of human nasal chondrocytes on macroporous microcarriers enhances redifferentiation. Biomaterials (2003) 24:5153-61.

[20] Seda Tigli R, Ghosh S, Laha MM, Shevde NK, Daheron L, Gimble J, et al. Comparative chondrogenesis of human cell sources in 3D scaffolds. J Tissue Eng Regen Med (2009) 3:348-60.

[21] Bigdeli N, Karlsson C, Strehl R, Concaro S, Hyllner J, Lindahl A. Coculture of human embryonic stem cells and human articular chondrocytes results in significantly altered phenotype and improved chondrogenic differentiation. Stem Cells (2009) 27:1812-21.

[22] Toh WS, Guo XM, Choo AB, Lu K, Lee EH, Cao T. Differentiation and enrichment of expandable chondrogenic cells from human embryonic stem cells in vitro. J Cell Mol Med (2009) 13:3570-90. 69

[23] Jukes JM, Moroni L, van Blitterswijk CA, de Boer J. Critical Steps toward a tissue- engineered cartilage implant using embryonic stem cells. Tissue Eng Part A (2008) 14:135-47.

[24] Hwang NS, Elisseeff J. Application of stem cells for articular cartilage regeneration. J Knee Surg (2009) 22:60-71.

[25] Schreml S, Babilas P, Fruth S, Orso E, Schmitz G, Mueller MB, et al. Harvesting human adipose tissue-derived adult stem cells: resection versus liposuction. Cytotherapy (2009) 11:947- 57.

[26] Steinert AF, Proffen B, Kunz M, Hendrich C, Ghivizzani SC, Noth U, et al. Hypertrophy is induced during the in vitro chondrogenic differentiation of human mesenchymal stem cells by bone morphogenetic protein-2 and bone morphogenetic protein-4 gene transfer. Arthritis Res Ther (2009) 11:R148.

[27] Weber C, Gokorsch S, Czermak P. Expansion and chondrogenic differentiation of human mesenchymal stem cells. Int J Artif Organs (2007) 30:611-8.

[28] Berry L, Grant ME, McClure J, Rooney P. Bone-marrow-derived chondrogenesis in vitro. J Cell Sci (1992) 101 (Pt 2):333-42.

[29] Vinatier C, Bouffi C, Merceron C, Gordeladze J, Brondello JM, Jorgensen C, et al. Cartilage tissue engineering: towards a biomaterial-assisted mesenchymal stem cell therapy. Curr Stem Cell Res Ther (2009) 4:318-29.

[30] Satija NK, Singh VK, Verma YK, Gupta P, Sharma S, Afrin F, et al. Mesenchymal stem cell-based therapy: a new paradigm in regenerative medicine. J Cell Mol Med (2009) 13:4385- 402.

[31] Kafienah W, Mistry S, Dickinson SC, Sims TJ, Learmonth I, Hollander AP. Three- dimensional cartilage tissue engineering using adult stem cells from osteoarthritis patients. Arthritis Rheum (2007) 56:177-87.

[32] Krinner A, Zscharnack M, Bader A, Drasdo D, Galle J. Impact of oxygen environment on mesenchymal stem cell expansion and chondrogenic differentiation. Cell Prolif (2009) 42:471- 84. 70

[33] Ahmed N, Stanford WL, Kandel RA. Mesenchymal stem and progenitor cells for cartilage repair. Skeletal Radiol (2007) 36:909-12.

[34] Matsuda C, Takagi M, Hattori T, Wakitani S, Yoshida T. Differentiation of Human Bone Marrow Mesenchymal Stem Cells to Chondrocytes for Construction of Three-dimensional Cartilage Tissue. Cytotechnology (2005) 47:11-7.

[35] Markway BD, Tan GK, Brooke G, Hudson JE, Cooper-White JJ, Doran MR. Enhanced chondrogenic differentiation of human bone marrow-derived mesenchymal stem cells in low oxygen environment micropellet cultures. Cell Transplant (2010) 19:29-42.

[36] Dominici M, Le Blanc K, Mueller I, Slaper-Cortenbach I, Marini F, Krause D, et al. Minimal criteria for defining multipotent mesenchymal stromal cells. The International Society for Cellular Therapy position statement. Cytotherapy (2006) 8:315-7.

[37] McCarty RC, Gronthos S, Zannettino AC, Foster BK, Xian CJ. Characterisation and developmental potential of ovine bone marrow derived mesenchymal stem cells. J Cell Physiol (2009) 219:324-33.

[38] Mrugala D, Bony C, Neves N, Caillot L, Fabre S, Moukoko D, et al. Phenotypic and functional characterisation of ovine mesenchymal stem cells: application to a cartilage defect model. Ann Rheum Dis (2008) 67:288-95.

[39] Rentsch C, Hess R, Rentsch B, Hofmann A, Manthey S, Scharnweber D, et al. Ovine bone marrow mesenchymal stem cells: isolation and characterization of the cells and their osteogenic differentiation potential on embroidered and surface-modified polycaprolactone-co-lactide scaffolds. In Vitro Cell Dev Biol Anim (2010) 46:624-34.

[40] Murdoch AD, Grady LM, Ablett MP, Katopodi T, Meadows RS, Hardingham TE. Chondrogenic differentiation of human bone marrow stem cells in transwell cultures: generation of scaffold-free cartilage. Stem Cells (2007) 25:2786-96.

[41] Chevrier A, Nelea M, Hurtig MB, Hoemann CD, Buschmann MD. Meniscus structure in human, sheep, and rabbit for animal models of meniscus repair. J Orthop Res (2009) 27:1197- 203. 71

[42] Taylor DW, Ahmed N, Gan L, Gross AE, Kandel RA. Proteoglycan and collagen accumulation by passaged chondrocytes can be enhanced through side-by-side culture with primary chondrocytes. Tissue Eng Part A (2010) 16:643-51.

[43] Walker E, Ohishi M, Davey RE, Zhang W, Cassar PA, Tanaka TS, et al. Prediction and testing of novel transcriptional networks regulating embryonic stem cell self-renewal and commitment. Cell Stem Cell (2007) 1:71-86.

[44] Pelttari K, Winter A, Steck E, Goetzke K, Hennig T, Ochs BG, et al. Premature induction of hypertrophy during in vitro chondrogenesis of human mesenchymal stem cells correlates with calcification and vascular invasion after ectopic transplantation in SCID mice. Arthritis Rheum (2006) 54:3254-66.

[45] Dickhut A, Pelttari K, Janicki P, Wagner W, Eckstein V, Egermann M, et al. Calcification or dedifferentiation: requirement to lock mesenchymal stem cells in a desired differentiation stage. J Cell Physiol (2009) 219:219-26.

[46] Allan KS, Pilliar RM, Wang J, Grynpas MD, Kandel RA. Formation of biphasic constructs containing cartilage with a calcified zone interface. Tissue Eng (2007) 13:167-77.

[47] Mueller MB, Tuan RS. Functional characterization of hypertrophy in chondrogenesis of human mesenchymal stem cells. Arthritis Rheum (2008) 58:1377-88.

[48] Zscharnack M, Hepp P, Richter R, Aigner T, Schulz R, Somerson J, et al. Repair of chronic osteochondral defects using predifferentiated mesenchymal stem cells in an ovine model. Am J Sports Med (2010) 38:1857-69.

[49] Swieszkowski W, Tuan BH, Kurzydlowski KJ, Hutmacher DW. Repair and regeneration of osteochondral defects in the articular joints. Biomol Eng (2007) 24:489-95.

[50] Im GI, Ahn JH, Kim SY, Choi BS, Lee SW. A hyaluronate-atelocollagen/β-tricalcium phosphate-hydroxyapatite biphasic scaffold for the repair of osteochondral defects: a porcine study. Tissue Eng Part A (2010) 16:1189-200. 72

[51] Barbero A, Grogan SP, Mainil-Varlet P, Martin I. Expansion on specific substrates regulates the phenotype and differentiation capacity of human articular chondrocytes. J Cell Biochem (2006) 98:1140-9.

[52] Kasten A, Muller P, Bulnheim U, Groll J, Bruellhoff K, Beck U, et al. Mechanical integrin stress and magnetic forces induce biological responses in mesenchymal stem cells which depend on environmental factors. J Cell Biochem (2010) 111:1586-97.

[53] Woodfield TB, Miot S, Martin I, van Blitterswijk CA, Riesle J. The regulation of expanded human nasal chondrocyte re-differentiation capacity by substrate composition and gas plasma surface modification. Biomaterials (2006) 27:1043-53.

[54] Kawazoe Y, Katoh S, Onodera Y, Kohgo T, Shindoh M, Shiba T. Activation of the FGF signaling pathway and subsequent induction of mesenchymal stem cell differentiation by inorganic polyphosphate. Int J Biol Sci (2008) 4:37-47.

Chapter 3 Sol gel-derived hydroxyapatite films over porous calcium polyphosphate substrates for improved tissue engineering of osteochondral-like constructs

This chapter is to be submitted to Acta Biomaterialia.

3.1 Abstract

Integration of in vitro-formed cartilage on a suitable substrate to form tissue-engineered implants for osteochondral defect repair is a considerable challenge. In healthy cartilage, a zone of calcified cartilage (ZCC) provides an intermediary for mechanical force transfer from soft to hard tissue, as well as an effective interlocking structure to better resist interfacial shear forces. We have developed biphasic constructs that consist of scaffold-free cartilage tissue grown in vitro on, and interdigitated with, porous calcium polyphosphate (CPP) substrates. However, as CPP degrades, it releases inorganic polyphosphates (polyP) that can inhibit local mineralization, thereby preventing the formation of a ZCC at the interface. To overcome this, we hypothesized that thin calcium hydroxyapatite films, formed using either inorganic or organic sol-gel processing methods, would act as barrier coating on CPP to inhibit polyP release. Both types of coating supported the formation of ZCC in direct contact with the substrate. Tissues formed on coated substrates accumulated comparable quantities of extracellular matrix and mineral, but tissues formed on organic sol-gel (OSG)-coated substrates accumulated less polyP than tissues formed on inorganic sol-gel (ISG)-coated substrates. Constructs formed with OSG-coated CPP substrates had greater interfacial shear strength than those formed with ISG-coated and non- coated substrates. These results suggest that the OSG coating method can be used to improve the mechanical integrity of tissue-engineered constructs formed on porous CPP substrates.

3.2 Introduction

Articular cartilage covers the articulating surfaces of bones in a joint. Joint injury or disease can damage cartilage, leading to pain and limited mobility. Cartilage has little, if any, intrinsic capacity for regeneration; therefore, tissue engineering has been widely viewed as a promising approach to replace damaged tissues [1]. Since joint loading puts articular cartilage under mechanical stress, strong integration of tissue-engineered cartilage to its underlying substrate is critical for successful repair.

In healthy cartilage, a zone of calcified cartilage (ZCC) exists at the articular cartilage- subchondral bone interface. This ZCC serves as a transition region for force transmission between soft tissue (cartilage) and hard tissue (subchondral bone), laterally dissipating forces. ZCC also enhances fixation between two tissues by mechanical interdigitation at the interface region, thereby providing improved resistance to applied shearing forces [2]. We have developed biphasic constructs that consisted of scaffold-free cartilage tissue formed on and integrated to a porous bone substitute material, calcium polyphosphate (CPP) [3, 4]. The porous CPP substrate serves as a replacement for subchondral bone, which is often abnormal in osteoarthritis [5, 6] and also susceptible to damage in symptomatic joint injuries [7, 8]. The substrate is also a means of fixation between tissue-engineered cartilage and native bone, which would be achieved by bone ingrowth into the porous CPP in the subchondral region and interdigitation of cartilage and the porous CPP substrate at the interface. Incorporating a ZCC at this interface would enhance its resistance to shearing forces. However, CPP is biodegradable, resulting in the release of inorganic polyphosphates (polyP) as a degradation by-product [9]. PolyP is a known inhibitor of apatite crystal growth, and biphasic constructs grown under mineralizing conditions form a ZCC in close proximity to, but separated from, the CPP substrate by a distinct non-mineralized layer [10]. Thus, a method to limit the polyP release must be devised.

Sol-gel thin film processing is a non-line-of-sight method, suitable for coating complex surfaces such as those presented by porous CPP. Calcium phosphate sol can be synthesized from either organic or inorganic precursors to deposit a crack-free, thin calcium hydroxyapatite (HAp) coating [11]. In a previous study, porous CPP substrates were coated with HAp using the inorganic sol-gel (ISG) method, in which the coating was intended to serve as a barrier to inhibit polyP release [12]. This approach generated constructs with a ZCC forming in close proximity to

75 the substrate. However, mechanical testing did not demonstrate increased interfacial shear strength of the samples compared to constructs with a ZCC formed on CPP substrates without the HAp barrier coating. A comparative study of coating films formed using the ISG method and the organic sol-gel (OSG) method on porous-surfaced titanium implants showed that the OSG- derived coating appeared less nanoporous than the ISG-derived coating [11]. Therefore, we hypothesized that HAp barrier coating formed using the OSG method on porous CPP substrates would inhibit the polyP release more effectively and further increase the interfacial shear strength of the construct.

In this study, we directly compared the formation of HAp barrier coatings by either OSG or ISG thin film coatings applied to porous CPP substrates in order to create biphasic constructs with an enhanced capacity for withstanding shear forces at the cartilage-CPP interface.

3.3 Materials and Methods

3.3.1 Sol-gel processing of porous CPP disks

Two different calcium phosphate sol formulations were prepared for forming thin film coating by dip-coating of porous CPP disks as previously described [11]. To prepare an organic sol, triethyl phosphite (Sigma-Aldrich, Oakville, Ontario, Canada) was first hydrolyzed in excess ddH2O for 24 h. Calcium nitrate tetrahydrate (Sigma-Aldrich) was dissolved in ethanol and added to hydrolyzed triethyl phosphite solution at a calcium-phosphate ratio of 1:1.67. The sol was sealed, aged at 40ºC for 4 days, then diluted to 67% with ethanol and further aged in room temperature for 2 days. The pH of the resulting sol was about 2. To prepare an inorganic sol, aqueous solutions of ammonium dihydrogen phosphate (Sigma-Aldrich) and calcium nitrate were mixed in the presence of ammonium hydroxide (Sigma-Aldrich) at the same calcium- phosphate ratio of 1:1.67. The sol was sealed with a porous membrane to allow for ventilation and aged in room temperature for 7 days. After removing the supernatant, the sol was diluted to 75% with distilled water. The pH of the resulting sol was about 10.

Porous CPP disks of 4 mm diameter and 2 mm height were prepared by gravity sintering 75–150µm CPP particles as previously described [13]. Disks were pre-incubated in phosphate- buffered saline for a week. Buffer was changed every other day. Each layer of film was formed by dip coating porous CPP disks with a specific withdrawal speed, followed by annealing for 15 min at 210ºC. With the organic sol, disks were sequentially coated twice using a withdrawal speed of 20 cm/min [11] while for the inorganic sol, disks were sequentially coated eight times using a withdrawal speed of 30 cm/min [12]. After the final dip coating, disks were annealed at 500ºC for 20 min (organic sol) or 60 min (inorganic sol) and gradually cooled to room temperature. The coated disks were placed in Tygon tubing to create a well-like structure and subsequently sterilized by γ-irradiation (2.5 MRad).

3.3.2 Characterization of HAp coating on CPP disks

Sol-gel-formed coatings on porous CPP disks were characterized as previously described [12]. The outer and fracture surfaces of non-coated and coated CPP substrates were sputter-coated with gold (Desk II, Denton Vacuum, Moorestown, NJ, USA) to make them electrically conducting for viewing by scanning electron microscopy (XL30, FEI, Portland, OR, USA). The

77 surface coverage and microstructure of films were examined using secondary electron imaging. Some coated porous CPP disks were purposely fractured in order to form exposed fracture surfaces across the coatings and micrographs of cross-sectional fractured regions were used to estimate the thickness of the coatings. The crystallographic form of the coatings was determined by electron diffraction (Tecnai 20, FEI) of portions of the thin coating (along with some inadvertently detached CPP particles) scraped from the coated disks. To survey the distribution of coating throughout the porous disks, coated disks were embedded in polymethyl methacrylate (Osteo-Bed bone embedding kit, Polysciences Inc., Warrington, PA), bisected and imaged using backscattered electron imaging (BSEI). Contrast and brightness were adjusted to show the coating and CPP at different signal intensities. To characterize the composition of coating, coated CPP disks were sonicated in 0.5 N hydrochloric acid for 20 min to selectively dissolve the coating. The dissolution products were neutralized with sodium hydroxide solution before analysis. Calcium content was determined using the o-cresolphthalein complexone assay and measuring absorbance at 570 nm. Phosphate content was determined using the heteropoly blue assay and measuring absorbance at 620 nm. Non-coated CPP disks were used as negative control.

3.3.3 Quantification of released polyP, calcium and phosphate by CPP disks

Coated and non-coated CPP disks were individually submerged in wells of a 24-well tissue culture plate containing 1 mL of calcium- and phosphate-free Tris-buffer saline buffer (TBS) and incubated at 37ºC in 5% CO2 for up to 3 weeks. The buffer was changed every 48 h. After each change, polyP, calcium and phosphate in the conditioned buffer were quantified. Calcium content was determined using the o-cresolphthalein complexone assay. Phosphate content was determined using malachite green assay and measuring the absorbance at 620 nm [14]. PolyP content was determined by adding 50 µg/mL DAPI to each sample and measuring its emission at 558 nm using an excitation wavelength of 415nm against a standard curve of known concentrations of polyP chain length 45 (Sigma-Aldrich, marketed as “sodium phosphate glass type 45”) [15]. The fluorescence was measured using Fluoroskan Ascent FL microplate fluorometer (Thermo Scientific, Waltham, MA, USA).

3.3.4 Tissue culture of biphasic constructs

Deep zone articular cartilage was aseptically harvested from the metacarpal-phalangeal joint of 9–12 month old calves. Chondrocytes were isolated by sequential enzymatic digestion of 0.5% proteinase (Sigma-Aldrich) and 0.1% collagenase (Roche Diagnostics, Indianapolis, IN, USA) in Ham’s F12 media (Wisent, St-Bruno, Québec, Canada) as previously described [10]. 1×106 chondrocytes were seeded on top of each porous CPP disk and cultured in Ham’s F12 media supplemented with 5% fetal bovine serum (FBS; Wisent). After 48 h, the constructs were placed in mineralization-inducing media that consisted of high-glucose Dulbecco’s modified Eagle media (Life Technologies, Burlington, ON, Canada), 20% FBS, 100 µg/mL ascorbic acid and 10 mM β-glycerophosphate (both Sigma-Aldrich). Cultures were maintained for up to 4 weeks, with media changes every other day.

3.3.5 Histological and radiological analysis of constructs

Calcified cartilage in constructs was imaged using micro-computed tomography (µCT) and histological analysis. For µCT, whole constructs were imaged using a SkyScan 1174v2 µCT scanner (Bruker, Belgium). Scanning was performed at 50 kV and 800 µA through a 0.25 mm aluminum shield with a voxel size of 6.9 µm. After reconstruction, cross-sectional tomographs were obtained with the software provided by the manufacturer. For histological analysis, whole constructs were fixed in 10% formalin, dehydrated and embedded in polymethyl methacrylate without decalcification. Sections were cut and ground to ~50 µm thickness, stained with toluidine blue and light green and visualized by light microscopy.

3.3.6 Biochemical analysis of extracellular matrix and mineral accumulation

Tissues were removed from their substrates and digested in 40 µg/mL papain (Sigma-Aldrich) and DNA, glycosaminoglycan and hydroxyproline content were quantified as previously described [16]. Tissues used to measure mineral accumulation were lyophilized, dry weights measured, and then digested in 3N hydrochloric acid at 90ºC for 2 h. The pH was adjusted to 4.0 with 1.5 M acetate buffer. Calcium and phosphate contents were determined using the o-cresolphthalein complexone assay and the heteropoly blue assay respectively.

3.3.7 Quantification of inorganic polyphosphates accumulated in tissue

Accumulation of polyP in tissues grown on coated and non-coated CPP was quantified over a 2-week time course. Tissues were removed from the substrates and weighed before digesting in 1mg/mL proteinase K (Life Technologies), 10 mM Tris pH 7.5 and 10 mM EDTA (Sigma- Aldrich). Buffer containing 5 M guanidine thiocyanate, 0.9 M sodium citrate, 25 mM EDTA, 1% β-mercaptoethanol and 50 mM Tris pH 6.8 (all Sigma-Aldrich), and then 100% ethanol (Commercial Alcohols, Toronto, Ontario, Canada), were sequentially added to the digested samples, passed through a DNA-binding silica column (Epoch Life Science, Sugar Land, TX, USA), washed with buffer containing 1 M guanidine thiocyanate in 80% ethanol, then washed twice with a buffer containing 150 mM sodium chloride, 80% ethanol and 10 mM Tris pH 7.5. PolyP was eluted with 10 mM Tris pH 7.2, and then incubated with DNase and RNase A (both Roche Diagnostics) for 60 min at 37ºC. Concentrated Tris buffer was added to a final Tris concentration of 0.5 M, and emission at 558 nm was measured with an excitation wavelength of 415 nm using the Fluoroskan Ascent FL microplate fluorometer (Thermo Scientific). The amount of polyP was estimated against a standard curve of known concentrations of polyP chain length 45.

3.3.8 Alkaline phosphatase (ALP) activity assay

To quantify ALP activity in tissues, cells were isolated by digestion in 0.5% collagenase for 2 h and lysed by sonication for 15 min in 0.2 M Tris pH 7.4 buffer. Activity levels were quantified with p-nitrophenol phosphate (Sigma-Aldrich) and measuring the absorbance at 405 nm against a standard curve of p-nitrophenol [10]. The values were normalized to the DNA content (see Biochemical Analysis).

3.3.9 Interfacial shear testing

Interfacial shear strength was determined by applying a force at the interface region of cartilage- CPP constructs using a specially designed sample holder attached to an Instron universal testing machine as previously described [10, 12, 17]. Constructs were placed in a tight-fitting copper sleeve, and a shear load was directly applied to the interface with a constant rate of displacement (1 mm/min) until a sudden decrease in resisting force was measured. The interface was considered to have failed at this point. The energy absorbed by the interface before failure was calculated by integrating the force-displacement curve.

3.3.10 Statistical analysis

All experiments were repeated at least three times and each condition done in triplicates. Student’s t test was used for comparing dissolved calcium and phosphate of the coating. The polyP, calcium and phosphate release in buffer data, as well as biochemical and biomechanical data were evaluated using univariate analysis of variance (ANOVA) and Tukey post-hoc test. The data from polyphosphate accumulation experiments were analyzed using repeated measures ANOVA and Tukey post-hoc test, as data were animal-matched. Statistical significance was assigned at p < 0.05.

3.4 Results

3.4.1 Characterization of sol-gel-derived coatings on porous CPP disks

Scanning electron micrographs (secondary electron imaging) of the disk surfaces are shown in Figure 3.1 A–C. Organic sol-gel (OSG)-derived and inorganic sol-gel (ISG)-derived coatings were observed on the outer surface of porous CPP disks (Figure 3.1 A & B), somewhat masking the surface morphology of the CPP particles (Figure 3.1C). The thickness of coating, as determined by viewing fractured disks, was in the order of 1 µm (Figure 3.1 D–F). Backscattered electron microscopy of the disk cross-section confirmed the presence of coating throughout (Figure 3.1 G–I). The sol-gel-derived films formed web-like layers between closely positioned surface regions of sintered CPP particles in both OSG-coated and ISG-coated disks; however, the web-like layers formed in ISG-coated disks appeared slightly larger. The rings seen on the electron diffraction patterns obtained from OSG-derived and ISG-derived coatings both corresponded to d-spacings representing different crystallographic planes for HAp (Figure 3.2 A & B). Analysis of dissolution products collected from coated CPP disks in hydrochloric acid indicated no significant difference in the calcium-to-phosphate ratios of OSG-derived and ISG-derived coatings (Figure 3.2C), which corresponded to that of pure calcium HAp (1.67).

3.4.2 Characterization of polyP and mineral release from coated CPP substrates

The amount of polyP released from coated and non-coated CPP disks into the buffered solution every 48 hours over 21 days is shown in Figure 3.3A. PolyP release was not detected (less than 1nmol) from either OSG-coated or ISG-coated CPP disks, while non-coated CPP disks released about 100–150 nmol phosphate units of polyP. The ISG-coated CPP disks released more calcium and phosphate compared with OSG-coated CPP disks, with the exception of the first 24 h (Figure 3.3 B & C). Non-coated CPP disks released significantly less calcium and phosphate than either OSG-coated or ISG-coated CPP disks, suggesting that the released calcium and phosphate from coated CPP disks originated mainly from the coatings.

Figure 3.1: Porous CPP disks were coated using two different sol-gel thin film processing methods. (A-C) Secondary electron (SE) images were taken from the outer surface of coated and non-coated porous CPP disks. 1500× magnification, scale bar = 50µm. (D-F) Coating thickness and surface topology were observed with high-magnification SE images of cross-sectional, fractured surfaces. 50,000× magnification, scale bar = 1µm, coating and CPP as denoted on images. (G-I) Distribution of coating was observed with backscatter SE imaging of bisected CPP disks at the middle of the cross-section. Coating accumulated between closely positioned CPP particles to form “necks” (arrows on G and H). 300× magnification, scale bar = 100µm.

Figure 3.2: Coating deposited by sol-gel thin film processing methods was HAp. (A-B) Representative electron diffraction patterns of OSG-derived (A) and ISG-derived (B) coating scraped off the CPP disks corresponded to d-spacings that represent different crystallographic planes for HAp. (C) Coating on porous CPP disks were dissolved in hydrochloric acid, and calcium (Ca) and phosphate (Pi) were measured to calculate the calcium-phosphate ratio. n = 6, data shown as mean ± SD. Difference was not statistically significant.

Figure 3.3: In an aqueous environment, both OSG-derived and ISG-derived coating inhibited polyP release while releasing calcium and phosphate. OSG-derived, ISG-coated and non-coated (NC) porous CPP disks were incubated in TBS and polyP (A), calcium (B), and phosphate (C) in buffer were quantified every 48 h. In (A), polyP release from OSG-derived and ISG-coated CPPs was not detected. In (C), phosphate release from non-coated CPPs was not detected. Results are expressed as mean ± SD. n = 5, p < 0.05 between any 2 conditions in the same time point, except a pair marked ns (not significant) in (B).

3.4.3 Characterization of cartilage tissue formed on coated CPP substrates

Deep zone articular chondrocytes were cultured on OSG-coated, ISG-coated and non-coated CPP substrates for 4 weeks to form biphasic constructs with a ZCC. The location of the mineral accumulated in cartilage tissues formed on coated and non-coated CPP substrates was observed using µCT imaging (Figure 3.4 A–C). Mineral in constructs formed with either OSG-coated or ISG-coated substrates were in direct contact with the substrate through the cross-section (Figure 3.4A and 4B), while the mineral only made intermittent contact with the substrate and a clear separation between the mineral and the substrate was observed elsewhere in constructs formed with non-coated substrates (Figure 3.4C). Histological examination of tissues grown on OSG-coated substrates confirmed the presence of a zonal distribution of mineral in direct contact with the substrate (Figure 3.4D). Tissues grown on ISG-coated substrates similarly accumulated mineral in contact with the substrate (Figure 3.4E); however, the mineral appeared to be more discontinuously distributed in tissues grown on ISG-coated substrates, as compared to a more continuous, sheet-like appearance found in tissues grown on OSG-coated substrates. Tissue grown on non-coated substrates also accumulated mineral that was discontinuously distributed, but only rarely in direct contact with the CPP (Figure 3.4F). There were no differences in the accumulated proteoglycans (Figure 3.5A) or total collagen (Figure 3.5B) of tissues formed on any of the three substrates. Tissues grown on either of the coated substrates accumulated less calcium and phosphate in the matrix compared with tissues on non-coated substrates (Figure 3.5 C & D).

Figure 3.4: Deep zone chondrocytes grown on OSG- and ISG-coated CPP substrates formed a ZCC in direct contact with the substrate. (A-C) Cross-sectional µCT tomographs of 4-week-old constructs revealed locations of tissue-accumulated mineral (arrows on inset) relative to the substrate. Scale bars = 1mm. (D-F) Constructs were processed undecalcified, sectioned and stained to reveal the cartilaginous tissue (purple), mineral (green) and CPP particles (brown). Scale bars = 200µm.

Figure 3.5: Sol-gel-derived coating did not affect the extracellular matrix accumulation of cartilage tissue grown on coated porous CPP substrates. Accumulated proteoglycans (A) and total collagen (B) were normalized by the DNA content. Tissues on coated CPP substrates accumulated less calcium (C) and phosphate (D) than those grown on non-coated (NC) substrates, which were normalized by the tissue dry weight. Results are expressed as mean ± SD, n = 3. * p < 0.05.

3.4.4 HAp barrier coating on the CPP substrate affected polyP accumulation in cartilage

To determine whether the HAp barrier coating on the CPP substrate could inhibit polyP release in tissue culture and prevent the subsequent accumulation of polyP in tissue, polyP present in cartilage grown on coated and non-coated substrates were quantified at days 6, 10 and 14 of culture (Figure 3.6A). Cartilage grown on ISG-coated CPP accumulated more polyP than those grown on OSG-coated CPP at days 6 and 14. The difference in polyP accumulation was not statistically significant at 10 days. There was no significant difference in the amount of polyP accumulated by cartilage formed on OSG and non-coated CPP. Over time, polyP levels decreased in all tissues. Since deep zone chondrocytes express ALP that can break down polyP [18], ALP activity levels in cartilage grown on coated and non-coated substrates were quantified (Figure 3.6B). There was no significant difference in ALP activity level of cartilage formed on either coated or non-coated CPP, but ALP activity levels decreased over time in all groups.

Figure 3.6: Net polyP accumulation (A) and alkaline phosphatase (ALP) activity (B) of tissues grown on coated and non-coated CPP substrates for up to 14 days. (A) polyP accumulation values are expressed in equivalent phosphate (Pi) units and normalized by corresponding tissue wet weight, while (B) ALP activity was normalized by corresponding tissue DNA content. Results are expressed as mean ± SEM, n = 6 (A), n = 3 (B). * p < 0.05 by repeated measures ANOVA and Tukey post-hoc test.

3.4.5 Interfacial shear strength of cartilage-CPP substrate constructs

Constructs formed with OSG-coated substrates withstood a greater maximum shear load applied to the interface before failure than constructs formed with ISG-coated or non-coated substrates (Figure 3.7A). Constructs formed with OSG-coated substrates also absorbed more energy before failure compared to constructs formed with ISG-coated or non-coated substrates (Figure 3.7B). Constructs with ISG-coated and non-coated substrates yielded no significant difference in either the maximum resistance to shear forces or the absorbed energy to failure. These results indicated that constructs formed with OSG-coated substrates displayed greater cartilage-CPP interfacial shear strength compared to those formed with either ISG-coated or non-coated substrates.

Figure 3.7: Constructs formed with OSG-coated CPP substrates withstood a greater interfacial shear load than those formed with ISG-coated or non-coated CPP substrates. Shear force was applied to the interface until failure at a displacement rate of 1mm/min. (A) The force at which the failure occurred was recorded as peak load. (B) The time-displacement curve up until failure was integrated to calculate the energy absorbed by the interface before failure. All porous CPP substrates were identical in shape. Results are expressed as mean ± SEM, n = 9, * p < 0.05, ** p < 0.01.

3.5 Discussion

This study demonstrated that coating porous CPP substrates with OSG enhanced the constructs’ interfacial shear strength. Both OSG and ISG processing methods resulted in submicron-thick hydroxyapatite coatings over the sintered CPP particles throughout the disks’ thickness. Direct examination of the films formed on CPP substrates by the two sol-gel processing methods indicated little difference up to 50,000× magnification. In a previous study comparing the two processing methods applied to porous sintered titanium alloy substrates using higher-resolution (105×) field-emission scanning electron microscopy, nanoscale porosity was observed only in ISG-derived coating, in contrast to the featureless appearance of the OSG-derived coating [11]. Therefore, although not directly observed, OSG-derived and ISG-derived coatings on porous CPP substrates are likely to have the same nanoscale differences in porosity as those on titanium substrates. While both OSG-derived and ISG-derived coatings effectively inhibited the polyP release from CPP in buffer, ISG-coated CPPs released more calcium and phosphate than OSG- coated CPPs. Since the ISG-derived coating would have a larger free surface area than the OSG- derived coating at which degradation would occur, the difference in their degradation rate is expected.

Cartilage formed on OSG-coated substrates exhibited greater resistance to interfacial shear forces than cartilage formed on ISG-coated or non-coated substrates. Although both OSG- coated and ISG-coated CPP supported the formation of a ZCC in direct contact with the porous CPP substrate, mineral in tissues grown on OSG-coated substrates formed in a more continuous pattern than that on ISG-coated substrates, suggesting that both location and distribution of calcification contribute to the shear strength of the cartilage-CPP interface. The thin film coating did not affect cartilage formation per se as tissues formed on OSG-coated and ISG-coated substrates accumulated comparable amounts of extracellular matrix and mineral. The observed lower mineral content of cartilage formed on either coated substrates compared to cartilage on non-coated substrate was consistent with the µCT image, as band of tissue-accumulated mineral appeared to be thicker on non-coated substrates. The difference in measured mineral content could also have been exaggerated by incomplete separation of tissues from their coated substrates for analysis, which may have resulted in leaving behind some mineralized tissues. Alternatively, as the polyP is released from CPP and degraded, it could contribute to calcium and phosphate levels in the tissue. In the first two weeks of culture, cartilage on ISG-coated

89 substrates accumulated more polyP compared to tissues on OSG-coated substrates. The accumulated polyP could have been produced by chondrocytes themselves as mammalian cells have been shown to contain small amounts of polyP [19], but extracellular matrix of tissues formed by deep-zone chondrocytes cultured in mineralization-inducing media was previously shown to be devoid of polyP [20]. Therefore, accumulated polyP in cartilage most likely originated from the CPP substrate in spite of the coating. Since both OSG and ISG coatings effectively inhibited polyP release from CPP in the absence of tissue, it may be that tissue culture exacerbated the degradation of coating sufficiently to cause the polyP to leak from the CPP substrates. This observation provides a possible explanation as to why mineral in tissues grown on ISG-coated substrates appeared to be more discontinuously distributed. PolyP inhibits mineral crystal growth by binding condensed phosphates to mineral crystal nuclei and “poisoning” them [21]. A recent study with mammalian cells demonstrated that the predominant effect of polyP on biomineralization is inhibitory at higher concentrations [22]. The relative abundance of polyP in tissue on ISG-coated substrates could have prevented the mineral crystals from growing and connecting to each other, giving rise to the discontinuous distribution.

The observation that cartilage on OSG-coated substrates had comparable polyP accumulation to cartilage on non-coated substrates was unexpected. There are two possible explanations for this. Firstly, deep-zone chondrocytes express alkaline phosphatase [23], which has been shown to act as a highly active exopolyphosphatase [18]. Calcium and various forms of phosphates can regulate alkaline phosphatase activity [24], which raises the possibility that a difference in the amount of calcium and phosphate released from the OSG coating may differentially modulate polyP degradation of tissues, adding to the complexity of the interpretation of polyP accumulation results. Secondly, the presence of the coating may affect the polyP production by chondrocytes themselves. Little is known about polyP in mammalian biology, due in part to limitations in current assay methods [25]. To better understand this phenomenon, additional studies are required to investigate the interplay of coating degradation and the polyP metabolism in cartilage tissues.

This study addresses a limitation of porous CPP as a substrate for tissue-engineered cartilage. Porous CPP was chosen for our design of osteochondral-like constructs because it possesses suitable characteristics as a bulk, biodegradable, porous substrate on which cartilage tissue can be formed in vitro and attach by mechanical interdigitation and into which bone can

90 grow [24]. The ingrowth of bone into porous CPP substrate fixes the multiphasic construct in place and provides resistance against forces acting on the resulting biphasic osteochondral implant. Formation of a ZCC in such implants was proposed as a means of forming an interface with increased interfacial strength properties. However, due to the release of polyP by CPP, attempts to strengthen the cartilage-substrate interface in this manner were unsuccessful due to inhibition of mineralization within the cartilage at the interface. The present study explored the use of barrier coatings to inhibit polyp accumulation at this interface and this resulted in some improvement through use of sol-gel-derived HAp coatings formed using an organic precursor. Although HAp sol gel coating appears to enhance interfacial shear strength, further studies to develop more complete coatings and in vivo implantation studies will be required to confirm the effectiveness of this strategy for forming tissue-engineered implants for repair of osteochondral defects as well as other more extensive reconstruction. Also, it is not known whether this strategy is applicable to other types of substrates.

In this study, we utilized chondrocytes isolated from the deep zone of articular cartilage, which undergo mineralization in vitro at the basal aspect of tissue in the presence of β-glycerophosphate [10]. Using deep zone chondrocytes obviated the need for using such strategies as delivery of microspheres [26], complex bioreactor conditions [27] or inducing osteogenesis at the interface [27-29] to manipulate the location of mineralized cartilage. However, articular chondrocytes are limited in their potential as a cell source for tissue engineering because during the process of monolayer culture to expand cell numbers they dedifferentiate [30], losing collagen type II expression and gaining collagen type I expression [31]. In a separate study, we found that using chondrocytes derived from bone marrow stromal cells, OSG-coated CPP substrates could be used successfully to form a mineralized interface with this cell source [32], demonstrating its use for tissue engineering strategies.

It remains to be determined whether the coatings will affect the osseointegration properties of porous CPP, and whether the effect will depend on the method of coating preparation. It is unlikely that HAp sol-gel-formed coatings will negatively influence integration as both OSG and ISG coating on porous sintered titanium dental implants enhanced osseointegration compared to the non-coated control [33]. Enhanced integration would be beneficial in enabling the generation of tissue-engineered multiphasic constructs that can withstand the mechanical stress after implantation in a joint.

3.6 Conclusion

Both OSG and ISG processing methods resulted in formation of thin HAp coatings over sintered CPP throughout the porous CPP disks. As a substrate for tissue-engineered cartilage, OSG- coated CPP resulted in greater cartilage-CPP interfacial shear strength compared to either ISG- coated or non-coated porous CPP substrates, even though both coated CPP types enabled the ZCC to form in apposition to the substrate. This was attributed to a higher net accumulation of polyP in tissues grown on ISG-coated CPP substrates compared to the OSG-coated substrates, resulting in formation of a more discontinuous distribution of mineralized cartilage. This underlines the importance of generating biphasic constructs with a zone of calcified tissue similar to the native cartilage architecture.

3.7 Author Contributions The authors of the study were Whitaik David Lee, Robert M. Pilliar, Rita A. Kandel and William L. Stanford.

• Study conception and design: WDL, RMP, RAK. • Provision of study materials: RMP, RAK. • Analysis and interpretation of data: WDL, RMP, RAK. • Drafting of article: WDL, WLS, RAK. • Critical revision of the article: WDL, RMP, WLS, RAK. • Final approval of the article: WDL, WLS, RAK.

3.8 Acknowledgements

The authors would like to thank Dr. Jean-Philippe St-Pierre, Eugene Hu, Doug Holmyard, Dr. Jian Wang, Nancy Valiquette and Dr. Thomas Willett for their technical assistance. This work

92 was supported by the Canadian Institutes for Health Research (to R.A.K) and the Natural Sciences and Engineering Research Council of Canada (to W.L.S.). W.L.S. is supported by a Canadian Research Chair.

3.9 References

[1] Swieszkowski W, Tuan BH, Kurzydlowski KJ, Hutmacher DW. Repair and regeneration of osteochondral defects in the articular joints. Biomol Eng (2007) 24:489-95.

[2] Hoemann CD, Lafantaisie-Favreau CH, Lascau-Coman V, Chen G, Guzman-Morales J. The cartilage-bone interface. J Knee Surg (2012) 25:85-97.

[3] Kandel RA, Grynpas M, Pilliar R, Lee J, Wang J, Waldman S, et al. Repair of osteochondral defects with biphasic cartilage-calcium polyphosphate constructs in a sheep model. Biomaterials (2006) 27:4120-31.

[4] Lee WD, Hurtig MB, Kandel RA, Stanford WL. Membrane culture of bone marrow stromal cells yields better tissue than pellet culture for engineering cartilage-bone substitute biphasic constructs in a two-step process. Tissue Eng Part C Methods (2011) 17:939-48.

[5] Sharma AR, Jagga S, Lee SS, Nam JS. Interplay between cartilage and subchondral bone contributing to pathogenesis of osteoarthritis. Int J Mol Sci (2013) 14:19805-30.

[6] Madry H, van Dijk CN, Mueller-Gerbl M. The basic science of the subchondral bone. Knee Surg Sports Traumatol Arthrosc (2010) 18:419-33.

[7] Curl WW, Krome J, Gordon ES, Rushing J, Smith BP, Poehling GG. Cartilage injuries: a review of 31,516 knee arthroscopies. Arthroscopy (1997) 13:456-60.

[8] Dore D, Quinn S, Ding C, Winzenberg T, Cicuttini F, Jones G. Subchondral bone and cartilage damage: a prospective study in older adults. Arthritis Rheum (2010) 62:1967-73.

[9] Pilliar RM, Filiaggi MJ, Wells JD, Grynpas MD, Kandel RA. Porous calcium polyphosphate scaffolds for bone substitute applications -- in vitro characterization. Biomaterials (2001) 22:963-72.

[10] Allan KS, Pilliar RM, Wang J, Grynpas MD, Kandel RA. Formation of biphasic constructs containing cartilage with a calcified zone interface. Tissue Eng (2007) 13:167-77.

[11] Gan L, Pilliar R. Calcium phosphate sol-gel-derived thin films on porous-surfaced implants for enhanced osteoconductivity. Part I: Synthesis and characterization. Biomaterials (2004) 25:5303-12.

[12] St-Pierre JP, Gan L, Wang J, Pilliar RM, Grynpas MD, Kandel RA. The incorporation of a zone of calcified cartilage improves the interfacial shear strength between in vitro-formed cartilage and the underlying substrate. Acta Biomater (2012) 8:1603-15.

[13] Waldman SD, Grynpas MD, Pilliar RM, Kandel RA. Characterization of cartilagenous tissue formed on calcium polyphosphate substrates in vitro. J Biomed Mater Res (2002) 62:323- 30.

[14] Baykov AA, Evtushenko OA, Avaeva SM. A malachite green procedure for orthophosphate determination and its use in alkaline phosphatase-based enzyme immunoassay. Anal Biochem (1988) 171:266-70.

[15] Aschar-Sobbi R, Abramov AY, Diao C, Kargacin ME, Kargacin GJ, French RJ, et al. High sensitivity, quantitative measurements of polyphosphate using a new DAPI-based approach. J Fluoresc (2008) 18:859-66.

[16] Taylor DW, Ahmed N, Gan L, Gross AE, Kandel RA. Proteoglycan and collagen accumulation by passaged chondrocytes can be enhanced through side-by-side culture with primary chondrocytes. Tissue Eng Part A (2010) 16:643-51.

[17] Hamilton DJ, Seguin CA, Wang J, Pilliar RM, Kandel RA, BioEngineering of Skeletal Tissues T. Formation of a nucleus pulposus-cartilage endplate construct in vitro. Biomaterials (2006) 27:397-405.

[18] Lorenz B, Schroder HC. Mammalian intestinal alkaline phosphatase acts as highly active exopolyphosphatase. Biochim Biophys Acta (2001) 1547:254-61.

[19] Kumble KD, Kornberg A. Inorganic polyphosphate in mammalian cells and tissues. J Biol Chem (1995) 270:5818-22.

[20] St-Pierre JP, Pilliar RM, Grynpas MD, Kandel RA. Calcification of cartilage formed in vitro on calcium polyphosphate bone substitutes is regulated by inorganic polyphosphate. Acta Biomater (2010) 6:3302-9.

[21] Fleisch H, Russell RG, Straumann F. Effect of pyrophosphate on hydroxyapatite and its implications in calcium homeostasis. Nature (1966) 212:901-3.

[22] Ariganello MB, Omelon S, Variola F, Wazen RM, Moffatt P, Nanci A. Osteogenic cell cultures cannot utilize exogenous sources of synthetic polyphosphate for mineralization. J Cell Biochem (2014) 115:2089-102.

[23] Miao D, Scutt A. Histochemical localization of alkaline phosphatase activity in decalcified bone and cartilage. J Histochem Cytochem (2002) 50:333-40.

[24] Farley JR, Hall SL, Tanner MA, Wergedal JE. Specific activity of skeletal alkaline phosphatase in human osteoblast-line cells regulated by phosphate, phosphate esters, and phosphate analogs and release of alkaline phosphatase activity inversely regulated by calcium. J Bone Miner Res (1994) 9:497-508.

[25] Rao NN, Gomez-Garcia MR, Kornberg A. Inorganic polyphosphate: essential for growth and survival. Annu Rev Biochem (2009) 78:605-47.

[26] Dormer NH, Singh M, Wang L, Berkland CJ, Detamore MS. Osteochondral interface tissue engineering using macroscopic gradients of bioactive signals. Ann Biomed Eng (2010) 38:2167- 82.

[27] Chen K, Ng KS, Ravi S, Goh JC, Toh SL. In vitro generation of whole osteochondral constructs using rabbit bone marrow stromal cells, employing a two-chambered co-culture well design. J Tissue Eng Regen Med (2013).

[28] Grayson WL, Bhumiratana S, Grace Chao PH, Hung CT, Vunjak-Novakovic G. Spatial regulation of human mesenchymal stem cell differentiation in engineered osteochondral constructs: effects of pre-differentiation, soluble factors and medium perfusion. Osteoarthritis Cartilage (2010) 18:714-23.

[29] Chen K, Teh TK, Ravi S, Toh SL, Goh JC. Osteochondral interface generation by rabbit bone marrow stromal cells and osteoblasts coculture. Tissue Eng Part A (2012) 18:1902-11.

[30] Holtzer H, Abbott J, Lash J, Holtzer S. The loss of phenotypic traits by differentiated cells in vitro, I. dedifferentiation of cartilage cells. Proc Natl Acad Sci U S A (1960) 46:1533-42.

[31] Darling EM, Athanasiou KA. Rapid phenotypic changes in passaged articular chondrocyte subpopulations. J Orthop Res (2005) 23:425-32.

[32] Lee WD, Hurtig MB, Pilliar RM, Stanford WL, Kandel RA. Engineering of hyaline cartilage with a calcified zone using bone marrow stromal cells. Osteoarthritis Cartilage (2015).

[33] Gan L, Wang J, Tache A, Valiquette N, Deporter D, Pilliar R. Calcium phosphate sol-gel- derived thin films on porous-surfaced implants for enhanced osteoconductivity. Part II: Short- term in vivo studies. Biomaterials (2004) 25:5313-21.

Chapter 4 Engineering of Hyaline Cartilage with a Calcified Zone Using Bone Marrow Stromal Cells

This chapter is reprinted under license from Osteoarthritis and Cartilage: Lee WD, Hurtig MB, Pilliar RM, Stanford WL and Kandel RA. Engineering of hyaline cartilage with a calcified zone using bone marrow stromal cells. Copyright © 2015 Osteoarthritis Research Society International, with permission from Elsevier Ltd.

4.1 Abstract

Objective: In healthy joints, a zone of calcified cartilage (ZCC) provides the mechanical integration between articular cartilage and subchondral bone. Recapitulation of this architectural feature should serve to resist the constant shear force from the movement of the joint and prevent the delamination of tissue-engineered cartilage. Previous approaches to create the ZCC at the cartilage-substrate interface have relied on strategic use of exogenous scaffolds and adhesives, which are susceptible to failure by degradation and wear. In contrast, we report a successful scaffold-free engineering of ZCC to integrate tissue-engineered cartilage and a porous biodegradable bone substitute, using sheep bone marrow stromal cells (BMSCs) as the cell source for both cartilaginous zones.

Design: BMSCs were predifferentiated to chondrocytes, harvested and then grown on a porous calcium polyphosphate substrate in the presence of triiodothyronine (T3). T3 was withdrawn, and additional predifferentiated chondrocytes were placed on top of the construct and grown for 21 days.

Results: This protocol yielded two distinct zones: hyaline cartilage that accumulated proteoglycans and collagen type II, and calcified cartilage adjacent to the substrate that additionally accumulated mineral and collagen type X. Constructs with the calcified interface

98 had comparable compressive strength to native sheep osteochondral tissue and higher interfacial shear strength compared to control without a calcified zone.

Conclusion: This protocol improves on the existing scaffold-free approaches to cartilage tissue engineering by incorporating a calcified zone. Since this protocol employs no xenogeneic material, it will be appropriate for use in preclinical large-animal studies.

4.2 Introduction

In a synovial joint, articular cartilage bears both compressive and shear forces to facilitate weight bearing and movement. Once damaged, these mechanical forces contribute to the progression of cartilage damage, resulting in loss of cartilage and subchondral bone remodelling. Tissue engineering is widely regarded as a promising technology to generate constructs that could replace damaged cartilage and bone [1]. A wide range of osteochondral tissue engineering strategies are under investigation [2], including our effort to produce biphasic constructs that consist of scaffold-free cartilage tissue formed by articular chondrocytes or bone marrow stromal cells (BMSCs) on porous bone substitutes (calcium polyphosphate) [3, 4]. Calcium polyphosphate is a biodegradable ceramic with excellent load-bearing and osseointegrative properties [5, 6]. In our design, ingrowth of bone into calcium polyphosphate holds the biphasic construct in place and provides support against compressive force on tissue-engineered cartilage. However, to resist shear force, a mechanically competent interface between the cartilage and its substrate is needed. Preventing cartilage delamination is a common challenge shared by many cartilage repair strategies [7, 8].

Healthy cartilage tissue interfaces with bone in vivo through a zone of calcified cartilage, which enhances mechanical integration by redistributing forces acting at the interface [9]. BMSCs are suitable for creating calcified cartilage, because they can be differentiated to produce cartilage tissues that mineralize their matrix [10]. While chondrocytes from the deep zone of articular cartilage can also yield calcified cartilage in vitro [11], BMSCs can be obtained without donor site morbidity and expanded to clinically relevant numbers. Furthermore, in vivo studies have shown that BMSC-derived cartilage can maintain their non-calcified phenotype after being implanted in a joint cavity [12, 13], demonstrating the potential of BMSCs to produce both

99 calcified and non-calcified cartilage. However, to use BMSCs as a single source of cells to create cartilage tissues with a calcified cartilaginous interface, limiting mineralization of BMSC- derived cartilage to the interface will be crucial for the success of these implants.

Using BMSCs as the cell source, we report the first successful engineering of an osteochondral-like construct that incorporates a ZCC at the cartilage-calcium polyphosphate interface without the use of a scaffold as the cartilage tissue forms on top of a substrate. A stepwise culturing strategy enabled the selective activation of BMSC-derived chondrocytes to induce mineralization only at the interface, while avoiding mineralization in the contiguous cartilage tissue, thus conferring localized calcified and hyaline zones to the engineered cartilage tissue. This approach is suitable for clinical use as the BMSCs were expanded in media supplemented with autologous serum and the construct cultured in serum-free, defined media, so this strategy can be directly translated to in vivo preclinical studies.

100

4.3 Materials and Methods

4.3.1 Chondrogenic predifferentiation of BMSCs in membrane cultures

Sheep BMSCs were isolated and expanded from bone marrow aspirates taken from ewes aged between 2 and 5 years as previously described [4] using autologous serum (see Supplementary Material). To differentiate BMSCs to chondrocytes, 2.0×106 BMSCs were seeded onto 12 mm cell culture inserts (0.2 µm pore size; Millipore, Billerica, MA, USA) that were coated with collagen type IV (50 µg/insert; Sigma-Aldrich, Oakville, ON, Canada). Cells were cultured in a defined chondrogenic media composed of high-glucose Dulbecco’s modified Eagle medium (DMEM; Life Technologies, Burlington, ON, Canada), 1× insulin-transferrin-selenium plus (ITS+) cell culture supplement (BD Biosciences, Bedford, MA, USA), 100 nM dexamethasone (Sigma-Aldrich), 100 µg/mL ascorbic acid (Sigma-Aldrich) and 10 ng/mL transforming growth factor-β3 (TGF-β3; R&D Systems, Minneapolis, MN, USA). For the first 72 h, chondrogenic media was additionally supplemented with 10 µM blebbistatin (Cayman Chemical, Ann Arbor, MI, USA) to inhibit spontaneous detachment of cells. Media was changed every 2–3 days. After 3 weeks of culture, cells (herein referred to as predifferentiated chondrocytes) were isolated by digesting the tissue in 0.5% (w/v) collagenase (Roche Diagnostics, Indianapolis, IN, USA) for 2 h at 37ºC.

4.3.2 Preparation of porous calcium polyphosphate substrates with a hydroxyapatite coating

Porous calcium polyphosphate disks of 4 mm diameter and 2 mm height were prepared from gravity-sintered 75–150 µm calcium polyphosphate particles as previously described [14]. Disks were incubated in phosphate-buffered saline (PBS) for 1 week. Then, a thin, submicron hydroxyapatite layer was deposited over the disks using an organic sol-gel processing method. This hydroxyapatite layer was intended to serve as a ‘barrier’ coating inhibiting the rate of release of calcium polyphosphate degradation products during in vitro cartilage formation [15]. The sol was prepared with triethyl phosphite and calcium nitrate tetrahydrate (Sigma-Aldrich) as previously described [16]. Disks were dipped into the sol gel sideways for 8 s and withdrawn at a speed of 20 cm/min using a stepper motor-driven actuator, air-dried for 10 min and annealed for 15 min at 210ºC. To ensure a continuous barrier coating, coated disks were dipped again,

101 annealed at 500ºC for 20 min and gradually cooled. The coated disks were placed in Tygon tubing to create a well-like structure and subsequently sterilized by γ-irradiation (2.5 MRad).

4.3.3 Optimizing the mineralizing culture condition for predifferentiated chondrocytes

To find a culture condition in which predifferentiated chondrocytes would form calcified cartilage tissue, predifferentiated chondrocytes were placed on the top surface of porous calcium polyphosphate substrates (1.5×106 cells/substrate) and cultured in basal construct media composed of DMEM, 1× ITS+, 100 µg/mL ascorbic acid, 50 µg/mL L-proline and 10 mM β-glycerophosphate (Sigma-Aldrich). To determine which condition would induce mineralization, 100 nM dexamethasone, 3 nM triiodothyronine (T3) or 100 nM retinoic acid (all Sigma-Aldrich) were added to cultures. The cultures were harvested at various times up to 21 days for further study.

4.3.4 Tissue culture of multiphasic constructs

Predifferentiated chondrocytes were placed on the top surface of porous calcium polyphosphate substrates (1.0×106 cells/substrate) and cultured in basal construct media, with or without 3 nM T3. T3 was withdrawn at day 4. At day 5, additional predifferentiated chondrocytes were placed on top of the existing constructs (1.5×106 cells/construct). The multi-layered constructs were maintained under the same culture conditions for up to 21 days. This protocol is shown in Figure 1.

4.3.5 Visualization and quantification of alkaline phosphate (ALP) activity

To visualize ALP activity, tissues were removed from their substrates and fixed in 10% neutral buffered formalin for 1 h. Then, samples were incubated in 30% (w/v) sucrose overnight at 4ºC and snap-frozen in Tissue-Tek OCT compound (Sakura Finetek, Torrance, CA, USA). 6 µm cross-section cryosections were cut and mounted on silanized glass slides. Sections were incubated in azo dye (Naphthol AS-MX phosphate and Fast Blue BB salt, both Sigma-Aldrich) for 10 min, counterstained with eosin, dried and mounted with coverslips. To quantify the ALP activity, cells were isolated from tissues by digestion in 0.5% collagenase for 2 h and lysed by sonication for 15 min in 0.2 M Tris pH 7.4 buffer. Activity levels were quantified with

102

Figure 4.1: Line diagram of the tissue culture protocol used for forming scaffold-free multiphasic osteochondral-like constructs. p-nitrophenol phosphate (Sigma-Aldrich) and measuring the absorbance at 405 nm against a standard curve of p-nitrophenol [11]. Values were normalized to the lysate’s DNA content (see Biochemical Analysis).

4.3.6 Gene expression analysis

Predifferentiated chondrocytes (1.5×106 cells/substrate) were cultured on porous calcium polyphosphate substrates in basal construct media with 4 days of 3 nM T3. T3 was removed after 4 days, and the cells were grown in basal construct media only for the remainder of the 21 days of culture. Controls were cells that were untreated and grown in the same media. Tissues were removed from their substrates at day 4 and 21 and homogenized in 1 mL TRIzol reagent (Life Technologies) by bead milling. Total RNA was then extracted and reverse-transcribed using

103

SuperScript VILO cDNA Synthesis Kit (Life Technologies). Quantitative PCR was performed using a LightCycler 96 Real-Time PCR System (Roche Diagnostics) and SYBR GreenER master mix (Life Technologies) with gene-specific primers (Table 4.1).

4.3.7 Micro-computed tomography (µCT) imaging

Constructs were imaged using a Skyscan 1174v2 µCT scanner (Bruker, Belgium). Scanning was performed at 50 kV and 800 µA through a 0.25 mm aluminum shield with a voxel size of 6.9 µm. After reconstruction, cross-sectional tomographs were obtained with the software provided by the manufacturer.

4.3.8 Histological analysis of whole constructs

Constructs were fixed in 10% neutral buffered formalin overnight and infiltrated using the Osteo-Bed bone embedding kit (Polysciences Inc., Warrington, PA). Sections were cut and ground to ~50 µm thickness, stained with toluidine blue and light green and visualized under light microscopy.

4.3.9 Histological analysis of cartilaginous tissues

Tissue mineral and proteoglycan accumulation was visualized by staining histological sections (see Supplementary Material) with von Kossa and toluidine blue. Orientation of collagen fibrils was visualized by polarized light microscopy of trichrome-stained sections. Visualization for collagens types I, II and X were visualized by immunofluorescence (see Supplementary Material).

4.3.10 Biochemical analysis of extracellular matrix and mineral accumulation

Tissues were digested in 40 µg/mL papain (Sigma-Aldrich) and DNA, glycosaminoglycan and hydroxyproline contents were quantified as previously described [17]. To measure mineral accumulation, the tissues were lyophilized, dry weights measured, and then digested in 3 N hydrochloric acid at 90ºC for 2 h. The pH was adjusted to 4.0 with 1.5 M acetate buffer. Phosphate content was determined using the heteropoly blue assay and measuring the absorbance at 620 nm. Calcium content was determined using the o-cresolphthalein complexone assay and measuring the absorbance at 570 nm.

104

4.3.11 Mechanical testing of multiphasic constructs

On day 21 of culture, the bulk equilibrium compressive modulus of the multiphasic construct’s cartilaginous tissue was determined by stress relaxation testing on the Mach-1 mechanical tester (Bio-Momentum, Laval, Canada) with a 0.65 mm-diameter indenter as previously described [4]. Interfacial shear strength was determined by applying a force at the interface region of these samples using a specially designed jig attached to an Instron universal testing machine as previously described [15] (see Supplementary Material).

4.3.12 Statistical testing

All values are expressed as mean ± 95% confidence intervals (CI). Each experiment was carried out with three donor animals (N = 3). Univariate analysis of variance and Tukey post-hoc testing were used to analyze ALP activity levels on days 0–4, accumulation of GAG, total collagen and mineral contents, as well as the compressive moduli of the constructs. The interfacial shear strength data of the constructs were analyzed using t test. To account for the wide animal-to- animal variability, ALP activity levels on days 7, 10 and 14 were each analyzed using paired t test. Similarly, gene expression levels on days 4 and 21 were log-transformed and analyzed using paired t test.

105

4.4 Results

4.4.1 Generation of calcified cartilage at the calcium polyphosphate interface with predifferentiated chondrocytes

To engineer a multiphasic construct that incorporates a ZCC at the cartilage-calcium polyphosphate interface using predifferentiated chondrocytes, we first sought to establish a culture condition in which predifferentiated chondrocytes could be cultured on porous calcium polyphosphate substrates to form mineralized cartilage tissue. Triiodothyronine (T3) [18], retinoic acid [19] and dexamethasone [10] have all been identified as potential inducers of chondrocyte mineralization in vitro. We cultured predifferentiated chondrocytes on calcium polyphosphate substrates with either 3 nM T3 or 100 µM retinoic acid treatment in the presence or absence of dexamethasone for 1 week, and then quantified the ALP activity of the cells as an indicator of chondrocyte mineralization potential [20, 21]. ALP activity was the highest in predifferentiated chondrocytes cultured in the absence of dexamethasone with the T3 treatment (Figure 4.2 A). While ALP activity was also increased with the retinoic acid treatment, continued culture did not yield cartilaginous tissues on calcium polyphosphate after 3 weeks (data not shown). Therefore, treatment of predifferentiated chondrocytes with T3 was selected for further study.

When 2×106 predifferentiated chondrocytes were cultured on porous calcium polyphosphate substrates for 3 weeks, the mineralized zone was found in the top of the tissue (Figure 4.2 B). Between this mineralized zone and the substrate was an intermediary zone of non-mineralized cartilage, composed of cells with round morphology and extracellular matrix rich in proteoglycans. Changing the concentration of T3 treatment did not change the location of the mineralized zone (data not shown). However, when the number of initially seeded predifferentiated chondrocytes was decreased, the distance between the cartilage-substrate interface and the zone of mineralized cartilage also decreased (Figure 4.2 C & D). At an initial seeding of 1×106 predifferentiated chondrocytes, the zone of mineralized cartilage formed consistently adjacent to the interface (Figure 4.2 C & D, arrows).

106

Figure 4.2: Predifferentiated chondrocytes formed cartilaginous tissues on porous calcium polyphosphate substrates with a zone of mineralized cartilage. (A) ALP activity of predifferentiated chondrocytes cultured for 7 days with 3 nM triiodothyronine (T3) or 100 nM retinoic acid (RA) with or without dexamethasone (Dex). Mean ± 95% CI, n = 3, nd: not detected. * p = 0.043 between +T3 –Dex and +T3 +Dex; p = 0.045 between +T3 –Dex and +RA –Dex. (B) Von Kossa and toluidine blue-stained sections of cartilaginous tissue formed by culturing 2×106 predifferentiated chondrocytes on the substrates with 3 nM T3 for 3 weeks, showing mineralized zone occurring at the superficial aspect of cartilage. Asterisk denotes the original location of the substrate. Scale bar = 200µm. (C) Cross-sectional µCT tomographs and (D) and histology of 3-week-old constructs with decreasing numbers of seeded cells: 2×106 (left), 1.5×106 (middle) and 1.0×106 (right) predifferentiated chondrocytes. In (D) vertical bars denote non-mineralized cartilage (purple), mineralized zone (green) and the substrate (grey). Arrows indicate the lack of separation between the mineralized zone and the substrate. Scale bar = 1 mm (C) and 200 µm (D).

107

4.4.2 Short-term treatment of predifferentiated chondrocytes with T3 was sufficient to stimulate terminal differentiation

ALP activity levels of predifferentiated chondrocytes were characterized over a 2-week culture period. T3-treated cells had a significantly increased ALP activity by day 4 (Figure 4.3 A). Even though the T3 treatment was withdrawn at day 4, the ALP activity level of cells at day 7 was significantly higher than those that had been continuously treated with T3, and were comparable at days 10 and 14 (Figure 4.3 B).

To confirm that the 4-day T3 treatment was sufficient to induce terminal differentiation of predifferentiated chondrocytes, gene expression levels were examined. As early as 4 days, collagen type X (Col10a1) expression was significantly higher in the T3-treated cells compared to the untreated control (Figure 4.3 G). By 21 days, T3-treated cells expressed less collagen type II (Col2a1), aggrecan (Agc1) and Sox9, but greater Col10a1 and osteopontin (Spp1) compared to the non-treated controls (Figure 4.3 C–J). This confirmed that the 4-day T3 treatment of predifferentiated chondrocytes had altered the differentiation of these cells.

4.4.3 T3-treated predifferentiated chondrocytes did not induce ALP activity in non-T3-treated predifferentiated chondrocytes

Based on this observation, a two-stage culture protocol was developed (Figure 4.1). To grow an interfacial zone of mineralized cartilage, 1.0×106 predifferentiated chondrocytes were first seeded on porous calcium polyphosphate substrates, which was the number of cells that had been previously determined to be optimal for creating the mineralized cartilage at the interface, and cultured in the presence of T3 (Figure 4.2 C & D). T3 was withdrawn at 4 days of culture, then 1.5×106 predifferentiated chondrocytes were seeded on top of the existing tissue 1 day later and cultured in the presence of β-glycerophosphate for up to 21 days to generate a zone of non- mineralized cartilage. To confirm that the T3-treated cells would not stimulate the overlaid, non- T3-treated cells to also mineralize, the tissues on calcium polyphosphate were harvested 5 days after the addition of the top layer of cells (10 days of culture in total) and cryosectioned. Histological analysis demonstrated that ALP activity was present only in the calcifying bottom layer (Figure 4.4), demonstrating the lack of mineralization potential in the top layer even though β-glycerophosphate was present.

108

Figure 4.3: Short term T3 treatment of predifferentiated chondrocytes was sufficient to stimulate terminal differentiation. Predifferentiated chondrocytes were treated with 3 nM T3 for the first 4 days. In selected cultures, T3 treatment was withdrawn after 4 days (WD). (A) ALP activity during the first 4 days. * p < 0.001 between days 0–4 & 2–4. (B) ALP activity of WD, compared to cells in which T3 treatment was continued (+) for up to 14 days. * p = 0.014. (C–J) Gene expression analysis of WD on days 4 and 21 compared to non-T3-treated controls (–). * p = 0.013 (Col2a1 on day 21); p = 0.031 (Agc1 on day 21); p = 0.009 (Sox9 on day 21); p = 0.003 (Col10a1 on days 4 & 21); p = 0.025 (Spp1 on day 4) and p = 0.002 (Spp1 on day 21). Data points are color-coded for each animal throughout. Results are expressed as mean ± 95% CI, n = 3.

109

Figure 4.4: Layer of predifferentiated chondrocytes treated with T3 did not induce the overlaid layer of non-treated predifferentiated chondrocytes to activate ALP activity. 1×106 cells were cultured on porous calcium polyphosphate substrates with (A) or without (B) 3 nM T3 for 4 days. On day 5, 1.5×106 cells were seeded on top of these cells/tissue and cultured without T3. Cryosections of tissues from day 10 multiphasic constructs were stained for ALP activity. The blue staining indicative of ALP was seen only in the mineralizing layer (vertical bar). Asterisk denotes the location of where the substrate would have been located prior to its removal, scale bar = 200µm.

4.4.4 Characterization of the multiphasic constructs

Constructs were grown for 21 days using the two-stage culture protocol with or without the 4- day T3 treatment of the interfacial layer and characterized using various methods. Histological examination of cartilage tissue on the calcium polyphosphate showed two tissue zones. The upper zone of non-mineralized cartilage was hyaline cartilage, rich in type II collagen and proteoglycan (Figure 4.5 A, B, D). The interfacial zone was calcified cartilage, the mineral staining with von Kossa staining and the extracellular matrix staining with toluidine blue and type X collagen (Figure 4.5 A, B, E). Ingrowth of tissue into upper parts of the porous calcium polyphosphate substrate was also observed, with the mineral clusters directly in contact with the calcium polyphosphate particles (Figure 4.5A). No type I collagen was detected in either zone (Figure 4.5C). In contrast, control constructs not treated with T3 did not show a bizonal composition as there was no ZCC, and there was an uneven distribution of extracellular matrix (Figure 4.5 F–I). Differential collagen alignment in the tissue was detected using polarized light microscopy. In the superficial aspect, collagen was aligned parallel to the surface. Elsewhere,

110 collagen was observed both pericellularly and aligned orthogonal to the surface. Collagen, parallel to the surface, was also observed in areas within the bottom layer of tissue (Figure 4.5J). T3-treated constructs accumulated less proteoglycans (Figure 4.6A) and total collagens (Figure 4.6B) compared to untreated constructs, but the proteoglycan content was comparable to native sheep cartilage. Tissues on T3-treated constructs accumulated calcium and phosphate, consistent with calcification, while untreated constructs did not (Figure 4.6C).

Figure 4.5. A two-step culture protocol of predifferentiated chondrocytes on porous calcium polyphosphate substrates produced cartilage tissue with a mineralized zone at the interface. (A) Day 21 constructs with 4-day T3 treatment were processed undecalcified, sectioned and stained to reveal the cartilaginous (purple), mineral (green) and substrate particles (brown). (B–I) Tissues were excised from the substrate and examined histologically for the distribution of mineral and proteoglycan accumulation (B, F), by immunofluorescent staining for the accumulation of different types of collagen (C–E, G–I). Constructs with T3-treated interfacial layer are denoted as +T3, and untreated (no T3) control constructs are denoted as –T3. Scale bars = 200 µm. (J) Differential collagen fiber orientations on tissues was seen by polarized light microscopy. Scale bar = 100 µm.

111

Figure 4.6: Accumulation of extracellular matrix in T3-treated (+T3) tissues was less compared to those in untreated (–T3) tissues, but comparable to native articular cartilage. (A) Accumulated proteoglycan and (B) total collagen were quantified and normalized by their DNA content. (A) * p = 0.001, (B) * p = 0.012 between +T3 and –T3; p = 0.037 between +T3 and Native. (C) Accumulation of mineral in tissue was quantified by measuring the calcium and phosphate contents, normalized by tissue dry weight. ** p < 0.001. Native denotes data from sheep cartilage explants of femoral condyles. Results are expressed as mean ± 95% CI, n = 6.

T3-treated tissues had a comparable bulk equilibrium compressive modulus to that of osteochondral tissue obtained from native sheep femoral condyles (Figure 4.7A, p = 0.15). The untreated tissues had a significantly higher compressive modulus compared to that of the T3- treated tissues, possibly due to their greater extracellular matrix accumulation (Figure 4.6 A & B). However, T3-treated constructs could withstand a stronger shear force (Figure 4.7B) and absorb more energy (Figure 4.7C) than untreated constructs. Histological assessment of the failed constructs confirmed that failure occurred at the interface (data not shown). This demonstrated that the mineralized tissue at the interface (Figure 4.5A) enhanced the interfacial shear strength.

112

Figure 4.7: T3-treated constructs (+T3) exhibited comparable compressive strength as native sheep osteochondral explants and stronger shear strength than untreated constructs (–T3). (A) Thickness of tissues on T3-treated and untreated constructs. * p < 0.05. (B) Bulk equilibrium compressive modulus of tissues was measured by stress relaxation test with a 0.65 mm-diameter indenter situated at the centre of the tissue. * p = 0.008, ** p < 0.001. (C, D) A shear load was applied to the cartilage-substrate interface until failure. All substrates were identical in shape. The interfacial shear strength testing could not be performed on osteochondral explants due to slippage of the cartilage tissue. (C) * p = 0.003; (D) * p = 0.016. Results are expressed as mean ± 95% CI, n = 9 for constructs and 6 for explants.

113

4.5 Discussion

A mechanically competent interface between the cartilage and its substrate is required to resist the shear force generated by the movement of the joint. Previous studies have sought to recreate the ZCC that exists between hyaline cartilage and bone to reinforce this interface with limited success [22]. To overcome this problem, we successfully tissue-engineered osteochondral-like constructs with BMSCs as the single cell source, in which cartilage tissue and a porous bone substitute substrate were formed with an interfacial ZCC without using any exogenous scaffolds in the cartilage tissue. The two-stage culture protocol enabled the selective stimulation of predifferentiated chondrocytes to form calcified cartilage only at the interface. Treatment of the initially seeded cells with T3 induced the terminal differentiation of cells. This resulted in a ZCC, which, in addition to mineral, also contained collagens type II and X and proteoglycans. These features were seen only in the part of the tissue that was treated with mineralization- inducing T3, which was adjacent to (and integrated with) the porous calcium polyphosphate substrate. The interfacial shear strength of constructs with the calcified cartilaginous interface was increased significantly compared to those without it, functionally validating its intended effect. The hyaline quality of non-mineralized cartilage was demonstrated by the accumulation of collagen type II and the lack of mineral or collagen type I. Further study is required to determine whether this osteochondral-like tissue formation will occur on substrates other than hydroxyapatite-coated calcium polyphosphate substrate.

Thus far, approaches to induce selective mineralization of BMSC-derived cartilage and create this interface include delivery of bioactive signals either locally via microspheres [23] or by using gradients generated by a bioreactor [24] to undifferentiated BMSCs seeded in scaffolds. In another approach, BMSCs predifferentiated to chondrocytes and osteoblasts are seeded into appropriate locations within scaffolds [25, 26]. For both strategies, use of exogenous scaffolds is necessary, which has the potential to hinder host integration and tissue regeneration [27]. Matching the rate of scaffold degeneration to the rate of tissue accumulation remains a challenge when employing scaffolds [22, 28]. Synthetic materials also accumulate wear over time, and may provide a focal point for shear failure if not replaced with new tissue. On the other hand, if the rate of scaffold degradation exceeds the rate of cellular remodeling, the mechanical integrity is compromised and the construct would fail under load. A scaffold-free approach obviates this

114 issue altogether; however, this necessitated an alternative strategy to specify the location of the calcified interface.

The development of our two-stage culture protocol was first enabled by the observation that BMSC-derived predifferentiated chondrocytes formed mineralized cartilage at the tissue’s superficial aspect with mineral-inducing stimulus. Given that T3 was delivered to the tissue through culture media, and that the location of calcified cartilage did not change with tissue thickness, this effect was likely due to T3 diffusion into the tissue. As there was no scaffold present, selective delivery of T3 to the interface was difficult to achieve with media alone. To circumvent this, we seeded a small number of cells on the calcium polyphosphate in the presence of T3. This generated thin cartilaginous tissue, forcing mineralization to take place juxtaposed to the porous calcium polyphosphate substrate. This enabled the subsequent formation of a continuous layer of non-mineralized cartilage formed by the same cell type in the absence of T3. The presence of collagen types II, X and proteoglycans in the calcified interface, as well as the absence of collagen type I, is an important finding in our study. Taken together with the finding that T3-treated cells expressed hypertrophic chondrocyte makers, this demonstrates that the interface is cartilaginous, not osseous, successfully recapitulating characteristics of the native articular cartilage-subchondral bone interface.

Histology showed that the layer of non-mineralized cartilage successfully fused to the underlying T3-treated tissue. Interestingly, fissures were observed if the intended interfacial tissue (i.e., the first layer) was not treated with T3, and the total thickness of cartilage tissues on these constructs was increased, with an uneven distribution of the extracellular matrix. There are two possible explanations for these results. One is that the manner that the interfacial layer grows may affect integration, as BMSC-derived chondrocytes share many characteristics with proliferating cartilage [29] that grows appositionally. In development, Indian hedgehog, FGF and BMP signaling pathways regulate this appositional growth [30]; however, in an in vitro system, absence of such complex regulation may allow the interfacial layer to expand laterally. The presence of tubing around the constructs prevented lateral expansion by confinement, which could have resulted in lateral mechanical stress and caused the tissue to buckle and shear. This was not the case for T3-stimulated interfacial layer, however, whose behaviour may more resemble terminal hypertrophic chondrocytes, as evidenced by their mineralized matrix. Terminal hypertrophic chondrocytes do not proliferate and eventually undergo apoptosis upon

115 vascularization of matrix and bone formation [31]. Alternatively, the T3 treatment may have affected tissue fusion. Pellets of human MSCs undergoing chondrogenesis have been shown to accumulate tenascin along the outer surface that prevented them from fusing with one another [32]. The fissure between the two layers of non-mineralizing cartilage may stem from a similar boundary-setting phenomenon, whether driven by tenascin accumulation or otherwise.

Sheep gait studies estimate the maximum anterioposterior shear stress in a stifle joint to be about 500 kPa [33, 34], which is equivalent to approximately 9 N of peak shear force in our study. However, using our methodology the peak shear force of native cartilage interface could not be measured, as it withstood the maximum load (80 N) without shear failure. In healthy cartilage, collagen fibrils run orthogonally through the tidemark plane, which would confer a high resistance against failure in our mode of study. Although our multiphasic construct possesses inferior interfacial shear strength compared to articular cartilage, it still represented a four-fold improvement over the non-mineralized control, and sheep gait studies suggest that this improvement is adequate for preventing delamination.

Importantly, this protocol can be expanded to generate large, anatomically shaped osteochondral-like constructs as replacements for damaged joint tissues. Protocols have been developed to generate ~108 BMSCs with chondrogenic potential in a good manufacturing practices-complaint manner [35, 36], which can be used to generate sufficiently large numbers of chondrocytes to cover a large articular surface. We have previously shown that calcium polyphosphate powders can be used in additive manufacturing methods to generate substrates of required shape and size [37]. Calcium polyphosphate formed in this way exhibited the same degree of osseointegration after implantation as those produced conventionally [38]. As such, preclinical studies are warranted to determine if these osteochondral-like constructs with a mineralized cartilaginous interface could repair joint defects in vivo.

116

4.6 Conclusion

Using BMSCs as a single cell source, an osteochondral-like construct comprised of hyaline cartilage and porous bone substitute was created with a ZCC interfacing them. The novel, two- step cell culture strategy employed no exogenous scaffolds, just a substrate on which to form tissue. The presence of calcified cartilage increased the shear load that the construct could withstand at the interface. As no xenogeneic material was used at any point during culture, this strategy can be directly translated into preclinical studies to determine whether these constructs can repair joint defects in vivo.

4.7 Author Contributions The authors of the study were Whitaik David Lee, Mark B. Hurtig, Robert M. Pilliar, Rita A. Kandel and William L. Stanford.

• Study conception and design: WDL, WLS, RAK. • Provision of study materials: MBH, RMP. • Analysis and interpretation of data: WDL, RMP, RAK. • Drafting of article: WDL, WLS, RAK. • Critical revision of the article: WDL, RMP, WLS, RAK. • Final approval of the article: WDL, WLS, RAK.

4.8 Acknowledgments

We thank Eugene Hu for preparing the porous calcium polyphosphate substrates. We also thank Dr. Jian Wang and Nancy Valiquette for their technical assistance. This work was supported by the United States Department of Defense (to R.A.K., W81XWH-10-1-0787) and the Natural Sciences and Engineering Research Council of Canada (to W.L.S., RGPIN 293170-11). W.L.S. is supported by a Canadian Research Chair.

117

4.9 Supplementary Material

4.9.1 Isolation and expansion of sheep BMSCs using autologous serum

Media for the isolation and expansion of sheep BMSCs were prepared by adding 10% (v/v) autologous serum, generated by collecting blood from each sheep and allowing them to clot for 30 minutes, to minimum essential media α (αMEM; Wisent, St-Bruno, QC, Canada). Sheep bone marrow was treated with red blood lysis buffer (0.144 M ammonium chloride in 17 mM Tris-HCl buffer, pH 7.7) and plated at a cellular density of 2.5×105 cells/cm2. Non-adherent cells were removed after 24 hours, and cultures were maintained until 90% confluence was achieved. BMSCs were harvested for passaging with TrypLE Select (Life Technologies, Burlington, ON, Canada). BMSCs were cryopreserved after first passage. After thawing, BMSCs were seeded at 5000 cells/cm2 and further expanded for two passages in prior to chondrogenic differentiation.

4.9.2 Immunofluorescence of histological sections

To prepare histological slides, tissues were removed from their substrates and fixed in 10% neutral buffered formalin for 1 hour, followed by incubation in 30% (w/v) sucrose overnight at 4ºC and snap-freezing in the Tissue-Tek OCT compound. 6µm cross-section cryosections were cut and placed on silanized glass slides. For collagen type I visualization, cryosections were treated sequentially with 2.5 mg/mL trypsin and 25 mg/mL hyaluronidase, whereas for visualizing collagen types II and X, they were treated with 2 mg/mL pepsin (all Sigma-Aldrich). Sections were then incubated with antibodies reactive with the specific collagen (Type I: CalBiolChem, La Jolla, CA; type II: Labvision, Fremont, CA; type X: Sigma-Aldrich). DAPI (Life Technologies) counterstain was applied to all immunostained sections.

4.9.3 Analysis of interfacial shear strength

Using a specially designed jig attached to an Instron universal testing machine [15], a shear load was directly applied to the interface with a constant rate of displacement (1 mm/min) until a sudden decrease in resisting force was measured. The interface was considered to have failed at this point. The energy absorbed by the interface before failure was calculated by integrating the force-displacement curve.

118

4.9.4 List of primers used

Table 4.1: List of Primers for Chapter 4 Gene Dir Sequence (5’ à 3’) Accession F GTA ACC CGT TGA ACC CCA TT 18s rRNA DQ013885.1 (sheep) R CCA TCC AAT CGG TAG TAG CG F GGC CTG TCT CTC CGC GTT CA Col2a1 XM_004006408.1 (sheep) R ACG GTG GAC GGG GTC TGA CT F GGA GAA CAG GAC TGG AGA GG Agc1 XM_004018048.1 (sheep) R CTC CTC AAA GGT CAG CGA GT F GTG GTC CTT CTT GTG CTG C XM_004013527.1 (sheep) Sox9 R GTA CCC GCA CTT GCA CAA C NM_000346.3 (human) F GAA GAA GAC ATC CCA CCA GT XM_004012773.1 (sheep) Col1a1 R GGA CTT TGG CGT TAG GAC AG NM_001034039.2 (cow) F AGC AAC AGC ATT ACG ACC CA Col10a1 XM_004011185.1 (sheep) R ATT ACA GGG GTG CCG TTC TT F ATG TGA TGG ATA AAG GCT ATC CC Mmp13 NM_001166179.1 (sheep) R CGA ACG ATA CGG TTA CTC CAG A F GTC TTC ACA AAT CCT CCC CA Runx2 XM_004023440.1 (sheep) R ATA CTG GGA TGA GGA ATG CG F TCT GAT GAA TCT GAT GAA GTT G NM_001009224.1 (sheep) Spp1 R TGA AGT CCT CCT CTG TGG NM_174187.2 (cow)

4.10 References

[1] Swieszkowski W, Tuan BH, Kurzydlowski KJ, Hutmacher DW. Repair and regeneration of osteochondral defects in the articular joints. Biomol Eng (2007) 24:489-95.

[2] Shimomura K, Moriguchi Y, Murawski CD, Yoshikawa H, Nakamura N. Osteochondral Tissue Engineering with Biphasic Scaffold: Current Strategies and Techniques. Tissue Eng Part B Rev (2014) 20:468-76.

119

[5] Grynpas MD, Pilliar RM, Kandel RA, Renlund R, Filiaggi M, Dumitriu M. Porous calcium polyphosphate scaffolds for bone substitute applications in vivo studies. Biomaterials (2002) 23:2063-70.

[6] Pilliar RM, Kandel RA, Grynpas MD, Hu Y. Porous calcium polyphosphate as load-bearing bone substitutes: in vivo study. J Biomed Mater Res B Appl Biomater (2013) 101:1-8.

[7] Vahdati A, Wagner DR. Implant size and mechanical properties influence the failure of the adhesive bond between cartilage implants and native tissue in a finite element analysis. J Biomech (2013) 46:1554-60.

[8] Harris JD, Siston RA, Brophy RH, Lattermann C, Carey JL, Flanigan DC. Failures, re- operations, and complications after autologous chondrocyte implantation--a systematic review. Osteoarthritis Cartilage (2011) 19:779-91.

[9] Hoemann CD, Lafantaisie-Favreau CH, Lascau-Coman V, Chen G, Guzman-Morales J. The cartilage-bone interface. J Knee Surg (2012) 25:85-97.

[10] Mueller MB, Tuan RS. Functional characterization of hypertrophy in chondrogenesis of human mesenchymal stem cells. Arthritis Rheum (2008) 58:1377-88.

[11] Allan KS, Pilliar RM, Wang J, Grynpas MD, Kandel RA. Formation of biphasic constructs containing cartilage with a calcified zone interface. Tissue Eng (2007) 13:167-77.

[12] Zscharnack M, Hepp P, Richter R, Aigner T, Schulz R, Somerson J, et al. Repair of chronic osteochondral defects using predifferentiated mesenchymal stem cells in an ovine model. Am J Sports Med (2010) 38:1857-69.

[13] Shimomura K, Moriguchi Y, Ando W, Nansai R, Fujie H, Hart DA, et al. Osteochondral repair using a scaffold-free tissue-engineered construct derived from synovial mesenchymal stem cells and a hydroxyapatite-based artificial bone. Tissue Eng Part A (2014) 20:2291-304.

120

[14] Waldman SD, Grynpas MD, Pilliar RM, Kandel RA. Characterization of cartilagenous tissue formed on calcium polyphosphate substrates in vitro. J Biomed Mater Res (2002) 62:323- 30.

[15] St-Pierre JP, Gan L, Wang J, Pilliar RM, Grynpas MD, Kandel RA. The incorporation of a zone of calcified cartilage improves the interfacial shear strength between in vitro-formed cartilage and the underlying substrate. Acta Biomater (2012) 8:1603-15.

[16] Gan L, Pilliar R. Calcium phosphate sol-gel-derived thin films on porous-surfaced implants for enhanced osteoconductivity. Part I: Synthesis and characterization. Biomaterials (2004) 25:5303-12.

[17] Taylor DW, Ahmed N, Gan L, Gross AE, Kandel RA. Proteoglycan and collagen accumulation by passaged chondrocytes can be enhanced through side-by-side culture with primary chondrocytes. Tissue Eng Part A (2010) 16:643-51.

[18] Alini M, Kofsky Y, Wu W, Pidoux I, Poole AR. In serum-free culture thyroid hormones can induce full expression of chondrocyte hypertrophy leading to matrix calcification. J Bone Miner Res (1996) 11:105-13.

[19] Iwamoto M, Shapiro IM, Yagami K, Boskey AL, Leboy PS, Adams SL, et al. Retinoic acid induces rapid mineralization and expression of mineralization-related genes in chondrocytes. Exp Cell Res (1993) 207:413-20.

[20] Miao D, Scutt A. Histochemical localization of alkaline phosphatase activity in decalcified bone and cartilage. J Histochem Cytochem (2002) 50:333-40.

[21] Jiang J, Leong NL, Mung JC, Hidaka C, Lu HH. Interaction between zonal populations of articular chondrocytes suppresses chondrocyte mineralization and this process is mediated by PTHrP. Osteoarthritis Cartilage (2008) 16:70-82.

[22] Huey DJ, Hu JC, Athanasiou KA. Unlike bone, cartilage regeneration remains elusive. Science (2012) 338:917-21.

121

[23] Dormer NH, Singh M, Wang L, Berkland CJ, Detamore MS. Osteochondral interface tissue engineering using macroscopic gradients of bioactive signals. Ann Biomed Eng (2010) 38:2167- 82.

[24] Chen K, Ng KS, Ravi S, Goh JC, Toh SL. In vitro generation of whole osteochondral constructs using rabbit bone marrow stromal cells, employing a two-chambered co-culture well design. J Tissue Eng Regen Med (2013).

[25] Cheng HW, Luk KD, Cheung KM, Chan BP. In vitro generation of an osteochondral interface from mesenchymal stem cell-collagen microspheres. Biomaterials (2011) 32:1526-35.

[26] Grayson WL, Bhumiratana S, Grace Chao PH, Hung CT, Vunjak-Novakovic G. Spatial regulation of human mesenchymal stem cell differentiation in engineered osteochondral constructs: effects of pre-differentiation, soluble factors and medium perfusion. Osteoarthritis Cartilage (2010) 18:714-23.

[27] Castro NJ, Hacking SA, Zhang LG. Recent progress in interfacial tissue engineering approaches for osteochondral defects. Ann Biomed Eng (2012) 40:1628-40.

[28] Dhote V, Vernerey FJ. Mathematical model of the role of degradation on matrix development in hydrogel scaffold. Biomech Model Mechanobiol (2014) 13:167-83.

[29] Dickhut A, Pelttari K, Janicki P, Wagner W, Eckstein V, Egermann M, et al. Calcification or dedifferentiation: requirement to lock mesenchymal stem cells in a desired differentiation stage. J Cell Physiol (2009) 219:219-26.

[30] Kronenberg HM. Developmental regulation of the growth plate. Nature (2003) 423:332-6.

[31] Farnum CE, Wilsman NJ. Cellular turnover at the chondro-osseous junction of growth plate cartilage: analysis by serial sections at the light microscopical level. J Orthop Res (1989) 7:654- 66.

[32] Bhumiratana S, Eton RE, Oungoulian SR, Wan LQ, Ateshian GA, Vunjak-Novakovic G. Large, stratified, and mechanically functional human cartilage grown in vitro by mesenchymal condensation. Proc Natl Acad Sci U S A (2014) 111:6940-5.

122

[33] Taylor WR, Poepplau BM, Konig C, Ehrig RM, Zachow S, Duda GN, et al. The medial- lateral force distribution in the ovine stifle joint during walking. J Orthop Res (2011) 29:567-71.

[34] Lee-Shee NK, Dickey JP, Hurtig MB. Contact mechanics of the ovine stifle during simulated early stance in gait. An in vitro study using robotics. Vet Comp Orthop Traumatol (2007) 20:70-2.

[35] Chase LG, Yang S, Zachar V, Yang Z, Lakshmipathy U, Bradford J, et al. Development and characterization of a clinically compliant xeno-free culture medium in good manufacturing practice for human multipotent mesenchymal stem cells. Stem Cells Transl Med (2012) 1:750-8.

[36] Gottipamula S, Muttigi MS, Chaansa S, Ashwin KM, Priya N, Kolkundkar U, et al. Large- scale expansion of pre-isolated bone marrow mesenchymal stromal cells in serum-free conditions. J Tissue Eng Regen Med (2013).

[37] Shanjani Y, De Croos JN, Pilliar RM, Kandel RA, Toyserkani E. Solid freeform fabrication and characterization of porous calcium polyphosphate structures for tissue engineering purposes. J Biomed Mater Res B Appl Biomater (2010) 93:510-9.

[38] Shanjani Y, Hu Y, Toyserkani E, Grynpas M, Kandel RA, Pilliar RM. Solid freeform fabrication of porous calcium polyphosphate structures for bone substitute applications: in vivo studies. J Biomed Mater Res B Appl Biomater (2013) 101:972-80.

123

Chapter 5 Discussion & Conclusion

5.1 Discussion

The Kandel Lab has previously developed a methodology to create cartilage-CPP biphasic constructs for osteochondral defect repair using bovine and ovine primary articular chondrocytes [1, 2]. Since primary articular chondrocytes are not suitable for tissue engineering due to their scarcity and their propensity to de-differentiate during in vitro expansion, an alternative cell source of articular chondrocytes was necessary to implement this tissue engineering strategy. The first part of the thesis was devoted to investigating the use of BMSCs as a cell source for growing cartilage in vitro on the porous CPP substrate and creating a biphasic construct. Implantation of the BMSC-derived biphasic constructs revealed insufficient mechanical integration between the cartilage and the porous CPP substrate. The second part of the thesis addressed this problem by creating a zone of calcified cartilage selectively at the cartilage-CPP interface. The formation of a multi-zonal architecture in BMSC-derived cartilage improved the resistance of the constructs to shear load.

In Chapter 2, BMSCs successfully replaced primary articular chondrocytes in forming the biphasic constructs by pre-differentiating BMSCs to chondrocytes, which were subsequently harvested and grown on porous CPP substrate. Compared to cells that conventionally underwent chondrogenic differentiation in pellet culture, cells predifferentiated to chondrocytes on collagen-coated porous membranes yielded tissues on the CPP substrate that were superior in two aspects: less accumulation of collagen type I and higher equilibrium compressive strength. The mechanism by which the different predifferentiation methods affected the characteristics of cartilage tissue remains unclear. The collagen coating on the membrane inserts may have influenced cell differentiation, as different extracellular matrix proteins were shown to affect MSC proliferation and multipotency [3]. Cells in membrane cultures and pellet cultures experience different substrate stiffness, which was also shown to influence cell differentiation [4, 5]. The geometry of the cell mass in the predifferentiation cultures (sphere vs. disk) is another possible factor, as it could modify the mass transport properties of the tissues. This is supported 124 by the observations that the size of the chondrogenic pellets has a limit, beyond which adding additional cells does not increase its size [6], and that the loss of cells observed early in pellet culture is most pronounced at the centre of the pellet due to limits in passive diffusion [7]. It is therefore likely that the membrane culture has more favourable mass transport properties suitable for chondrogenic differentiation of a larger number of cells. In addition, cells in pellet culture are heterogeneous, and miniaturizing the pellet size can reduce this heterogeneity [8]. Comparing the heterogeneity of predifferentiated chondrocyte populations in pellet and membrane cultures would reveal further insight as to whether this is a factor in influencing tissue formation on the CPP substrate. To do this, RNA would have to be isolated from specific zones of tissues formed by predifferentiation culture by serial depth-wise tissue sections (similar to [9]) or laser capture microscopy from tissue cross-sections, and expression levels of various chondrocyte markers from each zone compared. Alternatively, single-cell qPCRs could be performed on isolated cells from predifferentiation cultures to establish the percentage distribution of cells expressing distinct sets (which could be defined post-hoc by principal component analysis) of chondrogenic markers.

The use of fetal bovine serum (FBS) in expanding BMSCs and culturing predifferentiated chondrocytes on CPP raises the concern about the immunogenicity of the constructs after implantation. Even though undifferentiated BMSCs grown in FBS were previously used in clinical trials, BMSC transplant programs such as that of National Institute of Health aims to use xeno-free protocols for manufacturing BMSCs in the future, citing risks of disease transmission and hypersensitivity [10]. Chondrocytes differentiated from MSCs are not immunoprivileged [11], and even a short-term exposure to FBS can cause cells to present bovine antigens [12]. Substitution of FBS with autologous serum posed a challenge as cells altered their behaviour in culture depending on the serum origin. This has also been observed as others, as the phenotype of MSCs varies widely depending on the methodology used to isolate and expand the cells, even when they were derived from the same source [13, 14]. Further, variability in the composition of autologous serum led to significant animal-to-animal variability of autologously expanded BMSCs. Data from Koellensperger et al. showed that variations among chondrogenic cultures of human adipose-derived stromal cells cultured in human serum were greater than cells cultured in FBS, likely due to varying growth factor contents among different serum samples [15]. The use of defined, serum-free media for BMSC culture was explored as an alternative at the beginning 125 of the research project; however, cells failed to expand in the serum-free media commercially available at that time. Since these experiments were performed, others studies reported the successful use of xeno-free, clinically compliant human MSC production [16, 17]. Adapting these defined culture conditions for the initial isolation and expansion of sheep MSCs may reduce the animal-to-animal variability and improve the reproducibility of results in future studies.

Results from the pilot in vivo study are described in Appendix A. The pilot study revealed the loss of BMSC-derived cartilage from constructs created without a zone of calcified cartilage at the cartilage-substrate interface. Since cartilage grown on porous CPP substrate ex vivo interdigitated with the substrate through its pores, resistance against shear load was expected even in the absence of a calcified interface. Interestingly, in our previous study, biphasic constructs formed using primary articular chondrocytes possessed very low interfacial shear strength [18], yet 60% of the implants survived implantation in sheep joints [2]. This discrepancy may be explained in part due to the different locations of these implants. Constructs formed with primary articular chondrocytes were placed in the trochlear groove of the femur, whereas the BMSC-derived constructs were placed in the femoral condyle. In human knees, peak compressive stresses in normal gait are 11.5 ± 3.0 MPa in the patellofemoral joint and 14 ± 2.5 MPa in the tibiofemoral joint [19, 20]. No comparable data was found for sheep. Regardless, this highlighted the need for enhanced mechanical integration between the cartilage and the CPP substrate. Challenges in enhancing mechanical integration of this biphasic construct system were twofold. First, since CPP releases polyP as a biodegradation by-product that inhibits mineralization in its proximity, a method to limit polyP release by CPP while preserving the material properties of CPP itself was required. Second, a method to selectively mineralize the BMSC-derived cartilage at the interface while simultaneously preventing the mineralization of the rest of the tissue needed to be developed.

In Chapter 3 of the thesis, a methodology to limit the polyP release from the porous CPP substrate was developed by adapting the organic sol gel thin film processing technique [21, 22]. Hydroxyapatite films formed by sol gel processing on metal substrates have been shown to be both osteoconductive and susceptible to biodegradation by osteoclasts [23], suggesting that this might be an appropriate approach. The hydroxyapatite films on porous CPP inhibited polyP release from the substrate in vitro when there were no cells being cultured. On the other hand, 126 when deep zone chondrocytes were cultured on the substrate, tissue formed on coated CPP substrate accumulated polyP accumulation in tissue was observed, with tissue on ISG-coated substrate accumulating more polyP than tissue on OSG-coated substrate. We have not directly investigated how polyP accumulated on the tissues grown on the coated substrates. There are several possible explanations: first, it could be that the deep-zone chondrocytes degraded the hydroxyapatite film. As discussed in Chapter 4, previous studies have showed that inorganic sol gel-derived hydroxyapatite coating is nanoporous and therefore possesses more surface area where degradation would occur. This would explain how tissue formed on ISG-coated porous CPP substrates accumulated more polyP compared to tissue formed on OSG-coated porous CPP substrates. Another possibility is that the presence of hydroxyapatite coating may have stimulated the synthesis and accumulation of endogenous polyP by deep zone chondrocytes. While the mechanism of polyP synthesis in mammalian cells is unknown, different amounts of calcium and phosphate released from ISG and OSG coating, as well as their nanoscale topological differences, may have affected the level of accumulated endogenous polyP in tissues.

Immersion in simulated body fluid was considered as an alternate method for coating the surface of porous CPP substrates, as it has been used to deposit a thin hydroxyapatite film on a wide variety of materials, including those that are calcium phosphate-based [24-27]. However, in our studies, simulated body fluid failed to deposit hydroxyapatite on porous CPP (data not shown). This may be due to the lack of functional groups at the CPP surface that favour the nucleation of apatite crystals [28], or the inhibitory effect of polyP on apatite crystal growth in the same way it inhibits mineralization of tissues in close proximity.

Our study showed that OSG coating of porous CPP substrates significantly enhanced the interfacial shear strength of the cartilage-substrate constructs and ISG coating did not. As previously discussed, OSG-derived and ISG-derived coatings on porous CPP substrates are likely to have the same nanoscale differences in porosity as those on titanium substrates. This provided the rationale for our hypothesis, which was that the increase in interfacial shear strength was due to differential polyP accumulation in tissues grown on OSG versus ISG coated CPP substrates. Yet, testing our hypothesis proved to be difficult because of the lack of methods to quantify the polyP accumulation in tissues [29]. Appendix B describes a novel method to quantify polyP accumulated by tissues grown in vitro. Until now, studies of polyP accumulation in tissue were limited to qualitative analysis via histology using DAPI. However, DAPI cannot 127 be used to quantify polyP accumulation in cartilage because DAPI binds to proteoglycans and other polyanionic biopolymer molecules to produce the same fluorescence shift as polyP. In the novel assay, proteoglycans do not interfere with the signal – at least in quantities found in in vitro-grown cartilage tissue. Efforts to further refine the protocol to increase its sensitivity for quantifying the polyP in native cartilage are ongoing.

To further investigate the effect of sol-gel coating on tissue polyP accumulation, attempts were made to visualize the distribution of accumulated polyP in tissues. DAPI has been used for visualization of polyP-rich organelles such as acidocalcisomes and vacuoles using the dye’s fluorescence shift [30], but since DAPI binds to other polyanionic biopolymer molecules abundant in cartilaginous matrix, fluorescence in tissue sections could not be positively interpreted to be from polyP. Angelova et al. recently published a more specific dye for polyP that did not bind to nucleic acids, but the dye also produced signal when bound to heparin, a proteoglycan [31]. Enzymatic treatment of tissue samples to eliminate proteoglycans is impractical because of the wide variety of proteoglycans present in cartilage. A possible method for future studies is the use of a recombinant polypeptide containing the polyP-binding domain of E. coli exopolyphosphatase, which in turn is visualized by fluorescent immunostaining of the polypeptide [32]. This method was used to visualize polyP in yeast vacuoles and fungal cell walls [32, 33].

Chapter 4 of the thesis describes the method by which selective mineralization of BMSCs at the interface was achieved in a two-step process, using organic sol gel-coated CPP described in the previous Chapter. This is the first scaffold-free (in the cartilage phase) cartilage tissue engineering technology to incorporate a zone of calcified cartilage at the substrate interface to create an osteochondral-like constructs that possess high compressive and shear strength. However, polarized light microscopy revealed incomplete integration of collagen fibres between the mineralized and the hyaline cartilage layers. Indeed, a thin band of acellular cartilage between the two cartilage layers was observed in some samples. In vivo studies suggested that hyaluronidase and collagenase treatment of cartilage explants prior to transplantation improved lateral cartilage integration and interfacial strength, with accumulation of collagen fibres running across the interface to surrounding native cartilage [34]. Using this approach in our system failed to yield consistent results. Bhumiratana et al. demonstrated that chondrogenic pellets could fuse early on in culture before a peripheral boundary marked by tenascin expression emerged [35]. It 128 remains to be explored whether predifferentiated chondrocytes, rather than MSCs, also establish such a boundary, tenascin or otherwise; and, whether the interfacial, T3-treated layer culture period could be shortened so that cells comprising the hyaline cartilage layer could be seeded earlier to enhance the integration of the two layers.

In Chapter 4, some unexpected observations were made on the non-treated control constructs, in which the first layer of predifferentiated chondrocytes was cultured without T3 treatment before seeding the second layer. Compared to tissues on T3-treated constructs, the resulting tissues were thicker and had an uneven distribution of extracellular matrix, which accumulated significantly more GAG and collagens (normalized to DNA content). Further, the bulk equilibrium compressive modulus of tissues on non-T3-treated controls was higher than that of both T3-treated constructs and the native sheep cartilage. Higher GAG and collagen content could be explained by the fact that hypertrophic chondrocytes affect catabolic changes to its extracellular matrix [36], leading to lower GAG and collagen content. The mechanism by which tissues on non-treated constructs were thicker and had a higher bulk equilibrium compressive modulus cannot be easily explained; however, it should be noted that higher bulk equilibrium compressive modulus does not always indicate better tissue-engineered cartilage, as cartilage in older individuals deform less under load compared to younger individuals [37]. The uneven distribution of extracellular matrix indicated poor structural integrity, which suggests that it would have less resistance under shear load applied at its articular surface compared to the T3- treated constructs. Characterization of the interfacial layer in isolation would shed additional insight as to why the tissues on non-treated constructs had a higher compressive modulus, such as the collagen fibril orientation, matrix composition, etc. Nevertheless, even though the non-T3- treated constructs may have a higher bulk equilibrium compressive modulus, the poor structural integrity and the lack of interfacial shear load resistance make the T3-treated constructs a more useful result for cartilage repair.

Through these studies, a method was developed to form cartilage with a multi-zonal architecture on porous CPP substrate to produce osteochondral-like constructs with biomechanical properties that approach native articular cartilage using BMSCs as a cell source in a sheep model. While the calcified zone was purposefully formed by the two-step culture method, the rest of the cartilage architecture relied on the cells’ self-organization. Remarkably, the hyaline layer exhibited some characteristics of healthy articular cartilage architecture. At its 129 most superficial aspect, cells appeared elongated and the collagen fibres ran parallel to the surface. Since the superficial zone performs a critical role in maintaining the biological properties and the overall homeostasis of tissue in healthy articular cartilage, the construct’s superficial aspect warrants further characterization. To accomplish this, an immunofluorescence protocol for Prg4 in sheep tissues and a method to assess the coefficient of friction in tissues ([38] for example) are required.

Although constructs were grown for a relatively short period of time (3 weeks) devoid of mechanical stimuli, their bulk equilibrium compressive modulus was comparable to native sheep cartilage. This may appear as an indication that additional mechanical stimuli may not be needed for creating osteochondral-like constructs with physiologically relevant biomechanical properties; however, it has been demonstrated that dynamic compression stimulus can also induce changes in the collagen fibril arrangement [41] and suppress hypertrophic changes in BMSC-derived chondrocytes [42], which would be beneficial to the hyaline layer of the construct. In addition, application of shear and compressive loads on cartilage grown in vitro using articular chondrocytes resulted in the upregulation of Prg4 gene and protein expression as well as hyaluronan production compared to compressive load alone [43]. Therefore, dynamic compression could be employed to yet further improve both biochemical and biomechanical properties of the engineered construct ex vivo prior to implantation.

Although the biomechanical properties of the native osteochondral interface could not be measured due to the limitation in the methodology used to assess the interfacial shear strength, it is at least an order of magnitude higher than the interfacial shear strength of the multi-zonal osteochondral-like constructs (~7N), since cylindrical osteochondral explants of the same shape withstood the maximum load by our methodology (80N) without shear failure. However, this high level may not be required to prevent construct failure by delamination. Shear loads in a joint are applied at the surface of cartilage tissue. The maximum anterioposterior shear force on the sheep stifle joint contact force is approximately 13 ± 12% of its body weight [44], but the amount of shear load transmitted to the cartilage-substrate interface will be dependent on the biological properties of the tissue-engineered cartilage, primarily determined by the quality of the superficial zone and also affected by the synovial fluid [38]. Hence, in vivo validation of these constructs is required to determine the success of the new methodology. 130

Whether the hyaline, non-mineralized layer of BMSC-derived cartilage would remain non-mineralized after implantation is an important question to address whether these multi-zonal constructs are to be used to repair joint defects. Predifferentiated chondrocytes cultured on porous CPP substrates expressed genetic markers of terminal differentiation and appeared hypertrophic when examined histologically, which were consistent with the conventional understanding that BMSC-derived chondrocytes inevitably undergo terminal differentiation [45]. Yet, the predifferentiated chondrocytes were devoid of cellular ALP activity, and the cartilage tissue formed by these cells did not accumulate either collagen type X or mineral in the extracellular matrix, even in the presence of β-glycerophosphate, a condition that has been shown to be sufficient to cause mineralization in other systems [18, 46, 47]. Compared to other in vitro-formed cartilage described in the literature, a unique feature of cartilage in this study was that the tissue accumulated polyP from the CPP substrate, regardless of whether the substrate is coated. The inhibitory effect by PolyP on tissue mineralization has been demonstrated [48, 49], but the reasons why there were no cellular ALP activity and collagen type X in the extracellular matrix is unclear. This motivates further research on the effect of polyP on the phenotype of BMSC-derived cartilage. Also, it raises the possibility of the CPP substrate preventing tissue mineralization after implantation as it degrades in vivo. Intra-articular injection of polyP in an osteoarthritic guinea pig model slowed down the progression of cartilage degeneration, demonstrating its in vivo potential [50]. In the study, hydroxyapatite coating was applied on the CPP substrate to reduce the cartilage tissue’s polyP accumulation; however, given the possible benefits, the coating should be optimized in future studies to allow the polyP release once the mineralized cartilage-CPP interface is formed.

Validation of these methods developed in this thesis requires an in vivo implantation study. Currently, work is ongoing to create large, anatomically shaped implants that will replace the entire medial side of a trochlear groove in the patellofemoral joint.. Additive manufacturing techniques were used to form joint-shaped porous CPP substrates [51]. The articular surface of these substrates was an order of magnitude greater in area (~400mm2, compared to 28mm2 of the 6mm tapered plugs) and possessed a convex curve along the proximal-distal axis. A method to form a polydimethylsiloxane mould was devised to create a watertight 4cm-tall well around the substrate, which contained the suspension of predifferentiated chondrocytes on top of the CPP substrate until they attached to the surface. This method was validated with articular 131 chondrocytes and showed the formation of cartilage tissue of even thickness throughout the articular surface. Using this method, 80 million predifferentiated chondrocytes were seeded in two stages to form cartilage tissue on the articular surface with a calcified interface. To the author’s knowledge, only a single other report of MSC-derived osteochondral construct has been reported to date [35]. In that study, immature chondrogenic human MSC pellets were press- formed to fuse with each other and onto decellularized bone as the substrate. Our ongoing study represents an alternative that incorporates a calcified interface with a synthetic substrate. Since the work was done in a large animal model, the study can be directly tested in vivo.

Challenges remain in translating the autologous BMSC-based methodology presented in this thesis towards generating osteochondral-like constructs for clinical trials. First, acquisition of bone marrow aspirates is an invasive procedure. More easily accessible sources of MSCs such as infrapatellar fat pad [52] may be considered for use. Second, donor-to-donor variability will be even greater than what the sheep study presented, in which donors were selected from a limited range of age, sex (all ewes) and genetic variation (breed). Some studies showed that aging diminishes the proliferation and multipotency of MSCs [53, 54], and this may pose a limitation for the application of this technology to older patients. Both challenges may be obviated if cells from allogeneic sources could be used. As discussed previously, chondrocytes differentiated from MSCs are immunogenic [11] and they may not be suitable for direct implantation. On the other hand, osteochondral allografts do not elicit immune responses if the cartilage matrix is intact [55]. This can be extended to hypothesize that cartilage tissue formed ex vivo from allogeneic cells would not be immunogenic. However, tissue-engineered cartilage has much less accumulation of cartilage matrix compared to healthy articular cartilage, so further studies are necessary to determine whether the extracellular matrix in tissue-engineered cartilage is sufficient to afford such protection to allogeneic cells.

A significant advancement in using human pluripotent cells for cartilage tissue engineering was recently made by Craft et al., who demonstrated the in vitro generation of articular chondrocytes for the first time by using human pluripotent cells via paraxial mesoderm and chondroprogenitor intermediates [56]. Chondroprogenitors generated in the presence of either TGF-β3 gave rise to chondrocytes that did not mineralize in vivo, while those generated in the presence of BMP4 gave rise to chondrocytes that did mineralize: therefore, the two cell types could respectively replace the BMSC-derived chondrocytes at the hyaline and calcified layers to 132 generate the osteochondral-like constructs described in this thesis project. This is a very exciting advancement, as this would establish an autologous cell-sourced protocol that would overcome many drawbacks in the current method. Firstly, the concern for eventual mineralization of tissue- engineered cartilage noted in BMSC-based methods is averted altogether. Secondly, induced pluripotent cells can be generated from virtually any cell type, thereby obtaining the source cells by minimally invasive sampling. Thirdly, pluripotent cells can be clonally derived due to their self-renewal capacity and rapid proliferation rate, avoiding the heterogeneity of MSCs and enabling extensive and reliable characterization of the cell source that can satisfy the rigorous quality control necessary for clinics.

5.2 Future Experiments

The importance of superficial zone (SZ) to articular cartilage’s biomechanical function and homeostasis were previously discussed in the Introduction section. Therefore, it would be highly desirable for the tissue-engineered multi-zonal constructs to incorporate the SZ. There exists evidence in the literature that MSCs can be differentiated to chondrocytes that express SZ chondrocyte-specific genes [57]. Histological examination of the osteochondral-like constructs in Chapter 4 revealed features that suggest the existence of SZ at the superficial aspect of hyaline zone, such as the relatively flattened cell morphology, lower proteoglycan staining and the orientation of collagen fibres being parallel to the surface. Attempts have been made to further characterize this zone. For example, the superficial aspect was dissected from flash-frozen tissues using a cryotome; however, insufficient quantity of RNA was collected. Use of techniques for small RNA quantity – glycogen microcarriers for Trizol preparation or micro spin columns – as well as pooling numerous samples would improve the outcome of this study. Also, several Prg4 antibodies were tested for immunofluorescence using sections of sheep articular cartilage without success. A sheep-specific Prg4 antibody was not available, and multiple alignment analysis of Prg4 protein and mRNA sequences revealed that Prg4 is a poorly conserved protein across mammalian species, leading to further complications in identifying a cross-reacting Prg4 antibody. Also, the possibility of sheep native cartilage having relatively low Prg4 content in the SZ compared to other species cannot be ruled out. However, if the 133 osteochondral-like constructs were found to spontaneously establish a SZ, this would be the first demonstration of such kind in scaffold-free tissue-engineered cartilage tissue.

As discussed earlier, the mechanical integration between the hyaline and calcified zones of cartilage warrants further investigation. While its separation has not been observed in interfacial shear loading tests, the possibility of shear failure taking place at the interface of the two zones under shear loading applied to the top surface of the hyaline zone was not directly tested in the study. Application of shear loading to the top surface not only can provide further insight into its mechanical properties, but can also enhance the SZ phenotype [58]. Also, the enzymatic treatment of the calcified layer’s surface warrants further consideration as the evidence for enhanced cartilage integration exists in the literature [34].

The effect of polyP on chondrogenic differentiation and post-differentiation phenotype maintenance of BMSCs should be pursued further. This investigation could provide clues as to why BMSC-derived chondrocytes on CPP substrates did not readily mineralize its matrix given comparable stimuli for tissue mineralization in other systems as previously discussed. The availability of the new polyP quantification assay makes possible a more thorough characterization of tissues that are exposed to polyP, as well as measuring the endogenous polyP in tissues. Part of this investigation should include the formation of cartilage tissue with a calcified zone using substrates other than hydroxyapatite-coated porous CPP substrate, such as porous sintered titanium. While this would serve as an exogenous polyP-free control that would be crucial to investigating the effect of polyP on tissues, it would also demonstrate whether the technology developed in this project could be applied to other substrates.

Ongoing efforts to create large, anatomical-shaped implants using the methods developed in the study revealed challenges in controlling animal variability and scaling up predifferentiation of BMSCs. One challenge is that the process is time-consuming. By eliminating the use of autologous serum and adapting a defined media for the isolation and expansion of BMSCs, animal variability could be reduced. Use of larger membrane inserts for predifferentiation is currently being investigated, which would reduce both material and labour costs necessary to seed and harvest from the predifferentiation cultures. To reduce the length of time required to create the constructs, several ideas have been proposed. The use of allogeneic cells would not directly reduce the actual length of time required to create the constructs, but it 134 would simplify the logistics of in vivo implantation study. Seeding BMSCs directly on porous CPP substrate and inducing its chondrogenic differentiation at the intended articulating surfaces was previously attempted, but histological examination of the resulting tissue revealed large clusters of cell death proximal to the substrate. However, since CPP substrates are coated with hydroxyapatite to control for polyP release, this strategy may be worth a revisit as well.

Finally, the methods developed in this study must be validated on human BMSCs, as the eventual goal of this study is to contribute to the development of new methods to repair damaged joints in clinics. Investigating the use of allogeneic cells would be especially valuable for this goal. In vitro assays for assessing the immunogenicity of cells and tissues are established in the literature (for example, [11]). Use of human BMSCs would also enable a more in-depth study of the constructs since many more investigative methods are available for samples of human origin than sheep origin (e.g. microarrays and antibodies). The use of other MSC sources should also be explored, such as those from infrapatellar fat pad [52], while recognizing that the required stimuli for MSCs from other sources to achieve the same outcome may differ than those for BMSCs.

5.3 Conclusion

Using a two-step process in which BMSCs were predifferentiated to chondrocytes in membrane cultures and subsequently cultured on porous CPP substrates, osteochondral-like constructs were engineered for in joint defect repair in a sheep preclinical model. Formation of a ZCC at the cartilage-substrate interface enhanced the mechanical integration between the cartilage and the porous CPP substrates. The organic sol-gel processing method for applying a thin barrier coating of hydroxyapatite on porous CPP substrates was effective in controlling the polyP release from the substrate that inhibits the mineralization of cartilage at the interface. A novel method to quantify the accumulated polyP in cartilage tissues for the first time was developed during this investigation. On coated substrates, BMSC-derived chondrocytes were sequentially cultured – in the presence of T3 only for the first seeding – to selectively mineralize the interfacial region of the osteochondral-like constructs. By doing so, tissue engineering of multi-zonal, osteochondral- like constructs without the use of a scaffold was achieved for the first time. The constructs possessed comparable equilibrium compressive modulus and, compared to non-mineralized 135 control, enhanced interfacial shear strength. In vivo implantation study of these constructs will verify whether the enhanced mechanical integration is sufficient to withstand the forces present in a joint. The methodology is currently being used to generate large, anatomically shaped constructs for whole joint replacement.

136

5.4 References

[1] Waldman SD, Grynpas MD, Pilliar RM, Kandel RA. Characterization of cartilagenous tissue formed on calcium polyphosphate substrates in vitro. J Biomed Mater Res (2002) 62:323-30.

[2] Kandel RA, Grynpas M, Pilliar R, Lee J, Wang J, Waldman S, et al. Repair of osteochondral defects with biphasic cartilage-calcium polyphosphate constructs in a sheep model. Biomaterials (2006) 27:4120-31.

[3] Lindner U, Kramer J, Behrends J, Driller B, Wendler NO, Boehrnsen F, et al. Improved proliferation and differentiation capacity of human mesenchymal stromal cells cultured with basement-membrane extracellular matrix proteins. Cytotherapy (2010) 12:992-1005.

[4] McBeath R, Pirone DM, Nelson CM, Bhadriraju K, Chen CS. Cell shape, cytoskeletal tension, and RhoA regulate stem cell lineage commitment. Dev Cell (2004) 6:483-95.

[5] Engler AJ, Sen S, Sweeney HL, Discher DE. Matrix elasticity directs stem cell lineage specification. Cell (2006) 126:677-89.

[6] Shirasawa S, Sekiya I, Sakaguchi Y, Yagishita K, Ichinose S, Muneta T. In vitro chondrogenesis of human synovium-derived mesenchymal stem cells: optimal condition and comparison with bone marrow-derived cells. J Cell Biochem (2006) 97:84-97.

[7] Dexheimer V, Frank S, Richter W. Proliferation as a requirement for in vitro chondrogenesis of human mesenchymal stem cells. Stem Cells Dev (2012) 21:2160-9.

[8] Markway BD, Tan GK, Brooke G, Hudson JE, Cooper-White JJ, Doran MR. Enhanced chondrogenic differentiation of human bone marrow-derived mesenchymal stem cells in low oxygen environment micropellet cultures. Cell Transplant (2010) 19:29-42.

[9] Muller C, Khabut A, Dudhia J, Reinholt FP, Aspberg A, Heinegard D, et al. Quantitative proteomics at different depths in human articular cartilage reveals unique patterns of protein distribution. Matrix Biol (2014) 40:34-45. 137

[10] Stroncek DF, Sabatino M, Ren J, England L, Kuznetsov SA, Klein HG, et al. Establishing a bone marrow stromal cell transplant program at the National Institutes of Health Clinical Center. Tissue Eng Part B Rev (2014) 20:200-5.

[11] Ryan AE, Lohan P, O'Flynn L, Treacy O, Chen X, Coleman C, et al. Chondrogenic differentiation increases antidonor immune response to allogeneic mesenchymal stem cell transplantation. Mol Ther (2014) 22:655-67.

[12] Spees JL, Gregory CA, Singh H, Tucker HA, Peister A, Lynch PJ, et al. Internalized antigens must be removed to prepare hypoimmunogenic mesenchymal stem cells for cell and gene therapy. Mol Ther (2004) 9:747-56.

[13] Lennon DP, Haynesworth SE, Bruder SP, Jaiswal N, Caplan AI. Human and animal mesenchymal progenitor cells from bone marrow: Identification of serum for optimal selection and proliferation. In Vitro Cell Dev-An (1996) 32:602-11.

[14] Prockop DJ. Repair of tissues by adult stem/progenitor cells (MSCs): controversies, myths, and changing paradigms. Mol Ther (2009) 17:939-46.

[15] Koellensperger E, Bollinger N, Dexheimer V, Gramley F, Germann G, Leimer U. Choosing the right type of serum for different applications of human adipose tissue-derived stem cells: influence on proliferation and differentiation abilities. Cytotherapy (2014) 16:789-99.

[16] Pal R, Hanwate M, Jan M, Totey S. Phenotypic and functional comparison of optimum culture conditions for upscaling of bone marrow-derived mesenchymal stem cells. J Tissue Eng Regen Med (2009) 3:163-74.

[17] Chase LG, Yang S, Zachar V, Yang Z, Lakshmipathy U, Bradford J, et al. Development and characterization of a clinically compliant xeno-free culture medium in good manufacturing practice for human multipotent mesenchymal stem cells. Stem Cells Transl Med (2012) 1:750-8.

[18] Allan KS, Pilliar RM, Wang J, Grynpas MD, Kandel RA. Formation of biphasic constructs containing cartilage with a calcified zone interface. Tissue Eng (2007) 13:167-77.

[19] Thambyah A, Goh JC, De SD. Contact stresses in the knee joint in deep flexion. Med Eng Phys (2005) 27:329-35. 138

[20] Akbarshahi M, Fernandez JW, Schache AG, Pandy MG. Subject-specific evaluation of patellofemoral joint biomechanics during functional activity. Med Eng Phys (2014) 36:1122-33.

[21] Liu DM, Yang Q, Troczynski T. Sol-gel hydroxyapatite coatings on stainless steel substrates. Biomaterials (2002) 23:691-8.

[22] Gan L, Pilliar R. Calcium phosphate sol-gel-derived thin films on porous-surfaced implants for enhanced osteoconductivity. Part I: Synthesis and characterization. Biomaterials (2004) 25:5303-12.

[23] Gan L, Wang J, Tache A, Valiquette N, Deporter D, Pilliar R. Calcium phosphate sol-gel- derived thin films on porous-surfaced implants for enhanced osteoconductivity. Part II: Short- term in vivo studies. Biomaterials (2004) 25:5313-21.

[24] Duan YR, Zhang ZR, Wang CY, Chen JY, Zhang XD. Dynamic study of calcium phosphate formation on porous HA/TCP ceramics. J Mater Sci Mater Med (2005) 16:795-801.

[25] Hirakata LM, Kon M, Asaoka K. Evaluation of apatite ceramics containing α-tricalcium phosphate by immersion in simulated body fluid. Biomed Mater Eng (2003) 13:247-59.

[26] Kasuga T, Sawada M, Nogami M, Abe Y. Bioactive ceramics prepared by sintering and crystallization of calcium phosphate invert glasses. Biomaterials (1999) 20:1415-20.

[27] Kokubo T, Ito S, Huang ZT, Hayashi T, Sakka S, Kitsugi T, et al. Ca,P-rich layer formed on high-strength bioactive glass-ceramic A-W. J Biomed Mater Res (1990) 24:331-43.

[28] Ohtsuki C, Kamitakahara M, Miyazaki T. Coating bone-like apatite onto organic substrates using solutions mimicking body fluid. J Tissue Eng Regen Med (2007) 1:33-8.

[29] Rao NN, Gomez-Garcia MR, Kornberg A. Inorganic polyphosphate: essential for growth and survival. Annu Rev Biochem (2009) 78:605-47.

[30] Gomes FM, Ramos IB, Wendt C, Girard-Dias W, De Souza W, Machado EA, et al. New insights into the in situ microscopic visualization and quantification of inorganic polyphosphate stores by 4',6-diamidino-2-phenylindole (DAPI)-staining. European Journal of Histochemistry (2013) 57. 139

[31] Angelova PR, Agrawalla BK, Elustondo PA, Gordon J, Shiba T, Abramov AY, et al. In Situ Investigation of Mammalian Inorganic Polyphosphate Localization Using Novel Selective Fluorescent Probes JC-D7 and JC-D8. ACS Chem Biol (2014) 9:2101-10.

[32] Saito K, Ohtomo R, Kuga-Uetake Y, Aono T, Saito M. Direct labeling of polyphosphate at the ultrastructural level in Saccharomyces cerevisiae by using the affinity of the polyphosphate binding domain of Escherichia coli exopolyphosphatase. Appl Environ Microbiol (2005) 71:5692-701.

[33] Werner TP, Amrhein N, Freimoser FM. Specific localization of inorganic polyphosphate (poly P) in fungal cell walls by selective extraction and immunohistochemistry. Fungal Genet Biol (2007) 44:845-52.

[34] van de Breevaart Bravenboer J, In der Maur CD, Bos PK, Feenstra L, Verhaar JA, Weinans H, et al. Improved cartilage integration and interfacial strength after enzymatic treatment in a cartilage transplantation model. Arthritis Res Ther (2004) 6:R469-76.

[35] Bhumiratana S, Eton RE, Oungoulian SR, Wan LQ, Ateshian GA, Vunjak-Novakovic G. Large, stratified, and mechanically functional human cartilage grown in vitro by mesenchymal condensation. Proc Natl Acad Sci U S A (2014) 111:6940-5.

[36] Mackie EJ, Tatarczuch L, Mirams M. The skeleton: a multi-functional complex organ: the growth plate chondrocyte and endochondral ossification. The Journal of endocrinology (2011) 211:109-21.

[37] Hudelmaier M, Glaser C, Hohe J, Englmeier KH, Reiser M, Putz R, et al. Age-related changes in the morphology and deformational behavior of knee joint cartilage. Arthritis Rheum (2001) 44:2556-61.

[38] Schmidt TA, Sah RL. Effect of synovial fluid on boundary lubrication of articular cartilage. Osteoarthritis Cartilage (2007) 15:35-47.

[39] Grodzinsky AJ, Levenston ME, Jin M, Frank EH. Cartilage tissue remodeling in response to mechanical forces. Annu Rev Biomed Eng (2000) 2:691-713. 140

[40] Tran SC, Cooley AJ, Elder SH. Effect of a mechanical stimulation bioreactor on tissue engineered, scaffold-free cartilage. Biotechnol Bioeng (2011) 108:1421-9.

[41] Nagel T, Kelly DJ. Mechanically induced structural changes during dynamic compression of engineered cartilaginous constructs can potentially explain increases in bulk mechanical properties. J R Soc Interface (2012) 9:777-89.

[42] Bian L, Zhai DY, Zhang EC, Mauck RL, Burdick JA. Dynamic compressive loading enhances cartilage matrix synthesis and distribution and suppresses hypertrophy in hMSC-laden hyaluronic acid hydrogels. Tissue Eng Part A (2012) 18:715-24.

[43] Grad S, Lee CR, Gorna K, Gogolewski S, Wimmer MA, Alini M. Surface motion upregulates superficial zone protein and hyaluronan production in chondrocyte-seeded three- dimensional scaffolds. Tissue Eng (2005) 11:249-56.

[44] Taylor WR, Poepplau BM, Konig C, Ehrig RM, Zachow S, Duda GN, et al. The medial- lateral force distribution in the ovine stifle joint during walking. J Orthop Res (2011) 29:567-71.

[46] Farrell E, Both SK, Odorfer KI, Koevoet W, Kops N, O'Brien FJ, et al. In-vivo generation of bone via endochondral ossification by in-vitro chondrogenic priming of adult human and rat mesenchymal stem cells. BMC Musculoskelet Disord (2011) 12:31.

[47] Farrell E, van der Jagt OP, Koevoet W, Kops N, van Manen CJ, Hellingman CA, et al. Chondrogenic priming of human bone marrow stromal cells: a better route to bone repair? Tissue Eng Part C Methods (2009) 15:285-95.

[48] Fleisch H, Russell RG, Straumann F. Effect of pyrophosphate on hydroxyapatite and its implications in calcium homeostasis. Nature (1966) 212:901-3.

[49] Ariganello MB, Omelon S, Variola F, Wazen RM, Moffatt P, Nanci A. Osteogenic cell cultures cannot utilize exogenous sources of synthetic polyphosphate for mineralization. J Cell Biochem (2014) 115:2089-102. 141

[50] St-Pierre JP, De Croos JN, Theodoropoulos JS, Petrera M, Sharma P, Li S, et al. Inorganic polyphosphate exhibits anabolic effects on articular cartilage. Osteoarthritis and Cartilage (2012) 20, Supplement 1:S39.

[51] Shanjani Y, De Croos JN, Pilliar RM, Kandel RA, Toyserkani E. Solid freeform fabrication and characterization of porous calcium polyphosphate structures for tissue engineering purposes. J Biomed Mater Res B Appl Biomater (2010) 93:510-9.

[52] Liu Y, Buckley CT, Almeida HV, Mulhall KJ, Kelly DJ. Infrapatellar Fat Pad-Derived Stem Cells Maintain Their Chondrogenic Capacity in Disease and Can be Used to Engineer Cartilaginous Grafts of Clinically Relevant Dimensions. Tissue Eng Part A (2014).

[53] Sethe S, Scutt A, Stolzing A. Aging of mesenchymal stem cells. Ageing Res Rev (2006) 5:91-116.

[54] Stolzing A, Jones E, McGonagle D, Scutt A. Age-related changes in human bone marrow- derived mesenchymal stem cells: consequences for cell therapies. Mech Ageing Dev (2008) 129:163-73.

[55] Langer F, Gross AE. Immunogenicity of allograft articular cartilage. J Bone Joint Surg Am (1974) 56:297-304.

[56] Craft AM, Rockel JS, Nartiss Y, Kandel RA, Alman BA, Keller GM. Generation of articular chondrocytes from human pluripotent stem cells. Nat Biotechnol (2015).

[57] Coates EE, Fisher JP. Engineering superficial zone chondrocytes from mesenchymal stem cells. Tissue Eng Part C Methods (2014) 20:630-40.

[58] Li Z, Yao S, Alini M, Grad S. Different response of articular chondrocyte subpopulations to surface motion. Osteoarthritis Cartilage (2007) 15:1034-41.

142

Appendix A In Vivo Validation of Biphasic Constructs for Repair of Focal Osteochondral Defects

A.1 Development of FBS-free protocol for biphasic construct culture

Chapter 2 of the thesis described a method to form cartilage-porous CPP biphasic constructs using BMSCs that were predifferentiated to chondrocytes in membrane culture. This method was divided into isolation and expansion of BMSCs, predifferentiation in membrane culture and culturing predifferentiated chondrocytes on porous CPP substrates. Fetal bovine serum (FBS) was a component of the culture medium for isolating/expanding BMSCs and for growing predifferentiated chondrocytes on porous CPP substrates [1]. However, it has previously shown that exposure of cells to FBS could cause cells to present antigens of bovine origin [2]. Undifferentiated MSCs cultured in the presence of FBS were well tolerated when transplanted [3] due in part to their immunomodulative properties [4]; however, chondrocytes differentiated from MSCs do not possess such immunomodulative property and are capable of eliciting an immune response [5]. Therefore, to minimize the immunogenic potential of constructs and the risk of rejection, FBS was substituted with autologous serum.

BMSCs isolated and expanded in the presence of autologous serum showed several differences in culture compared to those cultured in the presence of FBS. BMSCs cultured in autologous serum on monolayer did not detach with trypsin; instead, they were successfully detached with TrypLE Select (Invitrogen). At the beginning of the membrane predifferentiation culture, BMSCs cultured in autologous serum failed to stay adhered on collagen-coated membranes and formed a cell aggregate instead. Blebbistatin is a small-molecule inhibitor of non-muscle myosin II, an actin-binding protein involved in both the attachment and motor activity of cells [6]. Addition of 10 µM blebbistatin to the chondrogenic media for the first 72 hours of membrane culture prevented the formation of cell aggregates. After 72 hours, blebbistatin was withdrawn and cells were cultured in the chondrogenic media as described in

143

Chapter 2 up to 21 days. Cells cultured with blebbistatin for the first 72 hours did not differ in their expression levels of Col2a1, Col1a1, Sox9 at days 1, 3, 7 or 21 (Figure A.1).

Figure A.1: Blebbistatin treatment did not affect chondrogenic differentiation of BMSCs in membrane predifferentiation cultures. Expression levels of blebbistatin-treated cells were normalized to levels of animal and time point-matched non-treated cells. Results are expressed as mean ± SEM, n = 3, p > 0.05 between treatment and control for each gene and time point.

Finally, predifferentiated chondrocytes did not form tissue on porous CPP substrate in the presence of autologous serum (Figure A.2). However, predifferentiated chondrocytes formed cartilaginous tissues when cultured in a serum-free media, composed of high-glucose Dulbecco’s modified Eagle’s media (DMEM), insulin-transferrin-selenium plus (ITS+) tissue culture supplement, 100 nM dexamethasone and 100 µg/mL ascorbic acid. The resulting tissues were found to accumulate a low amount of collagen. In the literature, proline is sometimes added to chondrogenic culture media [7-9] as collagens are abundant in proline and hydroxyproline.

Preliminary data suggested that addition of 50 µg/mL L-proline to the serum-free media could increase both collagen and glycosaminoglycan contents of cartilage tissue grown serum-free on porous CPP substrates (Figure A.3).

144

Figure A.2: Predifferentiated chondrocytes derived from FBS-free BMSCs gave rise to tissue on porous CPP substrates in serum-free media. Surface on which cells were seeded are shown facing right. Culture media, from left to right: DMEM/F12 (1:1) with 10% autologous serum, DMEM/F12 (1:1) with 20% autologous serum, Ham’s F12 with 20% autologous serum, DMEM with 20% autologous serum and serum-free BCM. 100 µg/mL ascorbic acid was added to all cultures.

Figure A.3: In a preliminary study, addition of L-proline in media increased extracellular matrix accumulation by 4 weeks of culture. BCM = basal chondrogenic medium. n = 1.

145

A.2 In vivo evaluation: repair of focal defects

A.2.1 Materials and Methods

4 rams of 2 ½ years of age were chosen as test subjects for the study. BMSCs were isolated and expanded through two passages in the presence of autologous serum and predifferentiated on collagen type IV-coated membrane inserts for 2 weeks. Predifferentiated chondrocytes were isolated from membrane cultures and cultured on porous CPP substrates (1.7×105 cells/mm2) for 4 weeks in serum-free media. The substrates were shaped as tapered cylinders with a diameter of 6.5 mm at the intended articulating surface and a height of 7.3 mm. After 4 weeks of culture, constructs were transported to the Large Animal Clinic at the Ontario Veterinary College and implanted into the right stifle (hindleg) of donor-matched animals. During surgery, a tapered hole matching the shape of the implant was drilled on the medial femoral condyle. Then, the implant was carefully tapped into the defect. Animals were ambulated and allowed to move about freely after surgery. After 3 months, animals were euthanized and their knee joints were harvested. The implants and the surrounding native tissues were fixed in 10% formalin buffer, dehydrated and embedded in polymethyl methacrylate (Osteo-Bed bone embedding kit, Polysciences Inc., Warrington, PA, USA) without decalcification. Sections were cut and ground to ~50 µm thickness, stained with toluidine blue and light green and visualized by light microscopy.

A.2.2 Results

During the study period, animals displayed normal gait, appetite and behaviour. During dissection of harvested joints, the synovial fluid was clear and colourless. The fluid was free of debris. Gross and histological images of femoral condyles with biphasic constructs retrieved 3 months after implantation are shown in Figure A.4. Gross examination of retrieved joints showed partial tissue loss and cavitation of articulating surface in 3 of 4 animals (Figure A.4 A, C, E). Cartilage tissue was completely lost in one joint, exposing the substrate surface (Figure A.4 G). Histological examination of implants and the surrounding tissue revealed incomplete cartilage tissue coverage of articulating surface on the substrate (Figure A.4 B, D, F). A thin layer of cartilage persisted on the substrate’s surface (Figure A.4 D, F). Mineralized tissue covered the articulating surface of the substrate that sunk below the articulating surface in one implant (Figure A.4 B). Mineralization of the remaining cartilage was observed in another implant,

146 which was separated from the substrate (Figure A.4 F). Loss of cartilage tissue was observed wherever the substrate surface was proud in relation to the adjacent osteochondral junction (Figure A.4 D, right; F, left; H, left and right). In one implant, the proud corner of the substrate was round, likely due to erosion (Figure A.4 H). Mineralization (green staining) was visible in the pores of the substrates at their lateral aspects interfacing bone, demonstrating bone ingrowth in all except one implant (Figure A.4 H, right edge).

Figure A.4: Gross and histological images of biphasic constructs after 3 month of implantation. All 4 implants are shown with matching gross and histology. (A, C, E, G) Arrows indicate the location of the implant. (B, D, F, H) Implants and the surrounding tissue were processed undecalcified, sectioned and stained to reveal cartilage (purple), mineral (green) and substrate particles (brown). Scale bars = 1 mm.

147

A.2.3 Discussion

The pilot in vivo implantation study revealed that, at 3 months, BMSC-derived cartilage tissue on porous CPP substrates failed to survive. The mechanism by which the cartilage tissues were lost was unknown; however, given the appearance of failed implants, it was hypothesized that the cartilage tissues sheared off the substrate under mechanical stress, supported by the near- complete absence of cartilage tissue covering the substrate surface. The round edges of the substrate surface in Figure A.4 H suggested that the shear force in the tibiofemoral joint was sufficient to debride porous CPP. This demonstrated that interdigitation of cartilage and the porous CPP substrate at the interface alone was not sufficient for resisting the shear forces in the sheep femoral joint.

In one implant, calcification of what appeared to be implanted tissue was observed (Figure A.4 B). BMSC-derived chondrocytes are predisposed to terminal differentiation [10], but it was hypothesized that the orthotopic environment in which the constructs were implanted would prevent terminal differentiation. Others have reported that tissue-engineered cartilage of osteochondral constructs in sheep model did not mineralize [11-13]. Exposure of CPP under tissue-engineered cartilage may have injured the joint, leading to a hostile environment of the synovium to the implanted cartilage, even though no overt signs of abnormal synovium were observed during the harvesting of knees. Osteoarthritis patients exhibit various degrees of synovitis, whose macrophages exhibit an activated phenotype and produce both proinflammatory and angiogenic cytokines [14, 15].

A.3 Conclusion

The method to create biphasic constructs was modified to reduce the immunogenic potential of constructs. Using this methodology, 4 constructs were created and implanted in the femoral condyle of sheep. 3 months later, the observed disappearance of implanted cartilage tissues indicated that interdigitation of cartilage and the porous CPP substrate at the interface was not sufficient to resist the shear loads of the joint. Therefore, reinforcement of this cartilage-substrate interface is needed.

148

A.4 References

[1] Lee WD, Hurtig MB, Kandel RA, Stanford WL. Membrane culture of bone marrow stromal cells yields better tissue than pellet culture for engineering cartilage-bone substitute biphasic constructs in a two-step process. Tissue Eng Part C Methods (2011) 17:939-48.

[2] Spees JL, Gregory CA, Singh H, Tucker HA, Peister A, Lynch PJ, et al. Internalized antigens must be removed to prepare hypoimmunogenic mesenchymal stem cells for cell and gene therapy. Mol Ther (2004) 9:747-56.

[3] Horwitz EM, Gordon PL, Koo WK, Marx JC, Neel MD, McNall RY, et al. Isolated allogeneic bone marrow-derived mesenchymal cells engraft and stimulate growth in children with osteogenesis imperfecta: Implications for cell therapy of bone. Proc Natl Acad Sci U S A (2002) 99:8932-7.

[4] Nasef A, Ashammakhi N, Fouillard L. Immunomodulatory effect of mesenchymal stromal cells: possible mechanisms. Regen Med (2008) 3:531-46.

[5] Ryan AE, Lohan P, O'Flynn L, Treacy O, Chen X, Coleman C, et al. Chondrogenic differentiation increases antidonor immune response to allogeneic mesenchymal stem cell transplantation. Mol Ther (2014) 22:655-67.

[6] Kovacs M, Toth J, Hetenyi C, Malnasi-Csizmadia A, Sellers JR. Mechanism of blebbistatin inhibition of myosin II. J Biol Chem (2004) 279:35557-63.

[7] Abrahamsson CK, Yang F, Park H, Brunger JM, Valonen PK, Langer R, et al. Chondrogenesis and mineralization during in vitro culture of human mesenchymal stem cells on three-dimensional woven scaffolds. Tissue Eng Part A (2010) 16:3709-18.

[8] Ahmed N, Dreier R, Gopferich A, Grifka J, Grassel S. Soluble signalling factors derived from differentiated cartilage tissue affect chondrogenic differentiation of rat adult marrow stromal cells. Cell Physiol Biochem (2007) 20:665-78.

[9] Mauck RL, Yuan X, Tuan RS. Chondrogenic differentiation and functional maturation of bovine mesenchymal stem cells in long-term agarose culture. Osteoarthritis Cartilage (2006) 14:179-89.

149

[10] Dickhut A, Pelttari K, Janicki P, Wagner W, Eckstein V, Egermann M, et al. Calcification or dedifferentiation: requirement to lock mesenchymal stem cells in a desired differentiation stage. J Cell Physiol (2009) 219:219-26.

[11] Zscharnack M, Hepp P, Richter R, Aigner T, Schulz R, Somerson J, et al. Repair of chronic osteochondral defects using predifferentiated mesenchymal stem cells in an ovine model. Am J Sports Med (2010) 38:1857-69.

[12] Marquass B, Somerson JS, Hepp P, Aigner T, Schwan S, Bader A, et al. A novel MSC- seeded triphasic construct for the repair of osteochondral defects. J Orthop Res (2010) 28:1586- 99.

[13] Mrugala D, Bony C, Neves N, Caillot L, Fabre S, Moukoko D, et al. Phenotypic and functional characterisation of ovine mesenchymal stem cells: application to a cartilage defect model. Ann Rheum Dis (2008) 67:288-95.

[14] Bondeson J. Are we moving in the right direction with osteoarthritis drug discovery? Expert Opin Ther Targets (2011) 15:1355-68.

[15] Haywood L, McWilliams DF, Pearson CI, Gill SE, Ganesan A, Wilson D, et al. Inflammation and angiogenesis in osteoarthritis. Arthritis Rheum (2003) 48:2173-7.

150

Appendix B Development of a High-Sensitivity Method for Isolating and Quantifying Inorganic Polyphosphates from Cartilage Tissues Using Silica Spin Columns

This chapter is to be submitted to Analytical Chemistry.

B.1 Abstract Inorganic polyphosphates (polyP) play a multitude of roles in mammalian biology. PolyP research is hindered by the lack of an easy-to-use, sensitive quantification method. We developed a benchtop method for quantifying polyP from biological samples, in which polyP was isolated from enzyme-treated samples using silica membrane spin columns and quantified fluorometrically. The silica spin column method can recover at least 80% of initially loaded polyP, which can be as little as 10-10 mol. The eluent was contaminated with nucleic acids and glysaminoglycans that cause extraneous fluorescence signal, which were minimized by nuclease treatment and addition of concentrated Tris buffer. Proteinase K pre-treatment reduced the inhibitory effect of serum on polyP recovery. PolyP was quantified from in vitro-formed and native cartilage, with a correction factor empirically determined by exogenously adding polyP to tissue samples. Although the protocol was developed for mammalian tissues, utility of this method can be extended to most biological sources. With other complementary assays to characterize the distribution of polyP chain length, this method could provide a useful research tool for future polyP investigations in mammalian biology.

151

B.2 Introduction

Inorganic polyphosphates (polyP) are linear polymers of phosphate residues linked by high- energy phosphoanhydride bonds. While polyP’s ubiquity and its functions in prokaryotes are well established [1], its function in eukaryotes remains relatively unexplored. Recent studies with mammalian cells have demonstrated polyP’s role in coagulation [2, 3], proliferation [4], signaling [5, 6], skeletal mineralization [7], cartilage mineralization [8] and glial cell function [9]. However, further research is limited by the lack of sensitive and simple analytical methods to assess polyP quantity [1].

Current existing methods for polyP quantification in mammalian tissues have significant limitations. Mass spectrometry or chromatographic methods have been used [10, 11], but those are limited to shorter polyP chains and cannot quantify biologically relevant, longer polyP chains, such as polyP of 45 residues that play a role in cartilage biomineralization [12], and polyP of 60 to 100 residues that are present in platelets [13]. Nuclear magnetic resonance (NMR) techniques can quantify both the quantity and the chain length of polyP [14, 15]. However, this method is too expensive for routine quantification of polyP in biological samples. Another approach is to hydrolyze polyP to orthophosphates by treating cell/tissue extracts with Saccharomyces cerevisiae exopolyphosphatase or by converting polyP to adenosine 5’- triphosphate by treating with Escherichia coli polyphosphate kinase [16]. The quantity of polyP is then measured in equivalent phosphate units, but these are indirect methods and the enzymes are not readily available. Aschar-Sobbi et al. described a fluorometric alternative that exploits the shift in the excitation-emission peaks of 4',6-diamidino-2-phenylindole (DAPI) when bound to polyP [17]. Like enzymatic assays, the results of the fluorometric quantification method is given in equivalent Pi units; however, this method requires a high purity of samples as many other biological molecules can similarly shift the fluorescence spectrum of DAPI, such as DNA and RNA [18], glycosaminoglycans (GAG) and proteins, and give false false readings.

Quantification of polyP from cells and tissues requires extraction, for which several different methods have been developed. To isolate polyP from mammalian cells and tissues, Kumble and Kornberg described a phenol-chloroform extraction method [16]. Silica adsorption- based polyP isolation techniques were described for E. coli and S. cerevisiae [19, 20], but Kulakova et al. reported that the polyP recovery from silica was poor and dependent on the

152 amounts loaded [21]. The study suggested that direct quantification of polyP (using the fluorometric quantification method) without a prior isolation step may be optimal for quantifying polyP in microbial cells. Martin and Van Mooy refined this approach by compensating for the “matrix effect,” which corrects for the extraneous signal [18]. However, correcting for extraneous fluorescence from mammalian tissues may be more challenging as the concentration of polyP in mammalian cells is reported to be in the order of 10-5 M, 100-fold less than prokaryotes [16].

In this study, we describe a new silica membrane spin column-based polyP isolation method from which sub-nanomole quantities of polyP, whose chain lengths range from 45 to 130 residues, can be recovered with minimal loss and quantified with fluorometry. The quantification step was also optimized to minimize the extraneous signal from residual nucleic acids and GAG that persisted after the isolation step. The method does not require specialized analytical equipment, thus it is cost effective and readily adaptable for use in biological assays where multiple sample analysis is required.

153

B.3 Materials and Methods

B.3.1 Preparation of polyP standard solutions

PolyP solution was prepared by dissolving sodium phosphate glass (45 phosphate residues: Sigma-Aldrich, Oakville, Ontario, Canada; 23 phosphate residues: SPER Chemicals, Clearwater,

FL, USA) in ddH2O at 4ºC overnight. PolyP solutions with 14, 60 and 130 phosphate residues were provided by Dr. Toshikazu Shiba (Regenetiss Inc., Tokyo, Japan). All stock polyP standard solutions were kept at –30ºC at a concentration of 5 mM until use. PolyP amounts and concentrations are expressed in equivalent phosphate units. Unless otherwise stated, all experiments were performed with polyP chain length 45.

B.3.2 Silica spin column isolation and fluorometric quantification of polyP

140 µL of samples were first mixed with 280 µL of binding buffer and incubated for 10 min in room temperature. Binding buffer consisted of 5.0 M guanidine thiocyanate (GuSCN), 0.9 M sodium citrate, 25 mM ethylenediaminetetraacetic acid (EDTA), 1% β-mercaptoethanol and 50 mM Tris (all Sigma-Aldrich). Then, 280 µL of 100% ethanol was added to the samples and incubated for 3 min in room temperature. Samples were loaded in EconoSpin silica membrane mini spin columns (Epoch Life Science, Sugar Land, TX, USA) and centrifuged for 30 s at 10,000 × g. Following this step, 550 µL of wash buffer I was loaded in each column and centrifuged for 1 min at 12,000 × g. Wash buffer I consisted of 1.0 M GuSCN in 80% ethanol (Commercial Alcohols, Toronto, ON, Canada) solution. Then, 550 µL of wash buffer II was loaded in each column and centrifuged for 1 min at 12,000 × g. Wash buffer II consisted of 150 mM sodium chloride (Sigma-Aldrich) and 10mM Tris in 80% ethanol solution, pH 7.5. Finally, 300 µL of wash buffer II was loaded in each column and centrifuged for 2 min at 12,000 × g. PolyP was eluted from the spin columns by loading 60 µL of 10 mM Tris buffer, pH 7.2, incubating for 3 min at room temperature and then centrifuging for 1 min at 12,000 × g. This step was performed three times and the eluents collected for a total volume of 180 µL.

PolyP in the eluent was quantified fluorometrically as previously described [17] with the following modifications. Aliquots of the eluent was dispensed into wells of a black 96-well microplate (Corning, Corning, NY, USA) in triplicates. Then, 50 µL of DAPI solution (50 µg/mL in 10mM Tris pH 7.5 buffer; Invitrogen, Oakville, ON, Canada) was dispensed in

154 each well and incubated for 5 min at room temperature to achieve steady-state fluorescence signal for all polyP chain lengths examined [17]. Fluorescence was measured at an excitation wavelength of 415 nm and an emission wavelength of 558 nm using Fluoroskan Ascent FL microplate fluorometer (Thermo Scientific, Waltham, MA, USA). Concentration of polyP in eluents was calculated using a standard curve of polyP solution (1µM–60µM).

B.3.3 Determination of the polyP recovery ratios from silica spin columns

The silica spin column’s capacity to bind and elute polyP was determined by loading known amounts of polyP (140 µL/sample). 415/558nm DAPI fluorescence was measured in the solution prior to loading and in the eluents from the silica spin columns. The polyP recovery ratio was calculated as the ratio of recovered polyP versus the initially loaded polyP. PolyP quantities listed in Table 2 were loaded to compare the recovery ratios of the method in this study to previous studies [21]. The lowest amount of polyP that can be reliably recovered using the study’s method was determined by measuring the polyP recovery ratios of 0.25–8 nmol (1.79– 57.1µM) polyP. Chain length-dependent recovery ratios were determined using 7 nmol (50 µM) polyP standards of different chain lengths.

B.3.4 Determination of potential sources of interference for the polyP quantification assay

To evaluate the effect of possible molecules known to contribute to DAPI fluorescence, solutions of calf thymus DNA, S. cerevisiae RNA, adenosine triphosphate (ATP), sodium pyrophosphate, chondroitin sulfate (the predominant GAG present in cartilage [22]) and bovine serum albumin (BSA; all Sigma-Aldrich) were prepared at concentrations listed in Table B.1. FBS (Wisent, St- Bruno, Québec, Canada) solution was used to study protein interference as FBS contains 35mg/mL protein (manufacturer’s analysis). 7 nmol (50 µM) polyP was then added to 140 µL of these different macromolecular solutions and the silica spin column isolation step was performed. The DAPI fluorescence of eluents was compared to samples that were treated similarly but were not passed through a silica column.

155

Table B.1: Potential sources of extraneous 415 nm/558 nm DAPI fluorescence present in mammalian tissues, per 106 chondrocytes in 140µL volume.

Species Concentration Amount in cells/tissue used

DNA 55 µg/mL 7.7 pg per bovine articular chondrocytes [33]

RNA 180 µg/mL 20–30 pg per mammalian cell [34]

ATP 6.0 µM 0.5–10 mM per mammalian cell [35]

Pyrophosphate 5.0 µM 655 ± 45 pmol / 106 chondrocytes [36]

Chondroitin sulfate 6.1 mg/mL 112 ± 14 µg GAG per µg DNA in bovine cartilage explants [24]

Protein 7.3 mg/mL 2.22 ± 0.17 mg protein per gram of articular cartilage [37], 1.68 × 106 cells per gram of articular cartilage [38]

B.3.5 Elimination of DAPI fluorescence from DNA and RNA

Combinations of 55 µg/mL DNA and 180 µg/mL RNA samples (Table B.1) were treated with 10 U DNase I and 2.5 U RNase A (both Roche Diagnostics, Indianapolis, IN, USA) per 180 µL sample with or without 7 nmol polyP (38.9 µM) at 37ºC for 60 min in a nuclease buffer (100 µM calcium chloride, 1 mM magnesium chloride, 50 µg/mL bovine serum albumin (all Sigma- Aldrich) and 10 mM Tris pH 7.2). After 60 min, 6 µL of 0.5 M EDTA was added to the samples and to arrest the enzymatic activity. DAPI fluorescence from enzyme-treated and untreated samples were measured. In selected samples, Tris solution (pH 7.5) was added to increase the Tris concentration of the samples and standards up to 500 mM in order to determine if this would further decrease fluorescence.

B.3.6 Elimination of DAPI fluorescence from chondroitin sulfate

Silica spin column isolation step was performed on 140 µL of chondroitin sulfate (6.1 mg/mL in ddH2O) with or without 7 nmol polyP (50 µM in ddH2O). GAG contents of eluents were quantified using dimethylmethylene blue dye and spectrophotometry (absorbance at 525 nm)

156

[23]. 415nm/558nm DAPI fluorescence of eluents, some of which Tris solution (pH 7.5) was added to increase the Tris concentration of the samples to 500 mM, were measured. A standard curve of 415 nm/558 nm DAPI fluorescence in the presence of chondroitin sulfate was generated by adding DAPI to serial dilutions of chondroitin sulfate made in either 10 mM or 500 mM Tris pH 7.5 and measuring the fluorescence.

B.3.7 Proteinase K pre-treatment of samples

To determine a method to extract polyP from protein containing tissues/solutions, 7 nmol polyP (50 µM) were mixed with 20% or 50% (v/v) FBS samples and incubated in the presence or absence of proteinase K (1 mg/mL, Life Technologies, Burlington, Ontario, Canada) in 10 mM Tris, pH 8.0, for 4 h at 56ºC with periodic agitation. 10 mM EDTA were also added to selected samples. Silica spin column isolation was performed on proteinase K-treated samples and Tris pH 7.5 solution was added to the eluents for a final Tris concentration of 500mM before 415nm/558nm DAPI fluorescence of the eluents was measured. Recovery ratio was calculated by dividing the DAPI fluorescence of eluted samples by the non-enzyme-treated control (50 µM polyP in 10 mM Tris-EDTA pH 8.0).

B.3.8 PolyP quantification in in vitro-grown and native cartilage samples

Full-thickness and superficial zone (approximately top 15% of thickness) articular cartilage were harvested from the metacarpal-phalangeal joint of 9–12 month old calves. To generate in vitro- formed cartilage tissues, articular chondrocytes were isolated from the full-thickness cartilage by sequential enzyme digestion and cultured on either porous calcium polyphosphate (CPP) substrates (2.0 × 106 chondrocytes per each 4mm diameter disk) for 2–4 weeks, or on collagen type II-coated membrane culture inserts (2.0 × 106 chondrocytes per insert) as previously described [12, 24]. Wet weights of harvested tissues were measured before storing at –80ºC.

280 µL of 1 mg/mL proteinase K solution in a buffer consisting of 10 mM EDTA and 10 mM Tris pH 8.0 were added to either full-thickness or superficial zone native cartilage (28 mg), as well as each in vitro-grown tissue (8–15 mg), and incubated for 2 h at 56ºC with periodic agitation. Each sample was divided in two aliquots of equal volume, and 7 nmol polyP was added to one of the aliquots (50 µM, “spiked”). Samples were then incubated further for 2 h at 56ºC. For positive control, 50 µM polyP alone in proteinase K solution in the absence of

157 tissues was incubated under identical conditions. Following incubation, the full polyP quantification assay was performed (Figure B.6). For each sample, the recovered amount of exogenously added polyP was determined from the difference in DAPI fluorescence between the spiked and non-spiked aliquots. The recovery rate of exogenously added polyP was calculated by dividing the recovered amount of exogenously added polyP by the measured polyP content of the positive control. Then, the polyP content measured in each non-spiked aliquot was divided by the recovery rate to compensate for the loss of polyP during the assay and normalized to corresponding wet weight to derive the final polyP content of tissues. The GAG content of tissue samples were quantified from non-spiked aliquots using the dimethylmethylene blue dye assay and normalized to corresponding wet weight as done previously [25].

B.3.9 Statistical Analysis

All experiments were repeated at least three times. Sample sizes are indicated for each figure. One-way analysis of variance and Tukey’s pairwise post-hoc tests were used to compare the recovery ratios of polyP standards in the absence of tissue, fluorescence of potentially interfering molecules in the presence of polyP, fluorescence of nuclease-treated samples in increasing Tris concentration, and the quantified polyP and GAG data from tissues. Two-way analysis of variance and Bonferroni post-tests were used to compare the fluorescence and recovery ratios of either untreated/treated or pre-/post-isolation samples, as well as the FBS-added samples. Statistical significance was assigned at p < 0.05.

B.3.10 Safety Considerations

GuSCN can produce a toxic gas, hydrogen cyanide, upon contact with acids. As precaution, it is recommended that GuSCN-containing buffers be prepared and handled in a fume hood. GuSCN- containing waste was collected with concentrated sodium hydroxide solution, such that the final concentration of sodium hydroxide solution would be no less than 0.5 N when discarded.

158

B.4 Results

B.4.1 Binding and recovery of polyP from silica spin columns

The polyP recovery ratios were comparable between the three initial quantities of polyP ranging between 15.5 and 77.6 nmol (p = 0.2714). These ratios were significantly higher than previously described recovery ratios at the same polyP quantities (Table B.2). When less than 8 nmol polyP were loaded, a statistically significant decrease in the recovery ratio was observed at 2 nmol or lower (Figure B.1A), consistent with the observation of others that the recovery ratio decreases with smaller quantities of polyP [18]. Still, the recovery ratio was consistently over 80% when 0.25–2 nmol polyP were loaded..

The effect of polyP chain length on the recovery ratio was evaluated by loading equivalent quantities of polyP by phosphate units of varying chain lengths, which revealed an increasing trend of the recovery ratio with the chain length (Figure B.1B). Compared to the recovery ratio of polyP chain length 45, recovery ratios of polyP chain length 14 and 23 were significantly lower (p < 0.001) while recovery ratios of polyP chain length 60 and 130 were significantly higher (p < 0.01). Although statistically significant, differences in recovery ratios of polyP chain length 45 vs. 60 and 130 were minor (11% and 15%). No statistical differences between the recovered and standard polyP were observed in polyP with chain lengths 60 and 130. Therefore, despite the lower recovery ratios of shorter polyP chains, polyP chain lengths of interest (between 45 and 130) could be recovered from silica spin columns with a high recovery ratio. Since we did not measure the recovery ratios for polyP chain lengths of less than 14 or greater than 130 Pi residues, this method may not be suitable for binding very small or very large polyP polymers. In systems that require the quantification of shorter polyP chains, short chain length-specific quantification methods such as high performance liquid chromatography [11] may be the better suited method.

159

Table B.2: PolyP recovery ratios of silica spin column isolation method were higher than those reported in other studies and independent of initially loaded polyP quantities.

Input PolyP Recovery polyP (µg/mL) polyP (nmol) This study Kulakova et al. [21] Martin et al. [18]

2.0 15.5 98 ± 3% 42% 48%

6.0 46.6 99 ± 3% 63% 71%

10.0 77.6 97 ± 2% 77% 77%

Results are represented as mean ± 95% CI, n = 5. No pairwise statistically significant difference was found.

Figure B.1: Quantity and chain length affected the polyP recovery ratios of the silica spin column protocol at nanomole and sub-nanomole quantities. (A) Silica spin columns bound and released nanomoles of polyP45 at a recovery ratio that ranged between 80% to 100%. Results are represented as mean ± SD, n = 3. * p < 0.05, ** p < 0.01 compared to either 4nmol or 8nmol. ns: not significant among denoted samples. (B) The recovery ratio of 7 nmol polyP increased with increasing chain length. Results are represented as mean ± SD, n = 3. p < 0.01 between any two values, except the pair marked ns.

160

Silica spin columns are ubiquitously used in laboratories for nucleic acid purification. Based on the structural similarities of polyP and nucleic acids, several other studies have investigated the use of silica for polyP isolation [18, 20, 21]. Investigations of DNA adsorption on silica have shown that shielding intermolecular electrostatic forces, dehydration of the DNA and silica surfaces, and intermolecular hydrogen bond formation in the DNA-silica contact layer are the major driving forces for adsorption [26], which can be manipulated with the buffer used to load samples onto silica spin columns. Through the optimization process of the loading buffer we learned that the choice and concentration of the chaotrophic reagent, total ionic strength, monovalent cations in the buffer, pH and concentration of ethanol all affected polyP binding as indirectly determined by the recovery ratio (data not shown).

B.4.2 Elimination of non-polyP interference for the polyP quantification assay

DNA, RNA, ATP, pyrophosphate, GAG and protein were identified as potential molecules in mammalian tissues that may also produce 415/558nm DAPI fluorescence signals and interfere with the fluorometric quantification of polyP (Table B.1). To verify this, polyP was mixed with each of the listed molecules and their DAPI fluorescence was measured. DAPI fluorescence of polyP samples containing DNA, chondroitin sulfate, RNA or BSA were significantly higher than that of polyP alone (Figure B.2A). The mixed samples were loaded in the columns to test whether the silica spin column method could isolate the polyP from the mixture. Eluents from samples with DNA and chondroitin sulfate had significantly higher DAPI fluorescence compared to polyP-only control (Figure B.2B). DAPI fluorescence of eluents from samples with RNA was comparable to polyP-only control, which could indicate that spontaneous RNA degradation may be sufficient to prevent its adsorption on silica columns. Finally, DAPI fluorescence of eluents from samples with FBS, in contrast to just BSA alone, was significantly less than polyP-only control. Together, in order to develop a polyP-specific quantification method, additional steps to reduce the interfering effects of DNA, chondroitin sulfate and FBS were required.

161

Figure B.2: DNA, chondroitin sulfate and FBS affected the DAPI fluorescence signal of polyP isolated by using silica spin columns. 7 nmol polyP was added to 140µL samples of potentially interfering molecules listed in Table 1. DAPI fluorescence signal was measured (A) before and (B) after the silica spin column isolation step and compared to the polyP alone (“none”). Results are expressed as mean ± SD, n = 3. *** p < 0.01 compared to polyP standard; CS: chondroitin sulfate; BSA: bovine serum albumin; pyroP: sodium pyrophosphate.

A previous study reported that enzymatic degradation of DNA and RNA reduced the intensity of shifted DAPI fluorescence to negligible levels [18], but nucleic acids were at a lower concentration (3 µg/mL) than the relevant range used of this study (101–102 µg/mL). Nuclease treatment of 55 µg/mL DNA and 180 µg/mL RNA (concentrations present in tissues) failed to eliminate the DAPI fluorescence (Figure B.3A). However, when concentrated Tris buffer was added to nuclease-treated DNA and RNA, the DAPI fluorescence decreased in a concentration- dependent manner, eliminating 96% and 98% of the signals from degraded DNA and RNA, respectively, with a Tris concentration of 500mM (Figure B.3 B, C). On the other hand, reduction of DAPI fluorescence from polyP was statistically significant but minor (Figure B.3D). Increasing the background Tris concentration did not eliminate DAPI fluorescence of intact nucleic acids (Figure B.3E), showing that both nuclease treatment and the subsequent addition of Tris were both required to eliminate the signals produced by the nucleic acids. Nuclease-treated DNA and RNA samples with polyP produced similar DAPI fluorescence than samples containing only polyP (Figure B.3F).

162

Figure B.3: Both nuclease treatment and Tris were required to eliminate DAPI fluorescence signals produced by DNA and RNA. (A) Samples were treated with nucleases and DAPI fluorescence was measured as described in the Materials and Methods. *** p < 0.001 compared to untreated. (B-D) Tris concentration of nuclease-treated RNA (B), DNA (C) and polyP (D) samples were increased before adding DAPI. *** p < 0.001 compared to all others. (E) 500mM Tris concentration by itself did not eliminate DAPI fluorescence of untreated DNA and RNA. *** p < 0.001 compared to untreated samples. (F) DNA and RNA were mixed with polyP before the nuclease treatment and DAPI fluorescence was measured in the presence of 500mM Tris. Results are expressed as mean ± SD, n = 3.

163

The dimethylmethylene blue assay confirmed that the silica spin column isolation step eliminated 99% of chondroitin sulfate in the samples, regardless of whether polyP was present in the initially loaded sample (Figure B.4A). A standard curve of DAPI fluorescence and chondroitin sulfate revealed a nonlinear relationship with a peak at 100 µg/mL (Figure B.4B). This was consistent with the previous observation that DAPI fluorescence of the eluent from samples containing polyP and chondroitin sulfate was higher (Figure B.2). In the presence of 500mM Tris, a large reduction of signal was observed at all chondroitin sulfate concentrations (50–100 µg/mL), eliminating 99.7% of signal (Figure B.4B). PolyP samples with added chondroitin sulfate produced statistically higher DAPI fluorescence than polyP-only control (p < 0.01, Figure B.4C).

Figure B.4: DAPI fluorescence signal produced by residual chondroitin sulfate (CS) in post-isolation samples were reduced by the addition of Tris. (A) Chondroitin sulfate in silica column eluents was quantified with the dimethylmethylene blue dye assay. Presence of polyP in samples did not affect the detection of GAGs. GAG: glycosaminoglycans; ns: not significant. (B) Semilog plot of DAPI fluorescence in the presence of chondroitin sulfate. In 10mM Tris, DAPI fluorescence peaked at 100 µg/mL and decreased at higher concentrations. In 500mM Tris, DAPI fluorescence increased linearly (r2 = 0.9938) with chondroitin sulfate. (D) DAPI fluorescence of CS and polyP samples were measured (A) before and (B) after the silica spin column isolation step and compared to the polyP standard alone. Results are expressed as mean ± SD, n = 3. For denoted pairs of data: *** p < 0.001; ns: not significant.

164

Previously, proteinase K pre-treatment in the presence of 10mM EDTA was demonstrated to be the most effective treatment for maximizing polyP signal from environmental samples [18]. The recovery ratios of FBS-containing polyP samples following proteinase K digestion prior to silica spin column isolation were comparable to polyP-only control (Figure B.5). Without EDTA, samples treated with proteinase K in prior to isolation produced significantly less DAPI fluorescence, indicating that EDTA was necessary for the proteinase K treatment step in order to prevent the loss of polyP signal. Since divalent cations were shown to cause polyP degradation [28], the reduced DAPI fluorescence was likely to the divalent cations, not protein, present in FBS.

Figure B.5: Proteinase K pre-treatment enhances polyP recovery. 7 nmol polyP samples mixed with FBS were incubated with or without 1 mg/mL proteinase K (PK) in the presence or absence of 10mM EDTA and incubated for 4 hours in 56ºC. DAPI fluorescence was measured after the silica spin column isolation step and compared to similarly incubated polyP standard (no proteinase K treatment) to calculate the recovery ratio. Loss of polyP signal was observed in samples without EDTA during proteinase K pre-treatment (+PK). Results are expressed as mean ± SD, n = 3. ** p < 0.01 compared to –PK without FBS, # p < 0.01 between one another as well as compared to –PK without FBS, ns: not significant.

165

B.4.3 Quantification of polyP added to in vitro-grown cartilage samples

We then analyzed tissues rich in proteoglycans, DNA and RNA that could potentially interfere with the polyP assay. The optimized polyP quantification assay is outlined in Figure B.6. Tissue samples are first digested with proteinase K in the presence of EDTA before performing the silica spin column polyP extraction. The eluted samples are treated with nucleases, and then concentrated EDTA, Tris and DAPI are sequentially added before measuring its fluorescence. First, the polyP recovery rate of each tissue type was measured by pairwise addition of exogenous polyP (Figure B.7A). A significantly lower (p < 0.05) recovery rate was measured from both in vitro-grown cartilage tissue samples grown on porous CPP substrates and full thickness articular cartilage samples. In vitro-formed cartilage tissue on porous CPP substrates accumulated a significantly higher polyP compared to tissue on membrane inserts or native articular cartilage (Figure B.7B). Full thickness articular cartilage had a significantly higher GAG than all other tissues (Figure B.7C). This tissue-dependent variation in recovery rate may be caused by an abundance of negatively charged macromolecules in samples that oversaturated the silica membranes and hinder polyP binding.

Figure B.6: Line diagram of the complete polyP quantification method.

166

No No tissue

Figure B.7: Quantification of polyP in the in vitro-formed cartilage and native cartilage tissues with correction for recovery rate. Cartilage tissues were grown in vitro on calcium polyphosphate (CPP) or collagen-coated membranes. Native full-thickness (FT) or superficial zone (SZ) cartilage tissues were also harvested for polyP quantification. (A) Recovered amounts of exogenously added polyP from each tissue sample were normalized to the recovered polyP from the positive control. Results are expressed as mean ± SD, n = 6. * p < 0.05; ** p < 0.01 compared to control. (B) Measured polyP content of tissues were corrected with the recovery rate and normalized to tissue wet weight. (C) GAG content of tissues were normalized to tissue wet weight. Data points are colour-coded for each donor animal, and each point was measured in triplicates. Results are expressed as mean ± SD, n = 6. * p < 0.05; ** p < 0.01; *** p < 0.001 compared to all others.

167

B.5 Discussion

We report the development of a simple and highly sensitive method to isolate and quantify polyP ranging from 14 to 130 polyP residues from mammalian tissues. DAPI-based fluorometric quantification is a highly sensitive, rapid method of polyP detection, but an effective polyP isolation method from mammalian tissues was required in order to be able to use the fluorometric quantification technique. The isolation method required a high recovery ratio at minute polyP amounts, as mammalian cells have been previously shown to have much less polyP than prokaryotic cells [16]. We demonstrate a significant improvement of silica spin column- based method in both the recovery ratio and the lowest recoverable amounts of polyP. While the isolation method was not entirely specific for polyP, enzymatic treatments and high Tris concentration reduced the extraneous signals generated by the contaminants. Quantifying polyP in in vitro-formed cartilage tissue and native cartilage tissue demonstrated that an abundance of negatively charged macromolecules adversely affected the polyP recovery rate, which could be corrected using recovery rates calculated from exogenously added polyP to samples. However, further studies are required to lessen the dependency of polyP recovery rate to the nature of tissues being quantified.

This method can be readily adapted for use in other laboratories as our protocol closely resembles nucleic acid purification, and the enzymes used in our study (proteinase K, RNase A and DNase I) are readily available. Since proteinase K pre-treatment can effectively dissolve virtually all proteinaceous matter, this method can be readily adapted for most biological sources, as well as for liquid samples such as, for example, platelet-activated blood that was previously estimated to contain 1–3µM polyP [30]. While the assay cannot provide insight on the distribution of chain length, isolated polyP samples could be analyzed with complementary assays such as PAGE [31], capillary gel electrophoresis [32] or NMR spectroscopy [15]. This new method provides an easy-to-use research tool for future polyP investigation in mammalian biology.

168

B.6 Author Contributions The authors of the study were Whitaik David Lee, Rahul Gawri, Toshikazu Shiba, William L. Stanford and Rita A. Kandel.

• Study conception and design: WDL, RAK. • Provision of study materials: TS, RAK. • Analysis and interpretation of data: WDL, RG, RAK. • Drafting of article: WDL, RAK. • Critical revision of the article: WDL, RG, WLS, RAK. • Final approval of the article: WDL, RG, TS, WLS, RAK.

B.7 Acknowledgements

The authors would like to thank Dr. Patrick Martin for helpful discussions. This work was supported by the Canadian Institutes for Health Research (to R.A.K., MOP 12611) and the Natural Sciences and Engineering Research Council of Canada (to W.L.S., RGPIN 293170-11). W.L.S. is supported by a Canadian Research Chair. R. G. was supported by a fellowship from the Arthritis Society of Canada.

169

B.8 References

[1] Rao NN, Gomez-Garcia MR, Kornberg A. Inorganic polyphosphate: essential for growth and survival. Annu Rev Biochem (2009) 78:605-47.

[2] Ruiz FA, Lea CR, Oldfield E, Docampo R. Human platelet dense granules contain polyphosphate and are similar to acidocalcisomes of bacteria and unicellular eukaryotes. J Biol Chem (2004) 279:44250-7.

[3] Dinarvand P, Hassanian SM, Qureshi SH, Manithody C, Eissenberg JC, Yang L, et al. Polyphosphate amplifies proinflammatory responses of nuclear proteins through interaction with receptor for advanced glycation end products and P2Y1 purinergic receptor. Blood (2014) 123:935-45.

[4] Wang L, Fraley CD, Faridi J, Kornberg A, Roth RA. Inorganic polyphosphate stimulates mammalian TOR, a kinase involved in the proliferation of mammary cancer cells. Proc Natl Acad Sci U S A (2003) 100:11249-54.

[5] Kawazoe Y, Katoh S, Onodera Y, Kohgo T, Shindoh M, Shiba T. Activation of the FGF signaling pathway and subsequent induction of mesenchymal stem cell differentiation by inorganic polyphosphate. Int J Biol Sci (2008) 4:37-47.

[6] Zakharian E, Thyagarajan B, French RJ, Pavlov E, Rohacs T. Inorganic polyphosphate modulates TRPM8 channels. PLoS One (2009) 4:e5404.

[7] Omelon S, Georgiou J, Henneman ZJ, Wise LM, Sukhu B, Hunt T, et al. Control of vertebrate skeletal mineralization by polyphosphates. PLoS One (2009) 4:e5634.

[8] St-Pierre J-P. The Role of Inorganic Polyphosphates in the Formation of Bioengineered Cartilage Incorporating a Zone of Calcified Cartilage In Vitro 2011.

[9] Holmstrom KM, Marina N, Baev AY, Wood NW, Gourine AV, Abramov AY. Signalling properties of inorganic polyphosphate in the mammalian brain. Nat Commun (2013) 4:1362.

[10] Choi BK, Hercules DM, Houalla M. Characterization of polyphosphates by electrospray mass spectrometry. Anal Chem (2000) 72:5087-91.

170

[11] Sekiguchi Y, Matsunaga A, Yamamoto A, Inoue Y. Analysis of condensed phosphates in food products by ion chromatography with an on-line hydroxide eluent generator. J Chromatogr A (2000) 881:639-44.

[12] St-Pierre JP, Pilliar RM, Grynpas MD, Kandel RA. Calcification of cartilage formed in vitro on calcium polyphosphate bone substitutes is regulated by inorganic polyphosphate. Acta Biomater (2010) 6:3302-9.

[13] Muller F, Mutch NJ, Schenk WA, Smith SA, Esterl L, Spronk HM, et al. Platelet polyphosphates are proinflammatory and procoagulant mediators in vivo. Cell (2009) 139:1143- 56.

[14] Hupfer M, Glöss S, Schmieder P, Grossart H-P. Methods for Detection and Quantification of Polyphosphate and Polyphosphate Accumulating Microorganisms in Aquatic Sediments. International Review of Hydrobiology (2008) 93:1-30.

[15] Momeni A, Filiaggi MJ. Synthesis and characterization of different chain length sodium polyphosphates. J Non-Cryst Solids (2013) 382:11-7.

[16] Kumble KD, Kornberg A. Inorganic polyphosphate in mammalian cells and tissues. J Biol Chem (1995) 270:5818-22.

[17] Aschar-Sobbi R, Abramov AY, Diao C, Kargacin ME, Kargacin GJ, French RJ, et al. High sensitivity, quantitative measurements of polyphosphate using a new DAPI-based approach. J Fluoresc (2008) 18:859-66.

[18] Martin P, Van Mooy BA. Fluorometric quantification of polyphosphate in environmental plankton samples: extraction protocols, matrix effects, and nucleic acid interference. Appl Environ Microbiol (2013) 79:273-81.

[19] Ault-Riche D, Fraley CD, Tzeng CM, Kornberg A. Novel assay reveals multiple pathways regulating stress-induced accumulations of inorganic polyphosphate in Escherichia coli. J Bacteriol (1998) 180:1841-7.

171

[20] Werner TP, Amrhein N, Freimoser FM. Novel method for the quantification of inorganic polyphosphate (iPoP) in Saccharomyces cerevisiae shows dependence of iPoP content on the growth phase. Arch Microbiol (2005) 184:129-36.

[21] Kulakova AN, Hobbs D, Smithen M, Pavlov E, Gilbert JA, Quinn JP, et al. Direct quantification of inorganic polyphosphate in microbial cells using 4'-6-diamidino-2-phenylindole (DAPI). Environ Sci Technol (2011) 45:7799-803.

[22] Knudson CB, Knudson W. Cartilage proteoglycans. Semin Cell Dev Biol (2001) 12:69-78.

[23] Lee WD, Hurtig MB, Kandel RA, Stanford WL. Membrane culture of bone marrow stromal cells yields better tissue than pellet culture for engineering cartilage-bone substitute biphasic constructs in a two-step process. Tissue Eng Part C Methods (2011) 17:939-48.

[24] Waldman SD, Grynpas MD, Pilliar RM, Kandel RA. Characterization of cartilagenous tissue formed on calcium polyphosphate substrates in vitro. J Biomed Mater Res (2002) 62:323- 30.

[25] Kandel RA, Grynpas M, Pilliar R, Lee J, Wang J, Waldman S, et al. Repair of osteochondral defects with biphasic cartilage-calcium polyphosphate constructs in a sheep model. Biomaterials (2006) 27:4120-31.

[26] Melzak KA, Sherwood CS, Turner RFB, Haynes CA. Driving forces for DNA adsorption to silica in perchlorate solutions. J Colloid Interf Sci (1996) 181:635-44.

[27] Tanious FA, Veal JM, Buczak H, Ratmeyer LS, Wilson WD. DAPI (4',6-diamidino-2- phenylindole) binds differently to DNA and RNA: minor-groove binding at AT sites and intercalation at AU sites. Biochemistry (1992) 31:3103-12.

[28] Smith SA, Choi SH, Davis-Harrison R, Huyck J, Boettcher J, Rienstra CM, et al. Polyphosphate exerts differential effects on blood clotting, depending on polymer size. Blood (2010) 116:4353-9.

[29] Omelon S, Grynpas M. A nonradioactive method for detecting phosphates and polyphosphates separated by PAGE. Electrophoresis (2007) 28:2808-11.

172

[30] Lee A, Whitesides GM. Analysis of inorganic polyphosphates by capillary gel electrophoresis. Anal Chem (2010) 82:6838-46.

[31] Baluyot ES, Hartford CG. Comparison of polyphosphate analysis by ion chromatography and by modified end-group titration. J Chromatogr A (1996) 739:217-22.

[32] Momeni A, Filiaggi MJ. Comprehensive study of the chelation and coacervation of alkaline earth metals in the presence of sodium polyphosphate solution. Langmuir (2014) 30:5256-66.

173

B.9 Additional Figures

B.9.1 Development of the silica spin column protocol

Figure B.8: Initial silca spin column trial. 4–400 nmol polyP standard solutions in 200µL volume were loaded onto silica spin columns of Ambion PureLink RNA Mini Kit (Life Technologies) according to the manufacturer’s instructions. GF-75 glass fibre columns were assembled by removing the silica membranes from PureLink cartridges and packing 4 punches of Sterlich 0.3µm pore size into each cartridge. GF-75 was previously used to bind and elute polyP [18], but recovery ratios of PureLink columns were superior. However, this may have been due to the use of PureLink’s proprietary buffers.

174

Figure B.9: Optimization of buffers used in the silica spin columns. Using a starting amount of 80nmol polyP in 200µL volume, the optimal reagent concentrations were determined for each step of the silica spin column protocol. Except (D), polyP was eluted using 100µL ddH2O twice. Results are expressed as mean ± SEM, n = 1.

(A) Ethanol in binding buffer: polyP samples were mixed with 200µL of the putative binding buffer (5.0M guanidine thiocyanate (GuSCN), 0.9M sodium acetate, 50mM Tris pH 6.2; adopted from OpenWetWare wiki page for silica-based DNA isolation) and 200–1000µL of 100% ethanol to test a final per-volume ethanol concentration ranging from 33% to 71%. Kit desalting and washing buffer were used. From this experiment, the optimal ethanol concentration was determined to be 33%.

(B) Desalting and wash buffers: polyP samples were loaded onto silica spin columns using the putative binding buffer and ethanol. Desalting buffer was comprised of 3.5M guanidine thiocyanate and 50mM Tris pH 6.2 in either ddH2O or 30% isopropanol (iPrOH). Wash buffer was composed of 10mM Tris pH 7.4 with or without 150mM sodium chloride in either 50% or

175

80% ethanol (EtOH). From this experiment, the desalting buffer with 30% iPrOH and the wash buffer with 150mM NaCl and 50% EtOH were determined to be optimal.

(C) Tris concentration in binding buffer and alcohol in desalting buffer: binding buffers consisting of 5.0M guanidine thiocyanate, 0.9M sodium acetate and either 50mM, 100mM or zero Tris pH 6.2 were prepared. 200µL polyP samples were loaded onto silica spin columns with 200µL of the three binding buffers and 200µL of 100% ethanol. Desalting buffer was comprised of 3.5M guanidine thiocyanate and 50mM Tris pH 6.2 in either 30% isopropanol, 30% ethanol or 50% ethanol solutions. Optimal wash buffer was used. From this experiment, binding buffer containing 100mM Tris and desalting buffer in 50% ethanol were determined to be optimal.

(D) Elution buffer: polyP was loaded onto silica spin columns using the optimal binding, desalting and wash buffers. polyP was eluted with either ddH2O, 10-100mM Tris pH 8.8 and 50mM Tris pH 7.4. Elution was performed either twice or four times. From this experiment, maximum elution was obtained with 50mM Tris pH 7.4 eluting four times.

176

Figure B.10: PolyP was consistently bound and eluted from Epoch EconoSpin DNA Mini columns. (A) Punches of GF-75 were packed into either PureLink (PL) cartridges, which had a thick white membrane, or columns with 0.2µm pore size polyvinylidene difluoride membrane as filter (PVDF; Analytical Sales & Services, Pompton Plains, NJ, USA), which was used by Martin and Van Mooy [18]. In addition, EconoSpin DNA Mini columns (Epoch Life Sciences, Sugar Land, TX, USA) was also tested for polyP binding and recovery. (B) Other filters from Analytical Sales & Services were also tested, but the Epoch EconoSpin DNA Mini columns were superior. MWCO: molecular weight cut-off. Results are expressed as mean ± SEM, n = 1

Figure B.11: Elution volume did not significantly affect the recovery ratio. 0.9 and 9.0 nmol polyP was loaded onto Epoch EconoSpin DNA Mini columns and eluted using 40–80 µL of 10mM Tris pH 7.2 three times. The source of error for the 0.9 nmol data points was the discrepancy in measuring the normalizing value. However, samples eluted with 60µL and 40µL had less particles pelleted in the samples compared to those eluted with 80µL; therefore, 60µL was determined to be the optimal elution volume. Results are expressed as mean ± SD, n = 3.

177

Figure B.12: Protocol optimization for minimizing the background signal. The working protocol had two issues: a relatively high background signal (0.4 fluorescence units, compared to 2.0 fluorescence units for 50µM polyP) and sedimentation of silica particles in the eluted samples. (A) To identify which buffer is responsible for adding the background signal, binding buffer (BB, in 33% ethanol), desalting buffer (DB) and wash buffer (WB) were sequentially passed through silica spin columns without polyP, omitting one or more buffers as indicated. Columns were eluted three times using the elution buffer (10mM Tris pH 7.2). Each eluent was then split into two equal volume aliquots, one of which was agitated to re-suspend the silica particles. Columns through which DB was passed had higher background signal, which indicated that DB must be further optimized to reduce the background signal. There was no difference between the particle-free and re-suspended samples, which indicated that the presence of silica particles had no effect in the background signal. (B) Lowering GuSCN and Tris concentration in the DB successfully reduced the fluorescent signal. Results are expressed as mean ± SEM, n = 1.

178

B.9.2 Optimizing the recovery ratios of different chain length polyP

Figure B.13: Minor differences in quantity were observed between polyP standards with different chain lengths. (A) PolyP standards with different chain lengths were converted to orthophosphates by treating the samples with calf intestinal alkaline phosphatase (CIAP) for 2 hours. Orthophosphates were quantified using heteropoly blue assay. Mean values were shown’ n = 1. (B) 415nm/558nm DAPI fluorescence signals in the presence of PolyP standards with different chain lengths also showed minor differences. Minor variation was also observed when the background Tris concentration was increased. Results are expressed as mean ± SEM, n = 1. These data indicated that difference in phosphate contents among polyP standards with varied chain lengths were negligible, allowing the use of standards’ nominal concentration values for subsequent experiments.

179

Figure B.14: the silica spin columns did not bind well short-chain polyP (<45 chain length) and falsely amplified the DAPI fluorescence signals given off by long-chain polyP (>45 chain length). 9nmol of polyP chain lengths 14–130 were loaded onto silica spin columns using the working protocol, eluted and quantified by either DAPI fluorescence or ALP treatment and orthophosphate quantification. Nonlinear regression using the first-order association model1 had a high goodness-of-fit value (r2 = 0.9889 for fluorescence and r2 = 0.9900 for orthophosphate quantification). This indicated two issues to address in the working protocol: to enhance the binding of short-chain polyP, and to eliminate the false positive signal apparent in the long-chain polyP.

1 !!" � = �� + � − �� (� − � )

180

Figure B.15: Short-chain polyP binding is enhanced by increasing the ethanol concentration at the binding step. (A) polyP chain length 14 was loaded onto columns with increasing quantities of ethanol. The columns were desalted, washed and eluted. Compared to 33% EtOH, recovery rate of polyP loaded with 50% EtOH or greater was ~100%. (B) Recovery rate plot by chain length indicated that 50% EtOH enhanced the binding of polyP chain lengths 14 and 22. The false positive signal from longer-chain polyP remained unchanged. Excellent goodness-of-fit were observed (r2 = 0.9958 for 33% EtOH, r2 = 0.9991 for 50% EtOH). Results are expressed as mean ± SEM, n = 1.

181

Figure B.16: Presence of abundant proteoglycans limited the maximum concentration of ethanol at 40%. (A) Deep-zone bovine native cartilage digests (200mg per mL of 1mg/mL proteinase K-EDTA solution) were mixed with binding buffer and various amounts of ethanol. Samples whose ethanol concentration was greater than 40% turned turbid, which can be pelleted down to yield white precipitates. The precipitate caused the silica membrane to clog, and when pelleted, the supernatant did not contain any polyP (data not shown). No visible pellets were present in samples whose ethanol concentration was 40% or less. Precipitation was not affected by the presence of detergents (1% SDS, 1% Tween20 and 1% Triton X-100; data not shown). (B) The supernatant was discarded and pellets were dissolved in 1% BSA solution. The solution turned DMMB dye purple, suggesting that proteoglycans were present in the precipitate.

182

Figure B.17: GuSCN concentration in the desalting buffer also affected the recovery ratios shorter chain length polyP. In this experiment, 50% EtOH in the binding was used to ensure full binding of polyP. (A) Titration of GuSCN in the DB. Interestingly, skipping DB altogether resulted in slightly lower but even recovery ratios for all polyP chains. (B) Recovery ratio of polyP chain length 22 was constantly high with even lower concentrations of GuSCN, but the presence of Tris in the DB resulted in erratic recovery ratios.

183

Figure B.18: Washing the silica spin columns before loading reduced the false positive signals in longer-chain polyP samples. (A) Before loading polyP chain length 130 samples, silica spin columns were washed by passing through 1% SDS, 1% Tween20, 1% Triton X-100,

10mM Tris pH 7.2 or ddH2O, followed by another pass of 10mM Tris pH 7.2 to eliminate residual detergents. Columns pre-washed with 10mM Tris pH 7.2 had the closest apparent recovery ratio to 100%. (B) Recovery rate plot by chain length indicated that pre-washing of columns with 10mM Tris pH 7.2 reduced the signals from polyP chain lengths 60 and 130. Singals from shorter-chain polyP were unaffected. Excellent goodness-of-fit were observed (r2 = 0.9984 for unwashed, r2 = 0.9989 for washed). Results are expressed as mean ± SEM, n = 1.

184

B.9.3 Nuclease protocol development

Figure B.19: Incubation of polyP with calcium in 37ºC for even 15 minutes results in signal degradation. DNase I requires the presence of calcium and either magnesium or manganese divalent cations for activity at 37ºC, but divalent cations also cause degradation of polyP [32]. (A) 10mM divalent cations were added to 50µM polyP samples and incubated at 37ºC for up to 1 h. Signal degradation of polyP was the most noticeable in the presence of 10mM Ca2+ alone, while signal did not degrade in the presence of 10mM Mg2+. Signal degradation of polyP in the presence of both 10mM Ca2+ and 10mM Mg2+ was less at 1 h compared to 10mM Ca2+ alone. Addition of Mn2+ to Ca2+ did not affect the signal degradation at 1 h. (B) Various concentrations of Ca2+ were added to 50µM polyP samples and incubated at 37ºC for an hour. Reducing the amount of Ca2+ also reduced the signal degradation by the divalent cation. Results are expressed as mean ± SEM, n = 1.

185

Figure B.20: Preliminary data on the use of DNase I. (A) Lyophilized DNase I was reconstituted with various buffers recommended for the long-term storage of the enzyme: (1)

20 mM Tris-HCl, 20 mM MgCl2, 5 mM CaCl2, 0.1 mM dithioerythritol (DTT), 0.1 mM EDTA, 50% (v/v) glycerol, pH 8; (2) 20 mM Tris-HCl, 50 mM NaCl, 1 mM DTT, 100 µg/mL BSA,

50% (v/v) glycerol, pH 7.6; (3) 20 mM Tris-HCl, 1 mM MgCl2, 50% (v/v) glycerol, pH 7.5; (4)

20 mM Tris-HCl, 1 mM MgCl2, pH 7.5. DNase I reconstitution in ddH2O is not recommended for long-term storage by the manufacturer. 50µM polyP was incubated with 10 U of DNase reconstituted in different buffers for 1 h at 37ºC. 25mM EDTA was added to stop the reaction and DAPI fluorescence was measured. This result indicated that reconstitution buffer (4) did not cause loss of polyP signal. (B) Titration of DNase I necessary in the samples, using 55µg/mL DNA (Table B.1). Even though 5U was sufficient, it was determined that 10U would be sufficient for any excess DNA present in the samples. Results are expressed as mean ± SEM, n = 1.

186

Figure B.21: Slope of the polyP standard curve is affected by the addition of DNase and EDTA. DAPI fluorescence of 2µM-80µM polyP standard solutions made with Tris pH 7.5, DNase buffer and DNase buffer plus EDTA were measured. DNase buffer alone increased the slope, while addition of EDTA reduced it. This indicated that the polyP standard curve used to convert fluorescence signals to polyP amount in samples must be established with DNase+EDTA background to better quantify the polyP in samples. Results are expressed as mean ± SEM, n = 1.

Figure B.22: Validation of nuclease treatments in polyP and DNA/RNA samples. 50µM polyP and DNA/RNA (concentrations as indicated in Table B.1) were treated with DNase I and RNase A respectively in the nuclease buffer, developed based on the above experiments. In both cases, signals from polyP and nucleic acid mixtures were identical to those from polyP alone, and signals from nucleic acids alone were almost nil. Results are expressed as mean ± SEM, n = 1.